Falling film heat transfer coefficient prediction method, system and equipment based on neural network and storage medium

Abstract

The invention discloses a falling film heat transfer coefficient prediction method, system and equipment based on a neural network, and a storage medium. The method comprises the following steps: S1, constructing a comprehensive database containing various data of single-phase relative flow, falling film evaporation and falling film boiling heat transfer mechanisms; s2, dividing data in the comprehensive database into a training set and a test set; s3, respectively setting input parameters of the three heat transfer mechanisms; s4, constructing an ANN model, adjusting ANN model parameters and structure of the ANN model by using the training set according to set input parameters, and optimizing to obtain an optimal ANN model under each heat transfer mechanism; and S5, using the optimal ANN model to predict the to-be-measured data of the three heat transfer mechanisms to obtain the falling film heat transfer coefficient. The falling film heat transfer coefficient is accurately predicted, and the method has higher universality, generalization ability and prediction precision.

Description

Technical Field

[0001] The present invention belongs to the field of falling film heat transfer, and relates to a method, system, device and storage medium for predicting the falling film heat transfer coefficient based on a neural network. Background Technique

[0002] The falling film heat transfer technology has wide applications in fields such as refrigeration, seawater desalination, petrochemical industry, food processing, and ocean thermal energy conversion (OTEC). As a potential alternative to traditional flooded evaporation, the efficient prediction of the heat transfer coefficient of the falling film evaporation technology is crucial for optimizing the evaporator design and improving the performance. However, the heat transfer mechanism of the falling film flow is complex and affected by many factors. Existing prediction methods are difficult to cover a wide range of design and operating parameters, and lack universality and computational efficiency. The current prediction methods relying on experimental data, analytical models, and numerical simulations have significant limitations in terms of application scope, universality, and computational efficiency.

[0003] Currently, the heat transfer coefficient prediction methods based on experimental data, analytical models, and numerical simulations have limitations in terms of application scope, universality, and computational efficiency. These methods usually require specific experimental conditions or design parameters, which limit their applications in a wide range of scenarios. In addition, they have low computational efficiency and it is difficult to derive a general prediction formula. Summary of the Invention

[0004] The purpose of the present invention is to overcome the above-mentioned disadvantages of the existing technologies, and provide a method, system, device and storage medium for predicting the falling film heat transfer coefficient based on a neural network, which can accurately predict the falling film heat transfer coefficient and has higher universality, generalization ability and prediction accuracy.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions: A method for predicting the falling film heat transfer coefficient based on a neural network includes the following processes: S1, constructing a comprehensive database covering various data of single-phase convection, falling film evaporation and falling film boiling heat transfer mechanisms; S2, dividing the data in the comprehensive database into a training set and a test set; S3, respectively setting the input parameters of the three heat transfer mechanisms; S4, constructing an ANN model, and according to the set input parameters, using the training set to adjust the ANN model parameters and structure, and optimizing to obtain the optimal ANN model under each heat transfer mechanism; S5, using the optimal ANN model to predict the data to be measured of the three heat transfer mechanisms, and obtaining the falling film heat transfer coefficient.

[0006] Preferably, the various data include experimental data, theoretical analysis data and numerical simulation data.

[0007] Preferably, in the comprehensive database, the single-phase convective heat transfer mechanism includes data on water and alcohols; the falling film evaporation heat transfer mechanism includes data on water, ammonia, and R22; the falling film boiling heat transfer mechanism includes data on water, R134a, R123, R11, R32, R290, R600a, R1234ze, and R245fa.

[0008] Preferably, the training set and test set of the single-phase convective heat transfer mechanism are divided in a ratio of 81% - 19%; the training set and test set of the falling film evaporation heat transfer mechanism are divided in a ratio of 70% - 30%; the training set and test set of the falling film boiling heat transfer mechanism are divided in a ratio of 78% - 22%.

[0009] Preferably, for the ANN model of the single-phase convective heat transfer mechanism, the input parameters are the liquid film Reynolds number, Prandtl number, and pipe diameter; for the ANN model of the falling film evaporation heat transfer mechanism, the input parameters are the fluid type, liquid film Reynolds number, Prandtl number, pipe diameter, fluid temperature, and heat flux density; for the ANN model of the falling film boiling heat transfer mechanism, the input parameters are the fluid type, liquid film flow rate, liquid film Reynolds number, Prandtl number, pipe diameter, fluid temperature, heat flux density, boiling number, Weber number, and Kapitza number.

[0010] Preferably, adjusting the ANN model parameters and structure of the ANN model includes the maximum number of iterations, the maximum number of validation failures, the target error accuracy, the activation function, the training function, and the number of neurons in each layer; the mean absolute error, mean square error, and coefficient of determination are used as the evaluation indicators for optimization.

[0011] Preferably, the Garson algorithm based on connection weights is used to obtain the sensitivity coefficients of the input parameters of the ANN model.

[0012] A falling film heat transfer coefficient prediction system based on neural network includes: A comprehensive database construction module for constructing a comprehensive database covering various data of single-phase convection, falling film evaporation, and falling film boiling heat transfer mechanisms; A data division module for dividing the data in the comprehensive database into a training set and a test set; An input parameter setting module for respectively setting the input parameters of the three heat transfer mechanisms; An ANN model construction and optimization module for constructing an ANN model, adjusting the ANN model parameters and structure according to the set input parameters, and optimizing to obtain the optimal ANN model under each heat transfer mechanism; A prediction module for using the optimal ANN model to predict the data to be measured for the three heat transfer mechanisms and obtaining the falling film heat transfer coefficient.

[0013] A computer 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 method for predicting the falling film heat transfer coefficient based on a neural network are implemented.

[0014] A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the steps of the method for predicting the falling film heat transfer coefficient based on a neural network are implemented.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The neural network prediction method adopted by the present invention has significant advantages. It can handle highly non-linear complex relationships and high-dimensional data, automatically extract features and achieve high-precision prediction. After being trained with a large amount of data, this method has broad application capabilities, can be applied to various working conditions and working fluids, and exhibits good generalization ability. Its parallel computing ability supports efficient prediction of the influence of multiple factors and can be extended to construct a hybrid system to improve prediction accuracy and robustness. With a reasonable number of data points and specific information of the fluid, it can accurately predict the falling film heat transfer coefficient under a wide range of design and operating parameter conditions, and has higher universality, generalization ability and prediction accuracy. This method provides practical guidance for the performance optimization design of evaporators and has significant practical application value. Under the three heat transfer mechanisms, the mean absolute errors ( MAE ) of the ANN model test set are 1.29%, 1.09% and 5.02% respectively. The corresponding coefficient of determination ( R 2 ) values are 0.9177, 0.9236 and 0.9637 respectively, which proves the effectiveness and high prediction accuracy of this method. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a flowchart of the method for predicting the falling film heat transfer coefficient based on a neural network according to Embodiment 2 of the present invention; Figure 2 is a schematic diagram of the principle of the method for predicting the falling film heat transfer coefficient based on a neural network according to Embodiment 3 of the present invention; Figure 3 is the prediction accuracy of the three heat transfer mechanism models under different numbers of hidden layer nodes according to Embodiment 3 of the present invention; Figure 4 is the sensitivity coefficient of each input parameter in each neural network model according to Embodiment 3 of the present invention; Figure 5 is the comparison diagram of the predicted values and actual values of the neural network model test set under the three heat transfer mechanisms according to Embodiment 3 of the present invention Nu ; DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application.

[0018] Embodiment 1: This embodiment provides a method for predicting the falling film heat transfer coefficient based on a neural network, including the following steps: S1. Construct a comprehensive database covering various data of single-phase convection, falling film evaporation, and falling film boiling heat transfer mechanisms.

[0019] S2. Divide the data in the comprehensive database into a training set and a test set.

[0020] S3. Set the input parameters of the three heat transfer mechanisms respectively.

[0021] S4. Construct an ANN model. According to the set input parameters, use the training set to adjust the ANN model parameters and structure, and optimize to obtain the optimal ANN model under each heat transfer mechanism.

[0022] S5. Use the optimal ANN model to predict the data to be measured for the three heat transfer mechanisms, and obtain the falling film heat transfer coefficient.

[0023] Embodiment 2: As Figure 1 shown, this embodiment provides a method for predicting the falling film heat transfer coefficient based on a neural network, including the following steps: A. Based on the experimental data, analysis methods, and numerical simulation results in the literature, construct a comprehensive database covering three heat transfer mechanisms of single-phase convection, falling film evaporation, and falling film boiling, and the data cover a wide range of parameter ranges.

[0024] B. Divide the comprehensive database into a training set and a test set according to a predetermined ratio, and perform normalization preprocessing on the sample data.

[0025] C. According to the characteristics of the three heat transfer mechanisms, select the corresponding neural network input and output parameters.

[0026] D. Construct the ANN (Artificial Neural Network) model structure, use the characteristic parameters in the training set of step B for model input and output, and optimize to obtain the optimal ANN model under each heat transfer mechanism by adjusting the ANN model parameters and structure.

[0027] E. Use sensitivity analysis to explore the influence of each input parameter on the falling film heat transfer coefficient.

[0028] F. Use the test set to test the ANN model optimized in step D and predict the falling film heat transfer coefficient.

[0029] The comprehensive database of the three heat transfer mechanisms involved in step A is as follows: Single-phase convection: includes 1,635 groups of water and alcohol data.

[0030] Falling film evaporation: includes 2,095 groups of water, ammonia, and R22 data.

[0031] Falling film boiling: 1,207 groups of water, R134a, R123, R11, R32, R290, R600a, R1234ze, and R245fa data.

[0032] The dataset splitting ratio and data preprocessing involved in step B are as follows: Single-phase convection: The training set and test set are divided in an 81% - 19% ratio.

[0033] Falling film evaporation: The training set and test set are divided in a 70% - 30% ratio.

[0034] Falling film boiling: The training set and test set are divided in a 78% - 22% ratio.

[0035] The maximum-minimum scaling method is used for data preprocessing.

[0036] The corresponding neural network input and output parameters involved in step C are as follows: Single-phase convection: The input parameters are Re , Pr , D , and the output is Nu ; Falling film evaporation: The input parameters are fluid type, Re , Pr , D , T , q , and the output is Nu ; Falling film boiling: The input parameters are fluid type, Γ , Re , T , Pr , D , q, Bo, We, Ka , and the output is Nu .

[0037] In step D, the optimized ANN model parameters include the maximum number of iterations, the maximum number of validation failures, the target error accuracy, the activation function, the training function, and the number of neuron nodes in each layer.

[0038] The optimized evaluation metrics include the mean absolute error ( MAE ), the mean square error ( MSE ), and the coefficient of determination ( R 2 ).

[0039] In step E, the sensitivity analysis uses the Garson algorithm based on connection weights to enhance the interpretability of the ANN model and analyze the sensitivity coefficients of each input parameter.

[0040] Example 3: In this example, a prediction method for the falling film heat transfer coefficient based on a neural network is provided. The specific implementation steps are as follows: 1. Establish a falling film heat transfer database under three different heat transfer mechanisms, as shown in Tables 1 to 3; The falling film heat transfer database includes a single-phase convection database, a falling film evaporation database, and a falling film boiling database. The above databases all contain falling film heat transfer sample data from experimental measurements, theoretical analyses, and numerical simulations in the open literature. The accurate prediction of the falling film heat transfer coefficient is affected by multiple factors such as fluid properties (density, viscosity), system design parameters (pipe diameter, spacing), and operating conditions (liquid film flow rate, heat flux density).

[0041] Therefore, the following three databases are established: Tables 1 to 3 detail the parameter information covered in these databases.

[0042] Table 1

[0043] Table 2

[0044] Table 3

[0045] 2. Division of the training set and the test set The sample data is divided into a training set and a test set. The training set is used for model training, while the test set is not seen by the model and is only used to evaluate the model's prediction performance. Specifically, the single-phase convection database is divided into 1324 groups of training data and 311 groups of test data according to 81% - 19%, the falling film evaporation database is divided into 1467 groups of training data and 628 groups of test data at 70% - 30%, and the falling film boiling database is divided into 941 groups of training data and 266 groups of test data according to the ratio of 78% - 22%.

[0046] 3. Normalization preprocessing of the data The sample data needs to be normalized to avoid model instability caused by differences in different feature dimensions or numerical ranges. In this implementation, the maximum-minimum scaling method is used, that is , the feature data is scaled to the interval [0,1] to eliminate the dimension difference, promote the convergence of the neural network, and improve the stability and accuracy of the model. Among them x max and x minThe maximum and minimum values of the respective features x * are the normalized data. The normalization is achieved through the mapminmax function in Matlab.

[0047] 4. Determination of the input and output parameters of the neural network A reasonable selection of the input and output parameters of the neural network can not only capture the key information of the data, but also reduce the model complexity, improve the training efficiency and generalization ability. In this embodiment, the corresponding neural network input and output parameters are selected for three different heat transfer mechanisms. Table 4 lists the prediction accuracies of the three finally determined models.

[0048] Table 4

[0049] 5. Building and optimizing the neural network model In this embodiment, the evaluation indexes of the neural network model performance include: mean square error ( MSE ), mean absolute error ( MAE ), and coefficient of determination ( R 2 ). In addition, the parameters β , θ and ξ are respectively used to represent the percentages of the predicted data within the error ranges of ±10%, ±20%, and ±30%.

[0050] (1) (2) (3) Among them, n is the number of samples, y ii and are the actual values and predicted values of the model, is the average value of the actual values of the model. MSE and MAE The lower the value, the higher the model accuracy. R 2 Evaluates the goodness of fit between the predicted value and the actual value. The closer the value is to 1, the stronger the prediction ability.

[0051] The optimal number of nodes in the hidden layer is determined through an empirical formula and iterative testing, that is . Among them, X, z, and m represent the number of nodes in the hidden layer, input layer, and output layer respectively, and a is a constant between 1 and 20.

[0052] Figure 3 Shows the prediction accuracies of each heat transfer mechanism model under different numbers of hidden layer nodes.Figure 3 In (a), for the single-phase convection model, the overall MSE decreases between nodes 6 and 19 and reaches the minimum at node 19. After that, overfitting occurs. Therefore, the optimal number of nodes is 19. Figure 3 As shown in (b), for the falling film evaporation model, the MSE is very high at node 11, probably because the network converges to a local optimum. While the MSE is the lowest at node 13. Therefore, the optimal number of nodes is 13. Figure 3 As shown in (c), for the falling film boiling model, the MSE is stable between nodes 12 and 16, and the fluctuation increases after node 16, and the MSE is the lowest at node 23. Therefore, the optimal number of nodes is 23.

[0053] The results of multiple experiments show that although the multi-hidden layer neural network can reduce the feature parameters in complex scenarios, adding unnecessary layers will increase complexity and training time, and the improvement of the prediction error is not significant. Selecting a single-hidden layer network can save time while maintaining accuracy.

[0054] The activation function is used to limit the output of neurons. In this embodiment, the tansig non-linear activation function is selected for the hidden layer, which is centered at zero and maps real numbers to [-1, 1], accelerating the network convergence and effectively alleviating the vanishing gradient. The purelin linear function is used for the output layer.

[0055] Different training functions have a significant impact on the results of the neural network model. The trainlm function is preferably used in the present invention because it implements the Levenberg-Marquardt (L-M) algorithm. As Figure 2 , the present invention adopts the BP algorithm, calculates the error gradient through the forward propagation of signals and the backpropagation of errors, and adjusts the network parameters with the L-M algorithm. The L-M algorithm combines the Newton method and the gradient descent method, converges quickly, is suitable for medium-scale networks, and has an adaptive learning rate built-in. However, it is still necessary to pay attention to carefully selecting and adjusting parameters such as the maximum number of iterations, the number of validation failures, and the target error to control the convergence and stop of the iteration.

[0056] 6. Sensitivity Analysis of Each Input Parameter of the Model Sensitivity analysis can quantify the influence of input variables on the model output and improve the interpretability of the model. The higher the sensitivity coefficient, the greater the influence. In this embodiment, the Garson algorithm in the sensitivity analysis based on connection weights is used to calculate the sensitivity coefficients of each input parameter, that is:

[0057] Among them, Q ik is the sensitivity coefficient of input variable i to output variable k, w ij is the connection weight value from input variable i to hidden layer node j, v jk is the connection weight from hidden layer node j to output variable k, and L and N are the number of neurons in the hidden layer and the number of input variables respectively.

[0058] Figure 4 The sensitivity coefficients of the input parameters in the three models are shown. In the single-phase convection model, the sensitivity coefficient of the Prandtl number Pr is the highest, the pipe diameter D significantly affects the falling film evaporation heat transfer, and the Reynolds number Re has the greatest influence on the falling film boiling heat transfer.

[0059] 7. Testing the prediction accuracy of the neural network model Table 5 lists the structural design and parameter settings of the optimized neural network model in this embodiment.

[0060] Table 5

[0061] Figure 5 The predicted values and actual values of Nu in the test set of the neural network model under three heat transfer mechanisms are compared. From Figure 5 (a), it can be seen that the single-phase convection model has accurate predictions. The MAE for 311 groups of test data is 1.29%, and R² is 0.9117. Among them, 94.86% of the data has a prediction error within ±20%, and 95.5% within ±30%. From Figure 5 (b), it can be seen that the falling film evaporation model has an MAE of 1.09% for 628 groups of test data, and R² is 0.9236. 96.82% of the data has an error within ±20%, and 98.25% within ±30%. From Figure 5 (c), it can be seen that the falling film boiling model has an MAE of 5.02% for 266 groups of test data, and R² is 0.9637. 86.84% of the data has an error within ±20%, and 92.86% within ±30%. This indicates that the falling film heat transfer prediction model proposed in the present invention effectively captures the complex non-linear relationship between the falling film heat transfer parameters and has excellent prediction performance under the three heat transfer mechanisms.

[0062] Example 4: In this embodiment, a falling film heat transfer coefficient prediction system based on a neural network is provided. The falling film heat transfer coefficient prediction system based on a neural network can be used to implement the above-mentioned falling film heat transfer coefficient prediction method based on a neural network. Specifically, the falling film heat transfer coefficient prediction system based on a neural network includes a comprehensive database construction module, a data division module, an input parameter setting module, an ANN model construction and optimization module, and a prediction module.

[0063] Among them, the comprehensive database construction module is used to construct a comprehensive database covering various data of single-phase convection, falling film evaporation, and falling film boiling heat transfer mechanisms.

[0064] The data division module is used to divide the data in the comprehensive database into a training set and a test set.

[0065] The input parameter setting module is used to set the input parameters of the three heat transfer mechanisms respectively.

[0066] The ANN model construction and optimization module is used to construct an ANN model. According to the set input parameters, the training set is used to adjust the parameters and structure of the ANN model, and the optimal ANN model under each heat transfer mechanism is optimized.

[0067] The prediction module is used to use the optimal ANN model to predict the data to be measured for the three heat transfer mechanisms, and obtain the falling film heat transfer coefficient.

[0068] Example 5: In this embodiment, a terminal device is provided. The terminal device includes a processor and a memory. The memory is used to store a computer program. The computer program includes program instructions. The processor is used to execute the program instructions stored in the computer storage medium. 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. It is the computing core and control core of the terminal, and is suitable for implementing one or more instructions. Specifically, it is suitable for loading and executing one or more instructions to implement the corresponding method flow or corresponding function. The processor described in the embodiment of the present invention can be used for the operation of the falling film heat transfer coefficient prediction method based on a neural network, including: S1, constructing a comprehensive database covering the data of single-phase convection, falling film evaporation and falling film boiling heat transfer mechanisms; S2, dividing the data in the comprehensive database into a training set and a test set; S3, respectively setting the input parameters of the three heat transfer mechanisms; S4, constructing an ANN model, and according to the set input parameters, using the training set to adjust the parameters and structure of the ANN model, and optimizing to obtain the optimal ANN model under each heat transfer mechanism; S5, using the optimal ANN model to predict the data to be measured for the three heat transfer mechanisms, and obtaining the falling film heat transfer coefficient.

[0069] Example 6: In this embodiment, a computer-readable storage medium (Memory) is provided. The computer-readable storage medium is a memory device in a terminal device and is used to store programs and data. It can be understood that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and, of course, the extended storage medium supported by the terminal device. The computer-readable storage medium provides a storage space, and this storage space stores the operating system of the terminal. Moreover, in this storage space, one or more instructions suitable for being loaded and executed by a processor are also stored. These instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory (Random Access Memory), or a non-volatile memory, such as at least one disk memory.

[0070] One or more instructions stored in the computer-readable storage medium can be loaded and executed by a processor to implement the corresponding steps of the method for predicting the falling film heat transfer coefficient based on a neural network in the above embodiment; the one or more instructions in the computer-readable storage medium are loaded and executed by the processor to perform the following steps: S1, construct a comprehensive database covering various data of single-phase convection, falling film evaporation, and falling film boiling heat transfer mechanisms; S2, divide the data in the comprehensive database into a training set and a test set; S3, respectively set the input parameters of the three heat transfer mechanisms; S4, construct an ANN model, and according to the set input parameters, use the training set to adjust the ANN model parameters and structure, and optimize to obtain the optimal ANN model under each heat transfer mechanism; S5, use the optimal ANN model to predict the data to be measured for the three heat transfer mechanisms to obtain the falling film heat transfer coefficient.

[0071] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, optical memories, etc.) containing computer-usable program codes.

[0072] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in one or more of the flows Figure 1 one or more of the flows and / or blocks Figure 1 or means for implementing the functions specified in one or more of the blocks.

[0073] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in one or more of the flows Figure 1 one or more of the flows and / or blocks Figure 1 or means for implementing the functions specified in one or more of the blocks.

[0074] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more of the flows Figure 1 one or more of the flows and / or blocks Figure 1 or means for implementing the functions specified in one or more of the blocks.

[0075] In the above embodiments of the present application, the descriptions of the respective embodiments have their own focuses. For parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0076] The above are only the preferred embodiments of the present application. It should be noted that for those of ordinary skill in the art in this technical field, without departing from the principle of the present application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present application.

[0077] It should be understood that the above description is for the purpose of illustration rather than limitation. By reading the above description, many embodiments and many applications other than the provided examples will be obvious to those skilled in the art.

Claims

1. A prediction method for the falling film heat transfer coefficient based on a neural network, characterized in that It includes the following processes: S1. Construct a comprehensive database covering various data of single-phase convection, falling film evaporation and falling film boiling heat transfer mechanisms; S2. Divide the data in the comprehensive database into a training set and a test set; S3. Set the input parameters of the three heat transfer mechanisms respectively; S4. Construct an ANN model. According to the set input parameters, use the training set to adjust the parameters and structure of the ANN model, and optimize to obtain the optimal ANN model under each heat transfer mechanism; S5. Use the optimal ANN model to predict the data to be measured for the three heat transfer mechanisms, and obtain the falling film heat transfer coefficient.

2. The method for predicting the falling film heat transfer coefficient based on a neural network according to claim 1, characterized in that The various data include experimental data, theoretical analysis data and numerical simulation data.

3. The method for predicting the falling film heat transfer coefficient based on a neural network according to claim 1, wherein In the comprehensive database, the single-phase convection heat transfer mechanism includes data of water and alcohols; the falling film evaporation heat transfer mechanism includes data of water, ammonia and R22; the falling film boiling heat transfer mechanism includes data of water, R134a, R123, R11, R32, R290, R600a, R1234ze and R245fa.

4. The method for predicting the falling film heat transfer coefficient based on a neural network according to claim 1, characterized in that The training set and test set of the single-phase convection heat transfer mechanism are divided according to the ratio of 81%-19%; the training set and test set of the falling film evaporation heat transfer mechanism are divided according to the ratio of 70%-30%; the training set and test set of the falling film boiling heat transfer mechanism are divided according to the ratio of 78%-22%.

5. The method for predicting the falling film heat transfer coefficient based on a neural network according to claim 1, characterized in that For the ANN model of the single-phase convection heat transfer mechanism, the input parameters are liquid film Reynolds number, Prandtl number and pipe diameter; for the ANN model of the falling film evaporation heat transfer mechanism, the input parameters are fluid type, liquid film Reynolds number, Prandtl number, pipe diameter, fluid temperature and heat flux density; for the ANN model of the falling film boiling heat transfer mechanism, the input parameters are fluid type, liquid film flow rate, liquid film Reynolds number, Prandtl number, pipe diameter, fluid temperature, heat flux density, boiling number, Weber number and Kapitza number.

6. The method for predicting the falling film heat transfer coefficient based on a neural network according to claim 1, characterized in that Adjusting the parameters and structure of the ANN model includes the maximum number of iterations, the maximum number of verification failures, the target error accuracy, the activation function, the training function and the number of neuron nodes in each layer; the mean absolute error, the mean square error and the coefficient of determination are used as the evaluation indexes for optimization.

7. The method for predicting the falling film heat transfer coefficient based on a neural network according to claim 1, wherein Adopt the Garson algorithm based on connection weights to obtain the sensitivity coefficients of the input parameters of the ANN model.

8. A falling film heat transfer coefficient prediction system based on a neural network, characterized in that, It includes: A comprehensive database construction module for constructing a comprehensive database covering various data of single-phase convection, falling film evaporation and falling film boiling heat transfer mechanisms; A data division module for dividing the data in the comprehensive database into a training set and a test set; An input parameter setting module for setting the input parameters of the three heat transfer mechanisms respectively; An ANN model construction and optimization module for constructing an ANN model. According to the set input parameters, use the training set to adjust the parameters and structure of the ANN model, and optimize to obtain the optimal ANN model under each heat transfer mechanism; A prediction module for using the optimal ANN model to predict the data to be measured for the three heat transfer mechanisms and obtain the falling film heat transfer coefficient.

9. A computer 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, it implements the steps of the neural network-based falling film heat transfer coefficient prediction method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the neural network-based falling film heat transfer coefficient prediction method according to any one of claims 1 to 7.

Images