Cavitation reactor cavitation intensity prediction method and system based on deep learning

Abstract

The invention provides a cavitation reactor cavitation intensity prediction method and system based on deep learning, and relates to the technical field of cavitation, and the method comprises the steps: obtaining the shape parameters of a to-be-predicted cavitation reactor; inputting the shape parameters into a trained strength prediction model, and predicting the cavitation strength; the prediction result is explained, and the influence of interaction among different shape parameters on the cavitation intensity is obtained; wherein the intensity prediction model is trained by using a training set obtained by numerical simulation, a construction method of the training set comprises the following steps: randomly sampling shape parameters of the cavitation reactor, constructing a shape space of the cavitation reactor, performing numerical simulation on the shape space of the cavitation reactor to obtain cavitation intensity, and taking the shape parameters and the cavitation intensity as a data item; forming a training set; according to the method, the improved neural network is trained according to the training set obtained by numerical simulation, so that the nonlinear relation between the cavitation reactor shape factor and the cavitation intensity is established, and the cavitation intensity prediction is realized.

Description

Technical Field

[0001] The present invention belongs to the field of cavitation technology, and in particular relates to a method and system for predicting cavitation intensity of a cavitation reactor based on deep learning. Background Art

[0002] The current global energy shortage and environmental pollution have led to an urgent need to explore innovative, energy-saving and environmentally friendly process intensification technologies; fluid dynamic cavitation technology is an emerging process intensification method that has been widely used in various fields, including wastewater treatment, emulsification, lignocellulosic biomass pretreatment, flotation, biodiesel production and food processing. Cavitation reactors are essential for modern industrial applications; cavitation technology refers to the phase change process from liquid to gas when the static pressure of the liquid is lower than its saturated vapor pressure; the collapse of cavitation bubbles will produce extreme conditions, including local high temperatures of up to 5000K, pressures of up to 1000bar, and strong oxidation and reduction effects; common types of cavitation reactors include orifice plates, Venturi reactors, eddy current diodes and rotary reactors.

[0003] The prediction of cavitation intensity of cavitation reactor and the influence of cavitation reactor shape factors on cavitation intensity are usually analyzed by control variable method or surface response method; the control variable method aims to eliminate the influence of other factors on cavitation intensity by controlling the shape parameter variables in the cavitation reactor, so as to obtain the relationship between the shape factor of the cavitation reactor and cavitation intensity; the response surface method mainly collects data by designing appropriate experimental points and factor levels, and then uses the multivariate quadratic regression equation to fit the functional relationship between shape factors and cavitation intensity; however, both methods cannot capture the nonlinear relationship between the shape factors of the cavitation reactor and cavitation intensity, and cannot form the influence law of specific shape factor combinations on cavitation intensity. Summary of the invention

[0004] In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a cavitation intensity prediction method and system for a cavitation reactor based on deep learning, establishes a nonlinear relationship between the shape factors of the cavitation reactor and the cavitation intensity, and realizes the cavitation intensity prediction.

[0005] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0006] The first aspect of the present invention provides a method for predicting cavitation intensity of a cavitation reactor based on deep learning.

[0007] A cavitation intensity prediction method for a cavitation reactor based on deep learning, comprising:

[0008] Obtaining shape parameters of the cavitation reactor to be predicted;

[0009] The shape parameters are input into the trained strength prediction model to predict the cavitation strength;

[0010] The prediction results are explained to obtain the influence of the interaction between different shape parameters on the cavitation intensity;

[0011] Among them, the strength prediction model is trained using a training set obtained by numerical simulation. The method for constructing the training set is to randomly sample the shape parameters of the cavitation reactor, construct the shape space of the cavitation reactor, perform numerical simulation on the shape space of the cavitation reactor to obtain the cavitation intensity, and the sampled shape parameters and the obtained cavitation intensity are used as a data item to constitute a training set.

[0012] Further, the cavitation reactor includes a venturi tube and a rotary cavitation reactor;

[0013] The shape parameters of the venturi tube include contraction angle, throat diameter, throat length, expansion angle and inlet and outlet diameters;

[0014] The shape parameters of the rotary cavitation reactor include cylinder diameter, rotor spacing, cylinder depth and rotor inclination.

[0015] Furthermore, the cavitation intensity parameters of the venturi tube include the pressure drop at the inlet and outlet of the venturi tube and the average cavitation volume in time;

[0016] The cavitation intensity parameters of the rotary cavitation reactor include the cavitation generation rate.

[0017] Furthermore, the intensity prediction model adopts a neural network, including an input layer, a hidden layer and an output layer;

[0018] The number of nodes in the input layer is consistent with the number of shape parameters, and each node represents a specific shape parameter;

[0019] The number of nodes in the output layer is consistent with the number of cavitation intensity parameters, and each node represents a specific cavitation intensity parameter.

[0020] Furthermore, the hidden layers are connected respectively according to the influence of a specific shape parameter combination on the cavitation intensity, and are activated by a ReLU function before the connection.

[0021] Furthermore, the construction of the cavitation reactor shape space is based on randomly sampled shape parameters to model the cavitation reactor;

[0022] The commercial software Fluent was used to perform numerical simulation on the modeled cavitation reactor to obtain the cavitation intensity.

[0023] Furthermore, the effect of the interaction between different shape parameters on the cavitation intensity is obtained by retaining the network structure of a specific shape parameter combination, deleting the others, and then randomly sampling the nodes of the input layer to obtain the effect of the interaction between different shape parameters on the cavitation intensity.

[0024] The second aspect of the present invention provides a cavitation intensity prediction system for a cavitation reactor based on deep learning.

[0025] A cavitation intensity prediction system for a cavitation reactor based on deep learning, including an acquisition module, a prediction module and an interpretation module:

[0026] An acquisition module is configured to: acquire shape parameters of a cavitation reactor to be predicted;

[0027] The prediction module is configured to: input the shape parameters into the trained strength prediction model to predict the cavitation strength;

[0028] An explanation module is configured to: explain the prediction results to obtain the influence of the interaction between different shape parameters on the cavitation intensity;

[0029] Among them, the strength prediction model is trained using a training set obtained by numerical simulation. The method for constructing the training set is to randomly sample the shape parameters of the cavitation reactor, construct the shape space of the cavitation reactor, perform numerical simulation on the shape space of the cavitation reactor to obtain the cavitation intensity, and the sampled shape parameters and the obtained cavitation intensity are used as a data item to constitute a training set.

[0030] The third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in a method for predicting cavitation intensity of a cavitation reactor based on deep learning as described in the first aspect of the present invention.

[0031] The fourth aspect of the present invention provides an electronic device, comprising a memory, a processor, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in a method for predicting cavitation intensity of a cavitation reactor based on deep learning as described in the first aspect of the present invention are implemented.

[0032] One or more of the above technical solutions have the following beneficial effects:

[0033] The present invention randomly samples shape parameters of a cavitation reactor, constructs a cavitation reactor shape space, performs numerical simulation on the cavitation reactor shape space to obtain cavitation intensity, thereby constructing a training set, training a neural network, thereby establishing a nonlinear relationship between the cavitation reactor shape factors and the cavitation intensity, and realizing cavitation intensity prediction.

[0034] The neural network used in the present invention is different from the traditional neural network structure. The input layer and the hidden layer are not fully connected, but are connected separately according to the influence of the specific shape parameter combination on the cavitation intensity, and the number of hidden layers is designed to be 2 layers, and the number of nodes in each layer is related to the input and output layers. On the one hand, the number of nodes in the hidden layer has a specific meaning, which is convenient for the influence of the subsequent specific shape parameter combination on the cavitation intensity; on the other hand, the structural characteristics are close to the actual situation, so that the neural network training can converge better.

[0035] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0037] Figure 1 It is a shape parameter diagram of the Venturi tube of the first embodiment.

[0038] Figure 2 This is a shape space diagram of the first embodiment.

[0039] Figure 3 This is a neural network structure diagram of the first embodiment.

[0040] Figure 4 A diagram of a neural network structure with a specific shape parameter combination for the first embodiment.

[0041] Figure 5 This is a diagram showing the influence of a specific shape parameter combination on cavitation intensity for the first embodiment. DETAILED DESCRIPTION

[0042] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0043] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0044] Embodiment 1

[0045] In one embodiment of the present disclosure, a deep learning-based cavitation intensity prediction method for a cavitation reactor is provided, which is used to predict the cavitation intensity of a venturi tube.

[0046] The Venturi tube is a commonly used cavitation reactor, and its shape parameters are as follows: Figure 1 As shown, the main features are the contraction angle β1, throat diameter d th , throat length l th , expansion angle β2 and inlet and outlet diameters d. For the Venturi tube, this embodiment only considers the first four shape parameters, and the cavitation intensity parameters select the inlet and outlet pressure drop △p and the average cavitation volume V of the Venturi tube.

[0047] The shape parameters of the venturi tube are randomly sampled, and the cavitation reactor is modeled based on the randomly sampled shape parameters, so as to obtain the following Figure 2 The shape space shown, Figure 2 It is about the contraction angle β1, expansion angle β2, throat length l th and throat diameter d th The relationship between.

[0048] 100 venturi flows were simulated through the fluent platform to obtain their cavitation intensity parameters. After the shape parameters and cavitation intensity parameters were standardized, the shape parameters and cavitation intensity were used as a data item to construct a data set. The data set was divided into a training set and a validation set, for example, 90 sets were used as training sets for the neural network and 10 sets were used as validation sets, and other conditions such as boundary conditions and mesh quality were consistent.

[0049] The intensity prediction model uses a neural network, such as Figure 3 As shown, it consists of an input layer, a hidden layer and an output layer, wherein the input layer has 4 nodes, representing 4 shape parameters respectively; the output layer has 2 nodes, representing 2 cavitation intensity parameters respectively.

[0050] The hidden layer consists of two layers, and the number of nodes in the first hidden layer is 2 n -1, the number of nodes in the second layer is m, where n represents the number of shape parameters and m represents the number of cavitation intensity parameters. The nodes in the first hidden layer are not fully connected to the input layer (shape parameters), but are connected to specific input nodes according to single factors, double factors, ..., n factors, respectively, representing the influence of single factors, double factors interaction, ..., n factors interaction on cavitation intensity. Each node in the first hidden layer is connected according to the influence of a specific shape parameter combination on cavitation intensity; each node in the second hidden layer is fully connected to the first hidden layer and connected to the output layer accordingly; the hidden layer of the two-layer structure has two layers of activation functions, which can fit nonlinear relationships.

[0051] In this embodiment, the number of shape parameters is 4, so the first hidden layer has 15 nodes, representing the influence of single shape parameter factors (4 nodes), dual shape parameters (6 nodes), triple shape parameters (4 nodes) and quadruple shape parameters (1 node) on cavitation intensity, wherein the node of the single shape parameter factor is connected to a shape parameter node corresponding to the input layer, the node of the dual shape parameter factor is connected to two shape parameter nodes corresponding to the input layer, and the same is true for the triple factor and quadruple factor; the activation function used in the hidden layer is ReLU, and the output that can be obtained by each connected node is:

[0052] Z [l] =W [l] a [l] +b [l]

[0053] Among them, W and b are the bias parameter matrices of the weights to be trained, l represents the number of neural network layers, and a [l] =δ(z [l -1] ), δ is the activation function, z [l-1] The second hidden layer has two nodes, which are fully connected to the nodes of the first hidden layer and are connected to the two cavitation intensity parameters of the output layer. The hidden layer of the two-layer structure has two layers of activation functions, so it can fit the nonlinear relationship.

[0054] The data of the training set is put into the neural network for training, and then the mean square error is used as the prediction performance criterion to judge the prediction performance, thereby forming a high-precision cavitation intensity prediction model; because this structure is more in line with actual impact, the prediction error is 22% lower than that of the traditional neural network structure, and the training time is reduced by 14%.

[0055] The trained cavitation intensity prediction model is interpreted to obtain the influence of the interaction between different shape parameters on the cavitation intensity. For example, the specific nodes of the hidden layer (i.e., the nodes with dual shape parameters) are retained, and the others are deleted to obtain the following: Figure 4 The model shown, Figure 4 (a), (b), and (c) are the model structure diagrams of retaining different dual shape parameter nodes respectively; then randomly sampling the nodes of the input layer, and finally performing cavitation intensity prediction, we can obtain the following Figure 5 The contraction angle β1, expansion angle β2, throat length l shown th and throat diameter d th The influence of the dual shape parameter combination composed of two of these four shape parameters on the cavitation intensity (i.e., the pressure drop △p at the inlet and outlet of the Venturi tube and the time-averaged cavitation volume V).

[0056] Embodiment 2

[0057] In one embodiment of the present disclosure, a deep learning-based cavitation intensity prediction method for a cavitation reactor is provided, which is used to predict the cavitation intensity of a rotary cavitation reactor.

[0058] The rotary cavitation reactor is a cavitation reactor with higher cavitation performance. The cavitation unit is a rotary cavitation reactor composed of a cylinder and a sphere. The shape parameters mainly include the cylinder diameter, rotor spacing, cylinder depth and rotor inclination. This embodiment only considers the first three shape parameters, and the cavitation intensity parameter selects the cavitation generation rate.

[0059] The shape parameters of the rotary cavitation reactor are randomly sampled, and the cavitation reactor is modeled based on the randomly sampled shape parameters, thereby obtaining a shape space.

[0060] The flow of 200 rotating cavitation reactors was simulated by the fluent platform to obtain their cavitation intensity parameters. After the shape parameters and cavitation intensity parameters were standardized, the shape parameters and cavitation intensity were used as a data item to construct a data set. The data set was divided into a training set and a validation set, for example, 200 sets were used as training sets for neural networks and 20 sets were used as validation sets, and other conditions such as boundary conditions and mesh quality were consistent.

[0061] The neural network used in the strength prediction model consists of an input layer, a hidden layer and an output layer. The input layer has three nodes, representing three shape parameters respectively, and the output layer has one node, representing one cavitation strength parameter respectively.

[0062] The hidden layer adopts a two-layer structure. The first hidden layer has 7 nodes, representing the influence of single shape parameter factors (3 nodes), dual shape parameters (3 nodes) and triple shape parameters (1 node) on cavitation intensity. The node of the single shape parameter factor is connected to a shape parameter node corresponding to the input layer, and the node of the dual shape parameter factor is connected to the two shape parameter nodes corresponding to the input layer. The same is true for the three factors. The activation function used in the hidden layer is ReLU. The hidden layer of the two-layer structure has two layers of activation functions, so it can fit a nonlinear relationship.

[0063] The data of the training set is put into the neural network for training, and then the mean square error is used as the prediction performance criterion to judge the prediction performance, thereby forming a high-precision cavitation intensity prediction model; the prediction error is 20% lower than that of the traditional neural network structure, and the training time is reduced by 12%.

[0064] The trained cavitation intensity prediction model is interpreted to obtain the influence of the interaction between different shape parameters on the cavitation intensity. For example, specific nodes in the hidden layer are retained and the others are deleted to obtain a model for a specific shape parameter combination. The nodes in the input layer are randomly sampled and the cavitation intensity is predicted. The influence of a specific shape parameter combination on the cavitation intensity can be obtained.

[0065] Embodiment 3

[0066] In one embodiment of the present disclosure, a cavitation intensity prediction system for a cavitation reactor based on deep learning is provided, including an acquisition module, a prediction module and an interpretation module:

[0067] An acquisition module is configured to: acquire shape parameters of a cavitation reactor to be predicted;

[0068] The prediction module is configured to: input the shape parameters into the trained strength prediction model to predict the cavitation strength;

[0069] An explanation module is configured to: explain the prediction results to obtain the influence of the interaction between different shape parameters on the cavitation intensity;

[0070] Among them, the strength prediction model is trained using a training set obtained by numerical simulation. The method for constructing the training set is to randomly sample the shape parameters of the cavitation reactor, construct the shape space of the cavitation reactor, perform numerical simulation on the shape space of the cavitation reactor to obtain the cavitation intensity, and the sampled shape parameters and the obtained cavitation intensity are used as a data item to constitute a training set.

[0071] Embodiment 4

[0072] The purpose of this embodiment is to provide a computer-readable storage medium.

[0073] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in a method for predicting cavitation intensity of a cavitation reactor based on deep learning as described in Embodiment 1 of the present disclosure.

[0074] Embodiment 5

[0075] The purpose of this embodiment is to provide an electronic device.

[0076] An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps in a method for predicting cavitation intensity of a cavitation reactor based on deep learning as described in Embodiment 1 of the present disclosure are implemented.

[0077] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. 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 cavitation intensity prediction method for a cavitation reactor based on deep learning, characterized in that: include: Obtaining shape parameters of the cavitation reactor to be predicted; The shape parameters are input into the trained strength prediction model to predict the cavitation strength; The prediction results are explained to obtain the influence of the interaction between different shape parameters on the cavitation intensity; Among them, the strength prediction model is trained using a training set obtained by numerical simulation. The method for constructing the training set is to randomly sample the shape parameters of the cavitation reactor, construct the shape space of the cavitation reactor, perform numerical simulation on the shape space of the cavitation reactor to obtain the cavitation intensity, and the sampled shape parameters and the obtained cavitation intensity are used as a data item to constitute a training set.

2. A method for predicting cavitation intensity of a cavitation reactor based on deep learning as claimed in claim 1, characterized in that: The cavitation reactor includes a venturi tube and a rotary cavitation reactor; The shape parameters of the venturi tube include contraction angle, throat diameter, throat length and expansion angle; The shape parameters of the rotary cavitation reactor include cylinder diameter, rotor spacing, cylinder depth and rotor inclination.

3. A method for predicting cavitation intensity of a cavitation reactor based on deep learning as claimed in claim 2, characterized in that: The cavitation intensity parameters of the venturi tube include the pressure drop at the inlet and outlet of the venturi tube and the average cavitation volume in time; The cavitation intensity parameters of the rotary cavitation reactor include the cavitation generation rate.

4. The method for predicting cavitation intensity of a cavitation reactor based on deep learning according to claim 1, characterized in that: The intensity prediction model adopts a neural network, including an input layer, a hidden layer and an output layer; The number of nodes in the input layer is consistent with the number of shape parameters, and each node represents a specific shape parameter; The number of nodes in the output layer is consistent with the number of cavitation intensity parameters, and each node represents a specific cavitation intensity parameter.

5. The method for predicting cavitation intensity of a cavitation reactor based on deep learning according to claim 4, characterized in that: The hidden layers are connected respectively according to the influence of a specific shape parameter combination on the cavitation intensity, and are activated by a ReLU function before the connection.

6. The method for predicting cavitation intensity of a cavitation reactor based on deep learning according to claim 1, characterized in that: The constructing of the cavitation reactor shape space is to model the cavitation reactor based on the randomly sampled shape parameters; The commercial software Fluent was used to perform numerical simulation on the modeled cavitation reactor to obtain the cavitation intensity.

7. The method for predicting cavitation intensity of a cavitation reactor based on deep learning according to claim 1, characterized in that: The effect of the interaction between different shape parameters on the cavitation intensity is obtained by retaining the network structure of a specific shape parameter combination, deleting the others, and then randomly sampling the nodes of the input layer to obtain the effect of the interaction between different shape parameters on the cavitation intensity.

8. A cavitation intensity prediction system for a cavitation reactor based on deep learning, characterized in that: Including acquisition module, prediction module and explanation module: An acquisition module is configured to: acquire shape parameters of a cavitation reactor to be predicted; The prediction module is configured to: input the shape parameters into the trained strength prediction model to predict the cavitation strength; An explanation module is configured to: explain the prediction results to obtain the influence of the interaction between different shape parameters on the cavitation intensity; Among them, the strength prediction model is trained using a training set obtained by numerical simulation. The method for constructing the training set is to randomly sample the shape parameters of the cavitation reactor, construct the shape space of the cavitation reactor, perform numerical simulation on the shape space of the cavitation reactor to obtain the cavitation intensity, and the sampled shape parameters and the obtained cavitation intensity are used as a data item to constitute a training set.

9. An electronic device, comprising: a memory for non-transitory storage of computer readable instructions; a processor for executing the computer-readable instructions; Wherein, when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 7 is executed.

10. A storage medium, characterized in that: The computer-readable instructions are non-transitorily stored, wherein when the computer-readable instructions are executed by a computer, the method of any one of claims 1 to 7 is performed.

Images