Electrical monitoring and control system

Electrical monitoring and control system
(12) United States Patent
(10) Patent N0.:
(45) Date of Patent:
Kao et al.
US 8,244,405 B2
Aug. 14, 2012
References Cited
(75) Inventors: Imin Kao, Stonybrook, NY (US);
Brenda Pomerance, New York, NY
Harford ........................ .. 361/56
(US); Robert P. Wong, Huntington, NY
6,373,257 B1*
(73) Assignee: BSafe ElectriX, Inc.
4,870,528 A *
Subject to any disclaimer, the term of this
patent is extended or adjusted under 35
U.S.C. 154(b) by 250 days.
... ... ...
Feb. 27, 2009
Prior Publication Data
US 2009/0222142 A1
Sep. 3, 2009
6,426,634 B1*
6,532,139 B2*
3/2003 Kim et a1.
Clunn et al. ..... ..
.. ... ...
. . . ..
. . . ..
6,839,208 B2*
1/2005 Macbeth et al.
Engel .......................... .. 374/141
6,948,846 B2*
7,253,637 B2
7,362,552 B2
8/2007 Dvorak
4/2008 Elms
7,499,250 B2
3/2009 Zhang
.. ... ...
2001/0033469 A1* 10/2001 Macbeth et al.
(21) Appl. No.: 12/380,460
. . . ..
4/2002 Macbeth et al.
2002/0008950 A1*
1/2002 Kim et a1.
2003/0058596 A1*
..... ..
.. 361/93.5
2004/0100274 A1*
5/2004 Gloster et a1. .
2005/0089079 A1*
. 361/42
Engel .......................... .. 374/141
* cited by examiner
Primary Examiner * Mohammad Ali
Assistant Examiner * Anthony Whittington
Related U.S. Application Data
Provisional application No. 61/067,693, ?led on Feb.
29, 2008.
A system for monitoring and controlling the electrical infra
Int. Cl.
G05D 11/00
(74) Attorney, Agent, or Firm * Thomas A. O’Rourke;
Bodner & O’Rourke, LLP
structure of a building includes at least one sensor for sensing
U.S. Cl. ........................................ .. 700/286; 361/42
an operating characteristic in the building, and a processor for
receiving information from the at least one sensor and pre
Field of Classi?cation Search ................ .. 700/286;
dicting a future operating characteristic.
361/42; 324/527
See application ?le for complete search history.
sensor 5
24 Claims, 10 Drawing Sheets
sensor 6
controller 20
display 30
processor 10
printer 31
comm 40
comm 41
tel net 101
comm net 102
sewer 1 1O
server 100
US. Patent
Aug. 14, 2012
Sheet 1 0f 10
US 8,244,405 B2
sensor 6
sensor 5
display 30
controller 20
processor 10
printer 31
comm 40
comm 41
tel net 101
comm net 102
server 110
server 100
US. Patent
Aug. 14, 2012
Sheet 2 0f 10
US 8,244,405 B2
Fig. 2
1 microsec
Fig. 3
1 microsec
Fig. 4
US. Patent
Aug. 14, 2012
Sheet 3 or 10
US 8,244,405 B2
Hg. 5
1 mlcrosec
Fig. 6
H W H ??l
US. Patent
Aug. 14, 2012
Sheet 4 0f 10
US 8,244,405 B2
red region
temp ___
yellow region
Fig. 8
green region
-. |
Fig. 9A
Fig. 9B
US. Patent
Aug. 14, 2012
Sheet 5 or 10
US 8,244,405 B2
Fig. 10
US. Patent
Aug. 14, 2012
Sheet 6 or 10
US 8,244,405 B2
Fig. 11
342 |_
__________________ a
""""""" ~-
3:74 377
375’ _______________________ " 370
US. Patent
Aug. 14, 2012
US 8,244,405 B2
403 405
431 411|
I 432
Flg. 12A
Fig. 128
US. Patent
Aug. 14, 2012
Sheet 8 or 10
US 8,244,405 B2
current sensor
light sensor
audio sensor
sensor 520
ADC 52
ADC 532
ADC 542
ADC 552
ADC 562
microprocessor 570
US. Patent
Aug. 14, 2012
Sheet 9 0f 10
US 8,244,405 B2
power-on or reset
set if 1 605
set fL1 606
set fA1 607
setli1 608
Store 5amp|e5in
circular buffer 610
store samples in
circular buffer 611
store samples in
circular buffer 612
store samples in
circular buffer 613
analyze Sample
analyze sample
analyze sample
analyze sample
window 615
window 616
window 617
window 618
set fT 2
set fL2
set fA2
Store Samples in
store samples in
store samples in
store samples in
circular buffer 631
circular butler 632
circular buffer 633
analyze sample
analyze sample
analyze sample
analyze sample
window 635
window 636
window 637
window 638
set i i2
Circular Duffel 630
no fault
55 Yes
@ yes
T '
@?ult ; 5
@ yes
redundant fault analysis 650
action 660A
action @605
combine 670 €
Fig. 14
action 660C
US. Patent
Aug. 14, 2012
Sheet 10 or 10
US 8,244,405 B2
serial arcing
norrnal (no arcing)
normal (no arcing)
serial arcing
normal (no arcing)
serial arcing
1 00
Fig. 17
US 8,244,405 B2
installation of combination-type AFCIs in all 15 and 20 amp
residential circuits With the exception of laundries, kitchens,
bathrooms, and garage, and un?nished basements.
Zigbee is a Wireless technology that does not have the
speed or bandWidth of Wi-Fi or Bluetooth, but is designed for
Wireless building controls. ZigBee is based on IEEE Standard
This application claims priority from US. provisional
patent application Ser. No. 61/067,693, ?led Feb. 29, 2008,
having common inventors herewith.
80215.4 and creates a self-organizing Wireless netWork
Where any ZigBee-compliant device introduced into the envi
ronment is automatically incorporated into the netWork as a
node. A number of manufacturers are currently developing
The present invention relates to building electrical system
monitoring and control.
devices that incorporate this technology, including sWitches,
thermostats and other common monitoring and control
devices. ZigBee devices are battery poWered, Which means
US. Patent Application Publication US 2007/0155349
(Nelson et al.) discloses a system for selectively controlling
electrical outlets using poWer pro?ling. An electrical outlet
that they do not need any interconnecting Wiring. These
includes a socket for receiving a plug, an outlet identi?cation
ing signal, so their batteries can last for months or even years
and a signal detector for detecting a signal from the plug, for
sending the signal and outlet identi?cation to a controller, and
Without replacement.
devices remain dormant until they are activated by an incom
ZigBee devices have the ability to form a mesh netWork
betWeen nodes. Meshing is a type of daisy chaining from one
for receiving a command from the controller, such as to
reduce or cut poWer to the device plugged into the outlet When
the device’s operation deviates from an operational pro?le for
device to another. This technique alloWs the short range of an
the device. The system may be used With motion sensors and
other environmental components.
65,000 nodes (active devices). The netWork they form in
Blemel and Furse, “Applications of Microsystems and Sig
nal Processing for Wiring Integrity Monitoring”, 2001 IEEE
Aerospace Symposium, 12 pages, the disclosure of Which is
individual node to be expanded and multiplied, covering a
much larger area. One ZigBee netWork can contain more than
cooperation With each other may take the shape of a star, a
branching tree or a net (mesh). There are three categories of
hereby incorporated by reference, discuss detection and pre
ZigBee devices: ZigBee NetWork Coordinator. Smart node
that automatically initiates the formation of the netWork. Zig
vention of Wiring related problems in aerospace vehicles.
Bee Router. Another smart node that links groups together
Blemel presents a system in Which sensors in an aircraft
interface With processors; the processors are netWorked
together on an aircraft and are able to communicate With a
and provides multi-hopping for messages. It associates With
other routers and end-devices. ZigBee End Devices. Sensors,
actuators, monitors, sWitches, dimmers and other controllers.
central Web server. The processors implement algorithms for
Z-Wave is an interoperable standard for residential and
fault detection, identi?cation, location, prediction and mes
light commercial devices, providing reliable, con?r'mable,
loW bandWidth, half duplex tWo Way control communications
US. Pat. No. 5,991,327 (Kojori) discloses a controller that
receives a plurality of sensor readings, including some extra
via Wireless mesh neWorking. The Z-Wave development plat
readings for diagnostic protection, and processes the readings
to predict and control voltages and currents in an electric arc
An arc fault circuit interrupter (AFCI) is a circuit breaker
designed to prevent ?res by detecting non-Working electrical
unit to one or more slave units. Slave units can forWard
commands to other slave units. The ZM3102N Z-Wave Mod
ule contains the ZW0301 Z-Wave Single Chip, system crystal
and RF front-end circuitry. The ZW0301 Single Chip
Memory for Z-Wave Protocol and OEM Application storage
softWare, Triac Controller, and various hardWare interfaces.
Motorola sells Home Monitoring and Control System
Wireless Temperature Sensors, namely model HMTS1050
and model HMSM4150, that are intended to be placed in a
room, and programmed With an upper and/or loWer limit.
When the limit is exceeded, the system sends a text alert to a
cell phone or e-mail.
While there is concern about the electrical infrastructure of
arcs and disconnect poWer before the arc starts a ?re. Arc
faults in a home are one of the leading causes for household
includes an RF transceiver, 8051 MCU core, SRAM, Flash
?res. AFCIs resemble a GFCI/RCD (Ground-Fault Circuit
Interrupt/Residual-Current Device) in that they both have a
test button. GFCIs are designed to protect against electrical
shock, While AFCIs are primarily designed to protect against
Starting With the 1999 version of the National Electrical
Code (NEC, also called NFPA 70) in the United States, AFCIs
are required in all circuits that feed receptacles in bedrooms of
form is described at WWW.Zen-sys.com. The Z-Wave Protocol
is for communicating short control messages from a control
dWelling units. This requirement is typically accomplished by
buildings, including residential and commercial, there is still
using a kind of circuit-breaker (de?ned by UL 1699) in the
breaker panel that provides combined arc-fault and overcur
rent protection. Not all U.S.A. jurisdictions have adopted the
room for improvement.
AFCI requirements of the NEC as Written. AnAFCI detects
sudden bursts of electrical current in milliseconds, long
In accordance With an aspect of this invention, there is
before a standard circuit breaker or fuse Would trip.
provided a system for monitoring and controlling the electri
In 2002, the NEC removed the Word “receptacle” leaving
“outlets”, in effect adding lights Within dWelling bedrooms to
cal infrastructure of a building.
It is not intended that the invention be summarized here in
the requirement. The 2005 code made it more clear that all
outlets must be protected. “Outlets” is de?ned in “Article 100
De?nitions” of the NEC as “A point on the Wiring system
Where current is taken to supply utiliZation equipment” and
this includes receptacles, light ?xtures, and smoke alarms,
among other things.
Beginning January 2008, only “combination type” AFCIs
Will meet the NEC requirement. The 2008 NEC requires
its entirety. Rather, further features, aspects and advantages of
the invention are set forth in or are apparent from the folloW
ing description and draWings.
FIG. 1 is a block diagram shoWing the elements of the
present system;
US 8,244,405 B2
FIG. 2 is a chart showing instantaneous arcing;
senses other devices, demonstrating that there are plural sen
sors in the present con?guration.
For a new installation, sensors 5, 6 may be built-in, and
their location optimiZed according to a procedure. For an
existing installation, sensors 5, 6 can be retro?t via plug-in or
FIG. 3 is a chart showing how a digital optical sensor reacts
to the signal of FIG. 2;
FIG. 4 is a chart showing continuous arcing;
FIG. 5 is a chart showing two instances of single arcing;
stick-on modules designed to minimize installation dif?culty.
Sensor 5 may be similar to Tmote Invent, a fully packaged
wireless sensing unit built on Moteiv’s Tmote Sky wireless
FIG. 6 is a chart showing how a digital optical sensor reacts
to the signal of FIG. 5;
FIG. 7 is a chart showing, as “curve ”, the current versus
time for an intermittent contact; as “curveB”, a sampling rate
module, the follow-on to Moteiv’s Telos sensor. Moteiv was
that is dynamically changed; and as “curveC”, the output of a
sensor whose sampling rate is dynamically changed;
purchased by Sentilla, and the Tmote Invent is no longer
offered. Tmote Invent, designed for industrial applications
FIG. 8 shows a vectoriZed map;
FIGS. 9A and 9B are curves referenced in explaining
including building monitoring and security control, included
integrated sensors for light, temperature, vibration (2-axis
accelerometer) and sound (microphone). Tmote Invent
included a speaker for auditory feedback, headphone jack for
FIG. 10 is a block diagram showing a faceplate for sensing
operation of a device plugged into an outlet;
FIG. 11 is a block diagram showing a faceplate for sensing
operation of a device plugged into an outlet, and for termi
nating power when an exception condition occurs;
FIGS. 12A and 12B are block diagrams showing a back
view and a side view of a current tap for sensing operation of
discrete applications, and LEDs for visual feedback. Included
with each Tmote Invent Application Kit was Moteiv’ s robust
distribution of the TinyOS open-source operating system.
Designed for low-power, long-lived mesh networking, the
device plugged into it;
FIG. 13 is a schematic of a circuit according to the present
FIG. 14 is a ?owchart for the circuit of FIG. 13;
FIG. 15 shows temperature vs. time curves for normal
Programming and data collection via USB, Light, Tempera
ture, Acceleration, and Sound sensors, Speaker, LEDs, and
User input buttons, rechargeable battery that charges through
any standard USB port, 250 kbps 2.4 GHZ IEEE 80215.4
operation and serial arcing;
Chipcon Wireless Transceiver, Interoperability with other
FIG. 16 shows current vs. time curves for normal operation
and serial arcing; and
FIG. 17 shows FFT frequency spectra for normal operation
and serial arcing.
distribution allows application developers to tune and con?g
ure the system through highly ?exible interfaces. The result
was a customiZable yet robust low-power sensing system.
Features included: Low power wireless mesh technology,
IEEE 80215.4 devices, 8 MHZ Texas Instruments MSP430
microcontroller (10 k RAM, 48 k Flash), Integrated antenna
with 50 m range indoors/125 m range outdoors, Ultra low
power consumption.
FIG. 1 shows sensors 5, 6 coupled to controller 20. Sensor
Sensor 5 is associated with one of a variety of devices (not
shown), such as an electrical receptacle, a faceplate, a circuit
breaker, an air-conditioning unit, a refrigerator, and so on.
5 uses a wireline connection, while sensor 6 uses a wireless
Sensor 5 may have local data analysis capability.
connection. Local communication network 25 couples pro
cessor 10, controller 20, display 30, printer 31, and commu
nication interfaces 40, 41. Each of processor 10, controller 20
sensed, to provide a variety of readings, possibly redundant,
Multiple characteristics of the same device or line may be
and communication interfaces 40, 41 may be one or more 40
general purpose computers programmed according to the
present invention.
Communications interface 40 is coupled via suitable
in a storage device (not shown). In cooperation with processor
10, controller 20 processes the sensor readings.
means, such as a wireline or wireless connection, to public
switched telephone network 101, which in turn is coupled to
third party server 110. The third party may be, e.g., a police
station, ?rehouse, or other service.
Communications interface 41 is coupled via suitable
so that processor 10 will process it in due course.
Processor 10 processes sensor readings to predict faults
and to detect faults. Processor 10 reports status of the devices
In some cases, processor 10 takes control action on its own,
Sensor 5 senses the operating condition of one or more
suitable for the device, and may include temperature read
such as isolating failed devices by eliminating power. In other
cases, processor 10 responds to instructions entered locally
by an input device (not shown), or received from a remote
controlling unit, such as server 100 or 110.
Processor 10 may also communicate with conventional
monitoring systems, such as home security systems via an
wall on/off switch, a wall dimmer, a fusebox, power-carrying
wires, communications wires and so on. The operating con
dition includes environmental conditions such as tempera
ture, humidity and so on. The sensing occurs in a manner
and wires being monitored to display 30, printer 31, and
possibly other noti?cation devices such as an audible alarm.
server 100, such as directly or via communication network
components, such as a wall receptacle, a plug-in group of
receptacles (also referred to as a current tap or power strip), a
For example, controller 20 may determine if a sensor read
ing is outside of a range, and then alert processor 10 that
emergency processing is needed. If the sensor reading is
within the range, then controller 20 simply stores the reading
means, such as a wireline or wireless connection, to commu
nications network 102, such as the Internet, which is also
coupled to server 100 and device 50. Server 110 is coupled to
discussed below.
Controller 20 receives the sensor readings and stores them
interface (not shown).
Processor 10 also reports status to server 100 and server
ings, voltage readings, power readings, image readings,
Server 100 and server 110 can query processor 10 for
status. For example, if sensor 5 is a camera, then server 100
can command processor 10 to obtain an image from sensor 5.
lines, sensing may include the name and identity of a device,
Server 100 functions in similar manner as processor 10,
such as its Internet Protocol (IP) address, and other network 65
monitoring functions. The sensing may occur in a passive or
except server 100 can run more sophisticated software, and
acoustic readings and so on. In the case of communication
an active mode. Sensor 6 is generally similar to sensor 5, but
can combine readings from a wider variety of sensors and
US 8,244,405 B2
systems. For example, server 100 may run image processing
“on” condition exhibits a particular pattern of current activity,
corresponding to a second ?ngerprint.
The analysis software determines whether a normal event
software that is beyond the processing capability of processor
Server 100 can poll processor 10 for its status, and receive
messages from processor 10, either periodically or in
response to events. Server 100 can download updates to pro
is occurring by comparing the duration of activity with the
duration of the known ?ngerprints; if the duration is outside
the ?ngerprint time bounds, then an abnormal event is occur
ring and a fault is likely.
The analysis software is further able to determine whether
cessor 10.
Since server 100 communicates with many systems, it
serves as a one-stop point of contact for a user, such as a
a normal event is occurring in a normal manner by comparing
homeowner away from home. The user can communicate
the pattern of actual activity with the patterns of the known
with server 100 via device 50, which may be a personal
?ngerprints, and determining abnormal operation exists
computer, cellphone with Internet browsing capability,
when the actual pattern is outside the threshold of the ?nger
vehicle telephone, or other suitable device. Server 100 may
prints, and so a fault is likely.
The system has the capability to learn from new informa
tion or patterns not previously de?ned, and can conduct sig
nal-based FDD, and can incorporate such new patterns with a
communicate alerts or other status information to device 50.
For example, a third party server (not shown) may be a
child location tracking service, and the user can readily deter
mine, via device 50, whether the child is near a failed recep
learning tool, such as an expert system, that expands the
knowledge database for FDD.
Energy conservation procedures are implemented when
the sensed information is used to control the devices being
In a fault detection and diagnosis (FDD) process, sensor
readings are obtained and analyZed to detect and/ or predict
faults. A “fault” refers to any operation outside a “normal”
operating range, that is, cessation of operations is not neces
sarily needed to be in a fault condition. For example, an
electrical outlet with serial arcing is likely to exhibit fault
characteristics. As another example, excessive energy use by
In another embodiment, event conditions are learned by the
analysis system.
For example, initially “normal” is de?ned as a particular
value, and “within normal range” is de?ned as a predeter
mined difference from the particular value. When the system
detects operation outside of normal range, it produces an
alarm; if a human then accepts the operation regularly, such as
three times, the system adjusts the predetermined difference
to a new amount, thereby learning the acceptable behavior of
the system.
a device may be considered a fault, and the device can be
automatically controlled to reduce its energy usage.
jump across a gap. Unwanted arcs in electrical circuits can
cause ?res.
Sampling sensors at a low rate may miss events such as
transients occurring between sampling points. However, gen
erally, it is not practical to continuously sample many sensors
at a high sampling rate, because (i) a large amount of data will
need to be transmitted, (ii) when comparing data from differ
ent sensors, synchronization becomes dif?cult with huge vol
umes of data, and (iii) storage limitations make it costly to
save huge amounts of data; typically only a prede?ned
The duration of a single arcing event is generally instanta
neous, such as less than one microsecond. The amplitude of
an initial decrease in current followed by an increase in cur
a series of spikes as one signal. FIG. 2 shows an instance ofa
FIG. 3 shows how an optical sensor detects the arcing event
sensors. The sensor readings from all sensors are analyZed,
Continuous arcing signals represent continuous arcing due
and when ?rst event conditions are met, the sampling rate of
are met.
Event-triggered sampling relies on bi-directional commu
resistance, and normally display repeatable patterns useful
nication. Generally, sensor readings are sent from sensor 5 to
20 to sensor 5.
Sensors can include current sensors (for example, Hall
effect sensors), temperature sensors such as a thermocouple,
humidity sensors, optical sensors, spatial thermal imaging
sensors (infrared cameras), other regional sensors, and so on.
Sensors can be analog or digital.
In one embodiment, the ?rst and second event conditions
are prede?ned, such as by a person. Prede?ned event condi
tions exhibiting a certain pattern over an approximate time
interval are sometimes referred to as a ?ngerprint or pro?le. 60
Analysis is typically called ?ngerprint analysis with pattern
For example, when a device is plugged into a receptacle,
there may be an immediate burst of current activity, possibly
corresponding to sparks on the blades of the plug; this corre
sponds to a ?rst ?ngerprint. As another example, a device that
is plugged into a receptacle and changed from an “off ’ to an
to, e.g., intermittent contact or an aging connection, and dis
play a ?ngerprint detectable using FDD and pattern recogni
tion. Continuous arcing signals are normally below the regu
lar signal amplitude because of the increase in contact
predetermined time interval or until second event conditions
controller 20, and control information is sent from controller
single arcing event.
of FIG. 2. Generally, the optical sensor’s digital output cor
responds to the spark that a human observes.
A ?rst sensor is sampled at a higher data rate than other
other sensors is adjusted, the adjustment occurring during a
the instantaneous spike in an arcing event is characterized by
rent, as shown in FIG. 2. It is helpful to distinguish between
the spikes in a series of single arcing events, rather than treat
amount of data is saved which limits historical analysis to
only the timeframe of predetermined amount of data.
An event-triggered sampling rate is determined as follows.
Use cases will now be discussed with respect to arcing.
Arcing refers to an electrical current in which electrons
for diagnosis. FIG. 4 shows a typical continuous arcing signal
pattern. The current initially is at a steady-state level. When
continuous arcing occurs, the current displays a quick drop in
values and jumps up and down with a ZigZag pattern. The time
duration is about 0.01-2 seconds.
Use Case 1: FDD Based on Fingerprint
FIG. 5 is a chart showing two instances of single arcing,
labeled “eventl” and “event2”. The abscissa (x-axis) shows
time while the ordinate (y-axis) shows current. In this
example, sensor 5 is an analog current sensor, whose output,
shown in FIG. 5, is provided to processor 10, which detects
when the signal amplitude has exceeded ?ve times the normal
signal amplitude for less than one microsecond, and deter
mines that this is an arcing event because such behavior ?ts
the characteristics of the arcing ?ngerprint.
After determining that an arcing event exists, in this
example, processor 10 continues to monitor the signal. One
single arcing event per 5 minutes is de?ned as acceptable.
US 8,244,405 B2
TWo or more single arcing events per 5 minutes is de?ned as
not, by de?nition, exceed the threshold of safety; hoWever, the
a possible fault, and processor 10 reports this condition to
combination of marginally safe operation based on the tWo
server 100. Four or more single arcing events per 5 minutes is
de?ned as a threat, and processor 10 sends a signal to a circuit
sensors can present a safety concern if the operating param
eters are maintained at this level for extended period of time.
breaker (not shoWn) to shut off the current supplied to sensor
5, and of course, any device that is draWing such current, and
The point “b” illustrates such a condition in this region, Which
may cause a fault in due time. The “green region” is the region
of safe operation, illustrated With a point “a.”
processor 10 also reports its automatic shut off to server 100.
Use Case 2: FDD With Redundancy
Such vectoriZed map of a pair of sensory data can be
applied to include more pairs of data. If there are n such pairs,
We effectively Will have n Ways of looking at the fault under
consideration. The information rendered by these n different
FIG. 6 is a chart shoWing hoW a digital optical sensor reacts
to the signal of FIG. 5, and generally corresponds to the
sparks that humans identify as arcing.
In this example, processor 10 receives the output of an
analog current sensor and a digital optical sensor. When the
sources can complement one another and corroborate to ren
der a more reliable diagnosis because of the redundant infor
current sensor senses an amplitude of at least ?ve times nor
An iFDD system is adopted, Which is capable of exploiting
both time-domain and frequency-domain analysis to comple
ment each other. Such system provides redundancy in data
mal and the optical sensor turns on, then processor 10 con
cludes that an arcing event has occurred, Without concern for
the duration of time that the analog signal amplitude exceeds
?ve times the normal. Subsequent processing of processor 10
analysis and can deliver more precise diagnosis because
information on both domains are utiliZed and corroborated.
may be as in use case 1.
Use Case 3: FDD With Change in Sampling Rate
Such kind of domain redundancy in analysis can be achieved
FIG. 7 is a chart shoWing, as “curveA”, the current versus
time for an intermittent contact; as “curveB”, a sampling rate
by signal processing algorithms such as Wavelet analysis.
that is dynamically changed; and as “curveC”, the output of a
sensor Whose sampling rate is dynamically changed.
Let it be assumed that curveA represents an analog signal
lished by using multiple pairs of sensory information. For
example, the “VectoriZed Map for iFDD” discussed above
Using multiple sensors: redundancy can also be estab
uses a pair of sensory information. When a different pair of
data is employed for FDD, the results can provide “redun
dancy” to the ?rst diagnosis. The additional diagnosis can
con?rm the results of the ?rst diagnosis, or to provide addi
tional insights to the ?rst diagnosis. In either case, the tWo
for a connection that is not secure, i.e., a connection that
exhibits an intermittent contact situation. Assume the sam
pling rate is as shoWn in curveB, and that curveC represents
What is sampled, such as a sensor that turns on When the
current is beloW a threshold value, shoWn as “threshold” 30 complements each other and can enhance the accuracy of the
relative to curveA.
diagnosis. As a result, the con?dence level of FDD is
The “redundancy” referred to above is a result of having
Beginning at the left side of curveC, no signal is produced,
so the sampling rate stays at a ?rst loWer rate. Then, When a
signal is produced, the sampling rate is shifted to a second
higher rate creating a smaller sampling interval that improves
the likelihood of capturing critical information. The sampling
rate remains at the higher rate until no signal is produced in
?ve successive sampling intervals, at Which time the sam
pling rate is shifted back to the ?rst loWer rate. Five intervals
is chosen for ease of illustration; in actuality, the number of
more sensors than the minimum set (in this case, one set of
Recognize ?ngerprint: for example, refrigeratoriturning
off and on of the compressor. The transition Will emit a
occurring, from a fault detection perspective.
A vectoriZed map for intelligent fault detection and diag
nosis (iFDD) and redundancy Will noW be discussed.
A vectoriZed map is akin to the Cartesian coordinates With
XY axes. Every tWo sets of data from tWo sensors can be used 50
to construct a vectoriZed map to represent the characteristics
of faults of a system, in a Way similar to that of XY Cartesian
coordinates With points on the XY space represented by the
coordinates (x, y). This type of vectoriZed map, after calibra
tion, can provide very useful information for diagnosis. FIG.
8 provides an illustration of hoW this type of vectoriZed map
When such tWo signals are plotted on the vectoriZed map, an
intermediate region indicated by the “yelloW region” does
from the operation Without fault. When a fault is devel
oped (in this a leakage in the system), the ?ngerprint of
the signal for the same operation changes and can be
ence is the same, but the response from different sources
of fault can be different. This can also be utiliZed as
information for FDD. The data shoWn are from Wavelet
analysis (for example, cD3 Wavelet data).
In FIG. 8, the current is in the horiZontal axis, and the
temperature measured at a prescribed location is in the verti
and the “temperature threshold”. For example, an operating
situation illustrated by the point “c” indicates such situation
that requires immediate attention due to the presence of fault.
signal Which has its oWn ?ngerprint. As an example,
When the ?ngerprint changes suddenly, a fault is likely to
be developed in the system. The recognition of such
?ngerprint and its changes is a reliable Way of imple
menting iFDD. As another example, shoWn in FIGS. 9A
and 9B, the reference ?ngerprint is shoWn in solid line
used for iFDD. The ?ngerprint (in solid lines) of refer
can be used.
cal axis. Each measurement has its oWn threshold value above
Which a fault can be triggered, as shoWn in the “red region”
area With the arroWs indicating the “max. alloWable current”
A system that is capable of learning and modifying the rec
ognition of events is desirable to intelligently adapt to differ
ent operating conditions. This can be done as folloWs:
intervals is about 50-500 intervals, or an interval amount
determined by a FDD algorithm.
Use of a dynamic sampling rate increases the chance that
events of interest Will be captured, While reducing the data
rate during “normal” operation, When no events of interest are
data pair) needed for FDD. Such redundancy provides addi
tional bene?ts in the process of FDD.
Learning Will noW be discussed.
Event-triggered iFDD depends on the detection of events.
Learning of behavior: This is best understood by an
example. Let’s take the air-conditioner as an example,
With a compressor for the heat exchange cycle. When the
compressor has been operating normally for a While, a
certain reference operating performance is established.
At some point, a WindoW in the room Was left open
Which causes the unit to operate harder and turn on for
longer period of time, deviating from the nominal oper
ating behavior. If this persists, a Warning can be issued to
the use to check the surrounding and check if WindoWs/
US 8,244,405 B2
Generally, Wires can be coupled to receptacles via the
doors are to be closed to bring the air-conditioner back to
normal operations. This learning behavior can save
side-Wire method, in Which Wire is Wrapped under a screW
energy by using the aforementioned iFDD algorithm.
head, the back-Wire method, in Which Wire is inserted from
A faceplate for an electrical receptacle Will noW be dis
cussed. The faceplate has a sensor for sensing operation of the
behind through a hole or slot and clamped under a clamping
plate as the screW is tightened, or the push-Wire method, in
device plugged into the electrical receptacle.
Which a Wire is simply pushed into a terminal and clamped by
a spring-loaded brass member inside the terminal. The push
FIG. 10 shoWs an embodiment Where the sensor is on the
backside of the faceplate. An embodiment Where the sensor is
on the front side of the faceplate is contemplated, but not
shoWn. As used herein, front side is the side that is visible
When the faceplate is installed over an electrical receptacle.
Wire method causes many loose connections, and is not
favored for this reason. FIG. 11 shoWs receptacle 380 adapted
for side-Wire connection. ScreWs 385, 386 are non-conduc
tive screWs, such as plastic, rather than the typical metallic
FIG. 10 shoWs faceplate 200 having apertures 201, 202. In
SWitch elements 360, 370 are placed betWeen the contacts
a standard receptacle, there are tWo receptacles, a top recep
tacle and a bottom receptacle. Aperture 201 surrounds the top
receptacle, While aperture 202 surrounds the bottom recep
on the side of receptacle 380 and the household Wiring (not
tacle. Faceplate 200 also has sensor strips 211, 212, 221, 222,
ing screWs 385, 386. ScreWs 385, 386 respectively insert
through the toroidal centers of sWitch elements 360, 370.
SWitch element 370 is has conductive plates 375, 376, such
shoWn), the household Wiring being held in place by tighten
each of Which is formed of a heat conducting material, such as
used to dissipate heat from computer chips, described for
example in European Patent EP0696630, “Heat conductive
material and method for producing same”, Feb. 14, 1996, the
disclosure of Which is hereby incorporated by reference. The
Width of each sensor strip is such that its inWard edge, relative
as brass or copper alloys, on either side, to ensure conduction
respectively. Plates 375, 376 are connected to Wires 377, 373.
Movable Wire 372 is controlled by control Wire 371 to con
nect betWeen Wire 377 and either Wire 373 or ground 374.
to the aperture is determined so that the inWard edge contact
a blade of a plug that is plugged into the receptacle.
Sensor strips 211, 212, 221, 222 are constructed similarly.
For brevity, only sensor strip 211 is discussed.
During normal operation, movable Wire 372 is connected
Sensor strip 211 is coupled to element 231 that converts its
temperature to an electrical signal. Coupling occurs, for
example, by connecting element 231 to sensor strip 211 via
epoxy. Suitable temperature sensors for element 231 include:
Analog Devices AD5 90 2-Terminal IC Temperature Trans
ducerian integrated circuit that produces an output
cient (PTC). NTC thermistors have a highly non-linear
change in resistance and reduce their resistance With
increases in temperature. Epoxy coatings are used for
temperatures from —50 to 1500 C. (—58 to 316° F). See
elements 411, 412, that function similarly to elements 211,
212 discussed above With regard to faceplate 200.
to 400 times the output of a thermocouple With no need
the ZM3102N Z-Wave Module, described in the background
section of this application. Transmitter 250 obtains poWer
from battery 255.
Although faceplate 200 has been described in an embodi
ment Wherein the temperature is sensed, other embodiments
Bottom prongs 404, 406 provide poWer to transformer 460.
During normal operation, transmitter element 450 derives
poWer through bottom prongs 404, 406. HoWever, When an
exception condition occurs, or When commanded by proces
sor 10, transmitter element 450 sWitches to battery 450 for its
In other embodiments, sensors are provided for each of
receptacles 470, 480, 490.
FIG. 13 is a schematic of circuit 500 according to the
present invention, shoWn as a receptacle. In another embodi
are contemplated, such as Where current or other character
istic is sensed.
ment, the circuit is a plug-in device, for retro?tting an existing
outlet, lacking sensors 550, 560.
Line current is supplied to load 590 via sWitch 510 con
trolled by controller 511.
FIG. 11 shoWs a faceplate that senses an operating condi
tion and, upon detection of an exception condition, cuts off
poWer to the electrical receptacle and noti?es an external
processor. FIG. 11 is similar to FIG. 10, and only the differ
having tWo slots for the prongs of a device plug, and also each
having an opening for a ground prong of a device plug.
Top prongs 403, 405 provide poWer to receptacles 470,
Betatherm BetaLinear 36K53Al thermistor, providing up
operating at 0 to 100° C.
The output of the temperature sensor element 231 is trans
mitted, at predetermined intervals, to a controller, such as
processor 10. Speci?cally, the output of the temperature sen
sor element is provided to transmitter element 250, such as
400 has a top pair of prongs 403, 405, top ground prong 407,
a bottom pair of prongs 404, 406, and bottom ground prong
408, for respectively plugging into a household outlet. Cur
rent tap 400 also provides receptacles 470, 480, 490, each
480, 490. Prongs 403, 405 are in conductive contact With
for junction temperature or lead Wire compensation,
betWeen Wire 377 and Wire 373, so that poWer ?oWs through
receptacle 380. When transmitter element 350 detects that the
sensed temperature exceeds a predetermined temperature,
transmitter element 250 sends a signal along control Wire 371
to sWitch movable Wire 372 to ground 374, thus cutting off
poWer to receptacle 380. In other embodiments, controller 10
commands transmitter element 350 to control sWitching ele
ments 360, 370 to cut off poWer to receptacle 380.
FIG. 12A shoWs a back vieW current tap 400. FIG. 12B
shoWs a side vieW of current tap 400. Generally, current tap
current proportional to absolute temperature, at tem
peratures up to 150° C. (316° F). See WWW.analog.com.
Omega.com 44000 Series ThermistoriA thermally sen
sitive resistor available in tWo types: negative tempera
ture coe?icient (NTC), or positive temperature coef?
betWeen the household Wiring and the side of receptacle 380,
Temperature sensor 520, such as a thermocouple (e. g.,
Omega part 5TC-TT-K-36-36) or thermistor (e.g., General
370 is discussed in detail. SWitch elements 360, 370 are torus
Electric part RL503-27.53K-l20-MS), is located next to the
screW (not shoWn) or other fastener for the line Wire, as this
location tends to become hot in a receptacle. Another hot spot
is the blades of a plug. In a plug-in retro?t device, sensor 520
is located near the blade that is plugged into the plug-in
retro?t device. Sensor 520 produces a sensed temperature
(doughnut) shaped.
signal, and supplies the sensed temperature signal to sample
ences are discussed for brevity.
FIG. 11 shoWs faceplate 300 that ?ts over receptacle 380.
Transmitter element 350 also is connected to sWitch elements
360, 370, Which are similar to each other; only sWitch element
US 8,244,405 B2
and hold data acquisition (SHDAC) element 521 (e.g.,
Process 680 Will noW be described. Process 680 includes
National Semiconductor part ADCl6l5626) that serves to
steps 605, 610, 615, 620, 625, 630, 635, 640A-640C.
At step 605, microprocessor 570 sets sampling frequency
latch (preserve in time) the sensed temperature signal, so that
its time reference can be synchronized With other sensed
signals. SHDAC element 521 supplies the latched sensed
temperature signal to analog-to-digital converter (ADC) 522
that operates to convert the analog sensed temperature signal
At step 610, microprocessor 570 receives values fromADC
522 at the rate determined by signal fT, in this example, 100
to digital data, and supplies the digital sensed temperature
signal to microprocessor 570.
samples per second, and stores them in an internal circular
buffer or storage such that only the most recent values are
Light sensor 530, such as a photodiode (e.g., Advanced
Photonix Inc. part PDB-Cl 58F) or infrared camera, is located
near the screW (not shoWn) or other fastener for the line Wire.
stored. For example, if the buffer siZe is 1000 samples, then
the most recent 10 seconds of sensed values are stored.
At step 615, Which occurs periodically such as every 30
Sensor 530 produces a sensed light signal, and supplies the
sensed light signal to SHDAC 531 that operates in similar
manner as SHDAC 521. ADC 532 operates in similar manner
as ADC 522 to supply a digitiZed sensed light signal to micro
processor 570.
Audio sensor 540, such as a microphone (e.g., Knowles
Acoustics part MD9745APA-l), is located at a convenient
place in or on the receptacle. Sensor 540 produces a sensed
Serial arcing is typically found in outlets Which have aging
Wire connections and/or degraded joints. The serial arcing is
541 that operates in similar manner as SHDAC 521. ADC 542
operates in similar manner as ADC 522 to supply a digitiZed
sensed audio signal to microprocessor 570.
Current sensor 550 is a Hall-effect current sensor (e.g., 25
Functional Devices Inc. part RIBX420 or Eaton Cutler-Ham
mer part EACl420SP) and senses the line current supplied to
load 590. Sensor 550 produces a sensed current signal, and
supplies the sensed current signal to SHDAC 551 that oper
load is applied to the outlet. The temperature can rise to over
2000 C. is 10 seconds for a continuous serial arcing connec
trip signal, and supplies the ground fault trip signal to SHDAC
ground fault trip signal value to microprocessor 570.
Microprocessor 570 is a general purpose microprocessor
programmed according to the present invention, and includes
suitable memory (not shoWn). Microprocessor 570 receives
the digitiZed values from ADCs 522, 532, 542, 552, 562, and
caused mainly by intermittent contact due to rusted Wire
and/or rusted screW and/or degraded junction Which causes
accelerated failure of connection. Once serial arcing begins,
the connection continues to degrade until it becomes haZard
ous because of the resulting sharp temperature rise When a
The problem analysis at step 615 tries to detect a rising
temperature in the sampled data, as folloWs.
current signal to microprocessor 570.
Optional ground fault sensor 560 produces a ground fault
561 that operates in similar manner as SHDAC 521. ADC 562
operates in similar manner as ADC 522 to supply a digitiZed
seconds, microprocessor 570 analyZes the stored values to
produce an analysis result.
The analysis at step 615 analyZes the sensed temperature
signals in vieW of the type of problem being detected to
produce a con?dence estimate of the problem. For example,
When the problem being detected is serial arcing, the tem
perature in the area surrounding the arcing characteristically
rises in a slanted saWtooth curve, as shoWn in FIG. 15.
audio signal, and supplies the sensed audio signal to SHDAC
ates in similar manner as SHDAC 521. ADC 552 operates in
similar manner as ADC 522 to supply a digitiZed sensed
control signal fT to a ?rst value fT 1, such as 100 HZ, and
supplies control signal fT to temperature sensor 520.
T0 be the oldest data sample in the circular buffer,
Tn be the most recent data sample in the circular buffer, and
Tavg be the average value of all samples in the circular
generates a control signal for controller 511 to interrupt the
line current, generates control signal 581 for status light(s)
then the samples probably are characteristic of a rising tem
580, Which may be one or more light emitting diodes (LEDs)
or other suitable device, generates control signal 586 for siren
585, Which may be a speaker or other device emitting an
audible signal, and generates a communication signal for
Wireless communication interface 505 that communicates
With controller 20 of FIG. 1. Microprocessor 570 also
receives communication signals via communication interface
505 from controller 20. Microprocessor 570 also generates
speci?c tests or values Will be apparent to those of ordinary
skill in the art; an important feature of the test is that it is to
perature and hence serial arcing may be occurring. Other
detect the behavior expected to be found When the problem
lem, processing proceeds to step 625.
sampling frequency control signals fT, fL, fA, ? and respec
tively supplies these control signals to sensors 520, 530, 540,
FIG. 14 is a ?owchart for the operation of microprocessor
570 of FIG. 13.
At step 600, microprocessor 570 is turned on, such as by
receiving poWer, or is reset by a reset button (not shoWn) in the
receptacle or plug-in retro?t device containing circuit 500.
Microprocessor 570 then simultaneously executes ?ve
processes 680-684 respectively corresponding to sensors
signals to fault analysis step 650 that determines Whether to
take one or more of a variety of actions, such as interrupting
activating/de-activating siren 585, sending a message to con
troller 20 and so on.
At step 625, Which occurs if a fault is determined at step
620 or can also occur if problem signal BB is received, micro
processor 570 sets sampling frequency control signal fT to a
second value fT 2 that is higher than the ?rst value fT 1, such as
fT2:l000 HZ, and supplies control signal fT to temperature
sensor 520. Microprocessor 570 also generates problem sig
nal BB and supplies it to steps 626, 627, 628 ofprocesses 681,
682, 683.
520, 530, 540, 550, 560. Processes 680-684 supply problem
the line current, activating/de-activating status light(s) 580,
At step 620, microprocessor 570 determines Whether the
analysis result produced at step 615 indicates there is a prob
lem. If not, processing returns to step 605. If there is a prob
It Will be appreciated that, in this manner, When any of
sensors 520, 530, 540, 550 generates data indicating a pos
sible fault, all of the sensors then begin to sample at a higher
frequency, enabling analysis in a more time- granular fashion.
That is, When things appear normal, a ?rst loWer sampling
rate is used to reduce poWer consumption, While When a
possible fault occurs, a second higher sampling rate is used
despite the higher poWer consumption to enable detection of
US 8,244,405 B2
transient faults. In embodiments Where power consumption is
not a concern, the second higher sampling rate may be used
non-arcing current is basically ?at With high frequency noise.
W-shaped period.
At step 630, microprocessor 570 receives data from ADC
522 at the higher rate, in this example, 1000 samples per
spectra of the current vs. frequency curves for normal opera
In contrast, the serial arcing curve has a pronounced
FIG. 17 shoWs exemplary Fast Fourier Transform (FFT)
tion and serial arcing. The normal non-arcing FFT spectrum
second, and stores them its internal circular buffer or storage
such that only the most recent values are stored. For example,
if the buffer siZe is 1000 samples, then the most recent 1000
milliseconds of sensed data are stored.
At step 635, Which occurs periodically such as every 500
has a lot of energy in loW frequencies and asymptotically
decreasing energy in higher frequencies, With pronounced
spikes at 60 HZ and 180 HZ. The serial arcing curve is similar
but also has pronounced spikes at 300 HZ and 420 HZ. In
different con?gurations, the spikes may occur at different
milliseconds or every 100 milliseconds or every 2 seconds,
frequencies, but there Will be additional spikes in the arcing
spectrum relative to the normal operation spectrum.
microprocessor 570 analyZes the stored values to produce an
analysis result. The analysis at step 635 is similar to the
analysis at step 615 but occurs With more temporal granular
ity due to the higher sampling rate, and instead of a binary
result (problem or normal), the result is a con?dence estimate
At step 618, a test corresponding to FIG. 16 is used. For the
stored current samples in the circular buffer, the folloWing
values are determined:
Imin?he minimum value of the current samples
Imax?he maximum value of the current samples
Iavg?he average value of the current samples
of Whether a fault exists. Let 771 be the CONFIDENCE value
for the temperature samples. An example analysis is:
If (Tn>2T0) then (111:1 .00)
else if (Tavg<32° F.) then (111:0)
The test is:
If (Imin<0.95*Iavg) then serial arcing exists, else no serial
arcing exists.
else (11l:0.97)
This analysis says that if the mo st recent temperature sample
is tWice the initial temperature sample, then there is de?nitely
a problem (CONFIDENCE:100%). If the average tempera
ture of the temperature samples in the circular buffer is at least
1500 F., then the CONFIDENCE that there is a problem is
97%. While the average temperature is betWeen 32° F. and
150° F., then the CONFIDENCE is the normaliZed tempera
ture difference betWeen the oldest and neWest samples. If the
average temperature is under 32° F., then there is de?nitely
At step 638, a more elaborate version of a test correspond
ing to FIG. 16 is used. In other embodiments, other tests are
used. Choose an upper bound Bu, such as [3u:0.3, and a loWer
bound [31, such as [31:003. The con?dence value for the
current samples is 114, computed as:
not a fault.
At step 640A, microprocessor 570 determines Whether the
analysis result produced at step 635 indicates there is a fault.
buffer are subjected to a Fast Fourier Transform.
If not, processing returns to step 605. If there is a fault,
processing proceeds to step 650. Continuing With the above
example, if 111 is at least 0.5, then there is a problem.
It Will be appreciated that the sensed temperature data can
be analyZed in different Ways to detect different types of fault
s. As described above, the fault of serial arcing is considered.
Let v?he magnitude of the frequency spectrum at an iden
ti?ed arcing frequency, such as 300 HZ.
Let Amax:a predetermined maximum value for the mag
nitude of the frequency spectrum at the identi?ed arcing
Let Amin:a predetermined minimum value for the magni
tude of the frequency spectrum at the identi?ed arcing fre
Other faults, such as appliance malfunctioning (the appliance
is load 590) or ?re haZard, may be considered by analyZing
the sensed temperature values in different manner. In some
In another embodiment, at step 618, a test corresponding to
FIG. 17 is used. The stored current samples in the circular
The speci?c values of Amax and Amin are determined by
embodiments, microprocessor 570 conducts multiple analy
the characteristics of the signals and arcing for speci?c loads
ses on the sensed data, and thus has steps 640B, 640C and so
and/or outlets.
on corresponding to the different types of fault analyses. In
The test is:
this embodiment, if one or more fault s are detected, process
ing proceeds to step 650, that is, only if no fault s are detected
does processing return to step 605.
If (v>Amin) then serial arcing exists, else no serial arcing
In another embodiment, at step 638, a more elaborate ver
sion of a test corresponding to FIG. 17 is used. The con?dence
value of a fault based on the current samples is 114, computed
Process 681 includes steps 606, 611, 616, 621, 626, 631,
636, 641A-641C. Process 681 is similar to process 680, and
for brevity, only differences Will be discussed. When serial
arcing occurs, the curve of light generated is similar to the
curve of current, discussed beloW, and the light samples are
analyZed similarly.
{ 0.5 * [(v — Amin)/(Amax — Amin)] + 0.4, Amin < v < Amax
Process 682 includes steps 607, 612, 617, 622, 627, 632,
T74 =
v z Amax
637, 642A-642C. Process 682 is similar to process 680, and
for brevity, only differences Will be discussed. When serial
Process 684 includes step 624, Wherein a ground fault trip
signal from ground fault sensor 560, as digitiZed, is simply
passed to step 650.
At step 650, triggered by arrival of a problem signal and
arcing occurs, the curve of sound generated is similar to the
curve of current, discussed beloW.
Process 683 includes steps 608, 613, 618, 623, 628, 633,
638, 643A-643C. Process 683 is similar to process 680, and
for brevity, only differences Will be discussed.
then occurring for one or more sub sequent periods to see if the
FIG. 16 shoWs typical current vs. time curves for serial
problem has cleared, microprocessor 570 performs redundant
arcing and for normal current Without arcing. The normal
fault analysis by combining the CONFIDENCE values of the
US 8,244,405 B2
problem signals, if any, from processes 680-684. A separate
is taken. For example, microprocessor 570 may be pro
grammed to With the folloWing rule:
analysis is performed for each type of problem being consid
ered. Additionally, if the ground fault trip signal indicates a
ground fault, then the fault analysis immediately indicates a
If (11famt>0.95) then (activate siren 585)
else if (0.80<11?,mt<0.95) then (send an alert message to
controller 20)
ground fault so that the current is interrupted to load 590.
Other actions Will be apparent to those of ordinary skill. For
example, if load 590 is an air-conditioner draWing too much
current, the action may be to regulate the voltage or current
In one embodiment, the fault con?dence signals are com
bined as folloWs:
and send an alert message to controller 20 to calibrate the
air-conditioner as it is running inef?ciently.
At step 655C, corresponding to another fault such as ?re
haZard, microprocessor 570 determines What action, if any, it
should take. If no action is to occur, processing continues at
In this example, n:4 since the results of four sensors are
step 670. OtherWise, at step 660C, the action is taken.
At step 670, if no faults are detected, then normality signal
provided at step 650.
111 corresponds to the CONFIDENCE of the sensed tem
perature from process 680.
112 corresponds to the CONFIDENCE of the sensed light
from process 681.
113 corresponds to the CONFIDENCE of the sensed audio
from process 682.
114 corresponds to the CONFIDENCE of the sensed current
from process 683.
The con?dence values 111,112,113,114 are betWeen 0.0 and 1.0.
AA is generated and provided to processes 680-683 so that all
processes return to a loWer sampling rate.
it is to be understood that the invention is not limited to this
precise embodiment and the described modi?cations, and
that various changes and further modi?cations may be
In another embodiment, the problem signals are Weighted;
effected therein by one skilled in the art Without departing
from the scope or spirit of the invention as de?ned in the
appended claims.
for instance, the sensed current may be deemed to be a better
indicator than the sensed light, so the sensed current in?u
ences the outcome of the redundant fault analysis more than
the sensed light.
Although an illustrative embodiment of the present inven
tion, and various modi?cations thereof, have been described
in detail herein With reference to the accompanying draWings,
What is claimed is:
1. A method of performing problem detection/prediction,
For example, assume con?dence values 11 1:062, 112:0.7 l ,
diagnosis, and resolution Within an electrical provision
and 114:0.82 are delivered to step 650; in this example, there
device While minimizing poWer consumption, said method
does not seem to be an audio problem perhaps because of loud
ambient noise so no value for 113 is delivered to step 650, i.e.,
113:0. The values are combined as:
using a microprocessor With a memory to simultaneously
set a loW, poWer-saving sampling frequency for one or
more sensor types,
using said one or more sensors to sense one or more respec
tive operating characteristics of the electrical current
provision device at said loW sampling frequency,
storing recent readings of each of said one or more sensors
in a buffer portion of said memory,
Thus, using three samples increases the overall level of con
?dence dramatically. That is, a fault that manifests simulta
neously in multiple domains, even if it is not so serious in each
domain, gives rise to strong con?dence that a problem truly
exists. It is preferred to use at least three domains, that is,
redundancy in tWo domains.
At step 655A, corresponding to the problem of serial arc
ing, microprocessor 570 determines What action, if any, it
periodically analyZing said readings in said buffer using
said microprocessor, by comparing said readings in said
buffer With an accepted normal range for each of said
respective sensor readings, Wherein each of said read
ings being Within said range comprises normal operation
and Wherein at least one of said sensor readings exceed
ing said range comprises abnormal operation,
instituting an event-triggered sampling frequency for each
of said one or more sensors When said at least one sensor
should take. If no action is to occur, processing continues at
reading in said buffer exceeds said accepted normal
step 670. OtherWise, at step 660A, the action is taken. For
example, microprocessor 570 may be programmed to With
the folloWing rule:
If (11fauZt>0.98) then (set signal 511 to open sWitch 510 and
make status light 580 red)
else if (0.95<11?,mt<0.98) then (make status light 580 blink
range, said event-triggering sampling frequency being at
a higher sampling frequency than said loW, poWer-sav
ing sampling frequency,
triggered sampling frequency in said buffer portion of
said memory,
frequently analyZing said event-triggered readings in said
ing and red)
buffer, said frequent analyZing comprising determining
else if (0.90<11?,mt<0.95) then (make status light 580 red)
else if (0.70<11?,mt<0.90) then (send an alert message to
controller 20)
Other rules for actions Will be apparent to those of ordinary
At step 655B, corresponding to another fault such as appli
ance malfunction, microprocessor 570 determines What
storing said one or more sensor readings at said event
a con?dence value, for said at least one sensor, indicat
ing an amount of con?dence that a fault exists, said
frequent analyZing of said event triggered sensor read
ings occurring at a frequency being greater than said
periodic analyZing of said sensor readings,
performing a redundant fault analysis, When said at least
action, if any, it should take. If no action is to occur, process
one abnormal sensor reading persists for tWo or more
ing continues at step 670. OtherWise, at step 660B, the action
occurrences of said frequent analyZing of said event
US 8,244,405 B2
triggered readings, said redundant fault analysis com
11. The method of claim 1, Wherein said loW, poWer-saving
prising a combined con?dence value for at least tWo
sampling frequency comprises sampling at approximately
100 HZ; and Wherein said higher event-triggered sampling
frequency comprises sampling at approximately 1000 HZ.
taking a control action to resolve said existing fault, When
said combined con?dence value exceeds a predeter
12. The method of claim 11, Wherein said periodically
mined threshold, by said microprocessor generating a
control signal that is received by a controller, thereby
analyZing comprises analyZing approximately every 10-30
seconds; and Wherein said frequently analyZing comprises
causing said controller to command said control action,
analyZing in the range of approximately every 500 millisec
onds to approximately every 2 seconds.
generating a normality signal if said combined con?dence
13. The method of claim 1, Wherein said one or more sensor
types comprise one or more of a photodiode or camera for
value does not exceed said threshold, or if said abnormal
sensing light; a 2-axis accelerometer for sensing vibrations; a
microphone for sensing sound; a thermocouple or infrared
sensor reading clears before said tWo or more occur
rences of said frequent analyZing, said normality signal
causing return to said loW, poWer-saving sampling fre
imager for sensing temperature; a Hall-effect sensor for sens
ing current.
2. The method of claim 1, Wherein each of said one or more
14. The method of claim 1 Wherein said storing recent
readings in said buffer portion of said memory comprises
operating characteristics is one of temperature, humidity,
light, sound, vibration, voltage, poWer, and line current.
3. The method of claim 1, Wherein said fault is one of serial
storing said one or more sensor readings until completing said
arcing, appliance malfunction, and ?re haZard.
15. The method of claim 14 Wherein said periodic analyZ
ing of said recent sensor readings for serial arcing comprises:
4. The method of claim 1, Wherein said control action
comprises one or more of
sending, by said processor, a control signal for commanded
control action by said controller in response to instruc
tions entered locally on an input device or instructions
received remotely over the intemet; and
sending, by said processor, a control signal for a predeter
mined control action, said predetermined control action
comprising one or more of:
determining a minimum current value, I min: of said recent
current sensor readings;
determining a maximum current value, lmax, of said recent
current sensor readings determining an average current
value, lavg of said recent current sensor readings; and
Wherein said at least one sensor reading exceeding said
accepted normal range is Where lmin<0.95(lavg).
(a) preventing line current from ?oWing in the electrical
16. The method of claim 15, Wherein said con?dence value
of said frequent analyZing of said event-triggered current
current provision device,
readings comprises:
(b) activating a status light,
(c) activating a sound generator, and
(d) sending a message to a controller.
5. The method of claim 1, Wherein said at least tWo sensors
periodic analyZing using said readings.
have con?dence values of 111, . . . , 11”, and Wherein said
combined con?dence value, 11fault, is given by the equation:
Where [3” is an upper bound, and [3Z is a loWer bound, and
Wherein [3” is approximately 0.3 and [3Z is approximately 0.03.
17. A method of performing intelligent problem detection/
prediction, diagnosis, and resolution Within an electrical cur
rent provision device, said method comprising:
using one or more sensors to detect a possible problem,
each of said one or more sensors sensing an operating
characteristic of one or more components of said elec
6. The method of claim 5, Wherein When said at least tWo
sensors comprises tWo sensors, 11 1 corresponds to a con?
dence value of a fault from a temperature reading sensed from
trical current provision device at a ?rst sampling rate,
a ?rst sensor, and 112 corresponds to a con?dence value of a
fault from a light reading sensed from a second sensor.
7. The method of claim 6 Wherein When said at least tWo
sensors comprises a third sensor, 113, corresponds to a con?
dence value of a fault from an audio reading sensed from said
said ?rst rate being a poWer-saving sampling rate;
Wherein When said present sensor reading is Within said
third sensor.
acceptable range, storing said present sensor reading in
8. The method of claim 7, Wherein When said at least tWo
sensors comprises a fourth sensor, 114 corresponds to a con?
dence value of a fault from a current reading sensed from said
possible problem, instituting an event-triggered sam
9. The method of claim 4 further comprising instructing
saidprocessor to ignore said at least one sensor reading in said
pling rate for said one or more sensors, said event-trig
said accepted normal range to be a neW normal range that
include said abnormal sensor reading, When regularly com
manded to ignore said abnormal sensor reading.
10. The method of claim 9, Wherein commanding said
processor to ignore said abnormal sensor reading regularly
comprises commanding said processor to ignore said abnor
mal sensor reading at least three times.
a controller;
Wherein When at least one of said present sensor readings is
outside said acceptable range indicating detection of a
fourth sensor.
buffer that exceeds said accepted normal range; and adjusting
comparing a present sensor reading of each of said one or
more sensors With an acceptable range for said respec
tive sensor reading;
gering sampling rate being at a higher frequency than
said ?rst sampling rate,
analyZing said event triggered readings to produce a con
?dence estimate to diagnose Whether the problem is a
taking a control action, by a processor, to resolve said fault
When said con?dence estimate exceeds a predetermined
threshold, and
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF