Method and apparatus for multi-stage fault diagnosis of fuel cell systems

Patent metadata

Field	Value
Publication number	US-11489179-B2
Application number	US-202016777492-A
Country	US
Kind code	B2
Filing date	Jan 30, 2020
Priority date	Sep 11, 2019
Publication date	Nov 1, 2022
Grant date	Nov 1, 2022

Abstract

A hierarchical fault classification method for a fuel cell system, a multi-stage fault diagnosis method therefor, and a fault diagnosis device therefor are disclosed. The fuel cell system is divided into a subsystem, a component, and an element step by step. The multi-stage fault diagnosis method includes detecting a subsystem, a fault of which occurs, in the fuel cell system composed of a plurality of subsystems and detecting an upper-level component, which causes the fault, among upper-level components included in the subsystem, the fault of which occurs, using measurement data and a control signal.

First claim

What is claimed is:

1. A multi-stage fault diagnosis method for a fuel cell system, performed by a multi-stage fault diagnosis device, the method comprising:
dividing the fuel cell system into a hierarchy including, in descending order, a system, a subsystem, an upper-level component, a lower-level component, and an element; and
diagnosing a fault of the fuel cell system in a sequence corresponding to the hierarchy;
wherein the subsystem includes a stack and a balance of plant (BOP); and
wherein the method further comprises, in response to detecting the fault, determining whether the fault is irreversible due to an internal part of the stack, or reversible due to the BOP.

2. The method of claim 1 , wherein the BOP includes an air supply device, a heat management device, a fuel supply device, a water management device, and a power inverter and controller.

3. A multi-stage fault diagnosis method for a fuel cell system, performed by a multi-stage fault diagnosis device, the method comprising:
detecting a subsystem, a fault of which occurs, in the fuel cell system composed of a plurality of subsystems; and
detecting an upper-level component, which causes the fault, among upper-level components included in the subsystem, the fault of which occurs, using measurement data and a control signal;
wherein the fuel cell system includes subsystems including a stack, an air supply device, a heat management device, a fuel supply device, a water management device, and a power inverter and controller;
wherein the detecting of the subsystem, the fault of which occurs, includes:
when it is detected that the fault occurs in the fuel cell system, determining whether the fault occurs in the stack or a balance of plant (BOP) including the air supply device, the heat management device, the fuel supply device, the water management device, and the power inverter and controller; and
wherein the determining whether the fault occurs includes:
determining whether the fault is irreversible due to an internal part of the stack or is reversible due to the BOP.

4. The method of claim 3 , further comprising:
detecting a lower-level component, which causes the fault, among one or more lower-level components included in the upper-level component which causes the fault.

5. The method of claim 4 , further comprising:
detecting an element, which causes the fault, among one or more elements included in the lower-level component which causes the fault.

6. The method of claim 3 , wherein the detecting of the subsystem, the fault of which occurs, includes:
detecting the subsystem, the fault of which occurs, based on measurement data and a control signal for the plurality of subsystems.

7. The method of claim 3 , wherein the multi-stage fault diagnosis device performs fault diagnosis for only the subsystem, the fault of which occurs, in the fuel cell system and does not perform the fault diagnosis for the other subsystems.

8. The method of claim 7 , wherein the performing of the fault diagnosis for the fuel cell system includes:
predicting a characteristic value for each of the plurality of subsystems constituting the fuel cell system;
calculating a residual value for each subsystem based on the characteristic value and a measurement value; and
detecting the subsystem, the fault of which occurs, among the subsystems using residual values for the subsystems and a classifier.

9. The method of claim 8 , wherein the detecting includes:
determining a residual pattern value for the subsystems based on a result of comparing the residual value for each subsystem with a threshold corresponding to each subsystem.

10. The method of claim 9 , wherein the detecting includes:
obtaining a classification pattern value for fault detection using the classifier which receives the residual pattern value; and
determining the subsystem, the fault of which occurs, among the subsystems using the obtained classification pattern value.

11. The method of claim 10 , wherein the classifier obtains the classification pattern value based on machine learning including at least one of an artificial neural network (ANN), a support vector machine (SVM), a linear regression equation, a general regression neural network (GRNN), and ensemble regression.

12. The method of claim 9 , wherein the threshold is set based on a standard deviation value of a residual value calculated in a state where each subsystem is normal.

13. The method of claim 8 , wherein the predicting of the characteristic value includes:
predicting the characteristic value using at least one of an artificial neural network (ANN), a support vector machine (SVM), a linear regression equation, a general regression neural network (GRNN), and ensemble regression.

14. A computer-readable storage medium storing instructions for executing the method of claim 1 .

15. A multi-stage fault diagnosis device for performing a multi-stage fault diagnosis method for a fuel cell system, the device comprising:
a subsystem fault detector to detect a subsystem, a fault of which occurs, in the fuel cell system composed of a plurality of subsystems;
a component fault detector to detect an upper-level component, which causes the fault, among upper-level components included in the subsystem, the fault of which occurs, using measurement data and a control signal;
a prediction unit to predict a characteristic value for each of the plurality of subsystems; and
a calculation unit to calculate a residual value for each subsystem, the residual value corresponding to a difference between the characteristic value and a measurement value;
wherein the multi-stage fault diagnosis device determines a residual pattern value for the plurality of subsystems based on a result of comparing the residual value for each subsystem with a threshold corresponding to each subsystem.

16. The device of claim 15 , wherein the component fault detector detects a lower-level component, which causes the fault, among one or more lower-level components included in the upper-level component which causes the fault,
further comprising:
an element fault detector to detect an element, which causes the fault, among one or more elements included in the lower-level component which causes the fault.

17. A multi-stage fault diagnosis device for performing a multi-stage fault diagnosis method, the device comprising:
a prediction unit to predict a characteristic value for each of a plurality of subsystems constituting a fuel cell system, the characteristic value corresponding to a formula for calculating a specific measurement value using a measurement value of a sensor which varies in characteristic or location;
a calculation unit to calculate a residual value for each subsystem, the residual value corresponding to a difference between the characteristic value and a measurement value; and
a fault detector to detect a subsystem, a fault of which occurs, among the subsystems using residual values for the subsystems and a classifier;
wherein the multi-stage fault diagnosis device determines a residual pattern value for each subsystem based on a pattern where the residual value for each subsystem deviates from a threshold.

Assignees

• Korea Inst Energy Res

Inventors

• Lee Won Yong
• Oh Hwan Yeong
• Kim Min Jin
• Park Seok Hee
• Park Gu Gon
• Yang Tae Hyun
• Yim Sung Dae
• Sohn Young Jun
• Bae Byung Chan
• Kim Seung Gon
• Shin Dong Won
• Woo Seung Hee
• Lee So Jeong
• Lee Hye Jin
• Choi Yoon Young

Classifications

• Y02E60/50

  Fuel cells · CPC title

• H01M8/04679 Primary

  of fuel cell stacks · CPC title

• H01M8/04992

  characterised by the implementation of mathematical or computational algorithms, e.g. feedback control loops, fuzzy logic, neural networks or artificial intelligence · CPC title

• H01M8/04686

  of auxiliary devices, e.g. batteries, capacitors · CPC title