Method for monitoring the state of a proton exchange membrane fuel cell based on two-point impedance measurements

Abstract

The application discloses a kind of proton exchange membrane fuel cell state monitoring method based on two-point impedance measurement. The impedance spectrum of proton exchange membrane fuel cell under different working conditions is obtained by using fast electrochemical impedance spectroscopy method;According to the impedance spectrum of fuel cell and corresponding working condition, the two frequencies of impedance spectrum are selected as characteristic frequencies by using the classification model based on linear discriminant analysis, and the trained classification model corresponding to the two characteristic frequencies is obtained;During online state monitoring, the impedance of two characteristic frequencies of fuel cell is collected in real time, and then the real-time collected impedance is input into the trained classification model, and the trained classification model outputs the working condition of current fuel cell.The application solves the problem that the real-time performance of proton exchange membrane fuel cell state monitoring is not enough, the accuracy is not high, and the migration is poor during operation, and can improve the reliability and durability of proton exchange membrane fuel cell working.

Description

Technical Field

[0001] The present invention relates to a proton exchange membrane fuel cell state monitoring method in the fuel cell application field, and in particular to a proton exchange membrane fuel cell state monitoring method based on two-point impedance measurement. Background Art

[0002] Proton exchange membrane fuel cell is an electrochemical device that efficiently utilizes hydrogen energy and has excellent performance such as high power density, high efficiency, cleanliness, pollution-free and low noise.

[0003] During the operation of proton exchange membrane fuel cells, due to complex power conditions and insufficient water and heat management, the stack is prone to failure states such as flooding, membrane drying, and air starvation. Working in a failure state for a long time will have a negative impact on the performance and structure of the battery. Due to the characteristics of the stack structure, it is difficult to install sensors inside it for real-time monitoring, which further increases the difficulty of control and diagnosis. Therefore, an efficient condition monitoring method is crucial to the durability and life of the fuel cell.

[0004] Traditional condition monitoring methods include model-based, voltage-based, data-based, and electrochemical impedance spectroscopy-based. Among them, the electrochemical impedance spectroscopy-based condition monitoring method can obtain internal fuel cell stack status information, but for real-time condition monitoring applications, the test time needs to be further shortened, and the accuracy of condition classification needs to be further improved. Summary of the Invention

[0005] In order to address the deficiencies of the prior art, the present invention proposes a proton exchange membrane fuel cell state monitoring method based on two-point impedance measurement.

[0006] The present invention adopts the following technical solutions:

[0007] 1) Using fast electrochemical impedance spectroscopy to obtain the impedance spectra of proton exchange membrane fuel cells under different operating conditions;

[0008] 2) Based on the impedance spectrum of the fuel cell and the corresponding operating state, a classification model based on linear discriminant analysis is used to select two frequencies of the impedance spectrum and record them as characteristic frequencies, and a trained classification model corresponding to the two characteristic frequencies is obtained;

[0009] 3) During online status monitoring, the impedance of the two characteristic frequencies of the fuel cell is collected in real time, and then the real-time collected impedance is input into the classification model trained in step 2), and the trained classification model outputs the current working status of the fuel cell.

[0010] The step 1) is specifically as follows:

[0011] Firstly, two segments of the longest linear shift register sequences with different bit numbers are used as current excitation signals. Then, the two current excitation signals are injected into the fuel cell in sequence and the current and voltage responses of the fuel cell are collected. Then, based on the two current and voltage responses of the fuel cell, the Morlet wavelet transform and maximum likelihood estimation are used in sequence to calculate the electrochemical impedance spectroscopy within the preset frequency range.

[0012] The step 2) is specifically as follows:

[0013] The preset frequency range of the impedance spectrum is divided into a low frequency band and a high frequency band, and a frequency is selected from each of the low frequency band and the high frequency band to form a frequency combination. The impedance and the corresponding working state under each frequency combination constitute a training set, and the training set is input into the classification model based on linear discriminant analysis for training to obtain the corresponding classification accuracy; each frequency combination is traversed to obtain the classification accuracy corresponding to each frequency combination, and the frequency combination with a classification accuracy greater than the accuracy threshold is retained. Among the retained frequency combinations, the frequency combination with the shortest measurement time is selected and the two frequencies of the frequency combination are used as characteristic frequencies. At the same time, the classification model based on linear discriminant analysis trained under this frequency combination is used as the trained classification model.

[0014] In step 3), the square wave signals of the two characteristic frequencies are both used as current excitation signals. The two current excitation signals are then injected into the fuel cell in sequence and the current and voltage responses of the fuel cell are collected. Based on the two current and voltage responses of the fuel cell, the impedance of the proton exchange membrane fuel cell at the two characteristic frequencies is calculated using fast Fourier transform.

[0015] The dividing point between low frequency and high frequency in the preset frequency range is determined according to the longest linear shift register sequence with a shorter number of bits.

[0016] The beneficial effects of the present invention are:

[0017] The present invention solves the problems of insufficient real-time performance, low accuracy and poor mobility of state monitoring during the operation of the proton exchange membrane fuel cell, and can improve the reliability and durability of the operation of the proton exchange membrane fuel cell. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] Figure 1 It is a structural diagram of the proton exchange membrane fuel cell state monitoring method in the present invention.

[0020] Figure 2 1 is the impedance spectrum of the proton exchange membrane fuel cell in different working states according to the embodiment of the present invention.

[0021] Figure 31 is a two-point impedance diagram of a proton exchange membrane fuel cell in different working states according to an embodiment of the present invention.

[0022] Figure 4 1 is a diagram of classification results of different working states in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0024] like Figure 1 As shown, the present invention includes the following steps:

[0025] 1) Using fast electrochemical impedance spectroscopy to obtain the impedance spectra of proton exchange membrane fuel cells under different operating conditions;

[0026] Step 1) is specifically as follows:

[0027] First, two segments of the longest linear shift register sequence (M sequence) with different bit numbers are used as current excitation signals. Then, the two current excitation signals are injected into the fuel cell and the current and voltage responses of the fuel cell are collected. Then, based on the two current and voltage responses of the fuel cell, the electrochemical impedance spectrum within the preset frequency range is calculated using Morlet wavelet transform and maximum likelihood estimation. In the specific implementation, the preset frequency range is 0.5Hz to 500Hz. Figure 2 As shown in Figure 1, fast electrochemical impedance spectroscopy can quickly measure the impedance information of proton exchange membrane fuel cells in the frequency range of 0.5Hz to 500Hz. Figure 2 (a) is the impedance spectrum of a 400W proton exchange membrane fuel cell. Figure 2 (b) is the impedance spectrum of a 3kW proton exchange membrane fuel cell.

[0028] 2) Based on the impedance spectrum of the fuel cell and the corresponding operating state, a classification model based on linear discriminant analysis is used to select two frequencies of the impedance spectrum and record them as characteristic frequencies, and a trained classification model corresponding to the two characteristic frequencies is obtained;

[0029] Step 2) is specifically as follows:

[0030] The preset frequency range of the impedance spectrum is divided into a low frequency band and a high frequency band. One frequency is selected from each of the low frequency band and the high frequency band to form a frequency combination. The impedance under each frequency combination and the corresponding working state constitute a training set. The real part and the imaginary part of the impedance under each frequency are respectively used as a classification feature. The training set is input into a classification model based on linear discriminant analysis for training to obtain the corresponding classification accuracy; each frequency combination is traversed to obtain the classification accuracy corresponding to each frequency combination, and the frequency combination with a classification accuracy greater than the accuracy threshold is retained. Among the retained frequency combinations, the frequency combination with the shortest measurement time is selected and the two frequencies of the frequency combination are used as feature frequencies. At the same time, the classification model based on linear discriminant analysis trained under this frequency combination is used as the trained classification model.

[0031] In a specific implementation, the low frequency band is 0.5Hz to 60Hz, and the high frequency band is 60Hz to 500Hz. The frequency dividing point is determined by the longest linear shift register sequence with a shorter bit number. The clock frequencies of the two longest linear shift register sequences selected are 200Hz and 1500Hz, respectively, and the corresponding effective bandwidths are 200 / 3Hz and 1500 / 3Hz, respectively. Therefore, 60Hz and 500Hz are taken as the highest frequencies of the two frequency bands within the bandwidth range.

[0032] 3) During online status monitoring, it is only necessary to collect the impedance of the two characteristic frequencies of the fuel cell in real time, and then input the real-time collected impedance into the classification model trained in step 2). The trained classification model can quickly output the current working status of the fuel cell. In specific implementation, the working status of the fuel cell is normal operation, air starvation, membrane dryness and water flooding. Figure 3 As shown, the two-point impedance measurement can measure the impedance information at any two frequencies in the frequency range of 0.5Hz to 500Hz. Figure 3 (a) is a two-point impedance sample of a 400W proton exchange membrane fuel cell under four states. Figure 3 Figure (b) shows two-point impedance samples of a 3kW PEMFC under four different states. If the trained classification model determines that the PEMFC is operating normally, regular monitoring continues; otherwise, fault handling is performed.

[0033] In step 3), the square wave signals of the two characteristic frequencies are used as current excitation signals. Then, the two current excitation signals are injected into the fuel cell in sequence and the current and voltage responses of the fuel cell are collected. Then, based on the two current and voltage responses of the fuel cell, the impedance of the proton exchange membrane fuel cell at the two characteristic frequencies is calculated using fast Fourier transform.

[0034] like Figure 4As shown in the figure, this condition monitoring method was validated through experiments conducted on a 400W and a 3kW proton exchange membrane fuel cell stack. During normal fuel cell operation, the operating conditions were varied to allow the cells to operate in normal, air-starved, membrane-dry, and flooded states, while rapid electrochemical impedance spectroscopy data were collected. This impedance spectroscopy data was used to train a classifier and identify the optimal two-point frequencies for each stack.

[0035] like Figure 4 As shown in the results of state classification, experimental verification was carried out on two proton exchange membrane fuel cell stacks with different power levels. The experimental results on the 400W proton exchange membrane fuel cell ( Figure 4 (a)) It can be seen that in 30 state classifications, 29 were correctly classified; the experimental results on a 3kW proton exchange membrane fuel cell ( Figure 4 As can be seen from (b), 31 of the 32 state classifications were correct. Furthermore, the measurement showed that the signal injection time via two-point impedance measurement was less than 5 seconds, and the total state monitoring cycle was less than 10 seconds.

[0036] The proton exchange membrane fuel cell state monitoring method based on two-point impedance measurement proposed in the present invention can realize rapid state monitoring of the proton exchange membrane fuel cell, thereby improving the rapidity and accuracy of state monitoring.

[0037] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above description and embodiments. The above description is merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications will fall within the scope of the present invention as claimed. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A proton exchange membrane fuel cell state monitoring method based on two-point impedance measurement, characterized in that: The following steps are involved: 1) Using fast electrochemical impedance spectroscopy to obtain the impedance spectra of proton exchange membrane fuel cells under different operating conditions; The step 1) is specifically as follows: First, two segments of the longest linear shift register sequence with different bit numbers are used as current excitation signals. Then, the two current excitation signals are injected into the fuel cell and the current and voltage responses of the fuel cell are collected. Then, based on the two current and voltage responses of the fuel cell, the electrochemical impedance spectroscopy within a preset frequency range is calculated using Morlet wavelet transform and maximum likelihood estimation. 2) Based on the impedance spectrum of the fuel cell and the corresponding operating state, a classification model based on linear discriminant analysis is used to select two frequencies of the impedance spectrum and record them as characteristic frequencies, and a trained classification model corresponding to the two characteristic frequencies is obtained; The step 2) is specifically as follows: The preset frequency range of the impedance spectrum is divided into a low frequency band and a high frequency band, and a frequency is selected from each of the low frequency band and the high frequency band to form a frequency combination. The impedance and the corresponding working state under each frequency combination constitute a training set, and the training set is input into the classification model based on linear discriminant analysis for training to obtain the corresponding classification accuracy; each frequency combination is traversed to obtain the classification accuracy corresponding to each frequency combination, and the frequency combination with a classification accuracy greater than the accuracy threshold is retained. Among the retained frequency combinations, the frequency combination with the shortest measurement time is selected and the two frequencies of the frequency combination are used as characteristic frequencies. At the same time, the classification model based on linear discriminant analysis trained under the frequency combination is used as the trained classification model; The dividing point between low frequency and high frequency in the preset frequency range is determined based on the longest linear shift register sequence with a shorter number of bits; 3) During online status monitoring, the impedance of the two characteristic frequencies of the fuel cell is collected in real time, and then the real-time collected impedance is input into the classification model trained in step 2), and the trained classification model outputs the current working status of the fuel cell.

2. A proton exchange membrane fuel cell state monitoring method based on two-point impedance measurement according to claim 1, characterized in that: In step 3), the square wave signals of the two characteristic frequencies are both used as current excitation signals. The two current excitation signals are then injected into the fuel cell in sequence and the current and voltage responses of the fuel cell are collected. Based on the two current and voltage responses of the fuel cell, the impedance of the proton exchange membrane fuel cell at the two characteristic frequencies is calculated using fast Fourier transform.

Images