PEM electrolytic cell modeling method, performance prediction method, related system and equipment

Abstract

The invention provides a PEM electrolytic cell modeling method, a performance prediction method, a related system and equipment, and belongs to the technical field of computers. The modeling method comprises the steps that a preset neural network model is trained according to a first data set obtained when the PEM electrolytic cell operates under a plurality of different experiment conditions, a voltage prediction model is obtained, the input of the voltage prediction model is the working current and the working temperature of the PEM electrolytic cell, and the output of the voltage prediction model is the working voltage of the PEM electrolytic cell; according to a second data set obtained when the PEM electrolytic cell operates under specific experimental conditions, a temperature prediction model is determined in a parameter fitting mode, and the temperature prediction model is used for representing the function relation between the working temperature of the PEM electrolytic cell and the working current, the working voltage, the inlet flow, the inlet temperature and the environment temperature; and integrating the voltage prediction model and the temperature prediction model to obtain a PEM electrolytic cell model. The method is beneficial for improving the prediction accuracy of the PEM electrolytic cell performance parameters.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular, to a PEM electrolyzer modeling method, a performance prediction method, related systems and devices. Background Art

[0002] In the prior art, the PEM electrolyzer is usually modeled by means of mathematical model and empirical curve fitting. However, the mathematical model lacks adaptability and the ability to describe non-linear relationships, making it difficult to accurately predict the relevant performance parameters of the PEM electrolyzer under different working conditions, and the application effect is not ideal. Summary of the Invention

[0003] The main purpose of this application is to propose a PEM electrolyzer modeling method, a performance prediction method, related systems and devices. By introducing a neural network model and training it, the finally obtained PEM electrolyzer model has adaptability, which is beneficial to improving the prediction accuracy of the performance parameters of the PEM electrolyzer.

[0004] To achieve the above object, one aspect of this application proposes a PEM electrolyzer modeling method, including:

[0005] Obtain a first data set, where the first data set includes a number of first data groups obtained when the PEM electrolyzer operates under a number of different experimental conditions. Each first data group corresponding to an experimental condition includes the working current, working voltage and working temperature of the PEM electrolyzer;

[0006] Train a preset neural network model according to the first data set to obtain a voltage prediction model; the input of the voltage prediction model is the working current and working temperature of the PEM electrolyzer, and the output is the working voltage of the PEM electrolyzer;

[0007] Obtain a second data set, where the second data set includes a number of second data groups corresponding to a number of sampling times obtained when the PEM electrolyzer operates under specific experimental conditions. Each second data group corresponding to a sampling time includes the working current, working voltage, working temperature, inlet flow rate, inlet temperature and ambient temperature of the PEM electrolyzer;

[0008] Determine a temperature prediction model by parameter fitting according to the second data set; the temperature prediction model is used to characterize the functional relationship between the working temperature of the PEM electrolyzer and the working current, working voltage, inlet flow rate, inlet temperature and ambient temperature;

[0009] Integrate the voltage prediction model and the temperature prediction model to obtain a PEM electrolyzer model.

[0010] Further, the first data set is obtained by the following method:

[0011] According to a preset temperature increment, values are taken within a preset water supply temperature range to obtain multiple water supply temperatures and form a first test set;

[0012] According to a preset current increment, values are taken within a preset power supply current range to obtain multiple power supply currents and form a second test set;

[0013] Starting from i = 1, obtain the i-th water supply temperature included in the first test set;

[0014] According to the i-th water supply temperature, control the heating device to heat the ultrapure water in the water tank, and then control the water pump to inject the heated ultrapure water in the water tank into the anode side of the PEM electrolyzer at a specific flow rate until the internal temperature of the PEM electrolyzer stabilizes at the i-th water supply temperature;

[0015] Starting from j = 1, obtain the j-th power supply current included in the second test set;

[0016] Control the DC power supply to supply the j-th power supply current to the PEM electrolyzer, and then after the PEM electrolyzer operates for a specific period of time, obtain the first data set corresponding to the current experimental conditions;

[0017] According to the first data set corresponding to the current experimental conditions, determine whether the current operating state of the PEM electrolyzer is reasonable;

[0018] If the current operating state of the PEM electrolyzer is unreasonable, then determine whether i < N holds, where N is the number of the multiple water supply temperatures;

[0019] If i < N holds, then assign i + 1 to i, and return to the step of obtaining the i-th water supply temperature included in the first test set;

[0020] If i < N does not hold, then determine the first data set according to the several first data sets corresponding to the obtained several different experimental conditions;

[0021] If the current operating state of the PEM electrolyzer is reasonable, then determine whether j < M holds, where M is the number of the multiple power supply currents;

[0022] If j < M holds, then assign j + 1 to j, and return to the step of obtaining the j-th power supply current included in the second test set;

[0023] If j < M does not hold, then return to the step of determining whether i < N holds.

[0024] Further, the preset neural network model includes an input layer, a hidden layer, and an output layer connected in sequence; training the preset neural network model according to the first data set to obtain a voltage prediction model includes:

[0025] Adjust the preset neural network model according to a plurality of preset hyperparameters, each of the hyperparameters including the number of neuron nodes in the hidden layer, to obtain a corresponding plurality of first neural network models;

[0026] Train the plurality of first neural network models respectively according to the first data set, and then select the trained first neural network model with the best prediction performance from them and denote it as the second neural network model;

[0027] Adjust the second neural network model according to a plurality of preset activation function combinations, each of the activation function combinations including the activation function between the input layer and the hidden layer and the activation function between the hidden layer and the output layer, to obtain a corresponding plurality of third neural network models;

[0028] Train the plurality of third neural network models respectively according to the first data set, and then select the trained third neural network model with the best prediction performance from them and denote it as the fourth neural network model;

[0029] Retrain the fourth neural network model according to the first data set to obtain the voltage prediction model.

[0030] Further, the second data set is obtained by the following method:

[0031] Control a heating device to heat ultrapure water in a water tank according to a specific water supply temperature, then control a water pump to inject the heated ultrapure water in the water tank into the anode side of the PEM electrolyzer at a specific flow rate, and then control a DC power supply to supply a specific supply current to the PEM electrolyzer;

[0032] Obtain a plurality of second data groups corresponding to a plurality of sampling times within a preset sampling time period according to a preset sampling time interval to determine the second data set.

[0033] Further, determining the temperature prediction model by parameter fitting according to the second data set includes:

[0034] Obtain a preset mathematical expression corresponding to the temperature prediction model, the preset mathematical expression including a plurality of unknown parameters, the plurality of unknown parameters including the lumped heat capacity, thermal resistance, and heat correction coefficient of the PEM electrolyzer;

[0035] According to the second data set and the preset mathematical expression, perform fitting and solution on the multiple unknown parameters to obtain the values of the multiple unknown parameters;

[0036] Substitute the values of the multiple unknown parameters into the preset mathematical expression to determine the temperature prediction model.

[0037] To achieve the above object, another aspect of the present application proposes a method for predicting the performance of a PEM electrolyzer, including:

[0038] Obtain the current operating data of the PEM electrolyzer, where the current operating data of the PEM electrolyzer includes the current working current, current inlet flow rate, current inlet temperature, current ambient temperature, and current initial working temperature of the PEM electrolyzer;

[0039] Input the current operating data of the PEM electrolyzer into the PEM electrolyzer model for iterative processing to obtain multiple sets of predicted operating data of the PEM electrolyzer, and each set of predicted operating data of the PEM electrolyzer includes the predicted working voltage and predicted working temperature of the PEM electrolyzer;

[0040] Wherein, the PEM electrolyzer model is obtained by the above-mentioned PEM electrolyzer modeling method.

[0041] To achieve the above object, another aspect of the present application proposes a PEM electrolyzer modeling system, including:

[0042] A first acquisition module for acquiring a first data set, where the first data set includes a plurality of first data groups obtained corresponding to when the PEM electrolyzer operates under several different experimental conditions, and each first data group corresponding to an experimental condition includes the working current, working voltage, and working temperature of the PEM electrolyzer;

[0043] A training module for training a preset neural network model according to the first data set to obtain a voltage prediction model; the input of the voltage prediction model is the working current and working temperature of the PEM electrolyzer, and the output is the working voltage of the PEM electrolyzer;

[0044] A second acquisition module for acquiring a second data set, where the second data set includes a plurality of second data groups corresponding to several sampling times obtained when the PEM electrolyzer operates under specific experimental conditions, and each second data group corresponding to a sampling time includes the working current, working voltage, working temperature, inlet flow rate, inlet temperature, and ambient temperature of the PEM electrolyzer;

[0045] A determination module, configured to determine a temperature prediction model by means of parameter fitting according to the second data set; the temperature prediction model is used to characterize the functional relationship between the operating temperature of the PEM electrolyzer and the operating current, operating voltage, inlet flow rate, inlet temperature, and ambient temperature;

[0046] An integration module, configured to integrate the voltage prediction model and the temperature prediction model to obtain a PEM electrolyzer model.

[0047] To achieve the above object, another aspect of the present application provides a PEM electrolyzer performance prediction system, including:

[0048] A third acquisition module, configured to acquire the current operating data of the PEM electrolyzer, where the current operating data of the PEM electrolyzer includes the current operating current, current inlet flow rate, current inlet temperature, current ambient temperature, and current initial operating temperature of the PEM electrolyzer;

[0049] A processing module, configured to input the current operating data of the PEM electrolyzer into the PEM electrolyzer model for iterative processing to obtain multiple sets of predicted operating data of the PEM electrolyzer, where each set of predicted operating data of the PEM electrolyzer includes the predicted operating voltage and predicted operating temperature of the PEM electrolyzer;

[0050] Wherein, the PEM electrolyzer model is obtained by the above-mentioned PEM electrolyzer modeling method.

[0051] To achieve the above object, another aspect of the present application provides an electronic device, where the electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the above-mentioned PEM electrolyzer modeling method or the above-mentioned PEM electrolyzer performance prediction method is implemented.

[0052] To achieve the above object, another aspect of the present application provides a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned PEM electrolyzer modeling method or the above-mentioned PEM electrolyzer performance prediction method is implemented.

[0053] The present application has at least the following beneficial effects: By training a preset neural network model according to a first data set obtained when a PEM electrolyzer operates under several different experimental conditions to determine a voltage prediction model, and fitting and determining a temperature prediction model according to a second data set obtained when the PEM electrolyzer operates under specific experimental conditions, and then constructing a PEM electrolyzer model based on the voltage prediction model and the temperature prediction model, the PEM electrolyzer model has an adaptive ability and is beneficial to improving the prediction accuracy of the performance parameters of the PEM electrolyzer. By analyzing the current operating data of the PEM electrolyzer using the constructed PEM electrolyzer model to predict the relevant performance parameters of the PEM electrolyzer, it can provide data support for the subsequent task of optimizing the operating parameters of the PEM electrolyzer to further improve the efficiency of electrolytic hydrogen production. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is a schematic flowchart of a method for modeling a PEM electrolyzer provided by an embodiment of the present application;

[0055] Figure 2 is a schematic diagram of a PEM electrolyzer experimental device provided by an embodiment of the present application;

[0056] Figure 3 is a schematic flowchart of a method for predicting the performance of a PEM electrolyzer provided by an embodiment of the present application;

[0057] Figure 4 is a schematic diagram of the composition of a PEM electrolyzer modeling system provided by an embodiment of the present application;

[0058] Figure 5 is a schematic diagram of the composition of a PEM electrolyzer performance prediction system provided by an embodiment of the present application;

[0059] Figure 6 is a schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] In order to make the objectives, technical solutions, and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present application. They are only examples of systems and methods that are consistent with some aspects of the embodiments of the present application as detailed in the appended claims.

[0061] It can be understood that the terms "first", "second", etc. used in this application may be used herein to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if", "when" as used herein may be interpreted as "when...", "while...", or "in response to determining".

[0062] The terms "at least one", "a plurality of", "each", "any one", etc. used in this application, at least one includes one, two or more than two, a plurality includes two or more than two, each refers to each one in the corresponding plurality, and any one refers to any one in the plurality.

[0063] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0064] Proton Exchange Membrane (PEM) electrolysis hydrogen production technology has become one of the important technologies for realizing low-carbon energy transformation due to its advantages such as high hydrogen production efficiency, high hydrogen purity, and rapid response to renewable energy fluctuations. In practical applications, accurately predicting the performance parameters of a PEM electrolyzer can help designers optimize the operating parameters of the PEM electrolyzer and improve the system design efficiency and economy.

[0065] In the prior art, a PEM electrolyzer is usually modeled by means of mathematical models and empirical curve fitting to describe the relationship between the operating parameters and performance of the PEM electrolyzer. However, the mathematical model lacks self-adaptive ability and has insufficient ability to describe non-linear relationships, making it difficult to accurately predict the relevant performance parameters of the PEM electrolyzer under different operating conditions. The application effect is not ideal, which limits its popularization in practical applications. In addition, this modeling process often requires a large amount of experimental data, resulting in high development costs, and the model prediction range is limited, making it difficult to adapt to complex and dynamic operating conditions.

[0066] In view of this, the embodiments of the present application provide a PEM electrolyzer modeling method, a performance prediction method, related systems and devices. This solution trains a preset neural network model according to a first data set obtained when the PEM electrolyzer operates under several different experimental conditions to determine a voltage prediction model, and fits and determines a temperature prediction model according to a second data set obtained when the PEM electrolyzer operates under specific experimental conditions. Then, a PEM electrolyzer model is constructed based on the voltage prediction model and the temperature prediction model, so that the PEM electrolyzer model has an adaptive ability and is conducive to improving the prediction accuracy of the performance parameters of the PEM electrolyzer. By analyzing the current operating data of the PEM electrolyzer using the constructed PEM electrolyzer model, relevant performance parameters of the PEM electrolyzer can be predicted, which can provide data support for the subsequent task of optimizing the operating parameters of the PEM electrolyzer to further improve the efficiency of electrolytic hydrogen production.

[0067] Figure 1 FIG. 4 is an optional flowchart of a PEM electrolyzer modeling method provided by an embodiment of the present application. Figure 1 The method in FIG. 4 may but is not limited to include steps S110 to S150:

[0068] Step S110: Obtain a first data set, which includes several first data groups corresponding to when the PEM electrolyzer operates under several different experimental conditions. Each first data group corresponding to an experimental condition includes the working current, working voltage, and working temperature of the PEM electrolyzer.

[0069] Step S120: Train a preset neural network model according to the first data set to obtain a voltage prediction model. The input of the voltage prediction model is the working current and working temperature of the PEM electrolyzer, and the output is the working voltage of the PEM electrolyzer.

[0070] Step S130: Obtain a second data set, which includes several second data groups corresponding to several sampling times when the PEM electrolyzer operates under specific experimental conditions. Each second data group corresponding to a sampling time includes the working current, working voltage, working temperature, inlet flow rate, inlet temperature, and ambient temperature of the PEM electrolyzer.

[0071] Step S140: Determine a temperature prediction model by parameter fitting according to the second data set. The temperature prediction model is used to represent the functional relationship between the working temperature of the PEM electrolyzer and the working current, working voltage, inlet flow rate, inlet temperature, and ambient temperature.

[0072] Step S150: Integrate the voltage prediction model and the temperature prediction model to obtain a PEM electrolyzer model.

[0073] Steps S110 to S150 illustrated in the embodiments of the present application, by introducing a neural network model and training it, enable the finally obtained PEM electrolyzer model to have an adaptive ability, which is beneficial to improving the prediction accuracy of the performance parameters of the PEM electrolyzer.

[0074] In some embodiments, both the first data set mentioned in the above step S110 and the second data set mentioned in the above step S130 can depend on Figure 2 the PEM electrolyzer experimental device shown in the figure to obtain. The PEM electrolyzer experimental device at least includes a PEM electrolyzer 210, a heating device 220, a water tank 230, a water pump 240, a DC power supply 250, a computer device 260, and multiple temperature sensors. The heating device 220 is preferably arranged at the bottom of the water tank 230. The water pump 240 preferably uses a high-precision peristaltic pump. The multiple temperature sensors include a first temperature sensor 271, a second temperature sensor 272, a third temperature sensor 273, and a fourth temperature sensor 274. Each temperature sensor preferably uses a PT100 temperature sensor, where:

[0075] The heating device 220 is used to adjust the temperature of the ultrapure water stored in the water tank 230 through the existing PID (Proportional-Integral-Derivative) control strategy;

[0076] The water pump 240 is used to adjust the flow rate of the ultrapure water stored in the water tank 230 when it is injected into the anode side of the PEM electrolyzer 210. In the present application, the flow rate adjustment range is preferably set to [0.1 L / min, 2.0 L / min];

[0077] The DC power supply 250 is used to supply power to the PEM electrolyzer 210 so that an electrolysis reaction occurs inside the PEM electrolyzer 210;

[0078] The first temperature sensor 271 is used to collect the temperature of the ultrapure water stored in the water tank 230. The second temperature sensor 272 is used to collect the anode side temperature of the PEM electrolyzer 210. The third temperature sensor 273 is used to collect the cathode side temperature of the PEM electrolyzer 210. The fourth temperature sensor 274 is used to collect the inlet temperature of the PEM electrolyzer 210, which can be understood as the temperature of the ultrapure water at the inlet of the PEM electrolyzer 210. The DC power supply 250 is also used to collect the working current and working voltage of the PEM electrolyzer 210. The heating device 220 is also used to receive the temperature of the ultrapure water collected by the first temperature sensor 271. The computer device 260 is used to receive the relevant temperatures collected by the other three temperature sensors and the relevant electrical data collected by the DC power supply 250.

[0079] In addition, the multiple temperature sensors further include a fifth temperature sensor (not shown in Figure 2 ), which is disposed near the PEM electrolyzer 210 for collecting the ambient temperature of the PEM electrolyzer 210; the PEM electrolyzer experimental device further includes a flow sensor (not shown in Figure 2 ), which is preferably disposed on the ultrapure water flow path formed between the water pump 240 and the inlet of the PEM electrolyzer 210 and is disposed near the PEM electrolyzer 210 for collecting the inlet flow rate of the PEM electrolyzer 210, which can be understood as the volume flow rate of ultrapure water at the inlet of the PEM electrolyzer 210; the computer device 260 is further configured to receive the ambient temperature collected by the fifth temperature sensor and the inlet flow rate collected by the flow sensor.

[0080] In step S110 of some embodiments, each experimental condition refers to defining the water supply temperature and the supply current required for the PEM electrolyzer during operation, and the water supply temperature and / or the supply current defined between every two different experimental conditions are different. The water supply temperature refers to the temperature of the ultrapure water injected into the anode side of the PEM electrolyzer; the process of obtaining the first data set may include, but is not limited to, the following steps S111 to S117:

[0081] Step S111: Obtain a plurality of water supply temperatures according to a preset temperature increment within a preset water supply temperature range, and form a first test set.

[0082] In this step, the temperature increment is preferably set to 4 °C, and the water supply temperature range is preferably set to [44 °C, 76 °C]; starting from the minimum allowable water supply temperature of 44 °C, incrementally take values step by step according to the temperature increment of 4 °C until the maximum allowable water supply temperature of 76 °C is reached. Then, arrange the 9 obtained water supply temperatures in ascending order to form a first test set. That is, the first water supply temperature included in the first test set is the minimum allowable water supply temperature of 44 °C, the last water supply temperature included in the first test set is the maximum allowable water supply temperature of 76 °C, and the difference between every two adjacent water supply temperatures included in the first test set is the temperature increment of 4 °C.

[0083] Step S112: Obtain a plurality of supply currents according to a preset current increment within a preset supply current range, and form a second test set.

[0084] In this step, the current increment is preferably set to 0.2 A, and the power supply current range is preferably set to [0.2 A, 75 A]; starting from the minimum allowable power supply current of 0.2 A, the value is gradually increased according to the current increment of 0.2 A until the maximum allowable power supply current of 75 A is reached. Then, the 375 obtained power supply currents are arranged in ascending order to form a second test set. That is, the first power supply current included in the second test set is the minimum allowable power supply current of 0.2 A, the last power supply current included in the second test set is the maximum allowable power supply current of 75 A, and the difference between every two adjacent power supply currents included in the second test set is the current increment of 0.2 A.

[0085] Step S113: Obtain the i-th water supply temperature included in the first test set. Subsequently, according to the i-th water supply temperature, control the heating device to heat the ultrapure water in the water tank, and then control the water pump to inject the heated ultrapure water in the water tank into the anode side of the PEM electrolyzer at a specific flow rate until the internal temperature of the PEM electrolyzer stabilizes at the i-th water supply temperature.

[0086] In this step, controlling the heating device to heat the ultrapure water in the water tank according to the i-th water supply temperature should be understood as: controlling the heating device to heat the ultrapure water in the water tank until the temperature of the ultrapure water in the water tank is close to the i-th water supply temperature; that is to say, a suitable allowable deviation range of the water supply temperature can be set for the i-th water supply temperature, and it is specified that the i-th water supply temperature falls within the allowable deviation range of the water supply temperature, and then control the heating device to heat the ultrapure water in the water tank until the temperature of the ultrapure water in the water tank falls within the allowable deviation range of the water supply temperature.

[0087] In this step, the specific flow rate is preferably set to 154 ml / min; the internal temperature of the PEM electrolyzer can be understood as the average value between the anode side temperature and the cathode side temperature when the PEM electrolyzer is not powered on and started.

[0088] Step S114: Obtain the j-th power supply current included in the second test set. Subsequently, control the DC power supply to supply the j-th power supply current to the PEM electrolyzer, and then, after the PEM electrolyzer operates for a specific period of time, obtain the first data set corresponding to the current experimental conditions.

[0089] In this step, the specific time period is preferably set to 60 seconds; when the DC power supply supplies the j-th supply current to the PEM electrolyzer, an electrolysis reaction occurs inside the PEM electrolyzer. After the PEM electrolyzer operates continuously for 60 seconds with the j-th supply current, the working voltage, working current, anode-side temperature, and cathode-side temperature of the PEM electrolyzer at the current moment are collected. After averaging the anode-side temperature and cathode-side temperature of the PEM electrolyzer, it is used as the working temperature of the PEM electrolyzer. Subsequently, according to the working voltage, working current, and working temperature of the PEM electrolyzer, the first data set corresponding to the current experimental conditions is determined.

[0090] Step S115: According to the first data set corresponding to the current experimental conditions, determine whether the current operating state of the PEM electrolyzer is reasonable; if so, execute step S116; if not, execute step S117.

[0091] In this step, determining whether the current operating state of the PEM electrolyzer is reasonable should be understood as: extracting the working voltage of the PEM electrolyzer from the first data set corresponding to the current experimental conditions, and determining whether the working voltage of the PEM electrolyzer is less than or equal to a preset voltage threshold, which is preferably set to 2.4V; if so, it means that the current operating state of the PEM electrolyzer is reasonable, and the supply current to the PEM electrolyzer can be continuously increased to conduct the next experiment; if not, it means that the current operating state of the PEM electrolyzer is unreasonable. At this time, the experimental process of the PEM electrolyzer at the i-th water supply temperature is terminated in advance to protect the service life of the PEM electrolyzer, and then the water supply temperature to the PEM electrolyzer is continuously increased to conduct the next experiment.

[0092] Step S116: Determine whether j < M holds, where M is the number of supply currents included in the first test set; if so, assign j + 1 to j, and then return to execute the above step S114; if not, it means that all the current-increasing experiments of the PEM electrolyzer at the i-th water supply temperature have been completed, and execute step S117.

[0093] Step S117: Determine whether i < N holds, where N is the number of water supply temperatures included in the first test set; if so, assign i + 1 to i, and then return to execute the above step S113; if so, according to the obtained several first data sets corresponding to several different experimental conditions, determine the first data set.

[0094] In this step, since there are 9 water supply temperatures in the first test set and 375 power supply currents in the second test set, in an ideal situation (i.e., ensuring that the current operating state of the PEM electrolyzer is reasonable in each current increase experiment), after executing the above steps S111 to S117, up to 3375 first data sets corresponding to 3375 different experimental conditions can be obtained.

[0095] It should be noted that the above step S113 starts from i = 1, and the above step S114 starts from j = 1.

[0096] Steps S111 to S117 illustrated in the embodiments of the present application can regularly obtain the first data sets corresponding to the PEM electrolyzer under different experimental conditions by relying on the PEM electrolyzer experimental device, which can provide effective and reliable data support for the subsequent task of training the required voltage prediction model.

[0097] In step S120 of some embodiments, the preset neural network model includes an input layer, a hidden layer, and an output layer connected in sequence. The preset neural network model preferably adopts an existing BP (Back Propagation) neural network model, and the output of the preset neural network model can be represented by the following function:

[0098]

[0099] In the formula, is the final output of the preset neural network model, g(·) is the activation function between the hidden layer and the output layer, f(·) is the activation function between the input layer and the hidden layer, h is the number of neuron nodes in the hidden layer, is the weight from the j-th neuron node in the hidden layer to the output layer, v is the number of all inputs received by the input layer, x i is the i-th input received by the input layer, is the weight from the i-th input received by the input layer to the j-th neuron node in the hidden layer, is the threshold of the j-th neuron node in the hidden layer, b (2) is the threshold of the output layer.

[0100] In step S120 of some embodiments, the training process of the preset neural network model can but is not limited to including the following steps S121 to S125:

[0101] Step S121: Adjust the preset neural network model according to a plurality of preset hyperparameters, each hyperparameter including the number of neuron nodes in the hidden layer, to obtain a corresponding plurality of first neural network models.

[0102] In this step, it is preferable to set the numerical interval to 1 and the value range to [1, 5]. First, according to this numerical interval, incremental values are taken within this value range to obtain multiple different values, where each value represents a hyperparameter, that is, each value represents the number of neuron nodes in this hidden layer; then, according to these multiple different values, the number of neuron nodes in the hidden layer inside the preset neural network model is set respectively to obtain multiple different first neural network models, as shown in Table 1 for details.

[0103] Table 1 Model Adjustment Based on Hyperparameters

[0104]

[0105]

[0106] In Table 1, Model_11 represents the first neural network model obtained by setting the number of neuron nodes in the hidden layer inside the preset neural network model to 1, Model_12 represents the first neural network model obtained by setting the number of neuron nodes in the hidden layer inside the preset neural network model to 2, and so on.

[0107] It should be noted that before adjusting the preset neural network model, the parameters of the preset neural network model are randomly initialized, and the relevant parameters include weights and thresholds.

[0108] Step S122: According to this first data set, train the multiple first neural network models respectively, and then select the trained first neural network model with the best prediction performance from them and denote it as the second neural network model.

[0109] In this step, it is preferable to set the preset ratio to 8:1:1. First, organize this first data set to obtain a data sample set; then, according to this preset ratio, divide the data sample set into a training sample set, a test sample set, and a validation sample set; then, according to this training sample set, train the multiple first neural network models to obtain multiple trained first neural network models; finally, according to this test sample set, evaluate the prediction performance of the multiple trained first neural network models to select the trained first neural network model with the best prediction performance from them and denote it as the second neural network model.

[0110] Among them, organizing this first data set to obtain a data sample set is specifically manifested as:

[0111] For each first data group included in the first dataset, where the first data group includes the working current, working voltage, and working temperature of the PEM electrolyzer, the working current and working temperature of the PEM electrolyzer are used as data samples, and the working voltage of the PEM electrolyzer is used as the true label carried by the data sample; arranging several first data groups included in the first dataset according to this implementation manner, several data samples and several corresponding true labels carried thereby can be obtained, thereby forming a data sample set.

[0112] Among them, for each first neural network model, the first neural network model is trained according to the training sample set to obtain a trained first neural network model, and the corresponding training process is as follows:

[0113] In the forward propagation stage, all data samples included in the training sample set are input into the first neural network model for analysis to obtain all corresponding predicted outputs;

[0114] In the error evaluation stage, according to a preset training loss function, all true labels carried by all data samples included in the training sample set, and all predicted outputs corresponding to all data samples included in the training sample set, the training loss is determined, and the training loss function is as follows:

[0115]

[0116] In the formula, E is the training loss, N 1 is the number of all data samples included in the training sample set, y k is the true label carried by the k-th data sample included in the training sample set, is the predicted output obtained by inputting the k-th data sample included in the training sample set into the first neural network model for analysis;

[0117] In the backpropagation stage, according to the training loss, the gradient descent method is used to update all weights and all thresholds included in the first neural network model, and the corresponding update formulas are as follows:

[0118]

[0119] In the formula, is the update result regarding the weight , is the update result regarding the weight , is the update result regarding the threshold , is the update result regarding the threshold b (2) , and η is the learning rate;

[0120] By iteratively executing the above forward propagation stage, error evaluation stage, and backpropagation stage until a preset maximum number of iterations is reached or the training loss reaches the desired error level, a trained first neural network model can be obtained.

[0121] Among them, the prediction performance of the trained multiple first neural network models is evaluated according to the test sample set to screen and determine the second neural network model. Specifically, it is manifested as follows:

[0122] For each trained first neural network model, all data samples included in the test sample set are input into the trained first neural network model for analysis to obtain corresponding all prediction outputs. Then, according to the predetermined prediction performance evaluation index, operations are performed on all prediction outputs corresponding to all data samples included in the test sample set to obtain the prediction performance evaluation result corresponding to the trained first neural network model. By evaluating the prediction performance of the trained multiple first neural network models in this implementation manner, multiple prediction performance evaluation results corresponding to the trained multiple first neural network models can be obtained. Finally, the trained first neural network model with the best prediction performance evaluation result is selected from the trained multiple first neural network models and denoted as the second neural network model.

[0123] It should be noted that the prediction performance evaluation index includes at least one of the mean absolute error, root mean square error, and goodness of fit. The following is an explanation of these three types of indexes:

[0124] (1) The mathematical expression of the mean absolute error is as follows:

[0125]

[0126] When the value of the mean absolute error is smaller, it indicates that the model prediction performance is better;

[0127] (2) The mathematical expression of the root mean square error is as follows:

[0128]

[0129] When the value of the root mean square error is smaller, it indicates that the model prediction performance is better;

[0130] (3) The mathematical expression of the goodness of fit is as follows:

[0131]

[0132] When the value of the goodness of fit is larger, it indicates that the model prediction performance is better;

[0133] In the formula, MAE is the mean absolute error, RMSE is the root mean square error, R 2is the goodness of fit, N 2 is the number of all data samples included in the test sample set, y t_s is the true label carried by the s-th data sample included in the test sample set, is the predicted output obtained by inputting the s-th data sample included in the test sample set into the trained first neural network model for analysis, is the average value of all true labels carried by all data samples included in the test sample set.

[0134] Step S123. According to a plurality of preset activation function combinations, each activation function combination includes the activation function between the input layer and the hidden layer and the activation function between the hidden layer and the output layer, adjust the second neural network model to obtain a corresponding plurality of third neural network models.

[0135] In this step, considering that the commonly used activation functions of the BP neural network model at least include logsig, tansig, and purelin, first pair these three types of activation functions in pairs to obtain a plurality of different activation function combinations; then, according to the plurality of different activation function combinations, respectively set the activation function between the input layer and the hidden layer and the activation function between the hidden layer and the output layer inside the second neural network model to obtain a plurality of different third neural network models, as specifically shown in Table 2.

[0136] Table 2 Model adjustment based on activation function combinations

[0137]

[0138] In Table 2, Model_31 represents the third neural network model obtained by setting both the activation function between the input layer and the hidden layer and the activation function between the hidden layer and the output layer inside the second neural network model to logsig, Model_32 represents the third neural network model obtained by setting the activation function between the input layer and the hidden layer inside the second neural network model to logsig and setting the activation function between the hidden layer and the output layer inside it to tansig, and so on.

[0139] The following is an explanation of the three types of activation functions appearing in Table 2:

[0140] (1) The logsig activation function is an S-shaped logarithmic function, and the corresponding function expression is as follows:

[0141]

[0142] (2) The tansig activation function is a hyperbolic tangent S-shaped logarithmic function, and the corresponding function expression is as follows:

[0143]

[0144] (3) The purelin activation function is a linear function, and the corresponding function expression is as follows:

[0145] f(x) = x;

[0146] In the formula, x is the input and f(x) is the output.

[0147] Step S124: According to the first data set, train the multiple third neural network models respectively, and then select the trained third neural network model with the best prediction performance from them and denote it as the fourth neural network model.

[0148] In this step, according to the training sample set obtained by sorting and partitioning the first data set, train the multiple third neural network models to obtain multiple trained third neural network models; then, according to the test sample set obtained by sorting and partitioning the first data set, evaluate the prediction performance of the multiple trained third neural network models, so as to select the trained third neural network model with the best prediction performance from them and denote it as the fourth neural network model.

[0149] Regarding the training and screening process of the multiple third neural network models, it has the same implementation principle as the training and screening process of the multiple first neural network models in the above step S122, and will not be elaborated here.

[0150] Step S125: According to the first data set, retrain the fourth neural network model to obtain a voltage prediction model.

[0151] In this step, according to the training sample set obtained by sorting and partitioning the first data set, perform M times of intermittent training on the fourth neural network model to obtain M fourth neural network models after intermittent training; then, according to the test sample set obtained by sorting and partitioning the first data set, evaluate the prediction performance of the M fourth neural network models after intermittent training, so as to select the fourth neural network model with the best prediction performance after intermittent training from them and denote it as the voltage prediction model; where M is preferably set to 5.

[0152] Among them, performing M times of intermittent training on the fourth neural network model according to the training sample set to obtain M fourth neural network models after intermittent training is specifically manifested as:

[0153] Perform the first intermittent training on the fourth neural network model according to the training sample set to obtain the fourth neural network model after the first intermittent training; perform the second intermittent training on the current fourth neural network model (i.e., the fourth neural network model after the first intermittent training) according to the training sample set to obtain the fourth neural network model after the second intermittent training; perform the third intermittent training on the current fourth neural network model (i.e., the fourth neural network model after the second intermittent training) according to the training sample set to obtain the fourth neural network model after the third intermittent training, and so on.

[0154] For each intermittent training process of the fourth neural network model, the implementation principle is the same as that of the training process of the first neural network model in step S122 above, and will not be elaborated here.

[0155] Steps S121 to S125 illustrated in the embodiments of the present application can improve the accuracy and reliability of the finally determined voltage prediction model by appropriately adjusting the number of neuron nodes in the hidden layer inside the preset neural network model and the activation function between different network layers and then performing training and evaluation and screening.

[0156] In step S130 of some embodiments, the specific experimental conditions refer to specifying the specific water supply temperature and specific power supply current required for the PEM electrolyzer during operation. The specific water supply temperature refers to the specific temperature of the ultrapure water injected into the anode side of the PEM electrolyzer. The specific water supply temperature is preferably set to 45 °C, and the specific power supply current is preferably set to 37.5 A; the acquisition process of the second data set may but is not limited to include the following steps S131 to S132:

[0157] Step S131: According to the specific water supply temperature, control the heating device to heat the ultrapure water in the water tank, and then control the water pump to inject the heated ultrapure water in the water tank into the anode side of the PEM electrolyzer at a specific flow rate. Subsequently, control the DC power supply to supply the specific power supply current to the PEM electrolyzer, so that an electrolysis reaction occurs inside the PEM electrolyzer.

[0158] In this step, controlling the heating device to heat the ultrapure water in the water tank according to the specific water supply temperature should be understood as: controlling the heating device to heat the ultrapure water in the water tank until the temperature of the ultrapure water in the water tank is close to the specific water supply temperature; that is to say, a suitable allowable deviation range of the specific water supply temperature can be set for the specific water supply temperature, and it is specified that the specific water supply temperature falls within the allowable deviation range of the specific water supply temperature, and then control the heating device to heat the ultrapure water in the water tank until the temperature of the ultrapure water in the water tank falls within the allowable deviation range of the specific water supply temperature.

[0159] In this step, the specific flow rate is preferably set to 154 ml / min.

[0160] Step S132: According to a preset sampling time interval, obtain a number of second data sets corresponding to a number of sampling times within a preset sampling time period to determine a second data set.

[0161] In this step, the sampling time interval is preferably set to 1 second, and the sampling time period is preferably set to 3500 seconds; starting from the moment when the specific supply current is provided from the DC power supply to the PEM electrolyzer, every 1 second, collect the working current, working voltage, anode side temperature, cathode side temperature, inlet flow rate, inlet temperature, and ambient temperature of the PEM electrolyzer at the current moment. After averaging the anode side temperature and cathode side temperature of the PEM electrolyzer, use it as the working temperature of the PEM electrolyzer. Subsequently, according to the working current, working voltage, working temperature, inlet flow rate, inlet temperature, and ambient temperature of the PEM electrolyzer, determine the second data set corresponding to the current sampling time; perform step-by-step incremental time acquisition according to this implementation method until the timing reaches the sampling time period, and then a number of second data sets corresponding to a number of different sampling times can be obtained.

[0162] It should be noted that this application can also start timing and perform relevant data acquisition after the PEM electrolyzer has been operating at the specific supply current for a period of time. This application does not make any limitations in this regard.

[0163] Steps S131 to S132 shown in the embodiments of this application can regularly obtain a number of second data sets corresponding to the PEM electrolyzer under specific experimental conditions by relying on the PEM electrolyzer experimental device, which can provide effective and reliable data support for the subsequent task of fitting to obtain the required temperature prediction model.

[0164] In step S140 of some embodiments, the temperature prediction model can be understood as a PEM electrolyzer thermal model. Since the heat transfer inside the PEM electrolyzer often involves multiple phenomena and variables, such as electrochemical reactions, polarization phenomena, and fluid dynamics, etc., accurately simulating these processes often requires a large amount of computing resources and time. In this application, it is preferred to use a lumped heat capacity model to simplify the calculation of the PEM electrolyzer thermal model, that is, assume that the temperature of the PEM electrolyzer is uniformly distributed, and assume that the PEM electrolyzer is a regular-shaped cuboid, which can reduce the computational complexity and improve the computational efficiency; the fitting determination process of the temperature prediction model can but is not limited to including the following steps S141 to S143:

[0165] Step S141: Obtain the preset mathematical expression corresponding to the temperature prediction model. The preset mathematical expression contains multiple unknown parameters, and the multiple unknown parameters include the lumped heat capacity, thermal resistance, and heat correction coefficient of the PEM electrolyzer.

[0166] In this step, considering that during the operation of the PEM electrolyzer, the input energy of the PEM electrolyzer mainly includes the electrical energy of the PEM electrolyzer and the energy brought by the ultrapure water flowing into the PEM electrolyzer, and the output energy of the PEM electrolyzer mainly includes the heat lost by the PEM electrolyzer to the external environment, the energy carried away by the ultrapure water flowing out of the PEM electrolyzer, and the energy for the PEM electrolyzer to generate hydrogen. According to the energy balance principle of the PEM electrolyzer, the preset mathematical expression corresponding to the temperature prediction model is determined as:

[0167]

[0168] P ele =U cell l cell ,

[0169]

[0170] In the formula, C is the lumped heat capacity of the PEM electrolyzer, T cell is the operating temperature of the PEM electrolyzer, t is time, which can be understood as the operating time of the PEM electrolyzer, P ele is the input power of the PEM electrolyzer, that is, the electrical energy of the PEM electrolyzer, is the energy brought by the ultrapure water flowing into the PEM electrolyzer, is the hydrogen production power of the PEM electrolyzer, that is, the energy for the PEM electrolyzer to generate hydrogen, is the energy carried away by the ultrapure water flowing out of the PEM electrolyzer, is the heat lost by the PEM electrolyzer to the external environment, U cell is the operating voltage of the PEM electrolyzer, I cell is the operating current of the PEM electrolyzer, Ut n is the thermal neutral voltage of the PEM electrolyzer, T a is the ambient temperature of the PEM electrolyzer, R is the thermal resistance of the PEM electrolyzer, ε is the emissivity and is preferably set to 0.8, σ is the Stefan-Boltzmann constant and is preferably set to 5.67×10 -8 W·m 2 ·K -4 , A cell is the total surface area of the PEM electrolyzer, is the inlet flow rate of the PEM electrolyzer, that is, the volume flow rate of the ultrapure water at the inlet of the PEM electrolyzer, is the density of ultrapure water and is preferably set to 1000 g / cm 3 , is the molar mass of ultrapure water and is preferably set to 2 g / mol, is the molar enthalpy value of the ultrapure water flowing into the PEM electrolyzer, is the molar enthalpy value of the ultrapure water flowing out of the PEM electrolyzer, is the inlet temperature of the PEM electrolyzer, and both k and b are heat correction coefficients;

[0171] Among them, regarding the molar enthalpy value of the ultrapure water flowing into the PEM electrolyzer and the molar enthalpy value of the ultrapure water flowing out of the PEM electrolyzer The corresponding mathematical expressions are as follows respectively:

[0172]

[0173] In the formula, H 0 is the molar enthalpy value of ultrapure water under reference conditions, and the preferred value is 1.8909 KJ / mol. The reference conditions refer to that the pressure of ultrapure water is 0.1 MPa (close to standard atmospheric pressure) and the temperature is 298 K (close to room temperature), and T 0 is the reference temperature of ultrapure water.

[0174] It should be noted that the above mathematical expression for the energy carried away by the ultrapure water when flowing out of the PEM electrolyzer is obtained by correcting the originally given first expression, and the first expression is as follows:

[0175]

[0176] In the first expression, it is assumed that the operating temperature of the PEM electrolyzer is equal to the temperature of the ultrapure water flowing out of the PEM electrolyzer. However, in actual situations, these two temperatures are not equal. Therefore, in this application, considering the complexity brought by the internal temperature distribution of the PEM electrolyzer, the structure of the PEM electrolyzer, and fluid dynamics, in order to simplify the calculation, the influence of the flow rate of ultrapure water on the heat transfer of the PEM electrolyzer is ignored here, and the operating temperature T of the PEM electrolyzer cell and the inlet temperature of the PEM electrolyzer are used to correct the first expression, so that the temperature prediction model can better assist in fitting the temperature curve of the PEM electrolyzer.

[0177] Step S142, according to the second data set and the preset mathematical expression, perform fitting and solution on the multiple unknown parameters to obtain the values of the multiple unknown parameters.

[0178] In this step, it is preferred to use the parameter estimator built in the Simulink software to complete the parameter fitting and solving task. By building a solution model for the preset mathematical expression in this parameter estimator, and then assigning basic parameters in this parameter estimator according to the actual application situation of the PEM electrolyzer, such as setting the molar mass of ultrapure water The density of ultrapure water The total surface area A of the PEM electrolyzer cell emissivity ε, Stefan-Boltzmann constant σ, etc. Subsequently, input the second data set into this parameter estimator for simulation and solution to obtain the values of the multiple unknown parameters.

[0179] Step S143: Substitute the values of the multiple unknown parameters into the preset mathematical expression to determine the temperature prediction model.

[0180] Steps S141 to S143 illustrated in the embodiments of the present application determine the preset mathematical expression corresponding to the temperature prediction model by considering the balance relationship between the input energy and output energy during the operation of the PEM electrolyzer, and then fit and solve the multiple unknown parameters included in the preset mathematical expression by means of the second data set obtained through experiments, which can improve the reliability of the finally determined temperature prediction model.

[0181] In step S150 of some embodiments, the voltage prediction model is provided with a working current input terminal, a working temperature input terminal, and a working voltage output terminal, and the temperature prediction model is provided with a working current input terminal, a working voltage input terminal, an inlet flow rate input terminal, an inlet temperature input terminal, an ambient temperature input terminal, and a working temperature output terminal. The PEM electrolyzer model can be integrated by connecting the working current input terminal of the voltage prediction model to the working current input terminal of the temperature prediction model, connecting the working voltage output terminal of the voltage prediction model to the working voltage input terminal of the temperature prediction model, and connecting the working temperature input terminal of the voltage prediction model to the working temperature output terminal of the temperature prediction model.

[0182] The PEM electrolyzer modeling method provided by the embodiments of the present application trains a preset neural network model according to the first data set obtained when the PEM electrolyzer operates under several different experimental conditions to determine the voltage prediction model, and fits and determines the temperature prediction model according to the second data set obtained when the PEM electrolyzer operates under specific experimental conditions, and then constructs the PEM electrolyzer model according to the voltage prediction model and the temperature prediction model, so that the PEM electrolyzer model has an adaptive ability and is conducive to improving the prediction accuracy of the performance parameters of the PEM electrolyzer.

[0183] Figure 3It is an optional process schematic diagram of a PEM electrolyzer performance prediction method provided by an embodiment of the present application. Figure 3 The method in may but is not limited to include steps S310 to S320:

[0184] Step S310: Obtain the current operating data of the PEM electrolyzer. The current operating data of the PEM electrolyzer includes the current working current, current inlet flow rate, current inlet temperature, current ambient temperature, and current initial working temperature of the PEM electrolyzer.

[0185] Step S320: Input the current operating data of the PEM electrolyzer into the PEM electrolyzer model for iterative processing to obtain multiple sets of predicted operating data of the PEM electrolyzer. Each set of predicted operating data of the PEM electrolyzer includes the predicted working voltage and predicted working temperature of the PEM electrolyzer. Among them, the PEM electrolyzer model is obtained by the above-mentioned PEM electrolyzer modeling method.

[0186] In step S310 of some embodiments, the current initial working temperature of the PEM electrolyzer can be obtained in the following way: Obtain the current anode-side temperature and current cathode-side temperature of the PEM electrolyzer, and then average the current anode-side temperature and current cathode-side temperature of the PEM electrolyzer as the current initial working temperature of the PEM electrolyzer.

[0187] In some embodiments, the implementation manner of the above step S320 may but is not limited to include the following:

[0188] Preset the number of iterative predictions of the PEM electrolyzer model in the current application to be K;

[0189] For the first prediction process, input the current working current and current initial working temperature of the PEM electrolyzer into the voltage prediction model for analysis to obtain the predicted working voltage of the PEM electrolyzer and record it as the first predicted working voltage. Then, input the current working current, current inlet flow rate, current inlet temperature, current ambient temperature, and the first predicted working voltage of the PEM electrolyzer into the temperature prediction model for analysis to obtain the predicted working temperature of the PEM electrolyzer and record it as the first predicted working temperature. Subsequently, use the first predicted working voltage and the first predicted working temperature of the PEM electrolyzer as the first set of predicted operating data of the PEM electrolyzer.

[0190] For the second prediction process, the current working current and the first predicted working temperature of the PEM electrolyzer are input into the voltage prediction model for analysis to obtain the predicted working voltage of the PEM electrolyzer, which is denoted as the second predicted working voltage. Then, the current working current, the current inlet flow rate, the current inlet temperature, the current ambient temperature, and the second predicted working voltage of the PEM electrolyzer are input into the temperature prediction model for analysis to obtain the predicted working temperature of the PEM electrolyzer, which is denoted as the second predicted working temperature. Subsequently, the second predicted working voltage and the second predicted working temperature of the PEM electrolyzer are used as the second set of predicted operating data of the PEM electrolyzer;

[0191] For the third prediction process, the current working current and the second predicted working temperature of the PEM electrolyzer are input into the voltage prediction model for analysis to obtain the predicted working voltage of the PEM electrolyzer, which is denoted as the third predicted working voltage. Then, the current working current, the current inlet flow rate, the current inlet temperature, the current ambient temperature, and the third predicted working voltage of the PEM electrolyzer are input into the temperature prediction model for analysis to obtain the predicted working temperature of the PEM electrolyzer, which is denoted as the third predicted working temperature. Subsequently, the third predicted working voltage and the third predicted working temperature of the PEM electrolyzer are used as the third set of predicted operating data of the PEM electrolyzer;

[0192] And so on. After performing the K - time iterative prediction process, K sets of predicted operating data of the PEM electrolyzer can be obtained, which is convenient for subsequent technicians to further draw the current working characteristic curve of the PEM electrolyzer using these K sets of predicted operating data.

[0193] The PEM electrolyzer performance prediction method provided by the embodiments of the present application analyzes the current operating data of the PEM electrolyzer by using the constructed PEM electrolyzer model to predict the relevant performance parameters of the PEM electrolyzer, which can provide data support for the subsequent task of optimizing the operating parameters of the PEM electrolyzer to further improve the efficiency of electrolytic hydrogen production.

[0194] Figure 4 It is an optional composition schematic diagram of a PEM electrolyzer modeling system provided by the embodiments of the present application, which can implement the above - mentioned PEM electrolyzer modeling method. The modeling system includes:

[0195] A first acquisition module 410, configured to acquire a first data set. The first data set includes a plurality of first data groups obtained when the PEM electrolyzer operates under several different experimental conditions. Each first data group corresponding to an experimental condition includes the working current, the working voltage, and the working temperature of the PEM electrolyzer;

[0196] A training module 420, configured to train a preset neural network model according to the first data set to obtain a voltage prediction model; the input of the voltage prediction model is the working current and working temperature of the PEM electrolyzer, and the output is the working voltage of the PEM electrolyzer;

[0197] A second acquisition module 430, configured to acquire a second data set, where the second data set includes a plurality of second data groups corresponding to a plurality of sampling times obtained when the PEM electrolyzer operates under specific experimental conditions, and each second data group corresponding to a sampling time includes the working current, working voltage, working temperature, inlet flow rate, inlet temperature, and ambient temperature of the PEM electrolyzer;

[0198] A determination module 440, configured to determine a temperature prediction model by means of parameter fitting according to the second data set; the temperature prediction model is used to characterize the functional relationship between the working temperature of the PEM electrolyzer and the working current, working voltage, inlet flow rate, inlet temperature, and ambient temperature;

[0199] An integration module 450, configured to integrate the voltage prediction model and the temperature prediction model to obtain a PEM electrolyzer model.

[0200] It can be understood that the content in the above embodiments of the PEM electrolyzer modeling method is applicable to the embodiments of this modeling system. The functions specifically implemented by the embodiments of this modeling system are the same as those specifically implemented by the above embodiments of the PEM electrolyzer modeling method, and the beneficial effects achieved by the embodiments of this modeling system are also the same as those achieved by the above embodiments of the PEM electrolyzer modeling method.

[0201] Figure 5 It is an optional composition schematic diagram of a PEM electrolyzer performance prediction system provided by an embodiment of this application, which can implement the above-mentioned PEM electrolyzer performance prediction method. The performance prediction system includes:

[0202] A third acquisition module 510, configured to acquire the current operating data of the PEM electrolyzer, where the current operating data of the PEM electrolyzer includes the current working current, current inlet flow rate, current inlet temperature, current ambient temperature, and current initial working temperature of the PEM electrolyzer;

[0203] A processing module 520, configured to input the current operating data of the PEM electrolyzer into the PEM electrolyzer model for iterative processing to obtain multiple groups of predicted operating data of the PEM electrolyzer, and each group of predicted operating data of the PEM electrolyzer includes the predicted working voltage and predicted working temperature of the PEM electrolyzer; wherein, the PEM electrolyzer model is obtained by the above-mentioned PEM electrolyzer modeling method.

[0204] It can be understood that the content in the above embodiments of the PEM electrolyzer performance prediction method is applicable to the embodiments of this performance prediction system. The functions specifically implemented by the embodiments of this performance prediction system are the same as those specifically implemented by the above embodiments of the PEM electrolyzer performance prediction method, and the beneficial effects achieved by the embodiments of this performance prediction system are also the same as those achieved by the above embodiments of the PEM electrolyzer performance prediction method.

[0205] An embodiment of the present application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the above PEM electrolyzer modeling method or the above PEM electrolyzer performance prediction method. The electronic device can include any intelligent terminal such as a tablet computer or an in-vehicle computer.

[0206] It can be understood that for any of the above method embodiments, the content in the method embodiments is applicable to the embodiments of this device. The functions specifically implemented by the embodiments of this device are the same as those specifically implemented by the method embodiments, and the beneficial effects achieved by the embodiments of this device are also the same as those achieved by the method embodiments.

[0207] Please refer to Figure 6 , Figure 6 which shows the hardware structure of an electronic device in another embodiment. The electronic device includes:

[0208] A processor 601, which can be implemented in ways such as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided by the embodiments of the present application;

[0209] A memory 602, which can be implemented in forms such as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 602 can store an operating system and other application programs. When implementing the technical solutions provided by the embodiments of the present application through software or firmware, the relevant program codes are stored in the memory 602 and are called by the processor 601 to execute the technical solutions provided by the embodiments of the present application;

[0210] An input / output interface 603, which is used to implement information input and output;

[0211] A communication interface 604 for implementing communication and interaction between this device and other devices, which can achieve communication through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.);

[0212] A bus 605 for transmitting information between various components of the device (such as a processor 601, a memory 602, an input / output interface 603, and a communication interface 604);

[0213] Among them, the processor 601, the memory 602, the input / output interface 603, and the communication interface 604 achieve communication connections with each other inside the device through the bus 605.

[0214] The embodiments of the present application also provide a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned PEM electrolyzer modeling method or the above-mentioned PEM electrolyzer performance prediction method.

[0215] It can be understood that for any of the above method embodiments, the content in the method embodiments is applicable to the storage medium embodiments. The functions specifically implemented by the storage medium embodiments are the same as those specifically implemented by the method embodiments, and the beneficial effects achieved by the storage medium embodiments are also the same as those achieved by the method embodiments.

[0216] As a non-transitory computer-readable storage medium, the memory can be used to store non-transitory software programs and non-transitory computer-executable programs. In addition, the memory can include high-speed random access memory, and can also include non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory may optionally include a memory remotely set relative to the processor, and these remote memories can be connected to the processor through a network. Examples of the above networks include, but are not limited to, the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0217] The embodiments described in the embodiments of the present application are for more clearly explaining the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. Those skilled in the art know that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0218] Those skilled in the art can understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than those shown, or combine some steps, or different steps.

[0219] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the objectives of the solution of this embodiment.

[0220] Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, and their appropriate combinations.

[0221] As used in the specification of this application and the above-mentioned drawings, the terms "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of this application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily limit to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0222] It should be understood that in this application, "at least one (item)" means one or more, and "a plurality" means two or more. "And / or" is used to describe the association relationship of associated objects and indicates that three relationships can exist. For example, "A and / or B" can mean: only A exists, only B exists, and both A and B exist at the same time. Among them, A and B can be singular or plural. The character " / " generally means that the associated objects before and after are in an "or" relationship. "At least one (one)" or a similar expression thereof refers to any combination of these items, including any combination of single item (one) or plural items (ones). For example, at least one (one) of a, b, or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0223] In several embodiments provided by the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of the above-mentioned units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of systems or units can be in electrical, mechanical or other forms.

[0224] The units described above as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0225] In addition, each functional unit in various embodiments of the present application 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-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0226] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in various embodiments of the present application. And the aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM for short), random access memory (RAM for short), magnetic disks or optical discs that can store programs.

[0227] The preferred embodiments of the embodiments of the present application have been described above with reference to the accompanying drawings. However, this does not limit the scope of the rights of the embodiments of the present application. Any modifications, equivalent replacements, and improvements made by those skilled in the art without departing from the scope and essence of the embodiments of the present application shall fall within the scope of the rights of the embodiments of the present application.

Claims

1. A PEM electrolyzer modeling method, characterized in that: include: Acquire a first data set, the first data set comprising a plurality of first data groups correspondingly acquired when the PEM electrolyzer is operated under a plurality of different experimental conditions, the first data group corresponding to each of the experimental conditions comprising an operating current, an operating voltage and an operating temperature of the PEM electrolyzer; According to the first data set, a preset neural network model is trained to obtain a voltage prediction model; the input of the voltage prediction model is the operating current and operating temperature of the PEM electrolyzer, and the output is the operating voltage of the PEM electrolyzer; Acquire a second data set, the second data set comprising a plurality of second data groups corresponding to a plurality of sampling times acquired when the PEM electrolyzer is operated under specific experimental conditions, the second data group corresponding to each sampling time comprising an operating current, an operating voltage, an operating temperature, an inlet flow rate, an inlet temperature and an ambient temperature of the PEM electrolyzer; According to the second data set, a temperature prediction model is determined by parameter fitting; the temperature prediction model is used to characterize the functional relationship between the operating temperature of the PEM electrolyzer and the operating current, operating voltage, inlet flow rate, inlet temperature and ambient temperature; The voltage prediction model and the temperature prediction model are integrated to obtain a PEM electrolyzer model.

2. The PEM electrolyzer modeling method according to claim 1, characterized in that: The first data set is obtained by: According to a preset temperature increment, taking values ​​within a preset water supply temperature range, obtaining multiple water supply temperatures and forming a first test set; According to a preset current increment, a value is taken within a preset power supply current range to obtain a plurality of power supply currents and form a second test set; Starting from i=1, obtaining the i-th water supply temperature included in the first test set; According to the i-th water supply temperature, controlling the heating device to heat the ultrapure water in the water tank, and then controlling the water pump to inject the heated ultrapure water in the water tank into the anode side of the PEM electrolyzer at a specific flow rate until the internal temperature of the PEM electrolyzer is stabilized at the i-th water supply temperature; Starting from j=1, obtaining the jth power supply current included in the second test set; Controlling a direct current power supply to provide a jth power supply current to the PEM electrolyzer, and then obtaining a first data set corresponding to a current experimental condition after the PEM electrolyzer runs for a specific period of time; Determining whether the current operating state of the PEM electrolyzer is reasonable according to the first data group corresponding to the current experimental condition; If the current operating state of the PEM electrolyzer is unreasonable, determine whether i<N holds, where N is the number of the multiple water supply temperatures; If i<N holds, assign i+1 to i, and return to the step of obtaining the i-th water supply temperature included in the first test set; If i<N does not hold, determining the first data set according to a plurality of first data sets corresponding to a plurality of different experimental conditions obtained; If the current operating state of the PEM electrolyzer is reasonable, then determining whether j<M holds, where M is the number of the multiple supply currents; If j<M holds, assign j+1 to j, and return to the step of obtaining the jth power supply current included in the second test set; If j<M is not true, then return to the step of determining whether i<N is true.

3. The PEM electrolyzer modeling method according to claim 1, characterized in that: The preset neural network model includes an input layer, a hidden layer and an output layer connected in sequence; the voltage prediction model obtained by training the preset neural network model according to the first data set includes: According to a plurality of preset hyperparameters, each of which includes the number of neuron nodes in the hidden layer, the preset neural network model is adjusted to obtain a plurality of corresponding first neural network models; According to the first data set, the plurality of first neural network models are trained respectively, and then the trained first neural network model with the best prediction performance is selected and recorded as the second neural network model; According to a plurality of preset activation function combinations, each of which includes an activation function between the input layer and the hidden layer and an activation function between the hidden layer and the output layer, the second neural network model is adjusted to obtain a plurality of corresponding third neural network models; According to the first data set, the plurality of third neural network models are trained respectively, and then the trained third neural network model with the best prediction performance is selected and recorded as the fourth neural network model; The fourth neural network model is retrained according to the first data set to obtain the voltage prediction model.

4. The PEM electrolyzer modeling method according to claim 1, characterized in that: The second data set is obtained by: According to a specific water supply temperature, controlling a heating device to heat the ultrapure water in a water tank, then controlling a water pump to inject the heated ultrapure water in the water tank into the anode side of the PEM electrolyzer at a specific flow rate, and then controlling a DC power supply to provide a specific power supply current to the PEM electrolyzer; According to a preset sampling time interval, a plurality of second data groups corresponding to a plurality of sampling times are acquired within a preset sampling time period to determine the second data set.

5. The PEM electrolyzer modeling method according to claim 1, characterized in that: Determining the temperature prediction model by parameter fitting according to the second data set includes: Obtaining a preset mathematical expression corresponding to the temperature prediction model, wherein the preset mathematical expression includes a plurality of unknown parameters, wherein the plurality of unknown parameters include a lumped heat capacity, a thermal resistance, and a heat correction coefficient of the PEM electrolyzer; According to the second data set and the preset mathematical expression, the multiple unknown parameters are fitted and solved to obtain values ​​of the multiple unknown parameters; The values ​​of the multiple unknown parameters are substituted into the preset mathematical expression to determine the temperature prediction model.

6. A method for predicting the performance of a PEM electrolyzer, characterized in that: include: Acquire current operating data of the PEM electrolyzer, wherein the current operating data of the PEM electrolyzer includes a current operating current, a current inlet flow rate, a current inlet temperature, a current ambient temperature, and a current initial operating temperature of the PEM electrolyzer; Inputting the current operating data of the PEM electrolyzer into the PEM electrolyzer model for iterative processing to obtain multiple groups of predicted operating data of the PEM electrolyzer, each group of predicted operating data of the PEM electrolyzer including a predicted operating voltage and a predicted operating temperature of the PEM electrolyzer; Wherein, the PEM electrolyzer model is obtained by the PEM electrolyzer modeling method according to any one of claims 1 to 5.

7. A PEM electrolyzer modeling system, characterized in that: include: A first acquisition module is used to acquire a first data set, wherein the first data set includes a plurality of first data groups correspondingly acquired when the PEM electrolyzer is operated under a plurality of different experimental conditions, wherein the first data group corresponding to each of the experimental conditions includes an operating current, an operating voltage and an operating temperature of the PEM electrolyzer; A training module, used to train a preset neural network model according to the first data set to obtain a voltage prediction model; the input of the voltage prediction model is the operating current and operating temperature of the PEM electrolyzer, and the output is the operating voltage of the PEM electrolyzer; A second acquisition module is used to acquire a second data set, wherein the second data set includes a plurality of second data groups corresponding to a plurality of sampling times acquired when the PEM electrolyzer is operated under specific experimental conditions, and each second data group corresponding to the sampling time includes an operating current, an operating voltage, an operating temperature, an inlet flow rate, an inlet temperature and an ambient temperature of the PEM electrolyzer; A determination module, used to determine a temperature prediction model by parameter fitting according to the second data set; the temperature prediction model is used to characterize the functional relationship between the operating temperature of the PEM electrolyzer and the operating current, operating voltage, inlet flow rate, inlet temperature and ambient temperature; The integration module is used to integrate the voltage prediction model and the temperature prediction model to obtain a PEM electrolyzer model.

8. A PEM electrolyzer performance prediction system, characterized in that: include: A third acquisition module is used to acquire current operating data of the PEM electrolyzer, wherein the current operating data of the PEM electrolyzer includes a current operating current, a current inlet flow rate, a current inlet temperature, a current ambient temperature and a current initial operating temperature of the PEM electrolyzer; a processing module, configured to input the current operation data of the PEM electrolyzer into a PEM electrolyzer model for iterative processing to obtain multiple groups of predicted operation data of the PEM electrolyzer, each group of predicted operation data of the PEM electrolyzer comprising a predicted operating voltage and a predicted operating temperature of the PEM electrolyzer; Wherein, the PEM electrolyzer model is obtained by the PEM electrolyzer modeling method according to any one of claims 1 to 5.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the PEM electrolyzer modeling method according to any one of claims 1 to 5 or the PEM electrolyzer performance prediction method according to claim 6 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the PEM electrolyzer modeling method according to any one of claims 1 to 5 or the PEM electrolyzer performance prediction method according to claim 6 is implemented.

Images