Working condition optimization method and device of fuel cell and terminal equipment

Abstract

The invention provides a working condition optimization method and device of a fuel cell and terminal equipment. The working condition optimization method of the fuel cell comprises the following steps: determining a fuel cell simulation model and a neural network model according to different application scenes of the fuel cell; training the neural network model based on simulation input and output of the fuel cell simulation model to obtain a fuel cell working condition optimization model; and based on the application scene of the fuel cell, through the fuel cell working condition optimization model, optimizing the working condition of the fuel cell. The working condition optimization method of the fuel cell can be suitable for various application scenes of the fuel cell, and the working condition optimization efficiency and the optimization prediction result are high in accuracy.

Description

Technical Field

[0001] The present invention relates to the field of fuel cells, and particularly to a method for optimizing the operating conditions of a fuel cell. The present invention also relates to an apparatus for optimizing the operating conditions of a fuel cell based on the above method, and a terminal device capable of implementing the above method. Background Art

[0002] The operating condition control of a fuel cell is an important means to control the fuel cell to operate efficiently and stably in a target state under specific scenarios. The operating conditions of a fuel cell include operating conditions such as fuel flow rate, pressure, temperature, humidity, etc. Under different operating conditions, the performance of the fuel cell will also change. Therefore, the optimization of the battery operating conditions generally refers to taking the performance of the fuel cell, that is, the voltage, as the optimization target, and at the same time controlling the operating conditions of the fuel cell to reach the optimal value, thereby improving the fuel efficiency of the fuel cell.

[0003] However, in the prior art, there is no method for optimizing specific operating conditions according to the specific scenarios of fuel cell operation. Currently, the optimization is all carried out for the voltage of the fuel cell, and in the optimization process, the simulation of the sensitivity of the operating conditions is completed through a fuel cell simulation model, and then the optimal operating conditions are selected. In this way of optimizing the operating conditions, since the fuel cell simulation model is established based on the physical model of the fuel cell, it is quite complex to describe multiple physical processes inside the fuel cell and their interactions. This results in that during the simulation, if the speed is fast, the accuracy is poor, and if the accuracy is high, the time consumption is long, and it is impossible to balance the efficiency and accuracy of the operating condition optimization. Summary of the Invention

[0004] In view of this, the present invention aims to propose a method for optimizing the operating conditions of a fuel cell to optimize the operating conditions of fuel cells in various application scenarios, and balance the efficiency and accuracy of the operating condition optimization.

[0005] To achieve the above object, the present invention is implemented by the following technical solutions:

[0006] A method for optimizing the operating conditions of a fuel cell, comprising:

[0007] Determine a fuel cell simulation model and a neural network model according to different application scenarios of the fuel cell;

[0008] Train the neural network model based on the simulation input and output of the fuel cell simulation model to obtain a fuel cell operating condition optimization model;

[0009] Optimize the operating conditions of the fuel cell through the fuel cell operating condition optimization model based on the application scenario of the fuel cell.

[0010] Further, determining the fuel cell simulation model according to different application scenarios of the fuel cell includes:

[0011] Determining the input parameters and output parameters for optimizing the fuel cell operating conditions according to different application scenarios of the fuel cell;

[0012] Constructing a fuel cell simulation model based on the input parameters and the output parameters;

[0013] Calibrating the fuel cell simulation model based on the actual test results of each parameter.

[0014] Further, training the neural network model based on the simulation input and output of the fuel cell simulation model to obtain a fuel cell operating condition optimization model, including:

[0015] Obtaining a first simulation input parameter value and a corresponding first simulation output parameter value of the fuel cell simulation model, as well as a second simulation input parameter value and a corresponding second simulation output parameter value;

[0016] Taking the first data set composed of the first simulation input parameter value and the first simulation output parameter value as the training set of the neural network model;

[0017] Taking the second data set composed of the second simulation input parameter value and the second simulation output parameter value as the validation set of the neural network model;

[0018] Training the neural network model based on the training set;

[0019] Validating the trained neural network model based on the validation set;

[0020] When the prediction accuracy of the neural network model meets the preset value, taking the current neural network model as the fuel cell operating condition optimization model.

[0021] Further, obtaining a first simulation input parameter value and a corresponding first simulation output parameter value of the fuel cell simulation model, as well as a second simulation input parameter value and a corresponding second simulation output parameter value, includes:

[0022] Determining the input parameters and output parameters for optimizing the fuel cell operating conditions according to different application scenarios of the fuel cell;

[0023] Taking points at a first sparsity level among the respective input parameters;

[0024] Based on the points taken at the first sparsity level, determining the first simulation input parameter value through the Taguchi method;

[0025] Input the first simulation input parameter value into the fuel cell simulation model to obtain the first simulation output parameter value corresponding to the first simulation input parameter value;

[0026] Among all the input parameters, sample points at the second sparsity level are taken respectively;

[0027] Based on the sample points at the second sparsity level, determine the second simulation input parameter value through the Taguchi algorithm;

[0028] Input the second simulation input parameter value into the fuel cell simulation model to obtain the first simulation output parameter value corresponding to the second simulation input parameter value; where the levels of the sample points at the first sparsity level are different from those of the sample points at the second sparsity level.

[0029] Further, the training of the neural network model based on the training set includes:

[0030] Divide the training set into N mutually exclusive training subsets evenly;

[0031] Denote the nth training subset among the N training subsets as the nth validation fold, and denote the training subsets other than the nth training subset among the N training subsets as the nth training fold; where, n ∈ [1, N] and is an integer;

[0032] Train the neural network model based on the nth training fold to obtain the nth sub - neural network model, and verify the nth sub - neural network model based on the nth validation fold to obtain the prediction accuracy of the nth sub - neural network model;

[0033] Select the sub - neural network model with the highest prediction accuracy among the first sub - neural network model to the Nth sub - neural network model as the trained neural network model.

[0034] Further, the optimization of the fuel cell operating conditions through the fuel cell operating condition optimization model based on the application scenario of the fuel cell includes:

[0035] Based on the application scenario of the fuel cell, determine the optimization input parameters and optimization output parameters;

[0036] Based on the optimization output parameters, determine the optimization objective;

[0037] Based on each optimization input parameter, select the optimization input parameter value;

[0038] Input the optimization input parameter value into the fuel cell operating condition optimization model to obtain the optimization output parameter value;

[0039] Calculate the optimized target value based on the optimized output parameter values; where when the optimized output parameters include multiple parameters, weight distribution is performed on each parameter to obtain the optimized target value.

[0040] Select the optimal operating condition of the fuel cell according to the optimized target value.

[0041] Further, determining the optimized input parameters and optimized output parameters based on the application scenario of the fuel cell includes:

[0042] When the application scenario of the fuel cell is the optimization of the performance operating condition of the fuel cell, the optimized input parameters are the stoichiometric ratio, pressure, and temperature, and the optimized output parameter is the average voltage;

[0043] When the application scenario of the fuel cell is the optimization of the cold start operating condition of the fuel cell, the optimized input parameters are the ambient temperature, the first-stage loading rate, the first-stage target current density, the second-stage loading rate, the second-stage target current density, and the condition of the water pump, and the optimized output parameters are the average voltage, the lowest voltage, the start-up success time, and the amount of ice formation;

[0044] When the application scenario of the fuel cell is the optimization of the durability operating condition of the fuel cell, the optimized input parameters are the variable load rate, the number of start-up and shutdown times, pressure, humidity, and temperature, and the optimized output parameters are the total start-up duration, the total output power, the average voltage decay rate, and the hydrogen consumption.

[0045] Further, selecting the optimized input parameter values based on each optimized input parameter includes:

[0046] In each of the optimized input parameters, sampling points are taken at a dense level;

[0047] Based on the sampling points, the optimized input parameter values are obtained in a fully orthogonal manner.

[0048] The present invention also provides a device for optimizing the operating condition of a fuel cell, including:

[0049] A first model generation module, configured to determine a fuel cell simulation model and a neural network model according to different application scenarios of the fuel cell;

[0050] A second model generation module, configured to train the neural network model based on the simulation input and output of the fuel cell simulation model to obtain a fuel cell operating condition optimization model;

[0051] An operating condition optimization module, configured to optimize the operating condition of the fuel cell through the fuel cell operating condition optimization model based on the application scenario of the fuel cell.

[0052] In addition, the present invention also provides a terminal device, including: a processor and a memory, wherein a computer program that can run on the processor is stored in the memory, and when the processor executes the computer program, the working condition optimization method of the fuel cell as described above is implemented.

[0053] Compared with the prior art, the present invention has the following advantages:

[0054] For the working condition optimization method of the fuel cell according to the present invention, the neural network model is trained based on the input and output of the fuel cell simulation model to obtain a fuel cell working condition optimization model, thereby achieving the effect of using the neural network model to replace the electrochemical model. Using this fuel cell working condition optimization model to optimize the working conditions of the fuel cell can take into account both the efficiency and accuracy of fuel cell working condition optimization. At the same time, the fuel cell simulation model and its input and output parameters are determined based on different application scenarios of the fuel cell. Therefore, the finally obtained fuel cell working condition optimization model can be applied to various application scenarios of fuel cells.

[0055] In addition, the fuel cell is actually tested using multiple parameters characterizing each working condition, and the fuel cell simulation model is calibrated according to the test results, which can ensure the simulation accuracy of the fuel cell simulation model in each working condition direction. And when obtaining the first simulation input parameter value, the method of taking points at the sparse level in each input parameter combined with the Taguchi algorithm can reduce the number of calculation examples while reflecting the orthogonality between each working condition parameter and its influence on the result.

[0056] In addition, by adopting a neural network training method in which the training set is evenly divided into N mutually exclusive training subsets, the stability and robustness of the trained neural network model can be improved. And according to the value of the numerical value N, overfitting of the training set data can be effectively avoided, thereby further improving the prediction accuracy of the obtained fuel cell working condition optimization model. And when selecting the optimized input parameter value, taking points at the dense level in each optimized input parameter characterizing each working condition factor and adopting a fully orthogonal method can more accurately reflect the influence relationship of each working condition factor on the result. Description of the Drawings

[0057] The drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0058] Figure 1 It is a schematic flowchart of the working condition optimization method of the fuel cell provided by the embodiment of the present invention;

[0059] Figure 2 It is a schematic flowchart of determining the fuel cell simulation model provided by the embodiment of the present invention;

[0060] Figure 3 Schematic flow chart for obtaining the fuel cell operating condition optimization model provided by an embodiment of the present invention;

[0061] Figure 4 Schematic diagram of the machine learning logic of the neural network provided by an embodiment of the present invention;

[0062] Figure 5 Schematic flow chart for obtaining simulation parameters provided by an embodiment of the present invention;

[0063] Figure 6 Schematic diagram of the logic of the Taguchi algorithm provided by an embodiment of the present invention;

[0064] Figure 7 Schematic flow chart for training the neural network model provided by an embodiment of the present invention;

[0065] Figure 8 Schematic diagram of the logic for training the neural network model provided by an embodiment of the present invention;

[0066] Figure 9 Schematic flow chart for optimizing the operating conditions of the fuel cell provided by an embodiment of the present invention;

[0067] Figure 10 Schematic flow chart for determining the optimization parameters provided by an embodiment of the present invention;

[0068] Figure 11 Schematic flow chart for selecting the optimized input parameter values provided by an embodiment of the present invention;

[0069] Figure 12 Schematic diagram of the prediction result and the test result provided by an embodiment of the present invention;

[0070] Figure 13 Schematic diagram of the structure of the fuel cell operating condition optimization device provided by an embodiment of the present invention;

[0071] Figure 14 Schematic diagram of the structure of the terminal device provided by an embodiment of the present invention.

[0072] Description of reference numerals:

[0073] 501, First model generation module; 502, Second model generation module; 503, Operating condition optimization module; 600, Terminal device; 610, Processor; 620, Memory; 621, Computer program. Detailed implementation manners

[0074] In the following description, specific details such as specific system architectures and technologies are presented for the purpose of illustration rather than limitation, in order to provide a thorough understanding of the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from obscuring the description of the present application.

[0075] It should be understood that when used in the specification and appended claims of the present application, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0076] It should also be understood that the term "and / or" used in the specification and appended claims of the present application refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0077] As used in the specification and appended claims of the present application, the term "if" can be interpreted as "when" or "once" or "in response to determining" or "in response to detecting" depending on the context. Similarly, the phrase "if determined" or "if [the described condition or event] is detected" can be interpreted as meaning "once determined" or "in response to determining" or "once [the described condition or event] is detected" or "in response to detecting [the described condition or event]" depending on the context.

[0078] In addition, in the description of the specification and appended claims of the present application, the terms "first", "second", "third", etc. are only used for differentiating descriptions and cannot be understood as indicating or implying relative importance.

[0079] The reference to "one embodiment" or "some embodiments" etc. described in the specification of the present application means that a specific feature, structure, or characteristic described in connection with that embodiment is included in one or more embodiments of the present application. Thus, statements such as "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc. that appear in different places in this specification do not necessarily all refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in another way. The terms "comprising", "including", "having", and their variants all mean "including but not limited to", unless otherwise specifically emphasized in another way.

[0080] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0081] Embodiment 1

[0082] This embodiment relates to a method for optimizing the operating conditions of a fuel cell. As Figure 1 shown in

[0083] Step S100: Determine a fuel cell simulation model and a neural network model according to different application scenarios of the fuel cell.

[0084] In this step S100, the application scenarios of the fuel cell are expanded, not limited to taking the fuel cell performance as the optimization goal in the prior art. For example, it can also be the self - humidifying operating condition optimization scenario, plateau operating condition optimization scenario, cold start operating condition optimization scenario, durability operating condition optimization scenario, etc. of the fuel cell. Thus, a suitable fuel cell simulation model is matched based on the application scenario of the fuel cell, and the operating condition optimization factors are determined, that is, the input parameters and output parameters for simulation.

[0085] It should be noted that the neural network model in this embodiment can be selected according to different application scenarios of the fuel cell and the hardware device used for calculating the operating condition optimization. For example, ANN (Artificial Neural Network) is used as the neural network model in each application scenario for subsequent training.

[0086] Step S200: Train the neural network model based on the simulation input and output of the fuel cell simulation model to obtain a fuel cell operating condition optimization model.

[0087] In this step S200, after obtaining the fuel cell simulation model and the neural network model to be trained, the neural network model can be trained according to the simulation input and output of the fuel cell simulation model. After training, this neural network model is used as the fuel cell operating condition optimization model.

[0088] Step S300: Optimize the operating conditions of the fuel cell based on the application scenario of the fuel cell through the fuel cell operating condition optimization model.

[0089] In this step S300, use this fuel cell operating condition optimization model to replace the simulation calculation process of the fuel cell simulation model in the application scenario to optimize the operating conditions of the fuel cell. Taking the ANN model as an example, this model is a 0D model, and its calculation speed is far faster than that of the fuel cell simulation model. Using the ANN model as a replacement model for the fuel cell simulation model can increase the calculation efficiency by 10,000 times.

[0090] Combined with the above description, the method for optimizing the operating conditions of the fuel cell provided in this embodiment can be applied to various application scenarios of the fuel cell, and can balance the efficiency and accuracy of fuel cell operating condition optimization.

[0091] In a preferred embodiment, as Figure 2 shown, according to different application scenarios of the fuel cell, determining a fuel cell simulation model includes:

[0092] Step S110, according to different application scenarios of the fuel cell, determining input parameters and output parameters for fuel cell operating condition optimization.

[0093] In step S110, the application scenario of the fuel cell can be performance operating condition optimization of the fuel cell, cold start operating condition optimization of the fuel cell, durability operating condition optimization of the fuel cell, etc. Specifically, for performance operating condition optimization of the fuel cell, the input parameters for fuel cell operating condition optimization can be stoichiometric ratio, pressure, and temperature, and the optimized output parameter can be the average voltage; for cold start operating condition optimization of the fuel cell, the input parameters for fuel cell operating condition optimization can be ambient temperature, first-stage loading rate, first-stage target current density, second-stage loading rate, second-stage target current density, and coolant pump condition, and the optimized output parameters can be the average voltage, minimum voltage, start success time, and ice formation amount; for durability operating condition optimization of the fuel cell, the input parameters for fuel cell operating condition optimization can be load change rate, number of start-stop cycles, pressure, humidity, and temperature, and the optimized output parameters can be total start duration, total output power, average voltage decay rate, and hydrogen consumption.

[0094] It should be noted that the above parameter selection depends on the corresponding calibration and prediction requirements. Based on the same principle, it can also be extended to other application scenarios of the fuel cell.

[0095] Step S120, constructing a fuel cell simulation model based on the input parameters and the output parameters.

[0096] Step S130, calibrating the fuel cell simulation model based on the actual test results of each parameter.

[0097] In this embodiment, the fuel cell simulation model is calibrated according to the test results under the determined application scenario. Generally speaking, in the actual use scenario, the calibration accuracy of the fuel cell simulation model calibrated according to the test results needs to be >90% in each operating condition direction.

[0098] In a preferred embodiment, as Figure 3 shown, training the neural network model based on the simulation input and output of the fuel cell simulation model to obtain a fuel cell operating condition optimization model includes:

[0099] Step S210, obtaining a first simulation input parameter value and a corresponding first simulation output parameter value of the fuel cell simulation model, and a second simulation input parameter value and a corresponding second simulation output parameter value.

[0100] Step S220: Use the first data set composed of the first simulation input parameter values and the first simulation output parameter values as the training set of the neural network model.

[0101] Step S230: Use the second data set composed of the second simulation input parameter values and the second simulation output parameter values as the validation set of the neural network model.

[0102] Step S240: Train the neural network model based on the training set.

[0103] Step S250: Validate the trained neural network model based on the validation set.

[0104] Step S260: When the prediction accuracy of the neural network model meets the preset value, use the current neural network model as the fuel cell operating condition optimization model.

[0105] In this embodiment, the above neural network model can be an ANN model, as Figure 4 shown. Figure 4 The Trainingdata and Validation data in can be the simulation data of two different fuel cell simulation models, and the PredictionInput is the parameter input for prediction. It can be understood that the combination of one input and one output constitutes one data, and multiple data constitute a data set. Specifically, taking the application scenario of optimizing the performance operating conditions of a fuel cell as an example, the modeling process of the ANN model is explained.

[0106] Training data, that is, the training set, consists of the first data set composed of the first simulation input parameter values and the first simulation output parameter values, that is, the data collection formed by multiple simulation results of the fuel cell simulation model. Among them, Input represents the first simulation input parameter values, and Output represents the first simulation output parameter values.

[0107] Validation data, that is, the validation set, consists of the first data set composed of the second simulation input parameter values and the second simulation output parameter values, that is, the data collection formed by multiple simulation results of the fuel cell simulation model. Among them, Input represents the second simulation input parameter values, and Output represents the second simulation output parameter values. It should be noted that due to the differences between the first simulation input parameter values and the second simulation input parameter values, the data in the validation set and the data in the training set are also different data.

[0108] Specifically, Input is composed of the stoichiometric ratio, pressure, and temperature of each simulation, and Output is composed of the average voltage of each simulation.

[0109] The parameters of the Prediction Input are the same as those of the Input of the Training data and the Validation data, which are stoichiometric ratio, pressure, and temperature. The input at this time is the operating condition of the fuel cell that the ANN model wants to predict. After being calculated by the ANN model, the prediction result Predicted Output, that is, the average voltage, is obtained.

[0110] The Training data is used to train the ANN model. When the model training is completed, the Input of the Validation data is input into the model for calculation, and the calculation result is compared with the Output of the Validation data to confirm the prediction accuracy of the model. When the accuracy of the ANN model meets the preset value, it can be considered that the training of the ANN model is completed, and the trained ANN model is used as the fuel cell operating condition optimization model. It should be noted that if the accuracy of the ANN model does not meet the preset value, the training set data or the ANN training model is debugged until the accuracy of the ANN model meets the preset value. It can be understood that the preset value is set according to actual requirements.

[0111] In this embodiment, a 0D neural network machine learning (ANN) model is used to replace the complex fuel cell simulation model, which greatly shortens the calculation time and improves the calculation efficiency.

[0112] It should be noted that in a preferred implementation form, as Figure 5 shown, the above-mentioned obtaining the first simulation input parameter values and the corresponding first simulation output parameter values, as well as the second simulation input parameter values and the corresponding second simulation output parameter values of the fuel cell simulation model, includes:

[0113] Step S211, according to different application scenarios of the fuel cell, determine the input parameters and output parameters for optimizing the fuel cell operating condition.

[0114] Step S212, in each of the input parameters, perform sampling at the first sparsity level.

[0115] Step S213, based on the sampling at the first sparsity level, determine the first simulation input parameter values through the Taguchi method.

[0116] Step S214, input the first simulation input parameter values into the fuel cell simulation model to obtain the first simulation output parameter values corresponding to the first simulation input parameter values.

[0117] Step S215, in each of the input parameters, perform sampling at the second sparsity level.

[0118] Step S216: Based on the points taken at the second sparsity level, determine the second set of simulation input parameter values through the Taguchi algorithm.

[0119] Step S217: Input the second set of simulation input parameter values into the fuel cell simulation model to obtain the first set of simulation output parameter values corresponding to the second set of simulation input parameter values; where the levels of the points taken at the first sparsity level are different from those of the points taken at the second sparsity level.

[0120] In this embodiment, Step S211 corresponds to Step S110 described above and will not be elaborated here. In Step S212, points are taken at the sparsity level for each operating condition factor. Through the Taguchi algorithm in Step S213, based on each operating condition factor and the levels of the points taken, obtain the first set of simulation input parameter values and formulate the simulation plan for the fuel cell simulation model. The simulation plan obtained using the Taguchi algorithm can, while reducing the number of calculation examples, well reflect the orthogonality among the operating condition factors and their influence on the results. It can be understood that the various input parameters described above represent the respective operating condition factors. Similarly, in Step S215, change the levels of the points taken to obtain the second set of simulation input parameter values.

[0121] Combined Figure 6 A specific description of the use of the Taguchi algorithm is as follows. First, based on the determined factor levels, the Taguchi calculation table for generating the experimental plan can be generated, as shown in the table in Figure 6 Taking the optimization of the performance operating conditions of a fuel cell as an example of the application scenario, input the input parameters: stoichiometric ratio, pressure, and temperature and their levels according to the table in the upper right of Figure 4 to obtain Table 1 below. There are a total of three factors, each with three levels. Output according to the table in the lower right of Figure 3 to obtain Table 2, getting a total of 9 test plans. The test plan formulated by the full orthogonal method with three levels for each of the three factors is as follows in Table 3, with a total of 27. It can be understood that these 27 sets of numerical values constitute the first set of simulation input parameter values, representing 27 test plans. In this embodiment, the Taguchi algorithm is used, which can effectively reduce the number of tests compared to the traditional orthogonal method while ensuring orthogonality, thereby improving the prediction accuracy of the model when training the neural network model.

[0122] Table 1

[0123]

[0124] Table 2

[0125]

[0126] Table 3

[0127]

[0128]

[0129] It should be noted that to obtain the second simulation input parameter value, only the level corresponding to each factor needs to be changed, and the remaining steps are the same as those for obtaining the first simulation input parameter value. It can be understood that the method in this embodiment is also applicable to obtaining the first simulation input parameter value of a fuel cell in other application scenarios.

[0130] In a preferred implementation form, in step S130, when actually testing the fuel cell based on each input parameter, the numerical value of the selected input parameter can also be obtained by the method in the above steps S212 - S213, so as to further improve the calibration accuracy of the fuel cell simulation model.

[0131] In a preferred implementation form, as Figure 7 shown, the training of the neural network model based on the training set includes:

[0132] Step S241, evenly divide the training set into N mutually exclusive training subsets.

[0133] Step S242, denote the nth training subset among the N training subsets as the nth validation fold, and denote the training subsets other than the nth training subset among the N training subsets as the nth training fold; where n ∈ [1, N] and is an integer.

[0134] Step S243, train the neural network model based on the nth training fold to obtain the nth sub - neural network model, and verify the nth sub - neural network model based on the nth validation fold to obtain the prediction accuracy of the nth sub - neural network model.

[0135] Step S244, select the sub - neural network model with the highest prediction accuracy among the first sub - neural network model to the Nth sub - neural network model as the trained neural network model.

[0136] In this embodiment, when using the training set to train the neural network model, the data set is randomly divided into N mutually exclusive subsets of the same size, that is, each time randomly select N - 1 portions as the training fold, and the remaining 1 portion is used as the test fold. After this round is completed, randomly select N - 1 portions again as the training fold. After several rounds (less than N), select the model with the best evaluation of the loss function as the trained neural network model.

[0137] Specifically, taking the performance condition optimization of a fuel cell as an application scenario as an example, combined with Figure 8, the process of training the neural network model is explained. Assume that N = 5, and there are 50 data in the training set, that is, the 50 data are divided into five training subsets, each training subset contains 10 data, where each data represents a set of input (stoichiometric ratio, pressure, temperature) and an output (average voltage). First, the second to fifth data subsets are used as training folds to train the model. After the training is completed, the first data subset is used as a validation fold to verify the trained model, and the first neural network model and its corresponding first model performance are obtained. 1 ), which is the prediction accuracy of the model.

[0138] The second to fifth data subsets are used as validation folds, and the remaining data subsets are used as training subsets. The second to fifth neural network models are obtained based on the same steps, and their corresponding second to fifth model performances (Performance 2 ~Performance 5 ). The average Performance of the five performances reflects whether the training set data meets the training requirements, such as quantity, accuracy, etc. By comparing the five performances, the sub-neural network model with the best prediction accuracy is selected as the trained neural network model.

[0139] The training method in this embodiment can improve the stability and robustness of the trained neural network model, and, depending on the value of the numerical value N, can effectively avoid overfitting of the training set data, thereby further improving the accuracy of the obtained fuel cell operating condition optimization model prediction.

[0140] In a preferred embodiment, Figure 9 As shown, the above-mentioned fuel cell operating condition optimization model based on the application scenario of the fuel cell is used to optimize the operating condition of the fuel cell, including:

[0141] Step S310: determining an optimization target based on the optimization output parameters.

[0142] Step S320: selecting an optimized input parameter value based on each optimized input parameter.

[0143] Step S330, inputting the optimized input parameter value into the fuel cell operating condition optimization model to obtain the optimized output parameter value.

[0144] Step S340, calculating the optimization target value based on the optimization output parameter value; wherein, when the optimization output parameter includes multiple parameters, weights are assigned to each parameter to obtain the optimization target value.

[0145] Step S350: Select the optimal operating condition of the fuel cell according to the optimized target value.

[0146] In this embodiment, the requirements in the application scenario are determined and converted into the output parameters of the fuel cell operating condition optimization model. The secondary parameters are removed, and weight distribution is carried out for the main parameters. After weighted summation, they are combined into an optimization target. According to the optimization target, the prediction results of the fuel cell operating condition optimization model are optimized and analyzed to delimit the range of optimized operating conditions. It should be noted that multiple optimization targets can be specified simultaneously. For example, by adjusting different parameters or adjusting different weight coefficients, after comparing their respective optimization results, the final optimization result is selected.

[0147] Specifically, for the performance operating condition optimization of the fuel cell, if the operating condition optimization targets the average voltage, then the predicted output value, the average voltage, is the optimization target. For the cold start operating condition optimization of the fuel cell, if the cold start needs to consider the start time and the amount of ice formation, then the predicted output values, the start time and the amount of ice formation, jointly form the optimization target. For example, if it is considered that the amount of ice formation is more important, a larger weight coefficient is assigned to it, and the optimization target = 0.2 * start time + 0.8 * amount of ice formation. For the durability operating condition optimization of the fuel cell, if the durability needs to consider the total output power, the average voltage decay rate, and the hydrogen consumption, then the predicted output values. The total output power, the average voltage decay rate, and the hydrogen consumption jointly form the optimization target. For example, if it is considered that the total output power and the decay rate are primarily considered, and the decay rate is more important, and the hydrogen consumption is considered as an auxiliary, then the optimization coefficient = 0.8 * (0.7 * decay rate + 0.3 * total output power) + 0.2 * hydrogen consumption. It can be understood that for the above various scenarios, due to different requirements, the considered parameters and optimization targets are also different.

[0148] As an implementable embodiment, as Figure 10 shown, based on the application scenario of the fuel cell, determining the optimized input parameters and the optimized output parameters includes:

[0149] Step S311: When the application scenario of the fuel cell is the performance operating condition optimization of the fuel cell, the optimized input parameters are the stoichiometric ratio, the pressure, and the temperature, and the optimized output parameter is the average voltage.

[0150] Step S312: When the application scenario of the fuel cell is the cold start operating condition optimization of the fuel cell, the optimized input parameters are the ambient temperature, the first-stage loading rate, the first-stage target current density, the second-stage loading rate, the second-stage target current density, and the condition of the transfer water pump, and the optimized output parameters are the average voltage, the lowest voltage, the start success time, and the amount of ice formation.

[0151] Step S313: When the application scenario of the fuel cell is the durability condition optimization of the fuel cell, the optimization input parameters are the variable load rate, the number of start-stop cycles, the pressure, the humidity, and the temperature, and the optimization output parameters are the total start-up duration, the total output power, the average voltage decay rate, and the hydrogen consumption.

[0152] It should be noted that the magnitudes of the serial numbers of the steps in this embodiment do not mean the order of execution.

[0153] As an implementable embodiment, as Figure 11 shown, the selection of the optimization input parameter values based on the respective optimization input parameters includes:

[0154] Step S331: In each of the optimization input parameters, perform sampling at a dense level.

[0155] Step S332: Based on the sampling, obtain the optimization input parameter values through a full orthogonal method.

[0156] In this embodiment, sampling at a dense level is performed on each working condition factor, and a full orthogonal method is adopted. According to the working condition conditions and the sampling, a prediction scheme for the fuel cell working condition optimization model is formulated, which can more accurately reflect the influence relationship of the working conditions on the results and further improve the accuracy of the prediction.

[0157] Based on the above overall introduction and the comparison of the prediction results and the test results as Figure 12 shown, the optimization effect of the method and process provided in this embodiment in a certain application scenario is > 70%, and the prediction error of the fuel cell working condition optimization model is < ±5%, which has good practicability.

[0158] The fuel cell working condition optimization method described in the embodiment of the present application trains the neural network model based on the input and output of the fuel cell simulation model to obtain a fuel cell working condition optimization model, thereby achieving the effect of using the neural network model to replace the electrochemical model. Using this fuel cell working condition optimization model to optimize the working conditions of the fuel cell can take into account both the efficiency and accuracy of the fuel cell working condition optimization. At the same time, the fuel cell simulation model and its input and output parameters are determined based on different application scenarios of the fuel cell, so the finally obtained fuel cell working condition optimization model can be applied to various application scenarios of the fuel cell.

[0159] It should be understood that the magnitudes of the serial numbers of the steps in the above embodiments do not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0160] Embodiment 2

[0161] This embodiment relates to a device for optimizing the operating conditions of a fuel cell, which corresponds to the method for optimizing the operating conditions of the fuel cell described in Embodiment 1. Among them, Figure 13 The structural block diagram of the device for optimizing the operating conditions of the fuel cell in this embodiment is shown. For the convenience of description, Figure 13 only the parts related to the embodiment are shown.

[0162] Still referring to Figure 13 , in terms of the overall composition, the device for optimizing the operating conditions of the fuel cell in this embodiment includes a first model generation module 501, a second model generation module 502, and an operating condition optimization module 503.

[0163] Among them, the above-mentioned first model generation module 501 is used to determine a fuel cell simulation model and a neural network model according to different application scenarios of the fuel cell. The second model generation module 502 is used to train the neural network model based on the simulation input and output of the fuel cell simulation model to obtain a fuel cell operating condition optimization model. The operating condition optimization module 503 is used to optimize the operating conditions of the fuel cell through the fuel cell operating condition optimization model based on the application scenario of the fuel cell.

[0164] In addition, the training of the neural network model based on the simulation input and output of the fuel cell simulation model to obtain a fuel cell operating condition optimization model includes: obtaining a first simulation input parameter value and a corresponding first simulation output parameter value, and a second simulation input parameter value and a corresponding second simulation output parameter value of the fuel cell simulation model; using the first data set composed of the first simulation input parameter value and the first simulation output parameter value as the training set of the neural network model; using the second data set composed of the second simulation input parameter value and the second simulation output parameter value as the verification set of the neural network model; training the neural network model based on the training set; verifying the trained neural network model based on the verification set; when the prediction accuracy of the neural network model meets the preset value, using the current neural network model as the fuel cell operating condition optimization model.

[0165] At this time, in an implementable manner, obtaining the first simulation input parameter values and the corresponding first simulation output parameter values, as well as the second simulation input parameter values and the corresponding second simulation output parameter values of the fuel cell simulation model includes: determining the input parameters and output parameters for optimizing the fuel cell operating conditions according to different application scenarios of the fuel cell; taking points at the first sparsity level for each of the input parameters; determining the first simulation input parameter values through the Taguchi method based on the points taken at the first sparsity level; inputting the first simulation input parameter values into the fuel cell simulation model to obtain the first simulation output parameter values corresponding to the first simulation input parameter values; taking points at the second sparsity level for each of the input parameters; determining the second simulation input parameter values through the Taguchi method based on the points taken at the second sparsity level; inputting the second simulation input parameter values into the fuel cell simulation model to obtain the first simulation output parameter values corresponding to the second simulation input parameter values; where the levels of the points taken at the first sparsity level are different from the levels of the points taken at the second sparsity level.

[0166] Meanwhile, in a preferred implementation form, training the neural network model based on the training set includes: evenly dividing the training set into N mutually exclusive training subsets; denoting the nth training subset among the N training subsets as the nth validation fold, and denoting the training subsets other than the nth training subset among the N training subsets as the nth training fold; where n ∈ [1, N] and is an integer; training the neural network model based on the nth training fold to obtain the nth sub-neural network model, and validating the nth sub-neural network model based on the nth validation fold to obtain the prediction accuracy of the nth sub-neural network model; selecting the sub-neural network model with the highest prediction accuracy among the first sub-neural network model to the Nth sub-neural network model as the trained neural network model.

[0167] In addition, in a preferred implementation form, optimizing the operating conditions of the fuel cell through the fuel cell operating condition optimization model based on the application scenario of the fuel cell includes: determining the optimization input parameters and optimization output parameters based on the application scenario of the fuel cell; determining the optimization objective based on the optimization output parameters; selecting optimization input parameter values for each of the optimization input parameters; inputting the optimization input parameter values into the fuel cell operating condition optimization model to obtain optimization output parameter values; calculating the optimization objective value based on the optimization output parameter values; where when the optimization output parameters include multiple parameters, weight distribution is performed on each parameter to obtain the optimization objective value. Selecting the optimal operating condition of the fuel cell according to the optimization objective value.

[0168] In a preferred embodiment, when actually testing a fuel cell based on various input parameters, the values of the selected input parameters can also be obtained through the following steps: for each of the input parameters, points are taken at the sparse level, and based on the points taken at the sparse level, the input parameter values are determined by the Taguchi algorithm.

[0169] In this embodiment, in a preferred implementation form, determining the optimized input parameters and the optimized output parameters based on the application scenario of the fuel cell includes: when the application scenario of the fuel cell is the optimization of the performance working condition of the fuel cell, the optimized input parameters are the stoichiometric ratio, pressure, and temperature, and the optimized output parameter is the average voltage; when the application scenario of the fuel cell is the optimization of the cold start working condition of the fuel cell, the optimized input parameters are the ambient temperature, the first-stage loading rate, the first-stage target current density, the second-stage loading rate, the second-stage target current density, and the coolant pump condition, and the optimized output parameters are the average voltage, the minimum voltage, the start-up success time, and the amount of ice formation; when the application scenario of the fuel cell is the optimization of the durability working condition of the fuel cell, the optimized input parameters are the load change rate, the number of start-up and shutdown times, pressure, humidity, and temperature, and the optimized output parameters are the total start-up duration, the total output power, the average voltage decay rate, and the hydrogen consumption.

[0170] In a preferred embodiment, selecting the optimized input parameter values based on each of the optimized input parameters includes: for each of the optimized input parameters, points are taken at the dense level; based on the points taken, the optimized input parameter values are obtained in a fully orthogonal manner.

[0171] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units, due to being based on the same concept as the method embodiment of the present application, for their specific functions and the technical effects brought, reference can be specifically made to the method embodiment part, and details are not described herein again.

[0172] It should be noted that those skilled in the art can clearly understand that, for the convenience and conciseness of description, only the above division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules as needed, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working process of the units and modules in the above system can refer to the corresponding process in the foregoing method embodiment and will not be elaborated here.

[0173] Embodiment 3

[0174] This embodiment relates to a terminal device. Refer to Figure 14 , the terminal device 600 may include: at least one processor 610 and a memory 620. The memory 620 is used to store a computer program 621, and the processor 610 is used to call and run the computer program 621 stored in the memory 620 to implement the steps in any method embodiment in Embodiment 1. Specifically, for example, it may be Figure 1 steps S100 to S300 in the embodiment shown, or when the processor 610 executes the computer program, it can implement the functions of each module / unit in the above device embodiments, for example Figure 13 the functions of each module shown.

[0175] Exemplarily, the computer program 621 can be divided into one or more modules / units. One or more modules / units are stored in the memory 620 and executed by the processor 610 to complete this application. The one or more modules / units can be a series of computer program segments capable of completing specific functions, and these program segments are used to describe the execution process of the computer program in the terminal device 600.

[0176] Those skilled in the art can understand that Figure 14 is only an example of a terminal device and does not constitute a limitation on the terminal device. It may include more or fewer components than shown in the figure, or combine some components, or different components, such as input / output devices, network access devices, buses, etc.

[0177] The processor 610 may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0178] The memory 620 may be an internal storage unit of the terminal device or an external storage device of the terminal device, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. The memory 620 is used to store the computer program and other programs and data required by the terminal device. The memory 620 may also be used to temporarily store data that has been output or is to be output.

[0179] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the buses in the drawings of the present application are not limited to only one bus or one type of bus.

[0180] The fuel cell operating condition optimization method provided in this embodiment can be applied to terminal devices such as computers, wearable devices, vehicle-mounted devices, tablet computers, laptop computers, and netbooks. The specific types of terminal devices are not limited in the embodiments of the present application.

Claims

1. A method for optimizing the operating conditions of a fuel cell, characterized in that, it includes: Determine a fuel cell simulation model and a neural network model according to different application scenarios of the fuel cell; Train the neural network model based on the simulation inputs and outputs of the fuel cell simulation model to obtain a fuel cell operating condition optimization model; Based on the application scenario of the fuel cell, optimize the operating conditions of the fuel cell through the fuel cell operating condition optimization model.

2. The method for optimizing the operating conditions of a fuel cell according to claim 1, characterized in that, The determining of the fuel cell simulation model according to different application scenarios of the fuel cell includes: Determine the input parameters and output parameters for optimizing the operating conditions of the fuel cell according to different application scenarios of the fuel cell; Construct a fuel cell simulation model based on the input parameters and the output parameters; Calibrate the fuel cell simulation model based on the actual test results of each parameter.

3. The method for optimizing the operating conditions of a fuel cell according to claim 1, characterized in that, The training of the neural network model based on the simulation inputs and outputs of the fuel cell simulation model to obtain a fuel cell operating condition optimization model includes: Obtain the first simulation input parameter values and the corresponding first simulation output parameter values of the fuel cell simulation model, as well as the second simulation input parameter values and the corresponding second simulation output parameter values; Use the first data set composed of the first simulation input parameter values and the first simulation output parameter values as the training set of the neural network model; Use the second data set composed of the second simulation input parameter values and the second simulation output parameter values as the verification set of the neural network model; Train the neural network model based on the training set; Verify the trained neural network model based on the verification set; When the prediction accuracy of the neural network model meets the preset value, use the current neural network model as the fuel cell operating condition optimization model.

4. The method for optimizing the operating conditions of a fuel cell according to claim 3, characterized in that, The obtaining of the first simulation input parameter values and the corresponding first simulation output parameter values of the fuel cell simulation model, as well as the second simulation input parameter values and the corresponding second simulation output parameter values includes: Determine the input parameters and output parameters for optimizing the operating conditions of the fuel cell according to different application scenarios of the fuel cell; Take points at the first sparsity level among the input parameters respectively; Determine the first simulation input parameter values through the Taguchi algorithm based on the points taken at the first sparsity level; Input the first simulation input parameter values into the fuel cell simulation model to obtain the first simulation output parameter values corresponding to the first simulation input parameter values; Take points at the second sparsity level among the input parameters respectively; Determine the second simulation input parameter values through the Taguchi algorithm based on the points taken at the second sparsity level; Input the second simulation input parameter values into the fuel cell simulation model to obtain the first simulation output parameter values corresponding to the second simulation input parameter values; where the levels of the points taken at the first sparsity level are different from the levels of the points taken at the second sparsity level.

5. The method for optimizing the operating conditions of a fuel cell according to claim 4, wherein, the training of the neural network model based on the training set includes: dividing the training set into N mutually exclusive training subsets evenly; denoting the nth training subset among the N training subsets as the nth validation fold, and denoting the training subsets other than the nth training subset among the N training subsets as the nth training fold; where n ∈ [1, N] and is an integer; training the neural network model based on the nth training fold to obtain the nth sub-neural network model, and validating the nth sub-neural network model based on the nth validation fold to obtain the prediction accuracy of the nth sub-neural network model; selecting the sub-neural network model with the highest prediction accuracy among the first sub-neural network model to the Nth sub-neural network model as the trained neural network model.

6. The method for optimizing the operating conditions of a fuel cell according to any one of claims 1-5, wherein, the optimization of the operating conditions of the fuel cell through the fuel cell operating condition optimization model based on the application scenario of the fuel cell includes: determining the optimization input parameters and the optimization output parameters based on the application scenario of the fuel cell; determining the optimization objective based on the optimization output parameters; selecting the optimization input parameter values based on each optimization input parameter; inputting the optimization input parameter values into the fuel cell operating condition optimization model to obtain the optimization output parameter values; calculating the optimization objective value based on the optimization output parameter values; wherein, when the optimization output parameters include multiple parameters, weight distribution is performed on each parameter to obtain the optimization objective value. selecting the optimal operating condition of the fuel cell according to the optimization objective value.

7. The method for optimizing the operating conditions of a fuel cell according to claim 6, wherein, the determination of the optimization input parameters and the optimization output parameters based on the application scenario of the fuel cell includes: when the application scenario of the fuel cell is the optimization of the performance operating condition of the fuel cell, the optimization input parameters are the stoichiometric ratio, pressure, and temperature, and the optimization output parameter is the average voltage; when the application scenario of the fuel cell is the optimization of the cold start operating condition of the fuel cell, the optimization input parameters are the ambient temperature, the first-stage loading rate, the first-stage target current density, the second-stage loading rate, the second-stage target current density, and the water pump condition, and the optimization output parameters are the average voltage, the lowest voltage, the start-up success time, and the ice formation amount; when the application scenario of the fuel cell is the optimization of the durability operating condition of the fuel cell, the optimization input parameters are the load change rate, the number of start-up and shutdown times, pressure, humidity, and temperature, and the optimization output parameters are the total start-up duration, the total output power, the average voltage decay rate, and the hydrogen consumption.

8. The method for optimizing the operating conditions of a fuel cell according to claim 6, wherein, the selection of the optimization input parameter values based on each optimization input parameter includes: performing point selection at a dense level for each optimization input parameter; obtaining the optimization input parameter values in a fully orthogonal manner based on the point selection.

9. A device for optimizing the operating conditions of a fuel cell, wherein, comprising: The first model generation module (501) is configured to determine a fuel cell simulation model and a neural network model according to different application scenarios of the fuel cell; The second model generation module (502) is configured to train the neural network model based on the simulation input and output of the fuel cell simulation model to obtain a fuel cell operating condition optimization model; The operating condition optimization module (503) is configured to optimize the operating condition of the fuel cell through the fuel cell operating condition optimization model based on the application scenario of the fuel cell.

10. A terminal device (500), comprising: a processor (510) and a memory (520), wherein a computer program (521) that can run on the processor (510) is stored in the memory (520), and is characterized in that when the processor (510) executes the computer program (521), the method for optimizing the operating condition of the fuel cell according to any one of claims 1 to 8 is implemented.

Images