Multi-stack solid oxide fuel cell system performance prediction model construction method and application

Abstract

The invention belongs to the technical field of SOFC performance prediction, and discloses a multi-stack solid oxide fuel cell system performance prediction model construction method and application, and the method comprises the steps: obtaining the output voltage of each stack of a multi-stack solid oxide fuel cell system under different working conditions at different moments; calculating the correlation between the output voltage of the current electric pile at different moments and the output voltage of other electric piles at different moments; taking the output voltages of other galvanic piles corresponding to the correlation exceeding a preset threshold value at different moments and the output voltages of the current galvanic pile at different moments as a data set of the current galvanic pile; and training the long-short-term memory neural network by using the training samples in the corresponding data set to obtain a performance prediction model of the current stack, the performance prediction model being used for predicting the future output voltage of the current stack, and the future output voltage being used for reflecting the performance of the current stack. According to the method, the performance prediction of the multi-stack SOFC system is realized, and the accuracy of performance prediction is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of solid oxide fuel cell performance prediction, and more specifically, relates to a method for constructing a multi-stack solid oxide fuel cell system performance prediction model and its application. Background Art

[0002] Solid oxide fuel cell (SOFC) systems inevitably experience performance degradation and shortened lifespan. Controlling the SOFC stack accordingly based on the current health status of the SOFC and its future degradation trend can effectively avoid unplanned downtime. Therefore, SOFC performance prediction is the key to extending its working life.

[0003] The performance prediction of SOFC system mainly focuses on the single stack performance prediction, but with the development of technology, high-power multi-stack system will become the main research object. Due to problems such as load sharing and fuel distribution between different stacks in multi-stack SOFC, the internal temperature of each stack will affect each other, which will lead to the performance degradation of some stacks in the multi-stack system. Therefore, when predicting the performance of multi-stack system, it is necessary to consider not only the impact of the performance degradation of the stack itself, but also the impact of the performance degradation caused by the mutual influence between the stacks, that is, the influence of the coupling between the stacks. This makes the existing method for single-stack performance prediction not suitable for the performance prediction of multi-stack SOFC system. Summary of the invention

[0004] In view of the above defects or improvement needs of the prior art, the present invention provides a method for constructing a performance prediction model for a multi-stack solid oxide fuel cell system and its application, which aims to realize the performance prediction of a multi-stack SOFC system and improve the accuracy of performance prediction.

[0005] To achieve the above object, according to a first aspect of the present invention, a method for constructing a performance prediction model for a multi-stack solid oxide fuel cell system is provided, comprising:

[0006] Obtaining the output voltage of each stack of a multi-stack solid oxide fuel cell system at different times under different operating conditions;

[0007] For each battery stack, calculate the correlation between the output voltage of the current battery stack at different times and the output voltages of other battery stacks at different times; take the output voltages of other battery stacks at different times corresponding to the correlation exceeding the preset threshold and the output voltage of the current battery stack at different times as the data set of the current battery stack;

[0008] For each fuel cell stack, a long short-term memory neural network is trained using training samples in the corresponding data set to obtain a performance prediction model for the current fuel cell stack; wherein the performance prediction model is used to predict the future output voltage of the current fuel cell stack, and the future output voltage is used to reflect the performance of the current fuel cell stack.

[0009] Furthermore, the long short-term memory neural network is trained using the training samples in the corresponding data set to obtain the performance prediction model of the current battery stack, including:

[0010] Input the current training sample into the long short-term memory neural network to obtain the predicted output voltage of the current battery stack at the next moment; with the goal of minimizing the feature loss between the predicted output voltage of the current battery stack at the next moment and the label, reversely adjust the network parameters of the long short-term memory neural network, and when the network converges, obtain the performance prediction model of the current battery stack;

[0011] The current training sample includes the output voltage of the current battery stack at the current moment and the output voltages of other battery stacks at the current moment, and the label is the output voltage of the current battery stack at the next moment.

[0012] Furthermore, a particle swarm algorithm is used to optimize the network parameters of the long short-term memory neural network; wherein the network parameters include the learning rate, the number of training times and the number of neurons of the long short-term memory neural network.

[0013] Furthermore, a feature selection algorithm is used to calculate the correlation between the output voltage of the current battery stack at different times and the output voltage of other battery stacks at different times;

[0014] Wherein, the feature selection algorithm is a random forest feature selection algorithm, a mutual information method or a Pearson correlation coefficient method.

[0015] Furthermore, the step of obtaining the output voltage of each stack of the multi-stack solid oxide fuel cell system at different times under different operating conditions includes:

[0016] Modeling of multi-stack solid oxide fuel cell systems;

[0017] Inputting data under different dynamic conditions into the model of the multi-stack solid oxide fuel cell system to obtain the output voltage of each stack at different times under different dynamic conditions;

[0018] The data under different dynamic working conditions include fuel flow, air flow and current with different Gaussian noise added.

[0019] Furthermore, for each battery stack, before using the training samples in the corresponding data set to train the long short-term memory neural network, the method further includes:

[0020] The data set is smoothed using an empirical mode decomposition algorithm, and the smoothed data is used as training samples.

[0021] According to a second aspect of the present invention, a method for predicting the performance of a multi-stack solid oxide fuel cell system is provided, comprising:

[0022] Input the output voltage of the current battery stack at the current moment and the output voltage of the battery stack at the current moment that has a high correlation with the current battery stack into the performance prediction model to obtain the output voltage of the current battery stack at the next moment, wherein the output voltage at the next moment is used to reflect the performance of the current battery stack;

[0023] The battery stacks with a high correlation with the current battery stack are other battery stacks corresponding to the output voltages of which the correlation with the current battery stack at different times exceeds a preset threshold;

[0024] The performance prediction model is constructed by the method for constructing a performance prediction model for a multi-stack solid oxide fuel cell system as described in any one of the first aspects.

[0025] According to a third aspect of the present invention, there is provided an electronic device comprising a computer-readable storage medium and a processor;

[0026] The computer-readable storage medium is used to store executable instructions;

[0027] The processor is used to read the executable instructions stored in the computer-readable storage medium to execute the method for constructing a multi-stack solid oxide fuel cell system performance prediction model as described in any one of the first aspects, or to execute the method for predicting the performance of a multi-stack solid oxide fuel cell system as described in the second aspect.

[0028] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, it implements the method for constructing a performance prediction model for a multi-stack solid oxide fuel cell system as described in any one of the first aspects, or implements the method for predicting the performance of a multi-stack solid oxide fuel cell system as described in the second aspect.

[0029] According to the fifth aspect of the present invention, there is provided a computer program product, comprising a computer program, which, when executed on a computer, enables the computer to execute the method for constructing a multi-stack solid oxide fuel cell system performance prediction model as described in any one of the first aspect, or to execute the method for predicting the performance of a multi-stack solid oxide fuel cell system as described in the second aspect.

[0030] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects:

[0031] (1) The method of the present invention obtains the output voltage of each stack at different times of the multi-stack solid oxide fuel cell system under different working conditions. When predicting the performance of each stack, in addition to considering the output voltage of the current stack at different times, it also considers the influence of the output voltage of the stack with high correlation with the current stack, which reflects the spatial coupling relationship between the current stack and other stacks related to its performance prediction. In this way, the performance prediction of the multi-stack solid oxide fuel cell system is converted into a spatiotemporal sequence prediction problem, and then the long short-term memory neural network is used to realize the prediction of the performance of each stack. The present invention realizes the performance prediction of the multi-stack SOFC system, and considers the coupling problem between the multi-stack systems during the prediction, thereby improving the accuracy of the performance prediction.

[0032] (2) Preferably, a particle swarm algorithm is used to optimize the network parameters of the long short-term memory neural network, which can further improve the prediction effect of the model.

[0033] (3) Preferably, when obtaining the output voltage of each battery stack at different times under different operating conditions, the input includes a current with different Gaussian noise added, which can simulate different dynamic operating conditions and is more consistent with actual data.

[0034] (4) Preferably, before each sample in the data set is input into the LSTM, the data set corresponding to each stack is smoothed using an empirical mode decomposition algorithm, which can retain the main information of the input voltage data while eliminating the impact of dynamic conditions on the multi-stack system, further improving the prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Schematic diagram of a method for constructing a performance prediction model for a multi-stack solid oxide fuel cell system in an embodiment of the present invention.

[0036] Figure 2 4 is a performance prediction flow chart in an embodiment of the present invention.

[0037] Figure 3 The output voltage data of 5 battery stack cells at different times collected in the embodiment of the present invention. DETAILED DESCRIPTION

[0038] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0039] In the present invention, the terms "first", "second", etc. in the present invention and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0040] Example 1

[0041] like Figure 1-Figure 2 As shown, an embodiment of the present invention provides a method for constructing a multi-stack solid oxide fuel cell system performance prediction model, comprising:

[0042] Obtaining the output voltage of each stack of a multi-stack solid oxide fuel cell system at different times under different operating conditions;

[0043] For each battery stack, a feature selection algorithm is used to calculate the correlation between the output voltage of the current battery stack at different times and the output voltages of other battery stacks at different times, and the output voltages of other battery stacks at different times corresponding to the correlation exceeding the preset threshold, as well as the output voltage of the current battery stack at different times are used as the data set of the current battery stack; a training sample in the data set is the output voltage of the battery stack at the current moment (including the output voltage of the current battery stack at the current moment and the output voltages of other battery stacks in the data set at the current moment), and the label is the output voltage of the current battery stack at the next moment;

[0044] For each fuel cell stack, a long short-term memory neural network (LSTM) is trained using the training samples in the corresponding data set to obtain a performance prediction model for the current fuel cell stack. The performance prediction model is used to predict the future output voltage of the current fuel cell stack, and the output voltage is used to reflect the performance of the current fuel cell stack.

[0045] During the training process, the output voltage of the current battery stack at the current moment and the output voltages of other battery stacks in the data set at the current moment are used as input to obtain the predicted output voltage of the current battery stack at the next moment. With the goal of minimizing the feature loss between the predicted result and the corresponding label, the network parameters of the long short-term memory neural network LSTM are adjusted in reverse. Preferably, the particle swarm algorithm is used to optimize the network parameters of the LSTM, including the learning rate, the number of training times and the number of neurons, to obtain a trained LSTM, that is, the required performance prediction model.

[0046] As a preferred implementation, obtaining the output voltage of each stack of a multi-stack solid oxide fuel cell system at different times under different operating conditions includes:

[0047] Modeling of multi-stack solid oxide fuel cell systems;

[0048] The data under different dynamic conditions are input into the model of the multi-stack solid oxide fuel cell system to obtain the output voltage of each stack at different times under different dynamic conditions. In an embodiment of the present invention, the data under different dynamic conditions include fuel flow, air flow, and current with different Gaussian noise added; wherein, the current with different Gaussian noise added is used to simulate different dynamic conditions. In a specific implementation, the fuel flow is 0.04626mol / s, the air flow is 0.5397mol / s, and Gaussian noise is used to simulate the dynamic conditions, so that the current fluctuates around 50A, and the fluctuation range is [-1,1]. Figure 3 The output voltage data of each stack at different times collected for a multi-stack solid oxide fuel cell system consisting of five series-connected stacks.

[0049] As a preferred implementation, the feature selection algorithm is a random forest feature selection algorithm, a mutual information method or a Pearson correlation coefficient method.

[0050] As a preferred implementation method, before each sample in the data set is input into the LSTM, it also includes: using the empirical mode decomposition algorithm to smooth the data set corresponding to each stack, which can retain the main information of the input voltage data, while eliminating the impact of dynamic conditions on the multi-stack system, and further improve the prediction accuracy.

[0051] As a preferred implementation method, an attention mechanism is added to the LSTM model, and LSTM performs performance prediction based on the attention mechanism.

[0052] In order to verify the prediction effect of the performance prediction model (PSO_LSTM) obtained by optimizing LSTM based on the particle swarm algorithm adopted in the present invention, in the embodiment of the present invention, under the same data set, the model of the present invention (PSO_LSTM) is compared with the existing models (CNN, ELM, GA_BP, LSSVM, LSTM, PSO_BP, SVM), and the evaluation indicators include the fitting index R 2 , mean absolute error MAE, mean root mean square error RMSE; where R 2 The closer it is to 1, the better the prediction performance is. The smaller the MAE and RMSE are, the better the prediction performance is. The comparison results are shown in Table 1:

[0053] Table 1 Comparison results of different prediction algorithms

[0054]

[0055] It can be seen that, overall, the performance prediction model of the present invention has the best prediction effect.

[0056] Example 2

[0057] An embodiment of the present invention provides a method for predicting the performance of a multi-stack solid oxide fuel cell system, comprising:

[0058] The output voltage of the current battery stack at the current moment and the output voltage of the battery stack at the current moment with a high correlation with the current battery stack are input into the corresponding performance prediction model to obtain the output voltage of the current battery stack at the next moment, and the output voltage at the next moment is used to reflect the performance of the current battery stack; wherein, the other battery stacks corresponding to the output voltages at different moments of which the correlation with the current battery stack exceeds a preset threshold are the battery stacks with a high correlation with the current battery stack;

[0059] The performance prediction model is constructed by the method for constructing a performance prediction model for a multi-stack solid oxide fuel cell system in the above-mentioned embodiment 1.

[0060] For the related technical solutions, please refer to the corresponding description in Example 1, which will not be repeated here.

[0061] The performance prediction method of a multi-stack SOFC system of the present invention not only considers the data of the target stack itself, but also considers the influence of other stacks in the system on the target stack, that is, the coupling between the stacks, when predicting the performance of the target stack of the multi-stack SOFC system. The stacks with high correlation with each stack in the multi-stack system are screened out through a feature selection algorithm to achieve the purpose of decoupling. The data of the target stack is time series data, and the data of the stack with high correlation with the target stack reflects the spatial coupling relationship between the target stack and other stacks related to the target stack. In this way, the performance prediction problem of the multi-stack SOFC system is converted into a spatiotemporal series prediction problem. After obtaining each target stack data set, the long short-term memory network (LSTM) is optimized by the particle swarm algorithm to predict the target stack voltage.

[0062] Example 3

[0063] An embodiment of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method in the above-mentioned embodiment 1 or embodiment 2 when executing the computer program.

[0064] The electronic device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The memory may be used to store computer programs and / or modules, and the processor may perform various functions of the electronic device by running or executing computer programs and / or modules stored in the memory, and calling data stored in the memory.

[0065] The relevant technical solutions are the same as above and will not be described in detail here.

[0066] Example 4

[0067] An embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method in the above-mentioned embodiment 1 or embodiment 2 are implemented.

[0068] Specifically, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, an internal memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0069] The relevant technical solutions are the same as above and will not be described in detail here.

[0070] Example 5

[0071] An embodiment of the present application provides a computer program product, including a computer program. When the computer program is run on a computer, the computer executes the steps of the method in the above-mentioned embodiment 1 or embodiment 2.

[0072] The relevant technical solutions are the same as above and will not be described in detail here.

[0073] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for constructing a performance prediction model for a multi-stack solid oxide fuel cell system, characterized in that: include: Obtaining the output voltage of each stack of a multi-stack solid oxide fuel cell system at different times under different operating conditions; For each battery stack, calculate the correlation between the output voltage of the current battery stack at different times and the output voltage of other battery stacks at different times; The output voltages of other battery stacks at different times corresponding to the correlation exceeding the preset threshold value and the output voltage of the current battery stack at different times are collectively used as the data set of the current battery stack; For each fuel cell stack, a long short-term memory neural network is trained using training samples in the corresponding data set to obtain a performance prediction model for the current fuel cell stack; wherein the performance prediction model is used to predict the future output voltage of the current fuel cell stack, and the future output voltage is used to reflect the performance of the current fuel cell stack.

2. The method for constructing a multi-stack solid oxide fuel cell system performance prediction model according to claim 1, characterized in that: The method of using the training samples in the corresponding data set to train the long short-term memory neural network to obtain the performance prediction model of the current battery stack includes: Input the current training sample into the long short-term memory neural network to obtain the predicted output voltage of the current battery stack at the next moment; with the goal of minimizing the feature loss between the predicted output voltage of the current battery stack at the next moment and the label, reversely adjust the network parameters of the long short-term memory neural network, and when the network converges, obtain the performance prediction model of the current battery stack; The current training sample includes the output voltage of the current battery stack at the current moment and the output voltages of other battery stacks at the current moment, and the label is the output voltage of the current battery stack at the next moment.

3. The method for constructing a multi-stack solid oxide fuel cell system performance prediction model according to claim 2, characterized in that: The particle swarm algorithm is used to optimize the network parameters of the long short-term memory neural network; wherein the network parameters include the learning rate, the number of training times and the number of neurons of the long short-term memory neural network.

4. The method for constructing a multi-stack solid oxide fuel cell system performance prediction model according to any one of claims 1 to 3, characterized in that: A feature selection algorithm is used to calculate the correlation between the output voltage of the current battery stack at different times and the output voltage of other battery stacks at different times; Wherein, the feature selection algorithm is a random forest feature selection algorithm, a mutual information method or a Pearson correlation coefficient method.

5. The method for constructing a multi-stack solid oxide fuel cell system performance prediction model according to claim 4, characterized in that: The method of obtaining the output voltage of each stack of the multi-stack solid oxide fuel cell system at different times under different operating conditions includes: Modeling of multi-stack solid oxide fuel cell systems; Inputting data under different dynamic conditions into the model of the multi-stack solid oxide fuel cell system to obtain the output voltage of each stack at different times under different dynamic conditions; The data under different dynamic working conditions include fuel flow, air flow and current with different Gaussian noise added.

6. The method for constructing a multi-stack solid oxide fuel cell system performance prediction model according to claim 5, characterized in that: For each battery stack, before using the training samples in the corresponding data set to train the long short-term memory neural network, it also includes: The data set is smoothed using an empirical mode decomposition algorithm, and the smoothed data is used as training samples.

7. A method for predicting the performance of a multi-stack solid oxide fuel cell system, characterized in that: include: Input the output voltage of the current battery stack at the current moment and the output voltage of the battery stack at the current moment that has a high correlation with the current battery stack into the performance prediction model to obtain the output voltage of the current battery stack at the next moment, wherein the output voltage at the next moment is used to reflect the performance of the current battery stack; The battery stacks with a high correlation with the current battery stack are other battery stacks corresponding to the output voltages of which the correlation with the current battery stack at different times exceeds a preset threshold; The performance prediction model is constructed by the method for constructing a performance prediction model for a multi-stack solid oxide fuel cell system according to any one of claims 1 to 6.

8. An electronic device, characterized in that: comprising a computer readable storage medium and a processor; The computer-readable storage medium is used to store executable instructions; The processor is used to read the executable instructions stored in the computer-readable storage medium to execute the method for constructing a multi-stack solid oxide fuel cell system performance prediction model as described in any one of claims 1-6, or to execute the method for predicting the performance of a multi-stack solid oxide fuel cell system as described in claim 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, it implements the method for constructing a performance prediction model for a multi-stack solid oxide fuel cell system as described in any one of claims 1 to 6, or implements the method for predicting performance of a multi-stack solid oxide fuel cell system as described in claim 7.

10. A computer program product, characterized in that It includes a computer program, which, when running on a computer, enables the computer to execute the method for constructing a multi-stack solid oxide fuel cell system performance prediction model as described in any one of claims 1 to 6, or execute the method for predicting the performance of a multi-stack solid oxide fuel cell system as described in claim 7.

Images