CN115084593A - Fuel cell fault diagnosis method based on nonlinear impedance spectrum - Google Patents
Fuel cell fault diagnosis method based on nonlinear impedance spectrum Download PDFInfo
- Publication number
- CN115084593A CN115084593A CN202210609220.9A CN202210609220A CN115084593A CN 115084593 A CN115084593 A CN 115084593A CN 202210609220 A CN202210609220 A CN 202210609220A CN 115084593 A CN115084593 A CN 115084593A
- Authority
- CN
- China
- Prior art keywords
- fault
- fuel cell
- frequency response
- response function
- fault diagnosis
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 239000000446 fuel Substances 0.000 title claims abstract description 54
- 238000003745 diagnosis Methods 0.000 title claims abstract description 47
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000001453 impedance spectrum Methods 0.000 title claims abstract description 15
- 238000005316 response function Methods 0.000 claims abstract description 54
- 230000005284 excitation Effects 0.000 claims abstract description 27
- 230000004044 response Effects 0.000 claims abstract description 20
- 230000009467 reduction Effects 0.000 claims abstract description 14
- 238000000605 extraction Methods 0.000 claims abstract description 10
- 238000000513 principal component analysis Methods 0.000 claims abstract description 8
- 230000036541 health Effects 0.000 claims abstract description 5
- 238000012549 training Methods 0.000 claims abstract description 5
- 239000011159 matrix material Substances 0.000 claims description 24
- 238000003066 decision tree Methods 0.000 claims description 20
- 238000007637 random forest analysis Methods 0.000 claims description 14
- 239000012528 membrane Substances 0.000 claims description 13
- 238000004422 calculation algorithm Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 6
- 230000001186 cumulative effect Effects 0.000 claims description 6
- 238000013461 design Methods 0.000 claims description 6
- 235000003642 hunger Nutrition 0.000 claims description 6
- 230000037351 starvation Effects 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 5
- 238000012706 support-vector machine Methods 0.000 claims description 4
- 238000001566 impedance spectroscopy Methods 0.000 claims description 3
- 238000013480 data collection Methods 0.000 claims 1
- 238000005070 sampling Methods 0.000 description 7
- 238000001228 spectrum Methods 0.000 description 3
- 238000002347 injection Methods 0.000 description 2
- 239000007924 injection Substances 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012952 Resampling Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000002238 attenuated effect Effects 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 239000000243 solution Substances 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01M—PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
- H01M8/00—Fuel cells; Manufacture thereof
- H01M8/04—Auxiliary arrangements, e.g. for control of pressure or for circulation of fluids
- H01M8/04298—Processes for controlling fuel cells or fuel cell systems
- H01M8/04313—Processes for controlling fuel cells or fuel cell systems characterised by the detection or assessment of variables; characterised by the detection or assessment of failure or abnormal function
- H01M8/04664—Failure or abnormal function
Landscapes
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Manufacturing & Machinery (AREA)
- Sustainable Development (AREA)
- Sustainable Energy (AREA)
- Chemical & Material Sciences (AREA)
- Chemical Kinetics & Catalysis (AREA)
- Electrochemistry (AREA)
- General Chemical & Material Sciences (AREA)
- Fuel Cell (AREA)
Abstract
The invention relates to a fuel cell fault diagnosis method based on a nonlinear impedance spectrum, which comprises the following steps: s1, designing fault conditions of different degrees under different types, applying alternating current disturbance with a set amplitude to the fuel cell, and collecting excitation current and response voltage; s2, performing fast Fourier transform to obtain each order of harmonic response under the target frequency, and calculating a first order frequency response function and a second order frequency response function; s3, forming a nonlinear impedance fault data set, and performing fault feature extraction and dimension reduction on the data set through principal component analysis; s4, according to the fault data set after dimensionality reduction, constructing a fault diagnosis model and training; and S5, applying alternating current disturbance with a set amplitude to the fuel cell, collecting voltage and current data, calculating a frequency response function, inputting the frequency response function into a trained fault diagnosis model after dimension reduction, and diagnosing the current health state of the fuel cell in real time. Compared with the prior art, the method provided by the invention has the advantages that the fault diagnosis is carried out based on the linear impedance, and the diagnosis precision is higher.
Description
Technical Field
The invention relates to the technical field of fuel cell fault diagnosis, in particular to a fuel cell fault diagnosis method based on a nonlinear impedance spectrum.
Background
The fuel cell automobile is an important branch of new energy automobiles, is considered as one of the final solutions of future automobiles due to the advantages of high filling speed, high efficiency, low noise, zero emission and the like, but is still limited by durability in large-scale commercial application, carries out fault diagnosis on a fuel cell system, can provide current health state information for system control, further carries out feedback control, and slows down the fault degree to prolong the service life of the fuel cell.
Chinese patent CN113359037A discloses a method for diagnosing a fault of a fuel cell based on a BP neural network, which performs normalization processing on fault data obtained from a sensor of a fuel cell system, and then performs feature extraction by using an improved linear discriminant analysis method, so as to perform fault diagnosis by using the BP neural network, but the method cannot determine different types of fuel cell faults and fault degrees of the fuel cell.
Chinese patent CN110676488A discloses an online proton exchange membrane fuel cell fault diagnosis method based on low-frequency impedance and electrochemical impedance spectrum, which compares the measured low-frequency impedance with a low-frequency impedance fault threshold on line, and classifies and diagnoses faults of the electrochemical impedance spectrum based on a fault diagnosis algorithm of fuzzy logic, but the method essentially adopts the traditional linear electrochemical impedance spectrum, and is difficult to distinguish individually when facing mixed faults.
Chinese patent CN109726452A discloses an online proton exchange membrane fuel cell fault diagnosis method based on impedance spectrum, which utilizes an equivalent circuit model to perform parameter identification on the electrochemical impedance spectrum of the fuel cell, then selects some of the parameters as classification features, and uses a fault diagnosis algorithm based on a binary tree support vector machine to perform classification processing on the proton exchange membrane fuel cell, so as to diagnose the faults such as membrane dryness, flooding, air starvation and the like which are easy to occur inside the proton exchange membrane fuel cell, but as with patent CN110676488A, it only adopts linear impedance information, and when facing a mixed fault, the accuracy is difficult to guarantee.
In summary, there is a need for a fuel cell barrier diagnostic method that is more accurate and suitable for hybrid faults.
Disclosure of Invention
The present invention is directed to a method for diagnosing a fault of a fuel cell based on a nonlinear impedance spectrum, which overcomes the above-mentioned drawbacks of the prior art.
The purpose of the invention can be realized by the following technical scheme:
a fuel cell fault diagnosis method based on nonlinear impedance spectroscopy comprises the following steps:
s1, designing fault conditions of different degrees under different types, applying alternating current disturbance with a set amplitude to the fuel cell, and collecting excitation current and response voltage;
s2, removing steady-state components in the response voltage data, carrying out fast Fourier transform to obtain each order of harmonic response under the target frequency, and calculating a first order frequency response function and a second order frequency response function;
s3, forming a nonlinear impedance fault data set by the first-order frequency response function and the second-order frequency response function together, and performing fault feature extraction and dimension reduction on the data set through principal component analysis;
s4, according to the fault data set after dimensionality reduction, constructing a fault diagnosis model and training;
and S5, applying alternating current disturbance with a set amplitude to the fuel cell, collecting voltage and current data, calculating a frequency response function, inputting the frequency response function into a trained fault diagnosis model after dimension reduction, and diagnosing the current health state of the fuel cell in real time.
In step S1, the different types of fault conditions include dry membrane, starvation, flooding, dry membrane and flooding, and flooding and starvation, each type including mild, moderate, and severe levels.
In step S1, different fault settings of different degrees and different types are implemented through different settings of operating parameters, and then:
on the basis of a standard working condition, the design of a membrane dry fault is realized by reducing the inlet air humidity and increasing the battery temperature; on the basis of standard working conditions, the flooding fault design is realized by increasing the inlet air humidity and reducing the battery temperature; on the basis of standard working conditions, the air metering ratio is reduced to realize the air-short fault design.
In the step S1, a sinusoidal current signal with a frequency range of 0.1Hz to 1000Hz is applied to the fuel cell by the ac excitation device, the frequency points are set in an exponential distribution type, and the response voltage and the excitation current are sampled by the data acquisition device to obtain an offline nonlinear impedance data set.
In the step S2, the first order frequency response function H 1 (omega) and a second order frequency response function H 2 The expression of (ω) is:
wherein A is AC excitation amplitude, H q,I (ω) is the first harmonic frequency response, H q,II (ω, ω) is the second order harmonic frequency response.
In step S1, the amplitude of the ac sinusoidal disturbance applied to the fuel cell is 15% of the dc, so as to ignore the third order frequency response function, ensure that the battery fault is not caused by the excessive current, and prevent the low signal-to-noise ratio from affecting the calculation accuracy of the frequency response function.
In step S3, the fault feature extraction and the dimension reduction through principal component analysis include the following steps:
s31, calculating the mean value and the standard deviation of the original data matrix according to columns, centralizing the mean value, obtaining a standardized matrix, and eliminating the difference of the characteristic components of different unit faults on the dimension and the magnitude;
s32, calculating a covariance matrix of the standardized matrix;
s33, acquiring eigenvectors and eigenvalues according to the covariance matrix;
s34, sorting the eigenvalues according to size, wherein the sorted eigenvalues are principal components;
and S35, calculating principal component contribution rate and cumulative contribution rate according to the eigenvalue, and selecting the eigenvalue and the corresponding eigenvector according to the requirement of the cumulative contribution rate to form a principal component matrix.
In step S35, principal components with an accumulated contribution rate exceeding 90% or 95% are selected to form a principal component matrix.
In step S4, the fault diagnosis model uses a support vector machine, a neural network, a gray scale model or a random forest.
In the step S4, when the fault diagnosis model adopts a random forest, the method specifically includes the following steps:
s41, randomly and repeatedly extracting samples with the same sample quantity from the fault data set by adopting a Bootstrap method, and using the samples to train each generated decision tree;
s42, establishing a plurality of decision trees by using the extracted sample subsets and a classification and regression algorithm, and combining to form a random forest;
s43, node splitting is carried out on each generated decision tree based on a random subspace, and an optimal characteristic variable and a splitting value are selected according to the Gini coefficient minimum principle to complete node splitting;
s44, repeating the steps S42-S43, constructing each decision tree in the random forest, and constructing a confusion matrix to test the classification result of all the decision trees;
and S45, voting the classification result of each decision tree, and taking the classification result with the largest vote number as a final classification.
Compared with the prior art, the invention has the following advantages:
firstly, the invention does not need to establish a complex or empirical fuel cell fault diagnosis model;
the invention can not only identify different fuel cell fault types, but also identify fault degrees;
compared with the existing electrochemical impedance spectrum diagnosis, the method adopts the nonlinear impedance information, additionally provides nonlinear fault characteristics for fault diagnosis, and has higher diagnosis precision.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
FIG. 2 shows the effect of different excitation amplitudes on the first and second order frequency response functions of the present invention, where FIG. 2a is the first order frequency response function obtained by applying different excitation amplitudes at a current density of 1.0A/cm2, FIG. 2b is the second order frequency response function obtained by applying different excitation amplitudes at a current density of 1.0A/cm2, FIG. 2c is the first order frequency response function obtained by applying different excitation amplitudes at a current density of 0.6A/cm2, and FIG. 2d is the second order frequency response function obtained by applying different excitation amplitudes at a current density of 0.6A/cm 2.
Fig. 3 is an example of the fault diagnosis results based on linearity and nonlinearity according to the present invention, where fig. 3a is a fault diagnosis result based on linear impedance and fig. 3b is a fault diagnosis result based on nonlinear impedance.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments.
Examples
As shown in fig. 1, the present invention provides a fuel cell fault diagnosis method based on nonlinear impedance spectroscopy, which includes the steps of:
s1, setting different types of fault conditions with different degrees, applying alternating current disturbance with a certain amplitude to the fuel cell, and collecting excitation current and response voltage;
in step S1, the fuel cell failure conditions should include common recoverable failures such as dry membrane, starvation, and flooding, while covering different degrees such as light, medium, and severe, and in this example, the fuel cell failure conditions are set as shown in table 1. For dry membrane failure, it was constructed by lowering inlet air humidity and raising cell temperature on a standard operating condition basis (temperature: 75 ℃; humidity: 50%/50%; metering ratio: 2/1.5; gas pressure: 120/100 kPa); aiming at the flooding fault, on the basis of a standard working condition, the structure is constructed by increasing the air inlet humidity and reducing the battery temperature; aiming at the air shortage fault, on the basis of the standard working condition, the air metering ratio is reduced to be constructed, and different degrees and different types of fault setting are realized through different working parameter settings.
TABLE 1 Fuel cell failure mode settings (cathode/anode)
Under fault conditions of different types and degrees, sinusoidal current signals with the frequency range of 0.1Hz-1000Hz are applied to the battery through the alternating current excitation device, the frequency points are set in an exponential distribution type, and meanwhile, the voltage and the current are sampled by the data acquisition device to obtain an offline nonlinear impedance data set.
In order to reduce the storage size of a data sample and the sampling duration, four frequency ranges are set, and in the high-frequency range of 100-1000Hz, each frequency excitation signal is injected with 0.5s of duration respectively; the period of the injection signal of the middle-low frequency band is set to be 10 in 10-100Hz and 1-10Hz, and the sampling frequency is respectively set to be 10kHz and 1 kHz; the fourth frequency range is a low frequency range of 0.1-1Hz, 6 frequencies are selected within the 10 frequency doubling range for sampling in order to further reduce the sampling time, 5 periodic signals are injected into each frequency, the sampling frequency is 100Hz, and the specific settings are shown in table 2.
TABLE 2 nonlinear impedance of fuel cell
| Excitation frequency range | Number of sampling points | Signal injection | Sampling frequency setting |
| 100-1000Hz | 11 | 0.5s | 100kHz |
| 10-100Hz | 11 | 10 cycles of | 10kHz |
| 1-10Hz | 11 | 10 cycles of | 1kHz |
| 1-0.1Hz | 6 | 5 cycles of | 100Hz |
In this example, the set of nonlinear impedance test data 180 includes 72 characteristic variables, which respectively represent the first order frequency response function value and the second order frequency response function value of the battery at 36 frequencies counted from 0.1 to 1000 Hz.
S2, removing steady-state components of the collected voltage data, then carrying out fast Fourier change to obtain each order of harmonic response under the target frequency, and calculating a first order frequency response function and a second order frequency response function;
in step S2, voltage data that is subjected to ac excitation in each fault state is collected, steady-state components are removed, and the voltage data is converted from time domain to frequency domain by using fast fourier transform algorithm, thereby obtaining each order harmonic frequency response of the fuel cell, wherein the first order, second order and third order harmonic frequency responses H q,I (ω,A)、H q,II (omega, A) and H q,III (ω, A) can be represented by:
wherein A is AC excitation amplitude, H 1 Is a first order frequency response function and is related to the linear response of the fuel cell; h 2 Is a second order frequency response function, H 3 Is a third order frequency response function, H 4 Is a fourth order frequency response function, H 5 In the present invention, for the convenience of frequency response function calculation, it is necessary to make the third order and above harmonic frequency responses negligible, and accordingly, the simplified first order and second order frequency response functions are:
therefore, the selection of a proper alternating current excitation amplitude is crucial, so that the third-order frequency response function can be ignored, battery faults caused by overlarge current can be avoided, and meanwhile, the selection of an undersize frequency response function can be avoided, so that the influence of the too low signal-to-noise ratio on the calculation precision of the frequency response function is prevented.
The invention is at 0.6A/cm 2 And 1.0A/cm 2 Under the current density, 3%, 5%, 10%, 15% and 20% of alternating current excitation is respectively applied to the fuel cell, and as a result, as shown in fig. 2, it can be seen that the first-order frequency response function has low dependence on the excitation amplitude and is less influenced by the excitation amplitude; the second order frequency response function is closely related to the excitation amplitude, and the current density is 0.6A/cm 2 When the amplitude is lower than 10%, the second-order frequency response function fluctuates greatly under the influence of measurement noise. However, excessive excitation amplitude is in practical fuel cellsThis is not easily achieved in the system and excessive current changes are prone to unwanted faults, for which the present invention selects an excitation amplitude of 15%, i.e., the amplitude of the ac sinusoidal disturbance applied to the fuel cell in step S1 is 15% of dc.
S3, forming a nonlinear impedance fault data set by the first-order frequency response function and the second-order frequency response function together, and performing fault feature extraction and dimension reduction on the data set by using principal component analysis;
the characteristic extraction of the diagnosis variable is a key step for constructing a fuel cell fault diagnosis model, the nonlinear impedance has more parameter variables, and partial linear correlation possibly exists, so that information redundancy is caused, and on the other hand, the prediction quality of the diagnosis model is possibly influenced by noise of partial data in a test sample. The original data is subjected to proper feature extraction, so that the complexity and the calculation difficulty of a model can be effectively reduced, and the extraction of key information and effective features in the original data is assisted.
In this example, the principal component analysis mainly includes the steps of preprocessing raw data, covariance calculation, eigenvalue arrangement, principal component extraction, and the like, and specifically includes the following steps:
(1) calculating the mean value and the standard deviation of the original data matrix according to columns, carrying out mean value centralization, obtaining a standardized matrix, and eliminating the difference of the characteristic components of different unit faults on the dimension and the magnitude;
(2) calculating a covariance matrix of the normalized matrix;
(3) acquiring eigenvectors and eigenvalues according to the covariance matrix;
(4) sorting the eigenvalues according to size, wherein the sorted eigenvalues are called principal components; the larger the eigenvalue of the covariance matrix is, the more information carried by the principal component is;
(5) and calculating principal component contribution rate and accumulated contribution rate according to the eigenvalue, and taking the eigenvalues of the specific number and corresponding eigenvectors according to the requirement of the accumulated contribution rate to form a principal component matrix.
In this example, the total variance interpretation of the fault data set is shown in table 3, which shows the eigenvalues, variance contribution rates, and variance cumulative contribution rates corresponding to the principal components. The total variance interpretation shows that the cumulative variance contribution rate of the first five principal components is 94.967%, the variance contribution rates of the 5 principal components are 62.205%, 23.307%, 5.974%, 1.992% and 1.488% respectively, the characteristic values of the five principal components are all larger than 1, and the characteristic root is obviously attenuated after the fifth principal component, and accordingly, the 5 principal components are selected as characteristic variables to construct a fault diagnosis model.
Table 3 total variance interpretation of failure data set
S4, training a diagnosis model based on the fault data set after dimension reduction;
the method comprises the following steps of training and learning a fault diagnosis model by using a fault data set after dimensionality reduction, wherein fault classification is performed by using a random forest algorithm, except for a random forest, other machine learning algorithms such as a support vector machine, a neural network and a gray model can be adopted, the random forest is a machine learning algorithm based on Bagging integrated learning and a random subspace, a plurality of decision trees are constructed by Bootstrap random resampling and node random splitting, and an optimal classification result is obtained by voting, and the method mainly comprises the following steps:
(1) randomly and repeatedly extracting samples with the same sample quantity from the original data set by using a Bootstrap method, and using the samples to train each generated decision tree;
(2) establishing a plurality of decision trees by using the extracted sample subsets and a classification and regression algorithm, and combining the decision trees to form a random forest;
(3) performing node splitting on each generated decision tree based on a random subspace, and selecting an optimal characteristic variable and a splitting value according to a Gini coefficient minimum principle to complete node splitting;
(4) building each decision tree in the random forest based on the steps, and building a confusion matrix to test classification results of all the decision trees;
(5) and voting the classification result of each decision tree, and taking the classification result with the largest vote number as a final classification.
And S5, applying alternating current disturbance with a set amplitude to the fuel cell, collecting voltage and current data, calculating a frequency response function according to S2, reducing the dimension, inputting the frequency response function into a trained diagnosis model, and diagnosing the current health state of the fuel cell in real time.
In the step S5, the fault diagnosis model is applied on line, when the fuel cell is in power generation operation, AC disturbances with certain amplitude and different frequencies are applied to the fuel cell in sequence, voltage and current data are collected on line, and first-order and second-order frequency response functions are calculated according to the method in the step S2; performing principal component analysis on the obtained frequency response function spectrum to make the frequency response function spectrum have the same dimension as the input dimension of the model; accordingly, the frequency response function spectrum after dimension reduction is input into the model to carry out fault diagnosis.
FIG. 3 shows the fault diagnosis result based on the random forest model, wherein the fault diagnosis based on the linear impedance spectrum only adopts a first-order frequency response function as an input; while the fault diagnosis based on the nonlinear impedance spectrum adopts the first-order and second-order frequency function as input, it can be seen that the accuracy of the fault diagnosis only by adopting the linear impedance information is 88.9%, and the accuracy of the fault diagnosis by adopting the nonlinear impedance is 97.8%.
The foregoing shows and describes the general principles and broad features of the present invention and advantages thereof. It should be apparent to those skilled in the art that the foregoing embodiments of the present invention are merely examples for clearly illustrating the present invention and are not to be construed as limiting the embodiments of the present invention. Other variations within the spirit of the invention will occur to those skilled in the art and are intended to be encompassed within the scope of the invention as claimed.
Claims (10)
1. A fuel cell fault diagnosis method based on nonlinear impedance spectroscopy is characterized by comprising the following steps:
s1, designing fault conditions of different degrees under different types, applying alternating current disturbance with a set amplitude to the fuel cell, and collecting excitation current and response voltage;
s2, removing steady-state components in the response voltage data, carrying out fast Fourier transform to obtain each order of harmonic response under the target frequency, and calculating a first order frequency response function and a second order frequency response function;
s3, forming a nonlinear impedance fault data set by the first-order frequency response function and the second-order frequency response function together, and performing fault feature extraction and dimension reduction on the data set through principal component analysis;
s4, according to the fault data set after dimensionality reduction, constructing a fault diagnosis model and training;
and S5, applying alternating current disturbance with a set amplitude to the fuel cell, collecting voltage and current data, calculating a frequency response function, inputting the frequency response function into a trained fault diagnosis model after dimension reduction, and diagnosing the current health state of the fuel cell in real time.
2. The method of claim 1, wherein in step S1, the different types of fault conditions include dry membrane, starvation, flooding, dry membrane and flooding, and flooding and starvation, each type including mild, moderate, and severe degrees.
3. The method of claim 2, wherein in step S1, different degrees and different types of fault settings are implemented through different operating parameter settings, and the following steps are performed:
on the basis of a standard working condition, the design of a membrane dry fault is realized by reducing the inlet air humidity and increasing the battery temperature; on the basis of standard working conditions, the flooding fault design is realized by increasing the inlet air humidity and reducing the battery temperature; on the basis of standard working conditions, the air metering ratio is reduced to realize the air-short fault design.
4. The method for diagnosing the fault of the fuel cell based on the nonlinear impedance spectrum as recited in claim 1, wherein in the step S1, a sinusoidal current signal with a frequency range of 0.1Hz to 1000Hz is applied to the fuel cell through an ac excitation device, and the frequency points are arranged in an exponential distribution type, and the response voltage and the excitation current are sampled through a data collection device to obtain an offline nonlinear impedance data set.
5. The method of claim 1, wherein in step S2, the first order frequency response function H is 1 (omega) and a second order frequency response function H 2 The expression of (ω) is:
wherein A is AC excitation amplitude, H q,I (ω) is the first harmonic frequency response, H q,II (ω, ω) is the second order harmonic frequency response.
6. The method as claimed in claim 5, wherein in step S1, the amplitude of the ac sinusoidal disturbance applied to the fuel cell is 15% of the dc, so as to ignore the third order frequency response function, and ensure that the cell fault is not caused by too much current, and at the same time, prevent the signal-to-noise ratio from being too low to affect the accuracy of the frequency response function calculation.
7. The method for diagnosing the fault of the fuel cell based on the nonlinear impedance spectrum as recited in claim 1, wherein the step S3 of extracting the fault feature and reducing the dimension by the principal component analysis comprises the steps of:
s31, calculating the mean value and the standard deviation of the original data matrix according to columns, centralizing the mean value, obtaining a standardized matrix, and eliminating the difference of the characteristic components of different unit faults on the dimension and the magnitude;
s32, calculating a covariance matrix of the standardized matrix;
s33, acquiring eigenvectors and eigenvalues according to the covariance matrix;
s34, sorting the eigenvalues according to size, wherein the sorted eigenvalues are principal components;
and S35, calculating principal component contribution rate and cumulative contribution rate according to the eigenvalue, and selecting the eigenvalue and the corresponding eigenvector according to the requirement of the cumulative contribution rate to form a principal component matrix.
8. The method as claimed in claim 7, wherein in step S35, the principal component with an accumulated contribution rate exceeding 90% or 95% is selected to form the principal component matrix.
9. The method as claimed in claim 1, wherein the fault diagnosis model in step S4 is a support vector machine, a neural network, a gray scale model or a random forest.
10. The method as claimed in claim 9, wherein in step S4, when the fault diagnosis model is a random forest, the method specifically comprises the following steps:
s41, randomly and repeatedly extracting samples with the same sample quantity from the fault data set by adopting a Bootstrap method, and using the samples to train each generated decision tree;
s42, establishing a plurality of decision trees by using the extracted sample subsets and a classification and regression algorithm, and combining to form a random forest;
s43, node splitting is carried out on each generated decision tree based on the random subspace, and the optimal characteristic variable and the splitting value are selected according to the Gini coefficient minimum principle to complete node splitting;
s44, repeating the steps S42-S43, constructing each decision tree in the random forest, and constructing a confusion matrix to test the classification result of all the decision trees;
and S45, voting is carried out on the classification result of each decision tree, and the classification result with the largest vote number is used as the final classification.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210609220.9A CN115084593B (en) | 2022-05-31 | 2022-05-31 | Fuel cell fault diagnosis method based on nonlinear impedance spectrum |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210609220.9A CN115084593B (en) | 2022-05-31 | 2022-05-31 | Fuel cell fault diagnosis method based on nonlinear impedance spectrum |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115084593A true CN115084593A (en) | 2022-09-20 |
| CN115084593B CN115084593B (en) | 2024-05-31 |
Family
ID=83248459
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210609220.9A Active CN115084593B (en) | 2022-05-31 | 2022-05-31 | Fuel cell fault diagnosis method based on nonlinear impedance spectrum |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115084593B (en) |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005063946A (en) * | 2003-07-24 | 2005-03-10 | Matsushita Electric Ind Co Ltd | Fuel cell system, fuel cell operation method, program, and record medium |
| US20070172708A1 (en) * | 2004-03-12 | 2007-07-26 | Yasuo Takebe | Fuel cell system failure diagnosis method, failure diagnosis device using same, and fuel cell system |
| US20140159738A1 (en) * | 2012-12-11 | 2014-06-12 | Hyundai Autron Co., Ltd. | Method and apparatus for diagnosing fault within fuel cell stack |
| CN109346745A (en) * | 2018-09-28 | 2019-02-15 | 北京机械设备研究所 | A kind of method and system judging fuel battery inside water state based on impedance |
| CN109726452A (en) * | 2018-12-12 | 2019-05-07 | 浙江大学 | A kind of online Proton Exchange Membrane Fuel Cells method for diagnosing faults based on impedance spectrum |
| US20190148750A1 (en) * | 2017-11-10 | 2019-05-16 | Hyundai Motor Company | Method for calculating voltage loss of fuel cell and system performing the same |
| US20200064406A1 (en) * | 2016-11-04 | 2020-02-27 | Avl List Gmbh | Method for diagnosing a technical system |
| CN111628195A (en) * | 2020-05-11 | 2020-09-04 | 中国科学技术大学 | Fuel cell stack real-time state identification method based on logic reasoning |
| WO2020183489A1 (en) * | 2019-03-08 | 2020-09-17 | INDIAN INSTITUTE OF TECHNOLOGY MADRAS (IIT Madras) | Method for impedance measurement using multiple phase shifted chirp signals |
| CN112415413A (en) * | 2020-11-05 | 2021-02-26 | 上海捷氢科技有限公司 | Diagnosis method and diagnosis system of fuel cell |
| CN113359037A (en) * | 2021-06-11 | 2021-09-07 | 国网新疆电力有限公司营销服务中心(资金集约中心、计量中心) | Fuel cell fault diagnosis method based on BP neural network |
| WO2021213989A1 (en) * | 2020-04-24 | 2021-10-28 | Robert Bosch Gmbh | Fuel cell diagnostic apparatus and corresponding diagnostic method, vehicle, and vehicle system |
| CN114361535A (en) * | 2021-12-16 | 2022-04-15 | 同济大学 | Fuel cell hydrogen permeation quantity evaluation method based on electrochemical impedance spectrum |
| KR102397652B1 (en) * | 2021-06-29 | 2022-05-13 | 지엘홀딩스 주식회사 | Fuel cell failure diagnosis system using variable frequency and the method using it |
-
2022
- 2022-05-31 CN CN202210609220.9A patent/CN115084593B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005063946A (en) * | 2003-07-24 | 2005-03-10 | Matsushita Electric Ind Co Ltd | Fuel cell system, fuel cell operation method, program, and record medium |
| US20070172708A1 (en) * | 2004-03-12 | 2007-07-26 | Yasuo Takebe | Fuel cell system failure diagnosis method, failure diagnosis device using same, and fuel cell system |
| US20140159738A1 (en) * | 2012-12-11 | 2014-06-12 | Hyundai Autron Co., Ltd. | Method and apparatus for diagnosing fault within fuel cell stack |
| US20200064406A1 (en) * | 2016-11-04 | 2020-02-27 | Avl List Gmbh | Method for diagnosing a technical system |
| US20190148750A1 (en) * | 2017-11-10 | 2019-05-16 | Hyundai Motor Company | Method for calculating voltage loss of fuel cell and system performing the same |
| CN109346745A (en) * | 2018-09-28 | 2019-02-15 | 北京机械设备研究所 | A kind of method and system judging fuel battery inside water state based on impedance |
| CN109726452A (en) * | 2018-12-12 | 2019-05-07 | 浙江大学 | A kind of online Proton Exchange Membrane Fuel Cells method for diagnosing faults based on impedance spectrum |
| WO2020183489A1 (en) * | 2019-03-08 | 2020-09-17 | INDIAN INSTITUTE OF TECHNOLOGY MADRAS (IIT Madras) | Method for impedance measurement using multiple phase shifted chirp signals |
| WO2021213989A1 (en) * | 2020-04-24 | 2021-10-28 | Robert Bosch Gmbh | Fuel cell diagnostic apparatus and corresponding diagnostic method, vehicle, and vehicle system |
| CN111628195A (en) * | 2020-05-11 | 2020-09-04 | 中国科学技术大学 | Fuel cell stack real-time state identification method based on logic reasoning |
| CN112415413A (en) * | 2020-11-05 | 2021-02-26 | 上海捷氢科技有限公司 | Diagnosis method and diagnosis system of fuel cell |
| CN113359037A (en) * | 2021-06-11 | 2021-09-07 | 国网新疆电力有限公司营销服务中心(资金集约中心、计量中心) | Fuel cell fault diagnosis method based on BP neural network |
| KR102397652B1 (en) * | 2021-06-29 | 2022-05-13 | 지엘홀딩스 주식회사 | Fuel cell failure diagnosis system using variable frequency and the method using it |
| CN114361535A (en) * | 2021-12-16 | 2022-04-15 | 同济大学 | Fuel cell hydrogen permeation quantity evaluation method based on electrochemical impedance spectrum |
Non-Patent Citations (7)
| Title |
|---|
| DU RB等: "A fault diagnosis model for proton exchange membrane fuel cell based on impedance identification with differential evolution algorithm", 《INTERNATIONAL JOURNAL OF HYDROGEN ENERGY》, vol. 46, no. 78, 11 November 2021 (2021-11-11), pages 38795 - 38808 * |
| KADYK, T等: "Nonlinear frequency response analysis for the diagnosis of carbon monoxide poisoning in PEM fuel cell anodes", 《JOURNAL OF APPLIED ELECTROCHEMISTRY》, vol. 41, no. 9, 5 April 2010 (2010-04-05), pages 1021 - 1032, XP019945655, DOI: 10.1007/s10800-011-0298-8 * |
| THOMAS KADYK等: "Nonlinear frequency response analysis of PEM fuel cells for diagnosis of dehydration, flooding and CO-poisoning", 《JOURNAL OF ELECTROANALYTICAL CHEMISTRY》, no. 630, pages 19 - 27 * |
| 孙腾飞: "质子交换膜燃料电池水淹/膜干研究", 《中国优秀硕士学位论文全文数据库 (工程科技Ⅰ辑)》, no. 3, pages 015 - 117 * |
| 张少哲等: "质子交换膜燃料电池电化学阻抗谱敏感性研究", 《机械工程学报》, vol. 57, no. 14, 31 July 2021 (2021-07-31), pages 40 - 51 * |
| 张雪霞等: "质子交换膜燃料电池水淹和膜干故障诊断研究综述", 《西南交通大学学报》, vol. 55, no. 04, 31 August 2020 (2020-08-31), pages 828 - 838 * |
| 王筱彤: "燃料电池EIS测试系统设计及故障诊断方法研究", 《 中国优秀硕士学位论文全文数据库 (工程科技Ⅰ辑)》, 15 July 2021 (2021-07-15), pages 015 - 219 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115084593B (en) | 2024-05-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113962259B (en) | Multi-mode double-layer fault diagnosis method for fuel cell system | |
| CN108664010A (en) | Generating set fault data prediction technique, device and computer equipment | |
| CN112257530B (en) | Rolling bearing fault diagnosis method based on blind signal separation and support vector machine | |
| CN104396073A (en) | Method for determining critical operating states in a fuel cell stack | |
| Huang et al. | Review of data-driven prognostics and health management techniques: lessions learned from PHM data challenge competitions | |
| CN111680726B (en) | Transformer fault diagnosis method and system based on neighbor component analysis and k neighbor learning fusion | |
| CN116226646B (en) | Method, system, equipment and medium for predicting health state and residual life of bearing | |
| CN113866642A (en) | Lithium ion battery fault diagnosis method based on gradient lifting tree | |
| CN111273125A (en) | RST-CNN-based power cable channel fault diagnosis method | |
| CN115563563A (en) | Fault diagnosis method and device based on transformer oil chromatographic analysis | |
| CN115331696A (en) | Multi-channel voiceprint signal blind source separation method for transformer abnormity diagnosis | |
| CN114118219A (en) | Data-driven real-time abnormal detection method for health state of long-term power-on equipment | |
| CN116089857A (en) | Transformer fault identification method based on CEEMDAN-DBN | |
| CN115496108A (en) | Fault monitoring method and system based on manifold learning and big data analysis | |
| CN113255591A (en) | Bearing fault diagnosis method based on random forest and fusion characteristics | |
| CN116010884A (en) | Fault diagnosis method of SSA-LightGBM oil-immersed transformer based on principal component analysis | |
| CN114325433A (en) | Lithium ion battery fault detection method and system based on electrochemical impedance spectrum test | |
| CN117117258B (en) | Fault monitoring method and device for hydrogen fuel cell system | |
| CN117056678B (en) | Machine pump equipment operation fault diagnosis method and device based on small sample | |
| Peng et al. | A hybrid health prognostics method for proton Exchange membrane fuel cells with internal health recovery | |
| CN117276600B (en) | PSO-GWO-DELM-based fault diagnosis method for proton exchange membrane fuel cell system | |
| CN104499001B (en) | Feature based subspace optimizes the aluminium cell condition diagnostic method of relative matrix | |
| CN116127398B (en) | Hydraulic pump fault diagnosis method based on mechanism model and multi-source data fusion | |
| CN115084593B (en) | Fuel cell fault diagnosis method based on nonlinear impedance spectrum | |
| CN116031453A (en) | On-line estimation method for characteristic frequency impedance of proton exchange membrane fuel cell |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |