A performance prediction method and system for a gasification integrated parallel three-electrode stack SOFC system based on GA-BP

Abstract

The application discloses a kind of based on GA-BP's gasification integrated parallel three electric pile SOFC system's performance prediction method and system, it is related to system performance prediction technical field, using the method of lumped modeling is modeled to gasification integrated parallel three electric pile SOFC system, the data set of gasification integrated parallel three electric pile SOFC system is used for training GA-BP model.According to this model, the electric efficiency and net voltage of each electric pile in gasification integrated parallel three electric pile SOFC system can be predicted, and the design parameters and operating conditions of the system are optimized.The application combines genetic algorithm with BP neural network to predict the performance parameters of gasification integrated parallel three electric pile SOFC system, solves the problem that traditional BP neural network often encounters local optimum, helps to carry out more extensive global search, finds out the optimal parameter combination, and improves the accuracy of system performance parameters predicted by the model.

Description

Technical Field

[0001] This invention relates to the field of performance prediction technology for gasification integrated parallel multi-stack SOFC systems, and more specifically to a performance prediction method and system for a gasification integrated parallel three-stack SOFC system based on GA-BP. Background Technology

[0002] Currently, to address the increasingly severe global energy challenges and environmental problems, biomass energy has emerged as a promising solution due to its renewability and low environmental impact. Biomass resources participate in the carbon cycle during conversion, playing a crucial role in mitigating climate change. Notably, compared to traditional fossil fuels, biomass gasification technology can convert biomass into syngas, a highly efficient energy carrier, thereby improving energy efficiency and significantly reducing environmental pollution. In particular, steam gasification technology can produce hydrogen-rich syngas, which is essential for SOFCs (Solar Oriented Fuel Cells).

[0003] SOFCs are considered a cutting-edge technology for clean energy conversion due to their high efficiency and ability to use a variety of fuels. By combining biomass gasification with SOFC technology, carbon emissions can be reduced and energy efficiency improved during the conversion of biomass into electricity. The high output of SOFCs makes them an ideal downstream application for gasification technology. However, to fully realize the potential of this integrated system, it is necessary to accurately model and predict the performance of the gasification-integrated SOFC system to optimize its design parameters and operating conditions.

[0004] Extensive research has been conducted on data-driven predictive modeling of SOFC systems. For example, based on SOFC degradation models, researchers have applied particle filtering algorithms to predict the long-term degradation trend of SOFCs, thereby assessing the system's lifespan and reliability. Simultaneously, other researchers have applied the Levenberg-Marquardt backpropagation algorithm to identify unknown parameters of SOFC systems, which is significant for improving system modeling accuracy and optimizing system performance. While these studies have made outstanding contributions to the development of SOFC system performance prediction and enhanced our understanding of system degradation and parameter identification, there is currently no literature analyzing and researching gasification integrated parallel multi-reactor systems. Furthermore, research in the field of SOFC system operating condition prediction remains insufficient. The bottleneck in operating condition prediction mainly lies in its tendency to get trapped in local optima, limiting the model's generalization ability and prediction accuracy.

[0005] Therefore, how to combine appropriate optimization algorithms to improve the accuracy and reliability of predictions, thereby promoting the widespread application and optimization of SOFC systems in practical applications, is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] In view of this, the present invention provides a performance prediction method and system for a gasification integrated parallel three-stack SOFC system based on GA-BP, in order to solve the problems existing in the background art.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] On the one hand, a performance prediction method for a gasification integrated parallel three-stack SOFC system based on GA-BP is provided, including:

[0009] Step 1: Model the gasification integrated parallel three-stack SOFC system to obtain the gasification integrated parallel three-stack SOFC system;

[0010] Step 2: The fuel obtained through biomass gas reforming is fed into a gasification integrated parallel three-pile SOFC system. The effects of system operating conditions on syngas composition and system power generation performance are recorded to obtain a training dataset.

[0011] Step 3: Design the structure of a BP neural network based on the training dataset generated by the operation of the gasification integrated parallel three-stack SOFC system;

[0012] Step 4: Using the system operation parameters in the training dataset as input and the stack voltage, electrical efficiency, and current density as output, train the system and use a genetic algorithm to optimize the initial weights and thresholds of the BP neural network to obtain a performance prediction model for a gasification integrated parallel three-stack SOFC system based on GA-BP.

[0013] Step 5: Obtain the actual test dataset collected from the gasification integrated parallel three-stack SOFC system. Use the system operation parameters in the actual test dataset as input parameters and input them into the prediction model of the gasification integrated parallel three-stack SOFC system based on GA-BP to obtain the predicted values ​​of stack voltage, electrical efficiency and current density.

[0014] Optionally, the system operating parameters in the training dataset include gasification temperature, the ratio of water vapor mass flow rate to biomass mass flow rate, and SOFC anode operating temperature.

[0015] Optionally, the BP neural network structure has the following configuration: 7 neurons in the input layer, 5 neurons in the hidden layer, and 1 neuron in the output layer.

[0016] Optionally, the system operating parameters in the actual test dataset are the mass content of O, C and H in biomass, the ratio of water vapor mass flow rate to biomass mass flow rate, gasification temperature, SOFC anode operating temperature and biomass supply flow rate.

[0017] Optionally, in step 4, the step of optimizing the initial weights and thresholds of the BP neural network using a genetic algorithm is as follows:

[0018] Step 4.1: Generate N initial strings as the initial population using a random method. Each initial string is called an individual. The individual is composed of the input layer weights, input layer thresholds, output layer weights, and output layer thresholds of the gasification integrated parallel three-electric stack SOFC performance prediction model based on BP network.

[0019] Step 4.2: Calculate the fitness value Z of each individual in the i-th generation population according to the fitness function;

[0020] Step 4.3, Selection Operation: Select subpopulations using a probabilistic selection method based on the magnitude of individual errors;

[0021] Step 4.4, Crossover operation: Select two parent individuals and exchange their gene loci according to the crossover probability to generate new offspring;

[0022] Step 4.5, Mutation Operation: Using the mutation operator with mutation probability, select individuals from the population to perform the mutation operation;

[0023] Step 4.6: Determine whether the genetic generation meets the termination condition. If it does, stop the operation and assign the initial weights and thresholds of the network corresponding to the optimal individual to the BP neural network; otherwise, return to step 4.2.

[0024] Optionally, in step 4, after obtaining the performance prediction model of the GA-BP-based gasification integrated parallel three-stack SOFC system, a model evaluation is also performed on the performance prediction model of the GA-BP-based gasification integrated parallel three-stack SOFC system. The evaluation indicators used are: root mean square error (MSE), root mean square error (RMSE), mean absolute percentage error (MAPE), and correlation coefficient R. 2 .

[0025] Optionally, the fitness value of the individual is defined as the sum of the absolute values ​​of the errors between the actual output and the expected output of the performance prediction model of the gasification integrated parallel three-stack SOFC system.

[0026] Optionally, the crossover operation is:

[0027] No. k Chromosomes and the l Chromosomes exist j The bit crossover operation method is as follows:

[0028] ;

[0029] In the formula, h is a random number between [0, 1].

[0030] Optionally, the mutation operation is as follows:

[0031] ;

[0032] In the formula, and Genes The upper and lower bounds; This represents the current iteration number; The maximum number of evolutions is given; r is a random number between [0, 1]. It is a random number.

[0033] On the other hand, a performance prediction system for a gasification integrated parallel three-stack SOFC system based on GA-BP is provided, including:

[0034] The system modeling module models the gasification integrated parallel three-fuel cell stack SOFC system, and obtains the gasification integrated parallel three-fuel cell stack SOFC system.

[0035] The dataset acquisition module feeds fuel obtained through biomass gas reforming into a gasification integrated parallel three-pile SOFC system, records the impact of system operating conditions on syngas composition and system power generation performance, and obtains a training dataset.

[0036] The neural network structure design module designs the structure of a BP neural network based on the training dataset generated by the operation of the gasification integrated parallel three-stack SOFC system.

[0037] The model training and optimization module takes the system operation parameters in the training dataset as input and the stack voltage, electrical efficiency, and current density as output for training. It uses a genetic algorithm to optimize the initial weights and thresholds of the BP neural network to obtain a performance prediction model for a gasification integrated parallel three-stack SOFC system based on GA-BP.

[0038] The system performance prediction module acquires the actual test dataset collected from the gasification integrated parallel three-stack SOFC system, and uses the system operation parameters in the actual test dataset as input parameters to input the prediction model of the gasification integrated parallel three-stack SOFC system based on GA-BP to obtain the predicted values ​​of stack voltage, electrical efficiency and current density.

[0039] As can be seen from the above technical solution, compared with the prior art, this invention discloses a performance prediction method and system for a gasification integrated parallel three-stack SOFC system based on GA-BP. It employs a lumped modeling method to model the gasification integrated parallel three-stack SOFC system, and then uses the dataset of the gasification integrated parallel three-stack SOFC system to train the GA-BP model. Based on this model, the performance parameters such as the electrical efficiency and net voltage of each stack in the gasification integrated parallel three-stack SOFC system can be predicted, thereby optimizing the system's design parameters and operating conditions. This invention uses the mass content of O, C, and H in the biomass of the gasification integrated parallel three-stack SOFC system, the S / B ratio, the gasification temperature, the SOFC anode operating temperature, and the biomass supply flow rate as input data to the model, and uses stack voltage, electrical efficiency, and current density as outputs. When training the model, it uses a genetic algorithm known for its global optimization capabilities, solving the problem of traditional BP neural networks often getting trapped in local optima. The GA-BP method in this invention combines genetic algorithms with BP neural networks, which helps to conduct a broader global search, find more optimal parameter combinations, and improve the prediction accuracy of the model. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0041] Figure 1 A schematic diagram illustrating the steps of the present invention's method for predicting the performance of a gasification integrated parallel three-stack SOFC system.

[0042] Figure 2 The flowchart of the genetic algorithm for optimizing the BP neural network model provided by this invention;

[0043] Figure 3a , Figure 3b , Figure 3c This is a comparison chart of the electrical efficiency, current density, and stack voltage curves obtained by the performance analysis of the training and test datasets of stack 1 based on MSE in this invention.

[0044] Figure 4a , Figure 4b , Figure 4c This is a comparison chart of the electrical efficiency, current density, and stack voltage curves obtained by the performance analysis of the training and test datasets of stack 2 based on MSE in this invention.

[0045] Figure 5a , Figure 5b , Figure 5c This is a comparison chart of the electrical efficiency, current density, and stack voltage curves obtained by the performance analysis of the training and test datasets of stack 3 based on MSE in this invention.

[0046] Figure 6a , Figure 6b , Figure 6c To achieve the present invention, the performance analysis of the fuel cell stack 1 based on the training and test datasets is performed to obtain error histograms for electrical efficiency, fuel cell stack voltage, and current density, respectively.

[0047] Figure 7a , Figure 7b , Figure 7c To achieve the present invention, the performance analysis of the fuel cell stack 2 based on the training and test datasets is performed to obtain error histograms for electrical efficiency, fuel cell stack voltage, and current density, respectively.

[0048] Figure 8a , Figure 8b , Figure 8c Based on the training and testing datasets, this invention performs performance analysis on fuel cell stack 3 to obtain error histograms for electrical efficiency, fuel cell voltage, and current density. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] One embodiment of the present invention discloses a performance prediction method for a gasification integrated parallel three-stack SOFC system based on GA-BP, such as... Figure 1 As shown, it includes:

[0051] Step 1: Model the gasification integrated parallel three-stack SOFC system using the lumped modeling method to obtain the gasification integrated parallel three-stack SOFC system;

[0052] Step 2: Fifteen kinds of fuels obtained from biomass gas reforming were introduced into a three-stack SOFC system. The effects of system operating conditions on syngas composition and the power generation performance of the multi-stack SOFC system were recorded. The system operating parameters included gasification temperature, water vapor mass flow rate to biomass mass flow rate ratio (S / B), and stack anode reaction temperature, thus obtaining a training dataset.

[0053] Step 3: Based on the training dataset generated by the operation of the gasification integrated parallel three-cell SOFC system, design the structure of the BP neural network; wherein, the designed BP neural network structure is: 7 neurons in the input layer, 5 neurons in the hidden layer, and 1 neuron in the output layer.

[0054] Step 4: Using the system operation parameters in the training dataset as input and the stack voltage, electrical efficiency, and current density as output, train the system and use a genetic algorithm to optimize the initial weights and thresholds of the BP neural network to obtain a performance prediction model for a gasification integrated parallel three-stack SOFC system based on GA-BP.

[0055] Among them, the genetic algorithm is used, see [link / reference]. Figure 2 The specific steps for optimizing the initial weights and thresholds of a BP neural network are as follows:

[0056] Step 4.1, Population Initialization: Since genetic algorithms cannot directly process the problem parameter space, it is necessary to pre-encode the parameters to generate an initial population solution. Here, to effectively measure the efficiency of the genetic algorithm, a real-number encoding mechanism is used to connect the network parameters. That is, each individual is composed of the input layer weights, input layer thresholds, output layer weights, and output layer thresholds of the BP-based gasification integrated parallel three-fuel-stacking SOFC performance prediction model. The formula for calculating the real-number encoding length is:

[0057] ;

[0058] In the formula, y is the encoding length; a b1 represents the number of input layer nodes in the BP neural network; b1 represents the number of hidden layer nodes; and c represents the number of output layer nodes.

[0059] Step 4.2: The fitness function is usually used as a metric to evaluate the quality of an individual. Here, the sum of the absolute values ​​of the errors between the actual output and the expected output of the performance prediction model of the gasification integrated parallel three-stack SOFC system is defined as the individual fitness. The smaller the error, the better the individual. The fitness function formula is as follows:

[0060] ;

[0061] In the formula, Z is the fitness value; n is the number of training samples; and These are the first BP neural networks. i The expected output value and the actual output value of each node.

[0062] Step 4.3, the selection operation, employs a probabilistic selection method based on the magnitude of individual error. The aim is to increase the probability of selecting individuals with smaller errors. As the fitness function shows, a smaller fitness value corresponds to a smaller error; therefore, the reciprocal of the fitness value is taken. The formula is as follows:

[0063] ;

[0064] In the formula, For the first i The fitness value of each individual; k is a coefficient; denoted as ... , where is the probability of an individual being selected; and N is the number of individuals in the population.

[0065] Step 4.4: Crossover operation. To achieve population diversity and global search capability, a crossover operation is performed on the population. Two parent individuals are selected, and their gene loci are exchanged and combined according to the crossover probability to generate new offspring. Since individuals use real-number encoding, the real-number crossover method is chosen. (The k-th chromosome...) and the lth chromosome exist j The bit crossover operation method is as follows:

[0066] No. k Chromosomes and the l Chromosomes exist j The bit crossover operation method is as follows:

[0067] ;

[0068] In the formula, h is a random number between [0, 1].

[0069] Step 4.5: Mutation operation, which involves changes at certain gene loci to enrich gene diversity. Select the i-th individual's... j One gene The mutation process is as follows:

[0070] ;

[0071] In the formula, and Genes The upper and lower bounds; This represents the current iteration number; The maximum number of evolutions is given; r is a random number between [0, 1]. It is a random number.

[0072] When constructing the GA-BP neural network, all relevant parameters, except for those specific to the genetic algorithm, are consistent with those of the BP neural network. The parameters of the GA-BP neural network are shown in Table 1.

[0073] Table 1 Parameters of GA-BP Neural Network

[0074]

[0075] Step 5: Obtain the actual test dataset collected from the gasification integrated parallel three-stack SOFC system. Use the system operation parameters in the actual test dataset as input parameters and input them into the prediction model of the gasification integrated parallel three-stack SOFC system based on GA-BP to obtain the predicted values ​​of stack voltage, electrical efficiency and current density.

[0076] In step 3 of this embodiment, a total of 21,870 sets of operational data for the gasification integrated parallel three-stack SOFC system were collected. The variables included were the mass content of O, C, and H in biomass, the S / B ratio, gasification temperature, SOFC anode operating temperature, biomass supply flow rate, voltage, current, and electrical efficiency. 70% of the 21,870 sets of data were allocated as a training dataset, and 30% as a test dataset.

[0077] The training dataset is used to train the BP neural network in step 4 above. During training, the input values ​​are the mass content of O, C and H in biomass, S / B ratio, gasification temperature, SOFC anode operating temperature and biomass supply flow rate in the first 15309 sets of data, and the output values ​​are the voltage, current density and electrical efficiency of the first 15309 sets of data.

[0078] The test dataset was used to test and evaluate the generated GA-BP solid oxide fuel cell voltage prediction model. During testing, the input values ​​were the mass content of O, C, and H in the biomass, the S / B ratio, the gasification temperature, the SOFC anode operating temperature, and the biomass supply flow rate from the last 6561 data sets. The output values ​​were the voltage, current density, and electrical efficiency from the last 6561 data sets.

[0079] Through the above training and testing process, the following test results were obtained:

[0080] Figure 3a , Figure 3b , Figure 3c The graphs show a comparison of the electrical efficiency, current density, and stack voltage curves obtained from the performance analysis of the training and testing datasets of stack 1 based on MSE.

[0081] Figure 4a , Figure 4b , Figure 4c The graphs show a comparison of the electrical efficiency, current density, and stack voltage curves obtained from the performance analysis of the training and testing datasets of stack 2 based on MSE.

[0082] Figure 5a , Figure 5b , Figure 5c The graphs show a comparison of the electrical efficiency, current density, and stack voltage curves obtained from the performance analysis of the training and testing datasets of stack 3 based on MSE.

[0083] Figure 6a , Figure 6b , Figure 6c To perform performance analysis on fuel cell stack 1 based on training and testing datasets, error histograms for electrical efficiency, fuel cell voltage, and current density were obtained.

[0084] Figure 7a , Figure 7b , Figure 7c To perform performance analysis on fuel cell stack 2 based on the training and testing datasets, error histograms for electrical efficiency, fuel cell voltage, and current density were obtained.

[0085] Figure 8a , Figure 8b , Figure 8c To perform performance analysis on fuel cell stack 3 based on the training and testing datasets, error histograms for electrical efficiency, fuel cell voltage, and current density were obtained.

[0086] Based on the test results, the prediction performance of the GA-BP-based gasification integrated parallel three-stack SOFC system was evaluated. The evaluation parameters included: mean square error (MSE), root mean square error (RMSE), mean absolute percentage error (MAPE), and correlation coefficient R. 2 Among them, MSE can reveal the deviation between the model's performance predictions and actual values. RMSE and MAPE are used to verify the predictive capability of the GA-BP model for gasification integrated parallel three-stack SOFC. 2 Typically used in regression models to assess the consistency between predicted and actual values, MSE, RMSE, MAPE, and R... 2 The evaluation is defined as follows:

[0087] ;

[0088] in, Represents the predicted value. The calculated value represents the model. m Represents the size of the data. and These represent the estimated value and the target value of the GA-BP model, respectively. This represents the average value of the target output.

[0089] On the other hand, a performance prediction system for a gasification integrated parallel three-stack SOFC system based on GA-BP is provided, including:

[0090] The system modeling module models the gasification integrated parallel three-fuel cell stack SOFC system, and obtains the gasification integrated parallel three-fuel cell stack SOFC system.

[0091] The dataset acquisition module feeds fuel obtained through biomass gas reforming into a gasification integrated parallel three-pile SOFC system, records the impact of system operating conditions on syngas composition and system power generation performance, and obtains a training dataset.

[0092] The neural network structure design module designs the structure of a BP neural network based on the training dataset generated by the operation of the gasification integrated parallel three-stack SOFC system.

[0093] The model training and optimization module takes the system operation parameters in the training dataset as input and the stack voltage, electrical efficiency, and current density as output for training. It uses a genetic algorithm to optimize the initial weights and thresholds of the BP neural network to obtain a performance prediction model for a gasification integrated parallel three-stack SOFC system based on GA-BP.

[0094] The system performance prediction module acquires the actual test dataset collected from the gasification integrated parallel three-stack SOFC system, and uses the system operation parameters in the actual test dataset as input parameters to input the prediction model of the gasification integrated parallel three-stack SOFC system based on GA-BP to obtain the predicted values ​​of stack voltage, electrical efficiency and current density.

[0095] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0096] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for performance prediction of a GA-BP based gasification integrated parallel three-stack SOFC system, characterized in that, The method comprises the following steps: Step 1, modeling a gasification integrated parallel three-stack SOFC system to obtain a gasification integrated parallel three-stack SOFC system; Step 2, feeding fuel obtained by reforming biomass gas into the gasification integrated parallel three-stack SOFC system, recording the influence of system operating conditions on the composition of the synthesis gas and the power generation performance of the system, and obtaining a training data set; Step 3, designing the structure of a BP neural network according to the training data set generated by the operation of the gasification integrated parallel three-stack SOFC system; Step 4, taking the system operating parameters in the training data set as inputs, and taking the stack voltage, electrical efficiency and current density as outputs, training, using a genetic algorithm to optimize the initial weights and thresholds of the BP neural network, and obtaining a GA-BP-based performance prediction model of the gasification integrated parallel three-stack SOFC system; Step 5, obtaining an actual test data set collected from the gasification integrated parallel three-stack SOFC system, taking the system operating parameters in the actual test data set as input parameters, inputting the actual test data set into the GA-BP-based performance prediction model of the gasification integrated parallel three-stack SOFC system, and obtaining predicted values of the stack voltage, electrical efficiency and current density; The system operating parameters in the actual test data set are the mass content of O, C and H in the biomass, the ratio of the steam mass flow rate to the biomass mass flow rate, the gasification temperature, the SOFC anode operating temperature and the biomass supply flow rate; In step 4, the step of using a genetic algorithm to optimize the initial weights and thresholds of the BP neural network is as follows: Step 4.1, generating N initial strings as an initial population by a random method, each initial string being referred to as an individual, and the individual being composed of the input layer weights, input layer thresholds, output layer weights and output layer thresholds of the GA-BP-based performance prediction model of the gasification integrated parallel three-stack SOFC system; Step 4.2, calculating the fitness value Z of each individual in the i-th generation population according to a fitness function; Step 4.3, selection operation: selecting a sub-population by a method of probabilistic selection according to the individual error size; Step 4.4, crossover operation: selecting two parent individuals, exchanging and combining the gene sites of the two parent individuals according to a crossover probability to generate new offspring; Step 4.5, mutation operation: selecting an individual from the population to perform a mutation operation by using a mutation operator with a mutation probability; Step 4.6, judging whether the number of generations satisfies a termination condition, if yes, stopping the operation, and assigning the network initial weights and thresholds corresponding to the optimal individual to the BP neural network; otherwise, returning to step 4.2; The fitness function formula is as follows: ； In the formula, Z is a fitness value; n is the number of training samples; and are respectively the expected output value and the actual output value of the BP neural network node i . The real number coding mechanism is used to connect the network parameters, and the real number coding length calculation formula is as follows: ； In the formula, y is the code length; is the number of input layer nodes set in the BP neural network; b1 is the number of hidden layer nodes; and c is the number of output layer nodes.

2. The performance prediction method of a GA-BP-based gasification integrated parallel three-stack SOFC system according to claim 1, characterized in that, The system operating parameters in the training data set include the gasification temperature, the ratio of the steam mass flow rate to the biomass mass flow rate and the SOFC anode operating temperature.

3. The method for performance prediction of a GA-BP based gasification integrated parallel three-stack SOFC system according to claim 1, characterized in that, The crossover operation is as follows: The first chromosome k and the second chromosome and the third chromosome l and the fourth chromosome The cross operation method at the position of j is as follows: ； In the formula, h is a random number in the interval [0, 1].

4. The method for performance prediction of a GA-BP based gasification integrated parallel three-stack SOFC system according to claim 3, characterized in that, The mutation operation is as follows: ； wherein and are the upper and lower bounds of the gene respectively; is the current iteration number; is the maximum number of evolutions; r is a random number in [0, 1]; is a random number.

5. A performance prediction system of a GA-BP based gasification integrated parallel three stack SOFC system, characterized in that, The application of the GA-BP-based performance prediction method of the gasification integrated parallel three-stack SOFC system according to any one of claims 1-4 comprises: The system modeling module models the gasification integrated parallel three SOFC stacks system to obtain the gasification integrated parallel three SOFC stacks system; The data set acquisition module introduces the fuel obtained through the biomass gas reforming into the gasification integrated parallel three SOFC stacks system, records the influence of the system operating conditions on the syngas components and the system power generation performance, and obtains a training data set; The neural network structure design module designs the structure of the BP neural network according to the training data set generated by the operation of the gasification integrated parallel three SOFC stacks system; The model training and optimization module takes the system operating parameters in the training data set as the input, takes the stack voltage, the electric efficiency and the current density as the output, performs training, uses the genetic algorithm to optimize the initial weight and threshold of the BP neural network, and obtains a GA-BP-based gasification integrated parallel three SOFC stacks system performance prediction model; The system performance prediction module obtains actual test data sets collected from the gasification integrated parallel three SOFC stacks system, takes the system operating parameters in the actual test data sets as the input parameters, inputs the actual test data sets into the GA-BP-based gasification integrated parallel three SOFC stacks system prediction model, and obtains the predicted values of the stack voltage, the electric efficiency and the current density.

Images