Real-time spatio-temporal coherence estimation for autonomous mode identification and invariance tracking
First Claim
Patent Images
1. A method for diagnosis and prognosis of faults in a physical system comprising:
- providing sensor data representative of measurements made on the physical system, the measurements being representative of values of signals produced by the physical system;
producing model enhanced sensor signals by fitting the sensor data to at least a partial physical model of the physical system;
identifying correlated signals from among the sensor data;
comparing the correlated signals with expected correlated signals to detect one or more occurrences of events, the expected correlated signals representative of known operating conditions of the physical system;
providing discrete data representative of system status variables and system command information;
detecting discrepancies among the discrete data;
identifying the one or more occurrences of events as unmodeled events based at least on the model enhanced sensor signals; and
verifying faults based on the discrepancies among the discrete data.
2 Assignments
0 Petitions
Accused Products
Abstract
A general method of anomaly detection from time-correlated sensor data is disclosed. Multiple time-correlated signals are received. Their cross-signal behavior is compared against a fixed library of invariants. The library is constructed during a training process, which is itself data-driven using the same time-correlated signals. The method is applicable to a broad class of problems and is designed to respond to any departure from normal operation, including faults or events that lie outside the training envelope.
-
Citations
16 Claims
-
1. A method for diagnosis and prognosis of faults in a physical system comprising:
-
providing sensor data representative of measurements made on the physical system, the measurements being representative of values of signals produced by the physical system;
producing model enhanced sensor signals by fitting the sensor data to at least a partial physical model of the physical system;
identifying correlated signals from among the sensor data;
comparing the correlated signals with expected correlated signals to detect one or more occurrences of events, the expected correlated signals representative of known operating conditions of the physical system;
providing discrete data representative of system status variables and system command information;
detecting discrepancies among the discrete data;
identifying the one or more occurrences of events as unmodeled events based at least on the model enhanced sensor signals; and
verifying faults based on the discrepancies among the discrete data. - View Dependent Claims (2, 3, 4, 5)
where Si and Sj are the signals produced by the physical system,
-
-
3. The method of claim 1 further including performing a training sequence to produce the expected correlated signals.
-
4. The method of claim 1 further including identifying suspect bad signals by detecting discrepancies among the sensor data based on a statistical model of the sensor data, wherein the step of identifying the unmodeled events is based on the suspect bad signals in addition to the model enhanced sensor data.
-
5. The method of claim 4 further including identifying statistical components of the sensor data, wherein the statistical model is based only on the statistical components of the sensor data.
-
6. A system health monitor for diagnosis and prognosis of faults in a physical system being monitored comprising:
-
a model filter having at least a partial model representation of the physical system, the model filter operable to produce a plurality of model enhanced signals based on sensor data, the sensor data representative of measurements made on the physical system;
a symbolic data model operable to produce predicted system states based on discrete data comprising system status variables and system command information, the symbolic data model further operable to detect discrepancies among the discrete data;
a first anomaly detector operable to identify unmodeled events by computing one or more coherence statistics from the sensor data and comparing the coherence statistics against expected coherence quantities indicative of known operating conditions of the physical system;
a predictive comparator module operable to confirm a failure based on detected discrepancies among the discrete data, and to distinguish the unmodeled events from modeled events based at least on the model enhanced signals;
a prognostic assessment module operable to produce predicted faults using a stochastic model of the sensor data to produce future values of the sensor data from the stochastic model; and
a presentation module for presenting information relating to the health of the system comprising detected discrepancies, a categorization of modeled and unmodeled events, and predicted faults, the information suitable for a human user or a machine process. - View Dependent Claims (7, 8, 9)
where Si and Sj are time-varying signals represented by the sensor data,
-
-
10. A computer program product effective operating a computer system for diagnosis and prognosis of faults in a physical system comprising:
-
computer-readable media; and
computer-executable instructions recorded on the computer-readable media comprising;
first executable program code effective to operate the computer system to receive sensor data representative of measurements made on the physical system and to receive discrete data, the measurements representative of values of signals produced by the physical system, the discrete data representative of system status variables and system command information;
second executable program code effective to operate the computer system to produce model enhanced sensor signals by fitting the sensor data to at least a partial physical model of the physical system;
third executable program code effective to operate the computer system to identify correlated signals from among the sensor data;
fourth executable program code effective to operate the computer system to compare the correlated signals with expected correlated signals to detect one or more occurrences of events, the expected correlated signals representative of known operating conditions of the physical system; and
fifth executable program code effective to operate the computer system to identify the one or more occurrences of events as unmodeled events based at least on the model enhanced sensor signals, to detect discrepancies among the discrete data, and to verify faults based on the discrepancies among the discrete data. - View Dependent Claims (11, 12, 13)
where Si and Sj are the signals produced by the physical system,
-
-
14. A system health monitor for detecting anomalies in a physical system being monitored comprising:
-
a model filter having at least a partial model representation of the physical system, the model filter operable to produce a plurality of model enhanced signals based on sensor data, the sensor data representative of measurements made on the physical system;
a symbolic data model operable to produce predicted system states based on discrete data comprising system status variables and system command information, the symbolic data model further operable to detect discrepancies among the discrete data;
means for identifying correlated signals from the sensor data;
data store comprising a plurality of expected coherence quantities representative of known operating conditions of the physical system;
means for selecting one or more of the expected coherence quantities based on the predicted system states; and
means for identifying an unmodeled event by comparing the correlated signals against one or more selected expected coherence quantities, wherein the unmodeled event constitutes a detected anomaly. - View Dependent Claims (15, 16)
where Si and Sj are time-varying signals represented by the sensor data,
-
Specification
-
- Resources
Litigation Campaign Assessment
Thank you for your request. You will receive a custom alert email when the Litigation Campaign Assessment is available.
×
-
Current AssigneeCalifornia Institute of Technology
-
Original AssigneeCalifornia Institute of Technology
-
InventorsZak, Michail, Park, Han G., Mackey, Ryan M. E., James, Mark L.
-
Primary Examiner(s)Hoff, Marc S.
-
Assistant Examiner(s)TSAI, CAROL S W
-
Application NumberUS10/092,471Publication NumberTime in Patent Office566 DaysField of Search702/84, 702/104, 702/183, 702/190, 702/191, 702/196, 700/29, 700/33, 700/44, 700/38, 708/300, 703/2, 703/7, 706/14, 706/24, 706/33, 706/52, 714/25US Class Current702/183CPC Class CodesG05B 17/02 electricG05B 23/0254 based on a quantitative mod...G06F 11/2263 using neural networksG06F 2218/12 Classification; Matching
