Prediction method for critical speed of cold spraying particles, electronic equipment and medium

Abstract

The invention relates to the field of cold spraying, in particular to a cold spraying particle critical speed prediction method, electronic equipment and a medium. The cold spraying particle critical speed prediction method comprises the steps of inputting cold spraying particle sample data into a gradient lifting regression tree model and a correlation analysis model to obtain target features; the cold spraying particle sample data comprises cold spraying particle material parameters and critical speed; the target features are cold spraying particle material parameters input into the prediction model; inputting the target characteristics of the cold spraying particles to be measured into a critical speed prediction model to obtain the critical speed of the cold spraying particles; the critical speed prediction model is a neural network model optimized by adopting target features in cold spraying particle sample data, a genetic algorithm and a Levenberg-Marquardt back propagation algorithm. According to the method, the cold spraying ion critical speed prediction efficiency and reliability can be improved.

Description

Technical Field

[0001] The present application relates to the field of cold spraying, and in particular to a method for predicting the critical velocity of cold spraying particles, electronic equipment and medium. Background Art

[0002] In the automotive manufacturing industry, die-casting castings occupy an important position in automotive structures and functional components. They are widely used in key components such as engine cylinders, transmission housings, suspension brackets, rear floors and battery housings. They are usually made of aluminum alloys or magnesium alloys and have excellent properties such as light weight, corrosion resistance and high strength. However, as the vehicle's operating time increases, these parts often have surface defects or failures such as cracks, wear or holes during use due to mechanical wear, corrosion or accidental damage. These defects not only affect the mechanical properties of die-castings, but also reduce the safety and service life of the vehicle.

[0003] As a new type of low-temperature material deposition process, cold spray technology has a wide range of application potential. Unlike traditional thermal spray methods, cold spray technology accelerates metal or alloy particles to a supersonic state by using high-speed airflow. When the particles collide with the substrate, they are physically combined to form a dense coating. Since the operating temperature of cold spray is low, it can effectively avoid thermal deformation of the substrate and degradation of material properties. It is particularly suitable for the repair of die castings, and can restore their original mechanical strength, wear resistance and corrosion resistance, thereby extending the service life of parts.

[0004] In the cold spray process, the powder impact velocity is one of the key parameters, and its value setting depends on the critical velocity of the particle, which refers to the minimum velocity that the particle needs to reach when effectively depositing on the substrate surface. The critical velocity depends on the physical and chemical properties of the particles, such as density, particle size, thermal properties of the material, etc. It determines whether the particles can successfully adhere to the substrate and form an effective coating. Therefore, in practical applications, accurate and rapid acquisition of the critical velocity parameters of the particles is crucial for optimizing the spraying process, controlling the coating quality, and process development efficiency.

[0005] At present, the prediction of the critical velocity of particles in cold spraying mainly relies on batch experimental methods. This method is time-consuming, costly, and has high requirements on equipment, personnel capabilities, and operating conditions. At the same time, with the application of new materials such as multi-component alloys, it is difficult to quickly adjust particle velocity parameters and conduct large-scale screening through traditional experimental methods, which limits the flexible application of cold spraying processes and the efficient development and optimization of parameters. When predicting the critical velocity of particles, existing theoretical models and calculation methods, such as finite element analysis (FEA), computational fluid dynamics (CFD), and discrete element method (DEM), often fail to fully consider the complex interactions between particles and the matrix, as well as the diversity of material properties, thus affecting the reliability of the prediction results.

[0006] In view of this, this application is hereby filed. Summary of the invention

[0007] The purpose of the present application is to provide a method, electronic device and medium for predicting the critical velocity of cold spray particles, so as to solve the problems of low efficiency and low reliability in predicting the critical velocity of cold spray ions in the prior art.

[0008] In order to achieve the above objectives, this application adopts the following technical solutions: In a first aspect, the present application provides a method for predicting the critical velocity of cold spray particles, comprising: Inputting the cold spray particle sample data into the gradient boosting regression tree model and the correlation analysis model respectively to obtain the target features; the cold spray particle sample data includes the cold spray particle material parameters and the critical velocity; the target features are the cold spray particle material parameters used to be input into the prediction model; The target features of the cold spray particles to be tested are input into a critical speed prediction model to obtain the critical speed of the cold spray particles; the critical speed prediction model is a neural network model optimized by using the target features in the cold spray particle sample data, a genetic algorithm and a Levenberg-Marquardt back propagation algorithm.

[0009] In some technical solutions, the cold spray particle sample data is input into the gradient boosting regression tree model and the correlation analysis model respectively to obtain the target features, including: Inputting the cold spray particle sample data into the gradient boosting regression tree model to be trained to obtain first prediction data of the cold spray particle sample data; updating the model parameters of the to-be-trained gradient boosted regression tree model to reduce the loss between the first prediction data and the critical velocity in the cold spray particle sample data, and obtaining a gradient boosted regression tree model, wherein the gradient boosted regression tree model is used to predict the critical velocity in response to the cold spray particle material parameters; According to the gradient boosting regression tree model, obtaining the weight of each of the cold spray particle material parameters in the gradient boosting regression tree model; Inputting the cold spray particle sample data into the correlation analysis model to obtain the correlation coefficient between each of the cold spray particle material parameters and the critical velocity; Determining the comprehensive weight of the cold spray particle material parameter according to the weight and the correlation coefficient; According to the comprehensive weight, the target feature is determined.

[0010] In some technical solutions, updating the model parameters of the to-be-trained gradient boosting regression tree model to reduce the loss between the first prediction data and the critical velocity in the cold spray particle sample data includes: According to the predicted values ​​of each decision tree, the residual between the predicted values ​​of each sample in each decision tree and the critical velocity in the cold spray particle sample data, and the number of decision trees, the model parameters of the gradient boosting regression tree model to be trained are updated to reduce the loss between the first predicted data and the critical velocity in the cold spray particle sample data.

[0011] In some technical solutions, the step of inputting the target characteristics of the cold sprayed particles to be tested into a critical velocity prediction model to obtain the critical velocity of the cold sprayed particles includes: Inputting the target feature in the cold spray particle sample data into the critical velocity prediction model to be trained to obtain second prediction data of the cold spray particle sample data; Based on the genetic algorithm and the Levenberg-Marquardt back propagation algorithm, the model parameters of the critical speed prediction model to be trained are updated to reduce the loss between the second prediction data and the critical speed in the cold spray particle sample data, and obtain the critical speed prediction model, wherein the critical speed prediction model is used to predict the critical speed in response to the cold spray particle material parameters; The target features of the cold sprayed particles to be tested are input into the critical velocity prediction model to obtain the critical velocity of the cold sprayed particles.

[0012] In some technical solutions, the critical speed prediction model structure to be trained includes: an input layer, a hidden layer and an output layer; the number of neurons in the hidden layer is and the number of neurons in the input layer The relationship between them is as follows: ; The number of nodes in the input layer is the number of target features; the hidden layer uses a hyperbolic tangent sigmoid transfer function; and the output layer uses a linear transfer function to output results.

[0013] In some technical solutions, the genetic algorithm includes a roulette selection operation, a real-valued crossover operation, and a real-valued mutation operation; The fitness function used before performing the roulette wheel selection operation for: ;in, Configure the weight The corresponding mean square error of the neural network.

[0014] In some technical solutions, the updating of the model parameters of the critical speed prediction model to be trained based on the genetic algorithm and the Levenberg-Marquardt back propagation algorithm includes: The weights and thresholds of the critical speed prediction model to be trained are optimized based on a genetic algorithm, and then the weights and thresholds are optimized again using a Levenberg-Marquardt back propagation algorithm.

[0015] In some technical solutions, the cold spray particle material parameters are normalized material parameters, and the normalization process includes: According to the cold spray particle material parameters, the maximum and minimum values ​​of each material parameter are determined; A normalized material parameter is determined according to the maximum value, the minimum value and the actual value of the material parameter.

[0016] In a second aspect, the present application provides an electronic device, including: at least one processor, and a memory communicatively coupled to at least one of the processors; The memory stores instructions that can be executed by at least one of the processors, and the instructions are executed by at least one of the processors so that at least one of the processors can perform the above method.

[0017] In a third aspect, the present application provides a computer-readable storage medium, on which computer instructions are stored, and the computer instructions are used to enable a computer to execute the above method.

[0018] Compared with the prior art, the beneficial effects of this application are: The prediction method for the critical velocity of cold spray particles provided in this application selects target features based on the gradient boosting regression tree model and the correlation analysis model, and uses the genetic algorithm and the Levenberg-Marquardt back propagation algorithm to optimize the neural network to obtain the critical velocity prediction model, thereby effectively improving the prediction accuracy and robustness of the model. The genetic algorithm is used to optimize the weights and thresholds of the neural network to ensure that the model avoids local optimal solutions and thus achieves global optimal performance. This method not only improves the accuracy of the prediction, but also reduces the computational complexity, providing a scientific basis and technical support for the optimization of the cold spray process. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the specific implementation methods of the present application or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 It is a schematic flow chart of a method for predicting the critical velocity of cold spray particles provided in the present application; Figure 2 It is a schematic diagram of the structure of a device for predicting the critical velocity of cold spray particles provided by the present application; Figure 3 It is a structural schematic diagram of the electronic device provided by this application. DETAILED DESCRIPTION

[0021] The following is a description of exemplary embodiments of the present application in conjunction with the accompanying drawings, including various details of the embodiments of the present application to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0022] Example 1 Figure 1 This is a flow chart of a method for predicting the critical velocity of cold spray particles provided in this embodiment. This embodiment is suitable for automatically predicting the critical velocity of cold spray particles when the material parameters of the cold spray particles are known. The method can be executed by a device for predicting the critical velocity of cold spray particles, which can be composed of software and / or hardware and is generally integrated in an electronic device, which can be an electronic computer, etc. For ease of understanding, each step in the method of this embodiment is performed by an electronic computer.

[0023] like Figure 1 As shown, this embodiment provides a method for predicting the critical velocity of cold spray particles, comprising the following steps: S110, inputting the cold spray particle sample data into the gradient boosting regression tree model and the correlation analysis model respectively to obtain target features; the cold spray particle sample data includes cold spray particle material parameters and critical velocity; the target features are cold spray particle material parameters used to be input into the prediction model.

[0024] Among them, the cold spray particle sample data refers to the data of the cold spray particle samples used for selecting target features and training all models in this embodiment, including cold spray particle material parameters and critical speed. Material parameters include but are not limited to density, elastic modulus, yield strength, tensile strength, thermal conductivity, specific heat capacity, melting point, body sound velocity, etc. In the cold spray process, the critical speed determines whether the particles can successfully adhere to the surface of the substrate, and this adhesion performance is closely related to material parameters such as the physical and mechanical properties of the particles. Therefore, this embodiment focuses on the influence of material parameters on the critical speed. However, since there are many material parameters, some have a greater impact on the critical speed, and some have a smaller impact on the critical speed. This step can obtain those parameters that have a greater impact on the critical speed through feature selection, and use them as target features as input features of the critical speed prediction model to improve prediction accuracy.

[0025] In an optional embodiment, the cold spray particle sample data is input into the gradient boosting regression tree model and the correlation analysis model respectively to obtain the target features, including: Inputting the cold spray particle sample data into the gradient boosting regression tree model to be trained to obtain first prediction data of the cold spray particle sample data; updating the model parameters of the to-be-trained gradient boosted regression tree model to reduce the loss between the first prediction data and the critical velocity in the cold spray particle sample data, and obtaining a gradient boosted regression tree model, wherein the gradient boosted regression tree model is used to predict the critical velocity in response to the cold spray particle material parameters; According to the gradient boosting regression tree model, obtaining the weight of each of the cold spray particle material parameters in the gradient boosting regression tree model; Inputting the cold spray particle sample data into the correlation analysis model to obtain the correlation coefficient between each of the cold spray particle material parameters and the critical velocity; Determining the comprehensive weight of the cold spray particle material parameter according to the weight and the correlation coefficient; According to the comprehensive weight, the target feature is determined.

[0026] In this embodiment, the cold spray particle sample data is first used to train the gradient boosting regression tree model to be trained, so as to obtain a gradient boosting regression tree model with higher prediction accuracy, and then the weights of each of the cold spray particle material parameters in the gradient boosting regression tree model are obtained based on the gradient boosting regression tree model; then the cold spray particle sample data is input into the correlation analysis model to obtain the correlation coefficient between each of the cold spray particle material parameters and the critical speed; then the weight and the correlation coefficient are combined to determine the comprehensive weight of the cold spray particle material parameters, and finally the target feature is obtained according to the comprehensive weight.

[0027] Optionally, a comprehensive weight may be obtained by taking a weighted average of the weight and the correlation coefficient.

[0028] Optionally, the comprehensive weights may be sorted from large to small, and the material parameters corresponding to the top five weights may be selected as target features.

[0029] In an optional embodiment, updating the model parameters of the to-be-trained gradient boosting regression tree model to reduce the loss between the first prediction data and the critical velocity in the cold spray particle sample data includes: According to the predicted values ​​of each decision tree, the residual between the predicted values ​​of each sample in each decision tree and the critical velocity in the cold spray particle sample data, and the number of decision trees, the model parameters of the gradient boosting regression tree model to be trained are updated to reduce the loss between the first predicted data and the critical velocity in the cold spray particle sample data.

[0030] Optionally, a gradient boosting regression tree is used to calculate the influence weight of material parameters on the critical velocity of cold spraying. The gradient boosting regression tree improves the prediction accuracy by superimposing multiple decision trees. The specific process includes the following steps: (1) Calculate the initial prediction value to minimize the value of the loss function. The formula is as follows: in, are the initial predictors (also called the initial regression tree) is the material parameter, is the number of all samples, is the critical speed corresponding to the ith sample, is the predicted value and L is the loss function.

[0031] (2) Calculate the residual of each sample, that is, the deviation between the actual critical speed and the predicted value. The residual of each sample is expressed as the deviation of the loss function relative to the predicted critical speed, which is determined by the following formula: in is the residual, is the number of trees, is the material parameter of the ith sample, is the m-1th regression tree.

[0032] (3) Fit the next decision tree based on the residuals, gradually reducing the loss. By modeling the pseudo residuals, we can obtain the The weighting factor , defined as follows: Finally, a new prediction is made for each sample based on the previous learning situation, and the predicted value The definition is as follows: .

[0033] Through the above steps, the required gradient boosting regression tree model can be obtained: Then the weights corresponding to each material parameter in the model are obtained.

[0034] Optionally, this embodiment uses correlation analysis to quantify the correlation between material parameters and critical speed. The correlation coefficient ranges from -1 to 1. The closer the value is to 1 or -1, the stronger the correlation between the variables; close to 0 means a weaker correlation. In this embodiment, the positive and negative correlations between material parameters and critical speed are ignored, and the focus is only on the quantification of the correlation strength. The correlation coefficient is calculated by the following formula: in, is the correlation coefficient between material parameters and critical speed, and represent material parameters and critical speed respectively, and Respectively represent and The mean of .

[0035] Optionally, in order to consider different situations as much as possible, a data set that accounts for 75% of the cold spray particle sample data is selected as a training set, and another 25% of the data set is selected as a validation set.

[0036] S120, inputting the target features of the cold sprayed particles to be tested into a critical velocity prediction model to obtain a critical velocity of the cold sprayed particles; the critical velocity prediction model is a neural network model optimized by using the target features in the cold sprayed particle sample data, a genetic algorithm, and a Levenberg-Marquardt back propagation algorithm.

[0037] In an optional embodiment, the step of inputting the target feature of the cold sprayed particles to be tested into a critical velocity prediction model to obtain the critical velocity of the cold sprayed particles comprises: Inputting the target feature in the cold spray particle sample data into the critical velocity prediction model to be trained to obtain second prediction data of the cold spray particle sample data; Based on the genetic algorithm and the Levenberg-Marquardt back propagation algorithm, the model parameters of the critical speed prediction model to be trained are updated to reduce the loss between the second prediction data and the critical speed in the cold spray particle sample data, and obtain the critical speed prediction model, wherein the critical speed prediction model is used to predict the critical speed in response to the cold spray particle material parameters; The target features of the cold sprayed particles to be tested are input into the critical velocity prediction model to obtain the critical velocity of the cold sprayed particles.

[0038] In an optional implementation, the critical speed prediction model structure to be trained includes: an input layer, a hidden layer and an output layer; the number of neurons in the hidden layer is and the number of neurons in the input layer The relationship between them is as follows: ; The number of nodes in the input layer is the number of target features; the hidden layer uses a hyperbolic tangent sigmoid transfer function; and the output layer uses a linear transfer function to output results.

[0039] This implementation constructs a three-layer neural network structure. The hidden layer uses the hyperbolic tangent sigmoid transfer function to map the input data to a higher-dimensional feature space. The activation function formula can be selected as follows: The output layer uses a linear transfer function to output the result, and the activation function formula can be selected as follows: In the above formulas, x is the input signal and is the weighted sum of the output values ​​of the previous layer.

[0040] In an optional embodiment, the genetic algorithm includes a roulette selection operation, a real-valued crossover operation, and a real-valued mutation operation; The fitness function used before performing the roulette wheel selection operation for: ;in, Configure the weight The corresponding mean square error of the neural network.

[0041] Genetic algorithms optimize the weights and thresholds of neural networks. In neural networks, weights are parameters that connect neurons and are used to control the impact of input signals on the next layer of neurons. Genetic algorithms find the global optimal solution through operations such as selection, crossover, and mutation, thereby avoiding falling into local optimality. Population initialization is to encode the weights and thresholds of the neural network as individuals. The population size is generally selected between 20 and 100.

[0042] In this implementation, mean square error is used as the evaluation indicator. This definition ensures that the lower the MSE, the higher the fitness value, so that individuals with better performance are preferentially selected.

[0043] Optionally, the roulette wheel selection operation performs a probability selection based on the fitness value of the individual, and the formula is as follows: ; In the formula, Is an individual The probability of being selected; Is an individual The fitness of is the result calculated by the fitness function; is the cumulative fitness of the population.

[0044] Optionally, the real-valued crossover operation generates new offspring individuals by weighted averaging the genes of the parent individuals. The formula is as follows: ; In the formula, For the offspring individuals; , is the gene vector of the parent individual; is a random number between [0,1].

[0045] Optionally, the real-valued mutation operation formula is as follows: ; In the formula is the weight or threshold after mutation; is the original weight or threshold; is the mutation rate; , is the maximum and minimum value of the weight or threshold; is a random number uniformly distributed in the interval [-1,1]. The weights and thresholds are randomly adjusted to avoid falling into the local optimum.

[0046] This embodiment repeats the selection, crossover and mutation process until the termination condition is met, and finally selects the individual with the highest fitness as the optimized weight and threshold to configure the neural network.

[0047] In an optional implementation, the updating of the model parameters of the critical speed prediction model to be trained based on the genetic algorithm and the Levenberg-Marquardt back propagation algorithm includes: The weights and thresholds of the critical speed prediction model to be trained are optimized based on a genetic algorithm, and then the weights and thresholds are optimized again using a Levenberg-Marquardt back propagation algorithm.

[0048] According to the previous implementation method, the weights and thresholds have been obtained, and the weights and thresholds are optimized again using the Levenberg-Marquardt back propagation algorithm. The update formula of the Levenberg-Marquardt back propagation algorithm is as follows: ;in, is the updated weight, is the current weight, is the Jacobian matrix, is the error vector, is a parameter that controls the convergence speed. The algorithm adjusts The value can flexibly switch between gradient descent and Newton's method when approaching the global optimal solution, thereby ensuring a faster convergence speed. The parameters in the training process include the number of training rounds, learning rate, and training target.

[0049] Optionally, the accuracy of the model prediction is evaluated on the test set, and the mean square error (MSE) is used to evaluate the performance of the model. The formula is: ;in, is the actual critical speed, is the predicted value, is the number of samples. The performance of the optimized neural network is verified on the test data set, and retrained to obtain the optimal performance to complete the prediction of the critical velocity of cold spray particles.

[0050] Optionally, when training the critical speed prediction model, the data set can be divided into an 80% training set and a 20% test set.

[0051] In an optional embodiment, the cold spray particle material parameters are normalized material parameters, and the normalization process includes: According to the cold spray particle material parameters, the maximum and minimum values ​​of each material parameter are determined; A normalized material parameter is determined according to the maximum value, the minimum value and the actual value of the material parameter.

[0052] In order to overcome the limitations of traditional thermodynamic coupling models, machine learning has become an effective complementary method in recent years. Unlike traditional models based on physical formulas, machine learning models can automatically discover complex patterns through big data training, and have stronger adaptability and predictive capabilities. Machine learning models can extract key features from a large amount of experimental data and do not rely on simple assumptions about material interactions, which makes it advantageous in dealing with complex nonlinear problems. However, existing machine learning methods often ignore the normalization of input data and feature selection when predicting critical speeds. This may cause deviations in the model during training, ultimately affecting the accuracy of the prediction. Especially when faced with a variety of material parameters, how to effectively screen and utilize these data is a key challenge. Therefore, this embodiment normalizes the material parameters before data screening to solve the problem of different measurements.

[0053] Optionally, the normalization formula is: ;in , Represent the original data, minimum data, maximum data and normalized data respectively.

[0054] The above-mentioned prediction method of the critical velocity of cold spray particles selects target features based on the gradient boosting regression tree model and the correlation analysis model, and uses the genetic algorithm and the Levenberg-Marquardt back propagation algorithm to optimize the neural network to obtain the critical velocity prediction model, thereby effectively improving the prediction accuracy and robustness of the model. The genetic algorithm is used to optimize the weights and thresholds of the neural network to ensure that the model avoids local optimal solutions and thus achieves global optimal performance. This method not only improves the accuracy of the prediction, but also reduces the computational complexity, providing a scientific basis and technical support for the optimization of the cold spray process.

[0055] In general, this application has the following advantages: 1. Improve prediction accuracy: This application uses a neural network method based on feature selection and genetic algorithm optimization, which can effectively improve the prediction accuracy of the critical velocity of cold spray particles. Through normalization processing and feature selection, the magnitude of different material attribute parameters is ensured to be consistent, thereby reducing the prediction error. The optimization of the genetic algorithm further improves the performance of the neural network model and reduces the prediction deviation.

[0056] 2. Reduce computational complexity: Through optimized neural network structure and effective genetic algorithm parameter setting, the present invention significantly reduces computational complexity in the prediction process. The crossover and mutation operations in the genetic algorithm enable the model to achieve better performance in fewer iterations, reducing the consumption of computing resources and computing time.

[0057] 3. Enhance the robustness of the model: Use feature selection methods to screen features that have a significant impact on critical speed prediction, avoiding the impact of irrelevant features on model performance, thereby enhancing the robustness of the model. Genetic algorithm optimization further improves the model's adaptability to data fluctuations and outliers, making the prediction results more stable and reliable.

[0058] 4. Provide detailed feature impact analysis: Through feature selection methods such as gradient boosted regression tree (GBRT) and correlation analysis, this application can evaluate in detail the impact of different material property parameters on particle critical velocity. This analysis helps to understand the specific impact of material properties on the cold spray process and provide a scientific basis for material selection and process optimization.

[0059] 5. Adaptable to a wide range of application scenarios: The method of this application is applicable to various cold spray materials and spraying processes, and can be flexibly adjusted to adapt to different material properties and spraying conditions. Its versatility and adaptability make this prediction method have a wide range of industrial application prospects.

[0060] 6. Improve industrial production efficiency: By accurately predicting the critical velocity of particles, the spraying quality and efficiency in the cold spray process can be better controlled, the trial and error costs in the experiment and production process can be reduced, and the overall efficiency of the production process can be improved.

[0061] Example 2 like Figure 2 As shown, this embodiment provides a prediction device for the critical velocity of cold spray particles, comprising: A target feature selection module 201 is used to input the cold spray particle sample data into the gradient boosting regression tree model and the correlation analysis model respectively to obtain target features; the cold spray particle sample data includes cold spray particle material parameters and critical velocity; the target features are cold spray particle material parameters used to be input into the prediction model; The critical speed prediction module 202 is used to input the target features of the cold spray particles to be tested into the critical speed prediction model to obtain the critical speed of the cold spray particles; the critical speed prediction model is a neural network model optimized by using the target features in the cold spray particle sample data, a genetic algorithm and a Levenberg-Marquardt back propagation algorithm.

[0062] The device is used to execute the above method, and thus has at least functional modules and beneficial effects corresponding to the above method.

[0063] Example 3 like Figure 3 As shown, this embodiment provides an electronic device, including: at least one processor; and a memory communicatively connected to at least one of the processors; wherein, The memory stores instructions executable by at least one of the processors, and the instructions are executed by at least one of the processors to enable at least one of the processors to perform the above method. At least one processor in the electronic device can perform the above method, and thus has at least the same advantages as the above method.

[0064] Optionally, the electronic device also includes interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are connected to each other using different buses and can be installed on a common mainboard or installed in other ways as needed. The processor can process instructions executed in the electronic device, including instructions stored in or on a memory to display graphical information of a GUI (Graphical User Interface) on an external input / output device (such as a display device coupled to an interface). In other embodiments, if necessary, multiple processors can be used together with multiple memories, and / or multiple buses can be used together with multiple memories. Similarly, multiple electronic devices can be connected (for example, as a server array, a group of blade servers, or a multi-processor system), and each device provides some necessary operations. Figure 3 A processor 301 is taken as an example.

[0065] The memory 302, as a computer-readable storage medium, can be used to store software programs, computer executable programs and modules, such as program instructions / modules corresponding to the method for predicting the critical velocity of cold spray particles in the embodiment of the present application (for example, the target feature selection module and the critical velocity prediction module in the device for predicting the critical velocity of cold spray particles). The processor 301 executes various functional applications and data processing of the device by running the software programs, instructions and modules stored in the memory 302, that is, realizes the above-mentioned method for predicting the critical velocity of cold spray particles.

[0066] The memory 302 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system and at least one application required for a function; the data storage area may store data created according to the use of the terminal, etc. In addition, the memory 302 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 302 may further include a memory remotely arranged relative to the processor 301, and these remote memories may be connected to the device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0067] The electronic device may further include: an input device 303 and an output device 304. The processor 301, the memory 302, the input device 303 and the output device 304 may be connected via a bus or other means. Figure 3 The example of connecting through bus is taken in the following.

[0068] The input device 303 can receive input digital or character information, and the output device 304 can include a display device, an auxiliary lighting device (e.g., LED), and a tactile feedback device (e.g., a vibration motor), etc. The display device can include, but is not limited to, a liquid crystal display (LCD), a light emitting diode (LED) display, and a plasma display. In some embodiments, the display device can be a touch screen.

[0069] Example 4 This embodiment provides a computer-readable storage medium, on which computer instructions are stored, and the computer instructions are used to enable a computer to execute the above method. The computer instructions on the computer-readable storage medium are used to enable a computer to execute the above method, and thus have at least the same advantages as the above method.

[0070] The medium in this application may adopt any combination of one or more computer-readable media. The medium may be a computer-readable signal medium or a computer-readable storage medium. The medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of the medium (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, the medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, device, or device.

[0071] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0072] The program code contained on the computer-readable medium can be transmitted using any appropriate medium, including but not limited to wireless, wire, optical cable, RF (Radio Frequency), etc., or any suitable combination of the above.

[0073] Computer program code for performing the operation of the present application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., using an Internet service provider to connect through the Internet).

[0074] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this application can be executed in parallel, sequentially or in different orders, as long as the expected results of the technical solution disclosed in this application can be achieved, and this document is not limited here.

[0075] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A method for predicting the critical velocity of cold spray particles, characterized in that: include: The cold spray particle sample data are input into the gradient boosting regression tree model and the correlation analysis model respectively to obtain the target features; The cold spray particle sample data includes cold spray particle material parameters and critical velocity; the target feature is the cold spray particle material parameter used to input into the prediction model; The target features of the cold spray particles to be tested are input into a critical speed prediction model to obtain the critical speed of the cold spray particles; the critical speed prediction model is a neural network model optimized by using the target features in the cold spray particle sample data, a genetic algorithm and a Levenberg-Marquardt back propagation algorithm.

2. The method for predicting the critical velocity of cold spray particles according to claim 1, characterized in that: The cold spray particle sample data is input into the gradient boosting regression tree model and the correlation analysis model respectively to obtain target features, including: Inputting the cold spray particle sample data into the gradient boosting regression tree model to be trained to obtain first prediction data of the cold spray particle sample data; updating the model parameters of the to-be-trained gradient boosted regression tree model to reduce the loss between the first prediction data and the critical velocity in the cold spray particle sample data, and obtaining a gradient boosted regression tree model, wherein the gradient boosted regression tree model is used to predict the critical velocity in response to the cold spray particle material parameters; According to the gradient boosting regression tree model, obtaining the weight of each of the cold spray particle material parameters in the gradient boosting regression tree model; Inputting the cold spray particle sample data into the correlation analysis model to obtain the correlation coefficient between each of the cold spray particle material parameters and the critical velocity; Determining the comprehensive weight of the cold spray particle material parameter according to the weight and the correlation coefficient; According to the comprehensive weight, the target feature is determined.

3. The method for predicting the critical velocity of cold spray particles according to claim 2, characterized in that: The updating of the model parameters of the to-be-trained gradient boosting regression tree model to reduce the loss between the first prediction data and the critical velocity in the cold spray particle sample data includes: According to the predicted values ​​of each decision tree, the residual between the predicted values ​​of each sample in each decision tree and the critical velocity in the cold spray particle sample data, and the number of decision trees, the model parameters of the gradient boosting regression tree model to be trained are updated to reduce the loss between the first predicted data and the critical velocity in the cold spray particle sample data.

4. The method for predicting the critical velocity of cold spray particles according to claim 1, characterized in that: The step of inputting the target characteristics of the cold sprayed particles to be tested into a critical velocity prediction model to obtain the critical velocity of the cold sprayed particles comprises: Inputting the target feature in the cold spray particle sample data into the critical velocity prediction model to be trained to obtain second prediction data of the cold spray particle sample data; Based on the genetic algorithm and the Levenberg-Marquardt back propagation algorithm, the model parameters of the critical speed prediction model to be trained are updated to reduce the loss between the second prediction data and the critical speed in the cold spray particle sample data, and obtain the critical speed prediction model, wherein the critical speed prediction model is used to predict the critical speed in response to the cold spray particle material parameters; The target features of the cold sprayed particles to be tested are input into the critical velocity prediction model to obtain the critical velocity of the cold sprayed particles.

5. The method for predicting the critical velocity of cold spray particles according to claim 4, characterized in that: The critical speed prediction model structure to be trained includes: an input layer, a hidden layer and an output layer; the number of neurons in the hidden layer and the number of neurons in the input layer The relationship between them is as follows: ; The number of nodes in the input layer is the number of target features; the hidden layer uses a hyperbolic tangent sigmoid transfer function; and the output layer uses a linear transfer function to output results.

6. The method for predicting the critical velocity of cold spray particles according to claim 4, characterized in that: The genetic algorithm includes a roulette selection operation, a real-value crossover operation, and a real-value mutation operation; The fitness function used before performing the roulette wheel selection operation for: ;in, Configure the weight The corresponding mean square error of the neural network.

7. The method for predicting the critical velocity of cold spray particles according to claim 4, characterized in that: The updating of the model parameters of the critical speed prediction model to be trained based on the genetic algorithm and the Levenberg-Marquardt back propagation algorithm includes: The weights and thresholds of the critical speed prediction model to be trained are optimized based on a genetic algorithm, and then the weights and thresholds are optimized again using a Levenberg-Marquardt back propagation algorithm.

8. The method for predicting the critical velocity of cold spray particles according to any one of claims 1 to 7, characterized in that: The cold spray particle material parameters are normalized material parameters, and the normalization process includes: According to the cold spray particle material parameters, the maximum and minimum values ​​of each material parameter are determined; A normalized material parameter is determined according to the maximum value, the minimum value and the actual value of the material parameter.

9. An electronic device, characterized in that: include: at least one processor, and a memory communicatively coupled to at least one of the processors; The memory stores instructions executable by at least one of the processors, and the instructions are executed by at least one of the processors so that the at least one processor can execute the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method according to any one of claims 1 to 8.

Images