A fuel cell fault diagnosis method, device and medium

Abstract

The embodiments of the present application disclose a fuel cell fault diagnosis method, device and medium, belonging to the technical field of fuel cells, and solving the problem of low accuracy of fuel cell fault diagnosis when historical data does not fully cover all faults. The method includes: performing a cyclic period dimension increase process on the two-dimensional impedance spectrum corresponding to the impedance data of the preset sample fuel cell to obtain a three-dimensional impedance field; performing data fitting on the three-dimensional impedance field to generate a mirror impedance field, and training the mirror impedance field by an interleaved learning method to obtain a cross-learning impedance field extension model; inputting the impedance spectrum of the fuel cell to be measured into the cross-learning impedance field extension model to generate a life cycle impedance field; performing coordinate plane projection processing on the life cycle impedance field of the fuel cell to be measured to obtain an impedance density map; classifying the impedance density map to determine the fault information of the fuel cell to be measured. The accuracy of fault diagnosis is improved by the above method.

Description

Technical Field

[0001] This application relates to the technical field of fuel cells, and particularly to a fuel cell fault diagnosis method, device, and medium. Background Art

[0002] Currently, common fuel cell state detection and fault diagnosis methods include: non-destructive testing technology based on electromagnetic field changes, non-destructive testing theory and methods based on magnetic field imaging, electrochemical impedance spectroscopy testing technology, data-driven fault diagnosis methods, and fault diagnosis based on deep learning, etc.

[0003] However, the fuel cell fault diagnosis in the prior art mostly relies on historical fault data, is easily affected by data quality, and has a low fault diagnosis accuracy when historical data does not fully cover all faults. Summary of the Invention

[0004] Embodiments of this application provide a fuel cell fault diagnosis method, device, and medium, which are used to solve the following technical problems: The fuel cell fault diagnosis in the prior art mostly relies on historical fault data, is easily affected by data quality, and has a low fault diagnosis accuracy when historical data does not fully cover all faults.

[0005] Embodiments of this application adopt the following technical solutions:

[0006] Embodiments of this application provide a fuel cell fault diagnosis method. The method includes: performing a cyclic period dimension increase process on a two-dimensional impedance spectrum corresponding to impedance data of a preset sample fuel cell to obtain a three-dimensional impedance field corresponding to the impedance data of the preset sample fuel cell; performing data fitting on the three-dimensional impedance field corresponding to the impedance data of the preset sample fuel cell through an equivalent circuit to generate a mirror impedance field, and training the mirror impedance field through an interleaved learning method to obtain a cross-learning impedance field expansion model; inputting an impedance spectrum of a to-be-tested fuel cell into the cross-learning impedance field expansion model, and generating a life cycle impedance field of the to-be-tested fuel cell through the cross-learning impedance field expansion model; performing coordinate plane projection processing on the life cycle impedance field of the to-be-tested fuel cell to obtain an impedance density map; classifying the impedance density map based on a preset neural network model to determine fault information of the to-be-tested fuel cell through a classification result.

[0007] In the embodiments of the present application, through knowledge transfer, on the basis of the two-dimensional impedance spectrum, the dimension of the fuel cell life cycle is added, and the two-dimensional impedance spectrum is transformed into a three-dimensional image. Then, based on the impedance field and impedance density, the fuel cell fault diagnosis is carried out, which improves the accuracy and efficiency of the fault diagnosis. At the same time, the impedance density image can also better display the abnormal performance gain caused by the oxygen supply interruption of the fuel cell, and assist in the mechanism analysis. By using the method of staggered forward and backward learning, the information in different time directions of the data can be fully mined, the utilization rate of the data can be improved, the understanding of the impedance change during the entire life cycle can be improved, the relationship between the impedance field and the fault can be better captured, and the internal fault mechanism of the fuel cell can be determined more accurately.

[0008] In one implementation manner of the present application, a cyclic period dimension increase process is performed on the two-dimensional impedance spectrum corresponding to the impedance data of the preset sample fuel cell to obtain a three-dimensional impedance field corresponding to the impedance data of the preset sample fuel cell, which specifically includes: generating a two-dimensional battery impedance spectrum based on the impedance data of the preset sample fuel cell; drawing an impedance curve waterfall diagram corresponding to the impedance curves of different cyclic periods based on the two-dimensional battery impedance spectrum to increase the cyclic period dimension of the two-dimensional battery impedance spectrum; performing a color gradient process on the impedance curve waterfall diagram; performing surface fitting on the impedance curve waterfall diagram after the color gradient process by using the cubic spline interpolation method, and performing smoothing processing on the interpolation result by using Gaussian filtering to obtain a three-dimensional impedance field corresponding to the impedance data of the preset sample fuel cell.

[0009] In one implementation manner of the present application, data fitting is performed on the three-dimensional impedance field corresponding to the impedance data of the preset sample fuel cell through an equivalent circuit to generate a mirror impedance field, which specifically includes: performing component parameter fitting based on the first cyclic period and the last cyclic period corresponding to the impedance data of the preset sample fuel cell; where the component parameters at least include one of a series resistance, a constant phase element, and a polarization resistance; determining the simulation circuit component parameters within the intermediate cyclic period based on the component parameter values corresponding to the first cyclic period and the last cyclic period respectively through a linear interpolation function to generate complete simulation equivalent circuit component parameter values; constructing a mirror impedance field based on the complete simulation equivalent circuit component parameter values; where the mirror impedance field is used to reflect the impedance change trend of the battery under different cyclic periods.

[0010] In one implementation manner of the present application, the simulation circuit component parameters within the intermediate cyclic period are determined based on the component parameter values corresponding to the first cyclic period and the last cyclic period respectively through a linear interpolation function, which specifically includes: based on the function:

[0011]

[0012] Obtain the slope between the component parameter values of the first cycle period and the last cycle period; based on the point - slope form equation of a straight line, obtain the component parameter values of the nth cycle period:

[0013]

[0014] Based on the component parameters of the first cycle period and the last cycle period, determine the simulation equivalent circuit component parameter values of any intermediate cycle period:

[0015]

[0016] Wherein, is the component parameter value of the last cycle period; is the component parameter value of the first cycle period; is the simulation equivalent circuit component parameter value of any intermediate cycle period; k is the slope; n is the cycle period sequence, where 1 < n < m .

[0017] In an implementation manner of the present application, the mirror impedance field is trained by an interleaved learning method to obtain a cross - learning impedance field expansion model, specifically including: obtaining mirror impedance field data, analyzing and modeling the mirror impedance field data through step - by - step linear regression, and optimizing the obtained step - by - step linear regression model through the interleaved learning method.

[0018] In an implementation manner of the present application, the step - by - step linear regression model is:

[0019]

[0020] Wherein, Z is the impedance value; CPE T is the first component parameter in the equivalent circuit; CPE P is the second component parameter in the equivalent circuit; Rs is the third component parameter in the equivalent circuit; Rp is the fourth component parameter in the equivalent circuit; n is the cycle period sequence; is the preset coefficient; is the coefficient corresponding to the first component parameter; is the coefficient corresponding to the second component parameter; is the coefficient corresponding to the third component parameter; is the coefficient corresponding to the fourth component parameter; is the coefficient corresponding to the fifth component parameter; is the product between the last component parameter and the corresponding coefficient.

[0021] In an implementation manner of the present application, the obtained stepwise linear regression model is optimized by an interleaved learning method, which specifically includes: performing Gaussian noise processing on the impedance density image corresponding to the preset sample fuel cell impedance data to obtain an augmented data set; dividing the augmented data set into a training set and a validation set according to a preset ratio; training based on the training set, the validation set, and the preset ResNet–50 neural network, and optimizing the stepwise linear regression model by using the interleaved learning method.

[0022] In an implementation manner of the present application, coordinate plane projection processing is performed on the impedance field of the fuel cell to be measured during its life cycle to obtain an impedance density map, which specifically includes: projecting the impedance field of the fuel cell to be measured during its life cycle onto the x-y, y-z, and x-z planes to obtain a scatter plot; optimizing the scatter plot into a contour map and performing gradient processing between the contour maps; filling the contour map through the contourf function and performing image boundary smoothing processing on the filled contour map through the pcolormesh function to obtain an impedance density map.

[0023] An embodiment of the present application provides a fuel cell fault diagnosis device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: perform cyclic period dimension increase processing on the two-dimensional impedance spectrum corresponding to the preset sample fuel cell impedance data to obtain a three-dimensional impedance field corresponding to the preset sample fuel cell impedance data; perform data fitting on the three-dimensional impedance field corresponding to the preset sample fuel cell impedance data through an equivalent circuit to generate a mirror impedance field, and train the mirror impedance field by using an interleaved learning method to obtain a cross-learning impedance field expansion model; input the impedance spectrum of the fuel cell to be measured obtained into the cross-learning impedance field expansion model, and generate an impedance field of the fuel cell to be measured during its life cycle through the cross-learning impedance field expansion model; perform coordinate plane projection processing on the impedance field of the fuel cell to be measured during its life cycle to obtain an impedance density map; classify the impedance density map based on the preset neural network model to determine the fault information of the fuel cell to be measured through the classification result.

[0024] A non-volatile computer storage medium provided by an embodiment of the present application stores computer-executable instructions, and the computer-executable instructions are configured to: perform a cyclic period dimension increase process on a two-dimensional impedance spectrum corresponding to preset sample fuel cell impedance data to obtain a three-dimensional impedance field corresponding to the preset sample fuel cell impedance data; perform data fitting on the three-dimensional impedance field corresponding to the preset sample fuel cell impedance data through an equivalent circuit to generate a mirror impedance field, and train the mirror impedance field through an interleaved learning method to obtain a cross-learning impedance field expansion model; input the impedance spectrum of the fuel cell to be measured obtained into the cross-learning impedance field expansion model, and generate a life cycle impedance field of the fuel cell to be measured through the cross-learning impedance field expansion model; perform coordinate plane projection processing on the life cycle impedance field of the fuel cell to be measured to obtain an impedance density map; classify the impedance density map based on a preset neural network model to determine the fault information of the fuel cell to be measured through the classification result.

[0025] The above at least one technical solution adopted by the embodiment of the present application can achieve the following beneficial effects: Through knowledge transfer, the embodiment of the present application adds the dimension of the fuel cell life cycle on the basis of the two-dimensional impedance spectrum, converts the two-dimensional impedance spectrum into a three-dimensional image, and then diagnoses the fuel cell fault based on the impedance field and impedance density, improving the accuracy and efficiency of fault diagnosis. At the same time, the impedance density image can also better display the abnormal performance gain caused by the oxygen supply interruption of the fuel cell, assisting in mechanism analysis. By using the method of interleaved forward and backward learning, the information in different time directions of the data can be fully mined, the utilization rate of the data can be improved, the understanding of the impedance change during the entire life cycle can be improved, the relationship between the impedance field and the fault can be better captured, and the internal fault mechanism of the fuel cell can be determined more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings. In the drawings:

[0027] Figure 1 It is a flowchart of a fuel cell fault diagnosis method provided by an embodiment of the present application;

[0028] Figure 2 It is a schematic diagram of the process of a fuel cell fault diagnosis method provided by an embodiment of the present application;

[0029] Figure 3 It is a schematic diagram of an impedance spectrum in the initial cyclic state provided by an embodiment of the present application;

[0030] Figure 4 A full - cycle impedance waterfall diagram provided by an embodiment of the present application;

[0031] Figure 5 A schematic diagram of the evolution process of the full - cycle impedance provided by an embodiment of the present application;

[0032] Figure 6 A schematic diagram of the preliminary generation of the impedance field provided by an embodiment of the present application;

[0033] Figure 7 A schematic diagram of the impedance field provided by an embodiment of the present application;

[0034] Figure 8 A comparison diagram of the impedance field and the mirror impedance field provided by an embodiment of the present application;

[0035] Figure 9 A schematic diagram of linear regression provided by an embodiment of the present application;

[0036] Figure 10 A schematic diagram of interleaved learning provided by an embodiment of the present application;

[0037] Figure 11 A schematic diagram of a 1% Gaussian noise image in the normal state provided by an embodiment of the present application;

[0038] Figure 12 A schematic diagram of a 10% Gaussian noise image in the normal state provided by an embodiment of the present application;

[0039] Figure 13 A schematic diagram of a 20% Gaussian noise image in the normal state provided by an embodiment of the present application;

[0040] Figure 14 A schematic diagram of a 1% Gaussian noise image in the dry - film state provided by an embodiment of the present application;

[0041] Figure 15 A schematic diagram of a 10% Gaussian noise image in the dry - film state provided by an embodiment of the present application;

[0042] Figure 16 A schematic diagram of a 20% Gaussian noise image in the dry - film state provided by an embodiment of the present application;

[0043] Figure 17 A schematic diagram of the generation of the impedance density image provided by an embodiment of the present application, where Figure 17 in (a) is a scatter plot, Figure 17 in (b) is a contour plot, Figure 17 in (c) is a gradient - added plot between the contours, Figure 17 in (d) is a plot for reducing the obvious boundaries between color blocks, Figure 17Figure (e) is a smooth color transition diagram, Figure 17 Figure (f) is an impedance density diagram;

[0044] Figure 18 This is a training progress diagram of ResNet-50 provided by an embodiment of the present application. Among them, Figure 18 Figure (a) is a loss schematic diagram, Figure 18 Figure (b) is a precision schematic diagram, Figure 18 Figure (c) is a comparison diagram of prediction and actual;

[0045] Figure 19 This is a structural schematic diagram of a fuel cell fault diagnosis device provided by an embodiment of the present application.

[0046] Reference numerals:

[0047] 200 Fuel cell fault diagnosis device, 201 Processor, 202 Memory. Detailed implementation manners

[0048] An embodiment of the present application provides a fuel cell fault diagnosis method, device and medium.

[0049] In order to enable those skilled in the art to better understand the technical solutions in the present application, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0050] The following will detail the technical solutions proposed in the embodiments of the present application through the accompanying drawings.

[0051] Figure 1 This is a flowchart of a fuel cell fault diagnosis method provided by an embodiment of the present application. As Figure 1 shown, the fuel cell fault diagnosis method includes the following steps:

[0052] Step 101: Perform a processing of increasing the cycle period dimension on the two-dimensional impedance spectrum corresponding to the impedance data of the preset sample fuel cell to obtain a three-dimensional impedance field corresponding to the impedance data of the preset sample fuel cell.

[0053] In one embodiment of the present application, a two-dimensional battery impedance spectrum is generated based on preset sample fuel cell impedance data. Based on the two-dimensional battery impedance spectrum, an impedance curve waterfall plot corresponding to impedance curves of different cycle periods is plotted to increase the cycle period dimension of the two-dimensional battery impedance spectrum. Color gradient processing is performed on the impedance curve waterfall plot. Cubic spline interpolation method is used to perform surface fitting on the impedance curve waterfall plot after color gradient processing, and Gaussian filtering is used to smooth the interpolation result to obtain a three-dimensional impedance field corresponding to the preset sample fuel cell impedance data.

[0054] Specifically, Figure 2 is a schematic flowchart of a fuel cell fault diagnosis method provided by an embodiment of the present application. As Figure 2 shown, taking a lithium battery as a preset sample fuel cell as an example, the states of the lithium battery are classified, a large amount of impedance data of the lithium battery under different cycle periods is collected, an impedance curve waterfall plot during the cycle is plotted, color gradient processing is performed on the impedance curves in different cycles, cubic spline interpolation method is used for fitting, and Gaussian filtering method is used to smooth the interpolation result to generate a three-dimensional "impedance field".

[0055] Furthermore, Figure 3 is a schematic diagram of an impedance spectrum at the initial state of a cycle provided by an embodiment of the present application, Figure 4 is a full-cycle impedance waterfall plot provided by an embodiment of the present application, Figure 5 is a schematic diagram of the evolution process of the full-cycle impedance provided by an embodiment of the present application, Figure 6 is a schematic diagram of the preliminary generation of an impedance field provided by an embodiment of the present application, Figure 7 is a schematic diagram of an impedance field provided by an embodiment of the present application. As Figure 3 、 Figure 4 、 Figure 5 、 Figure 6 and Figure 7 shown, the specific process of generating the impedance field is as follows:

[0056] Step 1.1: Generate a battery impedance spectrum. As Figure 3 shown, where the lower curve represents the impedance curve of the last cycle period, and the upper curve is the impedance curve of the first cycle period;

[0057] Step 1.2: According to the impedance curve waterfall plot generated from impedance curves in different cycle states, as Figure 4 shown, it is increased from a two-dimensional impedance spectrum to a three-dimensional waterfall plot, introducing the dimension of the cycle period of the lithium battery;

[0058] Step 1.3: Perform color gradient processing on the impedance data from the first cycle period to the last cycle period of the lithium battery to obtain an image as Figure 5 shown;

[0059] Step 1.4: Flip Re(z), perform cubic spline interpolation using the griddata function, and perform surface fitting on the impedance data of the lithium battery over the entire cycle to form a three-dimensional impedance field as Figure 6 shown. If abnormal points appear at the edge of the surface, Gaussian filtering is used to smooth the interpolation result, and finally a three-dimensional impedance field as Figure 7 shown is obtained.

[0060] Step 102: Perform data fitting on the three-dimensional impedance field corresponding to the impedance data of the preset sample fuel cell through an equivalent circuit, generate a mirror impedance field, and train the mirror impedance field through an interleaved learning method to obtain a cross-learning impedance field expansion model.

[0061] In an embodiment of the present application, component parameter fitting is performed based on the first cycle and the last cycle corresponding to the impedance data of the preset sample fuel cell; wherein, the component parameters include at least one of series resistance, constant phase element, and polarization resistance. Through a linear interpolation function, based on the component parameter values corresponding to the first cycle and the last cycle respectively, the simulation circuit component parameters in the intermediate cycle are determined to generate complete simulation equivalent circuit component parameter values. Based on the complete simulation equivalent circuit component parameter values, a mirror impedance field is constructed; wherein, the mirror impedance field is used to reflect the impedance change trend of the battery at different cycles.

[0062] Specifically, for fuel cell fault diagnosis, as Figure 2 shown on the right, since usually only one impedance curve can be obtained for a fuel cell, and a single impedance curve is difficult to reflect the impedance change of its entire life cycle and cannot provide sufficient information for accurate fault diagnosis. Therefore, machine learning is performed using the relationship between the "mirror impedance field" generated by the lithium battery and the original impedance field of the lithium battery. After training the model, the model is used to generate the impedance field of the fuel cell life cycle. In this process, the interleaved learning method is used to perform forward expansion of the impedance field from the historical data set to future prediction and backward expansion of the impedance field from the future data set to historical prediction, and stepwise linear regression training is used to capture the change process of the mirror impedance field to realize the expansion of the fuel cell impedance curve into an impedance field over the entire life cycle.

[0063] Furthermore, use Rs ( CPE - Rp ) equivalent circuit to fit the impedance data of the battery.

[0064] Select the first cycle and the 350th cycle (the last cycle) for component parameter fitting, wherein the component parameters include Rs (series resistance), CPE (constant phase element) and Rs(Polarization resistance), these parameter values reflect the chemical and physical characteristics inside the battery.

[0065] Analyze the equivalent circuit element parameters of the first and last cycles. Since the adopted cycle period is shorter than the overall service life of the battery, to a certain extent, the element parameters can be considered to change linearly. On this basis, use a linear interpolation function to process the element parameters within the cycle period, that is, calculate the simulation circuit element parameters within the intermediate cycle period according to the element parameter values of the first and last cycle periods, and automatically generate the complete simulation equivalent circuit element parameter values. Therefore, let the element parameter value of the 1st cycle period be P 1 and the element parameter value of the last cycle period be P m To calculate the element parameter value of the n th cycle period (1 < n < m ). P n .

[0066] In an embodiment of the present application, based on the function:

[0067]

[0068] Obtain the slope between the element parameter value of the first cycle period and the element parameter value of the last cycle period;

[0069] Based on the point-slope form equation of a straight line, obtain the element parameter value of the n th cycle period:

[0070]

[0071] Based on the element parameters of the first cycle period and the last cycle period, determine the simulation equivalent circuit element parameter values of any intermediate cycle period:

[0072]

[0073] Among them, is the element parameter value of the last cycle period; is the element parameter value of the first cycle period; is the simulation equivalent circuit element parameter value of any intermediate cycle period; k is the slope; n is the cycle period sequence, where 1 < n < m .

[0074] According to the above component parameters and algorithms, a virtual impedance field that can reflect the impedance change trend of the battery under different cycle periods is constructed, and this virtual impedance field is the mirror impedance field in the embodiment of the present application. Figure 8 A comparison graph of an impedance field and a mirror impedance field provided by an embodiment of the present application, Figure 8 It can be seen the change trends of the mirror impedance field and the impedance field under different periods.

[0075] In an embodiment of the present application, mirror impedance field data is obtained, the mirror impedance field data is analyzed and modeled by stepwise linear regression, and the obtained stepwise linear regression model is optimized by an interleaved learning method.

[0076] Specifically, based on the basic data provided by the mirror impedance field, stepwise linear regression analyzes and models the mirror impedance field data, and a model that can accurately describe the change law of the impedance field is obtained. Let Z be the impedance value, CPE - T , CPE - P , Rs , Rp etc. be the component parameters in the equivalent circuit, n be the independent variables such as the number of cycle periods, then the stepwise linear regression model can be expressed as:

[0077]

[0078] Among them, Z is the impedance value; CPE T is the first component parameter in the equivalent circuit; CPE P is the second component parameter in the equivalent circuit; Rs is the third component parameter in the equivalent circuit; Rp is the fourth component parameter in the equivalent circuit; n is the cycle period sequence; is the preset coefficient; is the coefficient corresponding to the first component parameter; is the coefficient corresponding to the second component parameter; is the coefficient corresponding to the third component parameter; is the coefficient corresponding to the fourth component parameter; is the coefficient corresponding to the fifth component parameter; is the product of the last component parameter and the corresponding coefficient.

[0079] Figure 9 A schematic diagram of linear regression provided by an embodiment of the present application, as Figure 9As shown, good accuracy was achieved when using stepwise linear regression in training the model, and the fuel cell data was successfully extended to an impedance field.

[0080] In one embodiment of the present application, Gaussian noise processing is performed on the impedance density image corresponding to the impedance data of the preset sample fuel cell to obtain an augmented data set. According to a preset ratio, the augmented data set is divided into a training set and a validation set. Based on the training set, the validation set, and the preset ResNet–50 neural network, training is carried out, and the stepwise linear regression model is optimized by using the interleaved learning method.

[0081] Specifically, Gaussian noise of 1% - 100% is added to the impedance density image of the obtained impedance data of the preset sample fuel cell for data set augmentation. The augmented data set of the battery impedance density image is divided into a training set and a validation set according to a ratio of 7:3. The ResNet–50 neural network is used for training, and interleaved learning is used to further optimize the training process of the stepwise linear regression model to ensure a high accuracy rate in the field of lithium battery state classification.

[0082] Figure 10 The following is a schematic diagram of interleaved learning provided by an embodiment of the present application. As Figure 10 shown, the data set is interleaved and moved, and the impedance field is extended forward and backward by using the historical and future data sets respectively, enabling the model to better learn the internal relationship of the data.

[0083] Step 103: Input the impedance spectrum of the measured fuel cell obtained into the cross-learning impedance field extension model, and generate the impedance field of the life cycle of the measured fuel cell through the cross-learning impedance field extension model.

[0084] In one embodiment of the present application, the impedance curve of the measured fuel cell is extended in the impedance field throughout its life cycle by using the model trained with the preset sample fuel cell that has been trained, and the impedance density images of the measured fuel cell in different states are obtained.

[0085] Figure 11 The following is a schematic diagram of a 1% Gaussian noise image in the normal state provided by an embodiment of the present application. Figure 12 The following is a schematic diagram of a 10% Gaussian noise image in the normal state provided by an embodiment of the present application. Figure 13 The following is a schematic diagram of a 20% Gaussian noise image in the normal state provided by an embodiment of the present application; Figure 14 The following is a schematic diagram of a 1% Gaussian noise image in the membrane dry state provided by an embodiment of the present application. Figure 15 The following is a schematic diagram of a 10% Gaussian noise image in the membrane dry state provided by an embodiment of the present application. Figure 16 The following is a schematic diagram of a 20% Gaussian noise image in the membrane dry state provided by an embodiment of the present application; Figure 11 andFigure 14 Compared with the Gaussian noise density image generated after adding 1% Gaussian noise, Figure 12 and Figure 15 compared with the Gaussian noise density image generated after adding 10% Gaussian noise, Figure 13 and Figure 16 compared with the Gaussian noise density image generated after adding 20% Gaussian noise, it can be seen by comparison that under the dry state of the membrane, as the number of life cycle increases, the impedance will continuously increase, which is completely different from the normal state.

[0086] Step 104: Perform coordinate plane projection processing on the impedance field of the fuel cell under test during its life cycle to obtain an impedance density map.

[0087] In an embodiment of the present application, project the impedance field of the fuel cell under test during its life cycle onto the x - y, y - z, and x - z planes to obtain a scatter plot. Optimize the scatter plot into a contour plot and perform a gradient processing between the contour plots. Fill the contour plot through the contourf function and perform image boundary smoothing processing on the filled contour plot through the pcolormesh function to obtain an impedance density map.

[0088] Specifically, Figure 17 is a schematic diagram for generating an impedance density image provided by an embodiment of the present application, where Figure 17 in (a) is a scatter plot, Figure 17 in (b) is a contour plot, Figure 17 in (c) is a gradient map added between the contour lines, Figure 17 in (d) is a map reducing the obvious boundary between color blocks, Figure 17 in (e) is a map for smoothing color transition, Figure 17 in (f) is an impedance density map; as Figure 17 shown, project the generated impedance field onto the x - y, y - z, and x - z planes to obtain a scatter plot. As shown in Figure 17 in (a), the colors of the scatter plot represent the z - value, x - value, and y - value respectively. Optimize the scatter plot into a contour plot, add a gradient between the contour plots, then use the contourf function to process the contour plot, and finally use the pcolormesh function to smooth the image boundary. This process is as shown in Figure 17 from (b) to Figure 17 in (e). The finally obtained impedance density image is as shown in Figure 17 in (d).

[0089] Step 105: Classify the impedance density map based on a pre - set neural network model to determine the fault information of the fuel cell under test through the classification result.

[0090] ​​​In one embodiment of the present application, a ResNet–50 neural network is used to perform fault diagnosis on the impedance density image of the fuel cell to be tested. Among them, the training process of the ResNet–50 neural network is as follows: taking the impedance density map of the historical fuel cell as the input sample, and taking the fuel cell fault information corresponding to the input sample as the output sample, training the preset neural network model to obtain the ResNet–50 neural network.

[0091] Among them, Figure 18 This is a training progress diagram of ResNet-50 provided by an embodiment of the present application. Among them, Figure 18 Figure (a) is a schematic diagram of loss, Figure 18 Figure (b) is a schematic diagram of precision, Figure 18 Figure (c) is a comparison diagram of prediction and actual; as Figure 18 shown, select the ResNet-50 neural network, the training set is 70%, the validation set is 30%, and the loss curve is as Figure 18 shown in Figure (a), where the abscissa is the epoch and the ordinate is the loss value; the accuracy curve is as Figure 18 shown in Figure (b), where the abscissa is the epoch and the ordinate is the precision; the confusion matrix is as Figure 18 shown in Figure (c), where the abscissa is the predicted value and the ordinate is the actual value. It can be Figure 18 seen that the trained ResNet-50 neural network has a high precision and meets the actual use requirements.

[0092] Figure 19 This is a schematic structural diagram of a fuel cell fault diagnosis device provided by an embodiment of the present application. As Figure 19 shown, the fuel cell fault diagnosis device 200 includes: at least one processor 201; and a memory 202 communicatively connected to the at least one processor 201; wherein, the memory 202 stores instructions executable by the at least one processor 201, and the instructions are executed by the at least one processor 201 so that the at least one processor 201 can: perform cyclic period dimension increase processing on the two-dimensional impedance spectrum corresponding to the preset sample fuel cell impedance data to obtain a three-dimensional impedance field corresponding to the preset sample fuel cell impedance data; perform data fitting on the three-dimensional impedance field corresponding to the preset sample fuel cell impedance data through an equivalent circuit to generate an image impedance field, and train the image impedance field through an interleaved learning method to obtain a cross-learning impedance field expansion model; input the impedance spectrum of the fuel cell to be tested into the cross-learning impedance field expansion model, and generate a life cycle impedance field of the fuel cell to be tested through the cross-learning impedance field expansion model; perform coordinate plane projection processing on the life cycle impedance field of the fuel cell to be tested to obtain an impedance density image; classify the impedance density image based on the preset neural network model to determine the fault information of the fuel cell to be tested through the classification result.

[0093] A non-volatile computer storage medium provided by an embodiment of the present application stores computer-executable instructions, and the computer-executable instructions are configured to: perform a cyclic period dimension increase process on a two-dimensional impedance spectrum corresponding to preset sample fuel cell impedance data to obtain a three-dimensional impedance field corresponding to the preset sample fuel cell impedance data; perform data fitting on the three-dimensional impedance field corresponding to the preset sample fuel cell impedance data through an equivalent circuit to generate an image impedance field, and train the image impedance field through an interleaved learning method to obtain a cross-learning impedance field expansion model; input the impedance spectrum of the to-be-detected fuel cell obtained into the cross-learning impedance field expansion model, and generate a life cycle impedance field of the to-be-detected fuel cell through the cross-learning impedance field expansion model; perform coordinate plane projection processing on the life cycle impedance field of the to-be-detected fuel cell to obtain an impedance density map; classify the impedance density map based on a preset neural network model to determine the fault information of the to-be-detected fuel cell through the classification result.

[0094] The embodiments in the present application are all described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the embodiments of the device, equipment, and non-volatile computer storage medium, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiments.

[0095] The above description is only for the embodiments of the present application and is not intended to limit the present application. For those skilled in the art, various changes and modifications can be made to the embodiments of the present application. These modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A fuel cell fault diagnosis method, characterized in that: The method comprises: Performing a cyclic dimension increase process on the two-dimensional impedance spectrum corresponding to the preset sample fuel cell impedance data to obtain a three-dimensional impedance field corresponding to the preset sample fuel cell impedance data; Performing data fitting on the three-dimensional impedance field corresponding to the preset sample fuel cell impedance data through an equivalent circuit to generate a mirror impedance field, and training the mirror impedance field through an interleaved learning method to obtain a cross-learning impedance field expansion model; Inputting the acquired impedance spectrum of the fuel cell to be tested into the cross-learning impedance field extension model, and generating the life cycle impedance field of the fuel cell to be tested through the cross-learning impedance field extension model; Performing coordinate plane projection processing on the life cycle impedance field of the fuel cell to be tested to obtain an impedance density map; Classifying the impedance density map based on a preset neural network model to determine fault information of the fuel cell to be tested through the classification result; The step of performing a cyclic dimension increasing process on the two-dimensional impedance spectrum corresponding to the preset sample fuel cell impedance data to obtain a three-dimensional impedance field corresponding to the preset sample fuel cell impedance data specifically includes: generating a two-dimensional battery impedance spectrum based on the preset sample fuel cell impedance data; Based on the two-dimensional battery impedance spectrum, an impedance curve waterfall diagram corresponding to the impedance curves of different cycle periods is drawn to increase the cycle dimension of the two-dimensional battery impedance spectrum; Performing color gradient processing on the impedance curve waterfall chart; Performing surface fitting on the impedance curve waterfall chart after color gradient processing by using a cubic spline interpolation method, and smoothing the interpolation result by using a Gaussian filter to obtain a three-dimensional impedance field corresponding to the preset sample fuel cell impedance data; The method of performing data fitting on the three-dimensional impedance field corresponding to the preset sample fuel cell impedance data through an equivalent circuit to generate a mirror impedance field specifically includes: Based on the first cycle and the last cycle corresponding to the preset sample fuel cell impedance data, element parameter fitting is performed; wherein the element parameter includes at least one of a series resistance, a constant phase element and a polarization resistance; By means of a linear interpolation function, based on the component parameter values ​​corresponding to the first cycle and the last cycle respectively, the simulation circuit component parameters in the intermediate cycle are determined to generate complete simulation equivalent circuit component parameter values; Based on the parameter values ​​of the complete simulation equivalent circuit elements, the mirror impedance field is constructed; wherein the mirror impedance field is used to reflect the impedance change trend of the battery under different cycle periods; The process of performing coordinate plane projection processing on the life cycle impedance field of the fuel cell to be tested to obtain an impedance density map specifically includes: Projecting the life cycle impedance field of the fuel cell to be tested onto the xy, yz and xz planes to obtain a scatter plot; Optimizing the scatter plot into a contour plot, and performing gradient processing between the contour plots; The contour map is filled using the contourf function, and the filled contour map is smoothed by the pcolormesh function to obtain the impedance density map.

2. A fuel cell fault diagnosis method according to claim 1, characterized in that: Determining the simulation circuit component parameters in the intermediate cycle period based on the component parameter values ​​corresponding to the first cycle period and the last cycle period by using a linear interpolation function specifically includes: Function-based: Obtaining the slope between the component parameter value of the first cycle and the component parameter value of the last cycle; Based on the point-slope equation of a straight line, we get n Component parameter values ​​for a cycle: Based on the component parameters of the first cycle and the last cycle, the simulation equivalent circuit component parameter values ​​of any intermediate cycle are determined: in, is the component parameter value of the last cycle; is the component parameter value of the first cycle; is the parameter value of the simulation equivalent circuit element of any intermediate cycle; k is the slope; n is a cyclic sequence, where 1< n < m .

3. A fuel cell fault diagnosis method according to claim 1, characterized in that: The method of training the mirror impedance field by an interleaved learning method to obtain a cross-learning impedance field extension model specifically includes: Mirror impedance field data are obtained, analyzed and modeled by stepwise linear regression, and the obtained stepwise linear regression model is optimized by an interleaved learning method.

4. A fuel cell fault diagnosis method according to claim 3, characterized in that: The stepwise linear regression model is: ； in, Z is the impedance value; CPE T is the first element parameter in the equivalent circuit; CPE P is the second element parameter in the equivalent circuit; Rs is the parameter of the third element in the equivalent circuit; R is the fourth element parameter in the equivalent circuit; n is a cyclical sequence; is the preset coefficient; is the coefficient corresponding to the first element parameter; is the coefficient corresponding to the second element parameter; is the coefficient corresponding to the third element parameter; is the coefficient corresponding to the fourth element parameter; is the coefficient corresponding to the fifth element parameter; is the product of the last component parameter and the corresponding coefficient.

5. A fuel cell fault diagnosis method according to claim 4, characterized in that: The stepwise linear regression model obtained by the interleaved learning method is optimized, specifically including: Performing Gaussian noise processing on the impedance density image corresponding to the preset sample fuel cell impedance data to obtain an expanded data set; Dividing the expanded data set into a training set and a validation set according to a preset ratio; Training is performed based on the training set, the validation set, and a preset ResNet-50 neural network, and an interleaved learning method is used to optimize a stepwise linear regression model.

6. A fuel cell fault diagnosis device, characterized in that: The device comprises a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the device is triggered to execute the method according to any one of claims 1 to 5.

7. A non-volatile computer storage medium storing computer executable instructions, characterized in that: The computer executable instructions can execute the method according to any one of claims 1 to 5.

Images