Solid-liquid rocket engine performance prediction method and system based on PSO-RBF neural network

Abstract

The invention belongs to the technical field of neural networks, and relates to a solid-liquid rocket engine performance prediction method and system based on a PSO-RBF neural network, and the method comprises the steps: obtaining the structural feature data and working condition feature data of a solid-liquid rocket engine, and obtaining engine performance parameters through a numerical simulation model; hyper-parameters of the PSO-RBF neural network model are trained based on the training set and a k-fold cross validation method, the PSO-RBF neural network model is tested through the test set, and performance parameters of the solid-liquid rocket engine are obtained through structural characteristics and working condition characteristic data of the engine with unknown performance. According to the method, the hyper-parameters of the radial basis function RBF neural network are optimized through the particle swarm optimization algorithm PSO, the optimal network structure and parameter configuration can be automatically searched, the engine performance is predicted through numerical simulation and a machine learning model, and the research and development efficiency is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of neural network and relates to a method and system for predicting the performance of a solid-liquid rocket engine based on a PSO-RBF neural network. Background Art

[0002] Rocket engines are the "heart" of rockets and spacecraft, and their development level indicates the scale and strength of a country's space activities. Solid-liquid rocket engines use solid fuel and liquid / gas oxidizer as propellants. Compared with liquid rocket engines, solid-liquid rocket engines have simpler structures, lower costs, larger thrust adjustment ratios and are easier to implement. Compared with solid rocket engines, solid-liquid rocket engines have higher specific impulse, better safety, and can achieve multiple starts and stops and thrust adjustment. Solid-liquid rocket engines have shown broad application prospects in the fields of sounding rockets, target missiles, upper stages, and attitude and orbit control engines, and have received widespread attention at home and abroad.

[0003] In the design process of solid-liquid rocket engines, in order to achieve the predetermined performance design indicators, it is necessary to iteratively improve the design scheme through performance prediction. Commonly used performance prediction methods include interior ballistic calculation, numerical simulation and engine ground hot test.

[0004] With the widespread application of high-precision numerical simulation methods, the level of refinement in the design of solid-liquid rocket engines has been greatly improved. By performing high-precision numerical simulation calculations on a variety of different working conditions, the credibility of the design results can be effectively improved. However, although the computational fluid dynamics method CFD can obtain high-precision engine performance solutions, it often requires the use of high-quality grids for calculations, resulting in several hours or even days to solve a case, and even longer time for cases with poor convergence. Moreover, for engines with the same design indicators, there are different aspect ratios, and multiple cases need to be calculated, which poses new challenges to the design efficiency of solid-liquid rocket engines. Summary of the invention

[0005] The purpose of the present invention is to solve the problem that the computational fluid dynamics method CFD in the prior art takes a long time to solve the high-precision engine performance and has low design efficiency of solid-liquid rocket engines, and to provide a solid-liquid rocket engine performance prediction method and system based on a PSO-RBF neural network.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] The performance prediction method of solid-liquid rocket engine based on PSO-RBF neural network includes:

[0008] Obtain the structural characteristic data and operating condition characteristic data of solid-liquid rocket engines, and calculate the corresponding engine performance parameters through numerical simulation models;

[0009] Construct a performance prediction dataset for a solid-liquid rocket engine and standardize the constructed performance prediction dataset;

[0010] Divide the standardized dataset to obtain a training set and a test set;

[0011] Based on the training set and the k-fold cross-validation method, train the hyperparameters of the PSO-RBF neural network model, and test the PSO-RBF neural network model through the test set to obtain the performance prediction results of the test data and verify the accuracy of the PSO-RBF neural network model;

[0012] Input the structural characteristics and operating condition characteristic data of the engine with unknown performance into the trained PSO-RBF neural network model to obtain the performance parameters of the solid-liquid rocket engine.

[0013] A further improvement of the present invention lies in:

[0014] Further, the structural characteristic parameters include the inner diameter D of the solid grain channel, the length L of the solid grain, and the nozzle throat diameter d; the operating condition characteristic data includes: oxidizer flow rate; the engine performance parameters include engine thrust, pressure, specific impulse, and solid grain burning rate.

[0015] Further, the solid-liquid rocket engine performance prediction dataset includes the structural characteristic data, operating condition characteristic data, and engine performance parameters of several solid-liquid rocket engines;

[0016] The specific method for standardizing the constructed performance prediction dataset is:

[0017] x′ = (x - μ) / σ

[0018] where x is the original data of the variable, x′ is the standardized value of the variable, μ = ∑x / n is the variable mean, is the variable standard deviation; after the original data is standardized, it becomes dimensionless input-output data with a mean of 0 and a standard deviation of 1.

[0019] Further, the RBF neural network includes an input layer, a hidden layer, and an output layer. The structural characteristic parameters and operating condition characteristic parameters of the solid-liquid rocket engine are used as input information and input into the input layer of the model, and the data is non-linearly processed through the hidden layer, and then the predicted data is output through the output layer;

[0020] The vector x of the input layer is expressed as:

[0021] x = (x 1 , x 2 , x 3 , …, x m )T ∈R m

[0022] Among them, m represents the number of input signals, and R m is the set of input layer vectors;

[0023] Using the Gaussian function as the basis function, the output G of the i-th neuron in the hidden layer i is expressed as:

[0024]

[0025] where c i is the center of the Gaussian activation function, and σ i is the variance of the Gaussian function;

[0026] The final output y of the RBF neural network is expressed as:

[0027]

[0028] where b is the bias value, k represents that there are k neurons in the hidden layer, ω represents the weights of the output layer, and i represents the i-th neuron.

[0029] Furthermore, based on the training set and the k-fold cross-validation method, the hyperparameters of the PSO-RBF neural network model are trained. Specifically:

[0030] Based on the particle swarm optimization (PSO) algorithm, the hyperparameters in the RBF neural network are optimized, and a particle swarm and the initial velocities and initial positions of each particle in the particle swarm are randomly generated; among them, each particle includes the variance of the Gaussian function of the initial solid-liquid rocket engine performance prediction model, the parameter values of each weight and each bias;

[0031] According to the PSO algorithm, the velocities and positions of each particle are updated. For each updated particle, the fitness value is calculated using 5-fold cross-validation;

[0032] Compare the current fitness value of each particle with its historical best fitness value. If the current fitness value is better, then update the best position of the particle; compare the current fitness values of all particles to find the global best position;

[0033] Update the parameters of the RBF neural network according to the positions of the particles. When the PSO algorithm reaches the convergence condition or the fitness value meets the termination condition, the optimal combination of RBF neural network parameters is obtained.

[0034] Further, for each updated particle, its fitness value is calculated using 5-fold cross-validation, specifically as follows: The training set is divided into five parts, four of which are used for training the RBF neural network inside the particle swarm optimization (PSO) algorithm, and the remaining one part of the data is used for prediction to obtain the root mean square error (RMSE) for this round. This round is repeated 5 times. If the RMSE obtained in the j-th round is RMSE j , then the fitness function of the RBF network corresponding to the current particle is:

[0035]

[0036] where J is the fitness function; n is the number of samples in the prediction set used for PSO optimization; s is the total number of repeated rounds; y ij are the predicted value and the true value of the current RBF neural network respectively.

[0037] Further, before randomly generating the particle swarm and the initial velocities and initial positions of each particle in the particle swarm, it also includes: setting the number of particle swarms, the maximum number of iterations, and determining the dimension of each particle; the termination condition for the fitness value is that the fitness value is less than the set threshold. If it is less than, the iteration stops, and the optimal RBF neural network parameter combination is obtained according to the particle position corresponding to the current fitness value.

[0038] A solid-liquid rocket engine performance prediction system based on PSO-RBF neural network includes:

[0039] A simulation module, which obtains the structural characteristic data and working condition characteristic data of the solid-liquid rocket engine, and calculates the corresponding engine performance parameters through a numerical simulation model;

[0040] A standardization module, which constructs a solid-liquid rocket engine performance prediction data set and performs standardization processing on the constructed performance prediction data set;

[0041] A partitioning module, which partitions the standardized data set to obtain a training set and a test set;

[0042] A training module, which trains the hyperparameters of the PSO-RBF neural network model based on the training set and the k-fold cross-validation method, and tests the PSO-RBF neural network model through the test set to obtain the performance prediction results of the test data and verify the accuracy of the PSO-RBF neural network model;

[0043] An acquisition module, which inputs the structural characteristics and working condition characteristic data of the engine with unknown performance into the trained PSO-RBF neural network model to obtain the performance parameters of the solid-liquid rocket engine.

[0044] A terminal device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above method are implemented.

[0045] A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the steps of the above method are implemented.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] Based on the structural characteristic data and operating condition characteristic data of a solid-liquid rocket engine, the present invention predicts the engine performance parameters of the solid-liquid rocket engine, which can provide relatively accurate data support for the performance evaluation of the solid-liquid rocket engine, and further reduce the performance prediction cost in the design process of the solid-liquid rocket engine. At the same time, the present invention optimizes the hyperparameters of the radial basis function (RBF) neural network through the particle swarm optimization (PSO) algorithm, and can automatically find the optimal network structure and parameter configuration, which is applicable to the modeling and prediction of complex nonlinear systems. The present invention predicts the engine performance through numerical simulation and machine learning models, reduces the need for physical tests, thereby shortening the R & D cycle and reducing the R & D cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0049] Figure 1 It is a schematic flow chart of the method for predicting the performance of a solid-liquid rocket engine based on a PSO-RBF neural network of the present invention;

[0050] Figure 2 It is a schematic structural diagram of a solid-liquid rocket engine;

[0051] Figure 3 It is a schematic structural diagram of an RBF neural network;

[0052] Figure 4 It is a flow chart for constructing a PSO-RBF neural network;

[0053] Figure 5 It is a schematic structural diagram of the system for predicting the performance of a solid-liquid rocket engine based on a PSO-RBF neural network of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. The components of the embodiments of the present invention usually described and illustrated in the accompanying drawings here can be arranged and designed in various different configurations.

[0055] Therefore, the detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0056] It should be noted that like reference numerals and letters denote like items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0057] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper", "lower", "horizontal", "inner", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the inventive product is customarily placed during use, it is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention. In addition, terms such as "first", "second", etc. are only used for descriptive distinction and cannot be construed as indicating or implying relative importance.

[0058] In addition, if the term "horizontal" appears, it does not mean that the component is required to be absolutely horizontal, but it can be slightly inclined. For example, "horizontal" only means that its direction is more horizontal relative to "vertical", and does not mean that the structure must be completely horizontal, but it can be slightly inclined.

[0059] In the description of the embodiments of the present invention, it should also be noted that unless otherwise clearly specified and limited, if terms such as "set", "installed", "connected", "connected" are understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0060] The present invention will be further described in detail below with reference to the accompanying drawings:

[0061] SeeFigure 1 , the present invention discloses a performance prediction method for a solid-liquid rocket engine based on a PSO-RBF neural network, including:

[0062] S101: Obtain the structural characteristic data and operating condition characteristic data of the solid-liquid rocket engine, and calculate the corresponding engine performance parameters through a numerical simulation model;

[0063] As Figure 2 shown, the structural characteristic parameters include the inner diameter D of the solid propellant grain channel, the length L of the solid propellant grain, and the nozzle throat diameter d; the operating condition characteristic data includes: oxidizer flow rate; the engine performance parameters include engine thrust, pressure, specific impulse, and solid propellant burning rate.

[0064] S102: Construct a performance prediction data set for the solid-liquid rocket engine, and perform standardization processing on the constructed performance prediction data set;

[0065] The performance prediction data set for the solid-liquid rocket engine includes the structural characteristic data, operating condition characteristic data, and engine performance parameters of several solid-liquid rocket engines;

[0066] The specific operation of performing standardization processing on the constructed performance prediction data set is as follows:

[0067] x′ = (x - μ) / σ

[0068] where x is the original data of the variable, x′ is the standardized value of the variable, μ = Σx / n is the variable mean, is the variable standard deviation; after the original data is standardized, it becomes dimensionless input and output data with a mean of 0 and a standard deviation of 1.

[0069] S103: Divide the standardized data set to obtain a training set and a test set;

[0070] S104: Based on the training set and the k-fold cross-validation method, train the hyperparameters of the PSO-RBF neural network model, and test the PSO-RBF neural network model through the test set to obtain the performance prediction results of the test data and verify the accuracy of the PSO-RBF neural network model;

[0071] See Figure 3 , the RBF neural network includes an input layer, a hidden layer, and an output layer. The structural characteristic parameters and operating condition characteristic parameters of the solid-liquid rocket engine are used as input information and input into the input layer of the model, and the data is non-linearly processed through the hidden layer, and then the predicted data is output through the output layer;

[0072] The vector x of the input layer is expressed as:

[0073] x = (x 1 ,x2 , x 3 , …, x m ) T ∈ R m

[0074] where m represents the number of input signals, and R m is the set of input layer vectors;

[0075] If the Gaussian function is used as the basis function, the output Gi of the i-th neuron in the hidden layer i is expressed as:

[0076]

[0077] where c i is the center of the Gaussian activation function, and σ i is the variance of the Gaussian function;

[0078] The final output y of the RBF neural network is expressed as:

[0079]

[0080] where b is the bias value; k represents that there are k neurons in the hidden layer, ω represents the weights of the output layer, and i represents the i-th neuron.

[0081] Based on the training set and the k-fold cross-validation method, the hyperparameters of the PSO-RBF neural network model are trained. Specifically:

[0082] See Figure 4 , based on the particle swarm optimization (PSO) algorithm, the hyperparameters in the RBF neural network are optimized, and a particle swarm and the initial velocities and initial positions of each particle in the particle swarm are randomly generated; among them, each particle includes the variance of the Gaussian function of the initial solid-liquid rocket engine performance prediction model, the parameter values of each weight and each bias;

[0083] According to the PSO algorithm, update the velocity and position of each particle. For each updated particle, calculate its fitness value using 5-fold cross-validation;

[0084] Compare the current fitness value of each particle with its historical best fitness value. If the current fitness value is better, update the best position of the particle; compare the current fitness values of all particles to find the global best position;

[0085] Update the parameters of the RBF neural network according to the position of the particle. When the PSO algorithm reaches the convergence condition or the fitness value meets the termination condition, obtain the optimal combination of RBF neural network parameters.

[0086] For each updated particle, its fitness value is calculated using 5-fold cross-validation, specifically as follows: The training set is divided into five parts, where four parts are used for training the RBF neural network inside the particle swarm optimization (PSO), and the remaining one part of the data is used for prediction to obtain the root mean square error (RMSE) for this round. This process is repeated 5 times. If the RMSE obtained in the j-th round is RMSE j , then the fitness function of the RBF network corresponding to the current particle is:

[0087]

[0088] where J is the fitness function; n is the number of samples in the prediction set used for optimizing the PSO; s is the total number of repeated rounds; y ij are the predicted value and the true value of the current RBF neural network respectively.

[0089] Before randomly generating the particle swarm and the initial velocities and initial positions of each particle in the particle swarm, it also includes: setting the number of particle swarms, the maximum number of iterations, and determining the dimension of each particle; the termination condition for the fitness value is that the fitness value is less than the set threshold. If it is less, the iteration stops, and the optimal RBF neural network parameter combination is obtained according to the particle position corresponding to the current fitness value.`

[0090] S105: Input the structural characteristics and operating condition characteristic data of the engine with unknown performance into the trained PSO-RBF neural network model to obtain the performance parameters of the solid-liquid rocket engine.

[0091] See Figure 5 , the present invention discloses a solid-liquid rocket engine performance prediction system based on a PSO-RBF neural network, including:

[0092] A simulation module, which obtains the structural characteristic data and operating condition characteristic data of the solid-liquid rocket engine and calculates the corresponding engine performance parameters through a numerical simulation model;

[0093] A standardization module, which constructs a solid-liquid rocket engine performance prediction data set and performs standardization processing on the constructed performance prediction data set;

[0094] A partitioning module, which partitions the standardized data set to obtain a training set and a test set;

[0095] A training module, which trains the hyperparameters of the PSO-RBF neural network model based on the training set and the k-fold cross-validation method, and tests the PSO-RBF neural network model through the test set to obtain the performance prediction results of the test data and verify the accuracy of the PSO-RBF neural network model;

[0096] An acquisition module, which inputs the structural characteristics and operating condition characteristic data of an engine with unknown performance into a trained PSO-RBF neural network model to obtain the performance parameters of a solid-liquid rocket engine.

[0097] The terminal device provided by an embodiment of the present invention. The terminal device of this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above-mentioned various method embodiments are implemented. Alternatively, when the processor executes the computer program, the functions of each module / unit in the above-mentioned various device embodiments are implemented.

[0098] The computer program can be divided into one or more modules / units, and the one or more modules / units are stored in the memory and executed by the processor to complete the present invention.

[0099] The terminal device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device may include, but is not limited to, a processor and a memory.

[0100] The processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0101] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the terminal device by running or executing the computer program and / or module stored in the memory, and by calling the data stored in the memory.

[0102] If the modules / units integrated in the terminal device are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, to implement all or part of the processes in the above-mentioned embodiment methods of the present invention, it can also be completed by a computer program instructing relevant hardware. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0103] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A solid-liquid rocket engine performance prediction method based on PSO-RBF neural network, characterized in that: include: Obtain the structural characteristic data and operating condition characteristic data of solid-liquid rocket engines, and calculate the corresponding engine performance parameters through numerical simulation models; Construct a performance prediction data set for solid and liquid rocket engines and standardize the constructed performance prediction data set; Divide the standardized data set into a training set and a test set; The hyperparameters of the PSO-RBF neural network model are trained based on the training set and the k-fold cross-validation method, and the PSO-RBF neural network model is tested using the test set to obtain the performance prediction results of the test data and verify the accuracy of the PSO-RBF neural network model. The structural characteristics and operating condition characteristics data of the engine with unknown performance are input into the trained PSO-RBF neural network model to obtain the performance parameters of the solid-liquid rocket engine.

2. The method for predicting the performance of solid-liquid rocket engines based on PSO-RBF neural network according to claim 1 is characterized in that: The structural characteristic parameters include the inner diameter D of the solid charge channel, the length L of the solid charge and the nozzle throat diameter d; the operating condition characteristic data include: oxidizer flow rate; the engine performance parameters include engine thrust, pressure, specific impulse and solid charge burning rate.

3. The method for predicting the performance of solid-liquid rocket engines based on PSO-RBF neural network according to claim 2 is characterized in that: The solid-liquid rocket engine performance prediction data set includes structural characteristic data, operating condition characteristic data and engine performance parameters of a plurality of solid-liquid rocket engines; The standardization process of the constructed performance prediction data set is specifically as follows: x′=(x-μ) / σ Among them, x is the original data of the variable, x′ is the standardized value of the variable, μ=∑x / n is the mean value of the variable, is the standard deviation of the variable; after standardization, the original data becomes dimensionless input and output data with a mean of 0 and a standard deviation of 1.

4. The method for predicting the performance of solid-liquid rocket engines based on PSO-RBF neural network according to claim 3 is characterized in that: The RBF neural network includes an input layer, a hidden layer, and an output layer. The structural characteristic parameters and operating condition characteristic parameters of the solid-liquid rocket engine are input into the input layer of the model as input information, and the data is processed nonlinearly through the hidden layer, and the predicted data is output through the output layer. The vector x of the input layer is represented as: x=(x1,x2,x3,..,x m ) T ∈R m Where m represents the number of input signals, R m is the set of input layer vectors; Using Gaussian function as the basis function, the output G of the i-th neuron in the hidden layer is i It is expressed as: Among them, c i is the center of the Gaussian activation function, σ i is the variance of the Gaussian function; The final output y of the RBF neural network is expressed as: Among them, b is the bias value, k means that the hidden layer has k neurons, ω represents the weight of the output layer, and i represents the i-th neuron.

5. The method for predicting the performance of solid-liquid rocket engines based on PSO-RBF neural network according to claim 4 is characterized in that: The hyper parameters of the PSO-RBF neural network model trained based on the training set and the k-fold cross validation method are specifically: Based on the particle swarm algorithm PSO, the hyperparameters in the RBF neural network are optimized, and a particle swarm and the initial velocity and initial position of each particle in the particle swarm are randomly generated; wherein each particle includes the variance of the Gaussian function of the initial solid-liquid rocket engine performance prediction model, the parameter values ​​of each weight and each bias; The speed and position of each particle are updated according to the particle swarm algorithm PSO. For each updated particle, its fitness value is calculated using 5-fold cross validation. Compare the current fitness value of each particle with its historical best fitness value. If the current fitness value is better, update the best position of the particle. Compare the current fitness values ​​of all particles to find the global best position. The parameters of the RBF neural network are updated according to the positions of the particles. When the particle swarm algorithm PSO reaches the convergence condition or the fitness value meets the termination condition, the optimal RBF neural network parameter combination is obtained.

6. The method for predicting the performance of solid-liquid rocket engines based on PSO-RBF neural network according to claim 5, characterized in that: For each updated particle, a 5-fold cross validation is used to calculate its fitness value. Specifically, the training set is divided into five parts, four of which are used for RBF neural network training within the particle swarm algorithm PSO, and the remaining one is used for prediction. The root mean square error of this round is obtained, and the round is repeated 5 times; if the root mean square error obtained in the jth round is RMSE j , then the fitness function of the RBF network corresponding to the current particle is: Among them, J is the fitness function; n is the number of prediction set samples used by the particle swarm algorithm PSO optimization; s is the total number of repeated rounds; are the predicted value and true value of the current RBF neural network respectively.

7. The method for predicting the performance of solid-liquid rocket engines based on PSO-RBF neural network according to claim 6 is characterized in that: Before randomly generating a particle swarm and the initial speed and initial position of each particle in the particle swarm, the method also includes: setting the number of particle swarms, the maximum number of iterations, and determining the dimension of each particle; the fitness value satisfies the termination condition that the fitness value is less than the set threshold value, if it is less than, the iteration is stopped, and the optimal RBF neural network parameter combination is obtained according to the particle position corresponding to the current fitness value.

8. The solid-liquid rocket engine performance prediction system based on PSO-RBF neural network is characterized by: include: A simulation module, wherein the simulation module obtains structural characteristic data and operating characteristic data of a solid-liquid rocket engine, and calculates corresponding engine performance parameters through a numerical simulation model; A standardization module, wherein the standardization module constructs a solid-liquid rocket engine performance prediction data set and performs standardization processing on the constructed performance prediction data set; A partitioning module, wherein the partitioning module partitions the standardized data set into a training set and a test set; A training module, wherein the training module trains the hyperparameters of the PSO-RBF neural network model based on the training set and the k-fold cross-validation method, and tests the PSO-RBF neural network model through the test set to obtain the performance prediction results of the test data and verify the accuracy of the PSO-RBF neural network model; The acquisition module inputs the structural characteristics and operating condition characteristics data of the unknown performance engine into the trained PSO-RBF neural network model to obtain the performance parameters of the solid-liquid rocket engine.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images