Electronic system condition monitoring and prognostics

• US 8,200,600 B2
• Filed: 03/20/2008
• Issued: 06/12/2012
• Est. Priority Date: 03/20/2007
• Status: Expired due to Fees

• Current Assignee: iRobot Corporation, Massachusetts Institute of Technology
• Original Assignee: iRobot Corporation
• Inventors: Rosenstein, Michael Todd, Adams, Bryan Paul, Roy, Nicholas George Dilip
• Primary Examiner(s): Gaffin, Jeffrey A
• Assistant Examiner(s): CHANG, LI WU
• Application Number:
  US12/052,726
• Publication Number:
  US 20080235172A1
• Time in Patent Office: 1,545 Days
• Field of Search: 706/46
• US Class Current: 706/46
• CPC Class Codes: G01R 31/367 Software therefor, e.g. for battery testing using modelling or look-up tables
• CPC Class Codes: G01R 31/392 Determining battery ageing or deterioration, e.g. state of health
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

A system for monitoring and predicting the condition of an electronic system comprises a component model, an inference engine based on the component model, and an action selection component that selects an action based on an output of the inference engine.

Claims

1. A system for monitoring and predicting the condition of a power system of a remote vehicle, the system including memory, and comprising:

a component model residing in the memory and constructed using power system modeling over both short time scales and long time scales, the short time scales being related to a state of charge of the power system and the long time scales being related to a likelihood of failure of the power system;

an inference engine utilizing the component model and sensor data including at least a temperature of the power system to make predictions about a state of charge of the power system and a likelihood of failure of the power system; and

an action selection component that selects an action to mitigate a potential failure based on an output of the inference engine.

2. The system of claim 1, wherein the power system is a battery of the remote vehicle.

3. The system of claim 2, wherein the component model is constructed using knowledge of a battery chemistry and a battery charging algorithm.

4. The system of claim 1, wherein the component model includes a dynamic Bayesian network model having current state variables that depend on state variables from previous time steps.

5. The system of claim 1, wherein the action selection component creates new data to improve prediction accuracy of the inference engine.

6. The system of claim 1, wherein the component model encodes known data about the power system, correlating the sensor data with beliefs about the state of charge of the power system.

7. The system of claim 1, wherein the sensor data comprises a stream of incoming data from one or more sensors of the remote vehicle.

8. The system of claim 7, wherein the one or more sensors comprise an internal battery impedance sensor.

9. The system of claim 7, wherein the stream of incoming data comprises a count of the number of power system charge cycles since manufacture.

10. The system of claim 1, wherein the sensor data comprises at least a voltage and a temperature of the power system.

11. The system of claim 10, wherein the inference engine also utilizes a rate of change of temperature of the power system to make predictions about the power system.

12. The system of claim 1, further comprising an active fault diagnosis system that uses an output of the inference engine to mitigate failure of the power system or refine a belief regarding a potential failure of the power system.

13. A method for monitoring and predicting the condition of a power system of a remote vehicle, the method comprising:

constructing a component model using power system modeling over both short and long time scales, the short time scales being related to a state of charge of the power system and the long time scales being related to a likelihood of failure of the power system;

predicting, using a processor, a state of charge of the power system and a likelihood of failure of the power system based on the component model and sensor data including at least a temperature of the power system; and

mitigating predicted potential failures of the power system by selecting an action to mitigate a potential failure.

14. The system of claim 13, wherein the power electronic system is a battery of the remote vehicle.

15. The system of claim 13, wherein the component model includes a dynamic Bayesian network model having current state variables that depend on state variables from previous time steps.

16. The system of claim 13, further comprising creating new data to improve prediction accuracy.

17. A method for monitoring and predicting the condition of a power system of a remote vehicle, the method comprising:

constructing a component model using power system modeling over both short time scales and long time scales, the short time scales being related to a state of charge of the power system and the long time scales being related to a likelihood of failure of the power system;

predicting, using a processor, a state of charge of the power system and a likelihood of failure of the power system based on the component model and sensor data including at least a temperature of the power system; and

creating new data to improve prediction accuracy.

18. The system of claim 17, wherein the power system is a battery of the remote vehicle.

19. The system of claim 17, wherein the component model includes a dynamic Bayesian network model having current state variables that depend on state variables from previous time steps.

20. The method of claim 17, wherein constructing the component model includes modeling and incorporating at least one failure mode into a dynamic Bayesian network model.

21. The method of claim 20, wherein the at least one failure mode comprises one or more of manufacturing differences and defects, long-term organic loss of capacity memory effects, and damage from improper charging.

22. A method for informing a user regarding the condition of a power system of a remote vehicle, the method comprising:

constructing a component model using power system modeling over both short time scales and long time scales, the short time scales being related to a state of charge of the power system and the long time scales being related to a likelihood of failure of the power system;

predicting, using a processor, a state of charge of the power system and a likelihood of failure of the power system based on the component model and sensor data including at least a temperature of the power system; and

displaying, to an operator of the remote vehicle, an identification of the power system, a predicted state of the power system, and a predicted likelihood of failure of the power system.