Fuel cell health diagnosis method based on internal running state of electric pile

Abstract

The invention discloses a fuel cell health diagnosis method based on the internal operation state of a galvanic pile, and the method comprises the steps: building a fuel cell voltage model based on the internal operation state of the galvanic pile through employing a vehicle-mounted electrochemical impedance detection unit and voltage and current data of the galvanic pile, and employing an artificial neural network; obtaining a quantitative relationship among the stack aging parameter, the stack operation parameter and the stack output voltage, reversely measuring and calculating the aging state by using the stack output voltage at any moment, and finally taking the relative quantity of the stack aging state at any moment deviating from the aging state at the end of the life of the stack as the health state percentage of the stack. The invention aims to utilize the internal state of the fuel cell stack to carry out health diagnosis on the stack in a short-period and long-period operation process, serve vehicle-mounted fuel cells, generate power and the like.

Description

Technical Field

[0001] The present invention belongs to the field of fuel cell testing, and in particular relates to a fuel cell health diagnosis method based on the internal operating status of a fuel cell stack, which is a universal method for identifying key parameters of a fuel cell thermal management system. Background Art

[0002] Proton exchange membrane fuel cells (PEMFCs) are clean and efficient power generation devices that are already in the early stages of industrialization. Currently, the technology for fuel cell stacks and fuel cell systems is "designable" and "controllable," but refined design and control methods are lacking. This is particularly evident in the health diagnosis and life prediction of fuel cell stacks and systems. Health diagnosis and life prediction are important technologies for improving fuel cell durability and operational reliability.

[0003] In terms of the data sources relied upon for diagnosis, there are currently two main technical approaches. One relies solely on cell voltage to determine whether the stack is faulty by setting a voltage threshold. The other relies on data on stack emissions, such as the ion release rate in wastewater (including fluoride ions, sulfate ions, iron ions, platinum ions, etc.), combined with the attenuation amplitude of the average cell voltage and the ion release rate to obtain the stack performance decay rate, and then perform health diagnosis.

[0004] There are three main methods for sorting data. One is the threshold method, which uses whether the average cell voltage is higher than a preset threshold to determine the state of the stack. One is a data-driven intelligent algorithm, which uses a large amount of operating data (mainly cell voltage and current data) to train a long-term and short-term neural network to determine the specific laws followed by the data over time to determine the operating state of the stack. Another method is to determine the state of the response after artificially interfering with the operating state, and then determine the operating state of the fuel cell based on the response.

[0005] The aforementioned methods either provide superficial analysis of stack data, resulting in incomplete diagnosis, or they rely on long-term analysis, resulting in poor diagnostic results during short periods of operation. In fact, effective health diagnosis relies on accurate modeling of the stack output voltage, which requires full utilization of the stack's internal state. Furthermore, effective diagnosis requires modeling, analysis, and judgment of the stack's characteristics during short periods of operation. Summary of the Invention

[0006] To address the problem of existing fuel cells lacking diagnostic methods that include the internal state of the stack, the present invention provides a fuel cell health diagnostic method based on the internal operating state of the stack. This method utilizes an on-board electrochemical impedance detection unit and stack voltage and current data, and uses an artificial neural network to construct a fuel cell voltage model based on the internal operating state of the stack. This model obtains a quantitative relationship between stack aging parameters, stack operating parameters, and stack output voltage. The stack output voltage at any moment is used to reversely calculate the aging state, and ultimately, the relative amount by which the stack aging state at any moment deviates from the aging state at the end of the stack life is used as the percentage of the stack's health state. The present invention aims to utilize the internal state of the fuel cell stack to perform health diagnosis of the stack during both short- and long-term operation, serving the fields of fuel cell applications in vehicles and power generation.

[0007] The purpose of the present invention is achieved through the following technical solutions:

[0008] A fuel cell health diagnosis method based on the internal operating status of the fuel cell stack includes the following steps:

[0009] S1. Use a neural network to construct a fuel cell voltage model based on the internal operating state of the fuel cell stack;

[0010] S2. Determine the quantitative relationship between stack aging parameters, stack operating parameters, and stack output voltage;

[0011] S3. Use the stack output voltage at any time to reversely measure the aging state and calculate the health status percentage of the stack.

[0012] Furthermore, the step S1 includes:

[0013] S11. Integrate the input data array Din of the neural network:

[0014] The input data consists of the stack aging parameter L, the operating condition parameter K, and the operating current I;

[0015] The stack aging parameter L includes a proton membrane aging state parameter L1, a cathode catalyst layer aging state parameter L2, and an anode catalyst layer aging state parameter L3;

[0016] The stack operating condition parameters K include the stack operating temperature K1, the stack cathode operating pressure K2, and the stack cathode gas flow K3;

[0017] The above-mentioned stack aging parameter L, operating condition parameter K and operating current I data are integrated into the input data group Din of the neural network:

[0018] D in,i =(L1 i ,L2 i ,L3 i ,K1i ,K2 i ,K3 i ,I i )

[0019] Where i represents the i-th array, i = 1, 2, ..., n, n is the number of sampling points, and n should correspond to different intervals of time during the operation of the battery stack;

[0020] S12. Integrate the output data array Dout of the neural network:

[0021] The output data is the output voltage V of the battery stack;

[0022] The voltage data is integrated into the output data array Dout of the neural network according to the following formula:

[0023] D out,i =(V i )

[0024] Where i represents the i-th array, i = 1, 2, ..., n, n is the number of sampling points, and n should correspond to different intervals of time during the operation of the battery stack;

[0025] S13. Train the neural network model using input data and output data to obtain a fuel cell voltage model based on the internal operating state of the fuel cell stack.

[0026] Furthermore, in step S2, the relationship between the stack aging parameter L, the operating condition parameter K, the operating current I and the stack output voltage V is:

[0027] V=f(I,L,K)

[0028] L=[L1,L2,L3]

[0029] K=[K1,K2,K3]

[0030] Furthermore, step S3 includes:

[0031] S31. By using the quantitative relationship between the stack aging parameters, stack operating parameters and stack output voltage, the stack output voltage at any time is used to reversely calculate the aging state L;

[0032] S32. Determine the stack life H corresponding to 10% voltage decay from the stack manufacturer. set , and calculate the stack aging state L when the voltage decays by 10% based on f end , L end That is, the aging state at the end of the life of the stack;

[0033] S33. Calculate the deviation of the stack aging state L measured at any time from the aging state L at the end of the stack life endThe relative amount Φ is used as the percentage of the health status of the battery stack:

[0034]

[0035] Furthermore, the proton membrane aging state parameter L1 is obtained by a hydrogen permeation current testing device.

[0036] Furthermore, the cathode catalyst layer aging state parameter L2 and the anode catalyst layer aging state parameter L3 are obtained as follows:

[0037] The frequency f, real impedance Re, and imaginary impedance Im of the stack are measured by an electrochemical impedance spectroscopy test device to form an impedance spectrum (f, Re, Im);

[0038] The impedance spectrum is calculated by a relaxation time distribution calculation unit to obtain the cathode catalyst layer activation resistance and the anode catalyst layer activation resistance at different relaxation times t. The relaxation time and activation resistance of the cathode are expressed as (t2, L2), and the relaxation time and activation resistance of the anode are expressed as (t3, L3).

[0039] Furthermore, the stack operating temperature K1, the stack cathode operating pressure K2, and the stack cathode gas supply flow K3 are respectively collected by sensors.

[0040] The present invention has the following advantages:

[0041] Considering that the output voltage of the fuel cell stack is mainly affected by the aging state of the membrane electrode, operating parameters and operating current, in order to determine the output current of the stack and overcome the drawback that the aging mechanism of the stack is unclear and cannot be described by physical laws, the present invention proposes to use an artificial neural network to train the input data composed of the stack aging parameters, operating parameters and operating current data and the output data composed of the stack output voltage, so as to determine the quantitative relationship between the stack output voltage and the stack aging parameters and the stack operating parameters, and use the stack output voltage at any time to reversely measure the aging state, and finally use the relative amount of the stack aging state at any time deviating from the aging state at the end of the stack life as the percentage of the stack health state. The present invention makes full use of the internal state parameters of the stack, accurately models the stack output voltage, and performs health diagnosis on the stack during short-term and long-term operation, serving the fields of fuel cells in vehicles and power generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings used in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the contents of the embodiments of the present invention and these drawings without any creative work.

[0043] Figure 1 A schematic diagram of a process for constructing a fuel cell voltage model based on the internal operating state of the fuel cell stack in an embodiment of the present invention;

[0044] Figure 2 Schematic diagram of a method for obtaining proton membrane aging state parameters according to an embodiment of the present invention;

[0045] Figure 3 Schematic diagram of a method for obtaining cathode / anode catalyst layer aging state parameters according to an embodiment of the present invention;

[0046] Figure 4 Schematic diagram of a method for acquiring fuel cell operating condition data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention.

[0048] Example

[0049] This embodiment provides a fuel cell health diagnosis method based on the internal operating status of the fuel cell stack, including the following steps:

[0050] S1. Use neural network to build a fuel cell voltage model based on the internal operating state of the fuel cell stack, such as Figure 1 As shown:

[0051] S11. Integrate the input data array Din of the neural network:

[0052] The input data consists of the stack aging parameter L, the operating condition parameter K, and the operating current I;

[0053] The stack aging parameter L includes a proton membrane aging state parameter L1, a cathode catalyst layer aging state parameter L2, and an anode catalyst layer aging state parameter L3;

[0054] The stack operating condition parameters K include the stack operating temperature K1, the stack cathode operating pressure K2, and the stack cathode gas flow K3;

[0055] The above-mentioned stack aging parameter L, operating condition parameter K and operating current I data are integrated into the input data group Din of the neural network:

[0056] D in,i =(L1 i ,L2 i ,L3 i ,K1 i ,K2 i ,K3 i ,I i )

[0057] Where i represents the i-th array, i = 1, 2, ..., n, n is the number of sampling points, and n should correspond to different moments in the operation of the battery stack.

[0058] S12. Integrate the output data array Dout of the neural network:

[0059] The output data is the output voltage V of the battery stack;

[0060] The voltage data is integrated into the output data array Dout of the neural network according to the following formula:

[0061] D out,i =(V i )

[0062] Where i represents the i-th array, i = 1, 2, ..., n, n is the number of sampling points, and n should correspond to different intervals of time during the operation of the battery stack.

[0063] S13. Train the neural network model by inputting and outputting data to obtain a fuel cell voltage model based on the internal operating state of the stack;

[0064] S2. Determine the quantitative relationship between stack aging parameters, stack operating parameters, and stack output voltage:

[0065] The relationship between the stack aging parameter L, operating condition parameter K, operating current I and stack output voltage V is:

[0066] V=f(I,L,K)

[0067] L=[L1,L2,L3]

[0068] K=[K1,K2,K3]

[0069] S3. Use the stack output voltage at any time to reversely measure the aging state and calculate the health status percentage of the stack:

[0070] S31. By using the quantitative relationship between the stack aging parameters, stack operating parameters and stack output voltage, the stack output voltage at any time is used to reversely calculate the aging state L;

[0071] S32. Determine the stack life H corresponding to 10% voltage decay from the stack manufacturer. set , and calculate the stack aging state L when the voltage decays by 10% based on f end , L end That is, the aging state at the end of the life of the stack;

[0072] S33. Calculate the deviation of the stack aging state L measured at any time from the aging state L at the end of the stack life end The relative amount Φ is used as the percentage of the health status of the battery stack:

[0073]

[0074] Furthermore, the proton membrane aging state parameter L1 is obtained by a hydrogen permeation current testing device, such as Figure 2 shown.

[0075] Furthermore, the cathode catalyst layer aging state parameter L2 and the anode catalyst layer aging state parameter L3 are obtained as follows: Figure 3 As shown:

[0076] The frequency f, real impedance Re, and imaginary impedance Im of the stack are measured by an electrochemical impedance spectroscopy (EIS) test device to form an impedance spectrum (f, Re, Im).

[0077] The impedance spectrum is calculated by a relaxation time distribution (DRT) calculation unit to obtain the cathode catalyst layer activation resistance and the anode catalyst layer activation resistance at different relaxation times t. The relaxation time and activation resistance of the cathode are expressed as (t2, L2), and the relaxation time and activation resistance of the anode are expressed as (t3, L3).

[0078] Furthermore, the stack operating temperature K1, the stack cathode operating pressure K2, and the stack cathode gas supply flow K3 are respectively collected by sensors.

[0079] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A fuel cell health diagnosis method based on the internal operating status of the fuel cell stack, characterized in that: The following steps are involved: S1. Use neural network to build a fuel cell voltage model based on the internal operating state of the fuel cell stack; S2. Determine the quantitative relationship between the stack aging parameters, stack operating parameters and stack output voltage; S3. Use the stack output voltage at any time to reversely measure the aging state and calculate the health status percentage of the stack.

2. A fuel cell health diagnosis method based on the internal operating state of the fuel cell stack as claimed in claim 1, characterized in that: The step S1 comprises: S11. Integrate the input data array Din of the neural network: The input data consists of the stack aging parameter L, the operating condition parameter K and the operating current I; The stack aging parameter L includes a proton membrane aging state parameter L1, a cathode catalyst layer aging state parameter L2, and an anode catalyst layer aging state parameter L3; The stack operating condition parameters K include stack operating temperature K1, stack cathode operating pressure K2 and stack cathode gas supply flow K3; The above-mentioned stack aging parameter L, operating condition parameter K and operating current I data are integrated into the input data group Din of the neural network: <h2 style=";text-align:left;direction:ltr">D<h2 style=";text-align:left;direction:ltr"> in,i <h2 style=";text-align:left;direction:ltr"> (L1)<h2 style=";text-align:left;direction:ltr"> i <h2 style=";text-align:left;direction:ltr"> L2<h2 style=";text-align:left;direction:ltr"> i <h2 style=";text-align:left;direction:ltr"> L3<h2 style=";text-align:left;direction:ltr"> i <h2 style=";text-align:left;direction:ltr"> K1<h2 style=";text-align:left;direction:ltr"> i <h2 style=";text-align:left;direction:ltr"> K2<h2 style=";text-align:left;direction:ltr"> i <h2 style=";text-align:left;direction:ltr"> K3<h2 style=";text-align:left;direction:ltr"> i <h2 style=";text-align:left;direction:ltr"> ,I<h2 style=";text-align:left;direction:ltr"> i <h2 style=";text-align:left;direction:ltr"> ) Where i represents the i-th array, i = 1, 2, ..., n, and n is the number of sampling points; S12. Integrate the output data array Dout of the neural network: The output data is the output voltage V of the battery stack; The voltage data is integrated into the output data array Dout of the neural network according to the following formula: D out,i =(V i ) Where i represents the i-th array, i = 1, 2, ..., n, and n is the number of sampling points; S13. Train the neural network model through input data and output data to obtain a fuel cell voltage model based on the internal operating state of the fuel cell stack.

3. A fuel cell health diagnosis method based on the internal operating state of the fuel cell stack as claimed in claim 2, characterized in that: In step S2, the relationship between the stack aging parameter L, the operating condition parameter K, the operating current I and the stack output voltage V is: V=f(I,L,K) L=[L1,L2,L3] K=[K1,K2,K3].

4. A fuel cell health diagnosis method based on the internal operating state of the fuel cell stack as claimed in claim 2, characterized in that: The step S3 comprises: S31. Based on the quantitative relationship between the stack aging parameters, the stack operating parameters and the stack output voltage, the stack output voltage at any time is used to reversely calculate the aging state L; S32. Determine the stack life H corresponding to 10% voltage decay from the stack manufacturer set , and calculate the stack aging state L when the voltage decays by 10% based on f end , L end That is, the aging state at the end of the life of the battery stack; S33. Calculate the deviation of the stack aging state L measured at any time from the aging state L at the end of the stack life end The relative amount Φ is used as the percentage of the health status of the battery stack:

5. A fuel cell health diagnosis method based on the internal operating state of the fuel cell stack as claimed in claim 2, characterized in that: The proton membrane aging state parameter L1 is obtained by a hydrogen permeation current testing device.

6. A fuel cell health diagnosis method based on the internal operating state of the fuel cell stack as claimed in claim 2, characterized in that: The cathode catalyst layer aging state parameter L2 and the anode catalyst layer aging state parameter L3 are obtained as follows: The frequency f, the real impedance Re, and the imaginary impedance Im of the battery stack are tested by an electrochemical impedance spectroscopy test device to form an impedance spectrum (f, Re, Im); The impedance spectrum is calculated by a relaxation time distribution calculation unit to obtain the cathode catalyst layer activation resistance and the anode catalyst layer activation resistance at different relaxation times t. The relaxation time and activation resistance of the cathode are expressed as (t2, L2), and the relaxation time and activation resistance of the anode are expressed as (t3, L3).

7. A fuel cell health diagnosis method based on the internal operating state of the fuel cell stack as claimed in claim 2, characterized in that: The stack operating temperature K1, the stack cathode operating pressure K2, and the stack cathode gas supply flow K3 are collected by sensors respectively.

Images