CN115936227A - Machine learning-based ion permeation power generation prediction system and method - Google Patents
Machine learning-based ion permeation power generation prediction system and method Download PDFInfo
- Publication number
- CN115936227A CN115936227A CN202211618751.0A CN202211618751A CN115936227A CN 115936227 A CN115936227 A CN 115936227A CN 202211618751 A CN202211618751 A CN 202211618751A CN 115936227 A CN115936227 A CN 115936227A
- Authority
- CN
- China
- Prior art keywords
- dimensionless
- power generation
- neural network
- ion
- prediction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The prediction system comprises a data acquisition module, a neural network module and a prediction module, wherein the data acquisition module is configured to acquire sample data of ion permeation power generation, the neural network module is configured to construct a prediction neural network and neural network training of the ion permeation power generation based on an artificial neural network, the neural network module connected with the data acquisition module comprises a plurality of network layers, the substructure of the neural network module comprises an input layer, an output layer and four full-connection layers, neurons of the input layer receive training data, each neuron of a hidden layer and each neuron of the output layer are in full connection with all neurons of the adjacent layers, and the neural network training is performed based on the sample data to form the prediction neural network of the ion permeation power generation; the model testing module is configured to adjust the hyper-parameters in the prediction neural network of the ion permeation power generation to obtain the ion permeation power generation prediction model. The accuracy of the prediction model reaches 97.6%.
Description
Technical Field
The invention relates to the field of ion permeation power generation, in particular to a prediction system and a prediction method of ion permeation power generation based on machine learning.
Background
The ion permeation power generation is an energy conversion mode that ion carrier energy is driven by concentration gradient to migrate and selectively and directionally pass through the porous nano-film to form ion current, and salt difference energy is directly converted into electric energy. The energy conversion efficiency in the ion directional migration process is high (the theoretical efficiency can reach 50 percent), and a new way can be provided for the utilization of low-grade energy sources (salt difference energy, solar energy and the like). The physical essence of the ion permeation power generation is the coupling ion migration, heat transfer and energy conversion process by taking electrolyte ions and a nano porous medium as an energy carrier and a thermal mass transmission carrier respectively.
At present, the related research of ion permeation power generation mainly focuses on the aspect of novel nano-channel materials, the performance of a porous nano-ion selective membrane material under the condition of salinity difference is researched, and the geometric shape of a channel is changed by changing the concentration ratio so as to obtain better power generation capacity. However, there is still a lack of a unified temperature-dependent performance theory, and there is no comprehensive empirical formula to relate these factors to conversion efficiency. The ion permeation energy conversion mechanism needs to be researched, the physical essence of the internal thermal mass transport and energy conversion process is clarified, and finally, an ion permeation energy prediction system is formed to accurately evaluate the ion permeation power generation efficiency.
The information disclosed in this background section is only for enhancement of understanding of the background of the invention and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art.
Disclosure of Invention
Aiming at the defects or defects of the prior art, a prediction system of ion permeation power generation based on machine learning is provided, a simplified dimensionless control parameter set of ion permeation power generation is obtained through a similar principle, samples are expanded to obtain a large number of experimental samples to be used as a data set of machine learning, a machine learning artificial neural network is used for establishing a prediction model among a nano-channel geometric structure, a concentration difference, ion physical properties and energy conversion performance by taking a dimensional variable of a main factor physical parameter of normalization preprocessing as an input and taking the efficiency of output power and energy conversion efficiency as an output parameter. After the model parameter training is completed, the ion permeation power generation prediction model with high precision and high acceleration ratio based on machine learning is finally obtained, and judgment and efficiency prediction of the dominant parameters of the permeation power generation system can be guided.
The purpose of the invention is realized by the following technical scheme.
A predictive system for machine learning-based ion-permeation power generation includes;
a data acquisition module configured to acquire sample data of ion permeation power generation, the data acquisition module comprising;
a measurement unit that measures a plurality of physical parameters of the ion permeation power generation;
a dimensionless transformation unit connected to the measurement unit and generating a plurality of dimensionless parameters based on the plurality of physical parameters and forming a dimensionless array based on the dimensionless parameters;
the sample expansion unit is connected with the dimensionless conversion unit and is used for obtaining an expansion sample by adopting a similar principle based on the dimensionless array, the expansion sample and the dimensionless array form sample data, and the sample data is divided into training data and test data according to a preset proportion;
a normalization processing unit connected to the sample expansion unit and normalizing the sample data;
the neural network module is configured to construct a prediction neural network for ion permeation power generation and neural network training based on an artificial neural network ANN, the neural network module connected with the data acquisition module comprises a plurality of network layers, the substructure of the neural network module comprises an input layer, an output layer and four full-connection layers, neurons in the input layer receive training data, all the neurons in the hidden layer and the output layer are in full-connection with all the neurons in the adjacent layer, and the neural network training is carried out based on the sample data to form the prediction neural network for ion permeation power generation;
a model testing module configured to adjust hyper-parameters in a predictive neural network of ion osmotic power generation to obtain an ion osmotic power generation predictive model, the model testing module comprising;
the hyper-parameter optimization unit is connected with the neural network module, adopts a Graphic Processing Unit (GPU) to perform parallel acceleration, activates the combination optimization of function nonlinearity and back propagation algorithm, and updates the weight w and the bias b to optimize the prediction neural network of the ion permeation power generation;
and the testing unit is connected with the data acquisition module and the neural network module so as to test the prediction neural network of the ion permeation power generation based on the test data until the error reaches the expectation, and a model is saved as an ion permeation power generation prediction model.
In the prediction system for ion-permeation power generation based on machine learning, the measurement unit comprises a first measurement part for obtaining dielectric constant, concentration, ion diffusion coefficient, nano-channel length, radius and surface charge density, a temperature sensor for measuring temperature data, a second measurement part for measuring the power level of output power P and energy conversion efficiency eta.
A prediction method using the prediction system for ion-permeation power generation based on machine learning comprises the following steps,
s100: the measuring unit measures a plurality of physical parameters of the ion permeation power generation, an ion permeation power generation dimensional equation is constructed based on the physical parameters, and the dimensionless conversion unit generates an ion permeation power generation dimensionless equation and a dimensionless array corresponding to the dimensionless equation based on the ion permeation power generation dimensional equation;
s200: the sample expansion unit adopts a similarity principle to expand based on the dimensionless array to obtain an expansion sample, the expansion sample and the dimensionless array form sample data, and the normalization processing unit normalizes the sample data;
s300: the sample data is divided into training data and test data according to a predetermined proportion, and the neural network module is configured to construct ion penetration based on the artificial neural networkThe prediction neural network of the power generation is formed by carrying out neural network training based on the sample data, wherein the output power P and the energy conversion efficiency eta are output parameters, the concentration C and the high-temperature end temperature T are output parameters h Low temperature end temperature T l Dimensional variables of an ion diffusion coefficient D, a dielectric constant epsilon, a nano-channel length L, a nano-channel radius R and a surface charge density sigma are taken as input;
s400: the hyper-parameter optimization unit adopts a Graphic Processing Unit (GPU) for acceleration, activates the combination optimization of function nonlinearity and back propagation algorithm, and updates the weight w and the bias b to optimize the prediction neural network of the ion permeation power generation;
s500: and the test unit tests the optimized prediction neural network of the ion permeation power generation based on the test data, adjusts the hyper-parameters in the network according to the training result, repeats iteration until the error reaches the expectation, and saves the model as the prediction model of the ion permeation power generation.
In the prediction method, the dimensionless control equation and the dimensionless array corresponding to the dimensionless control equation are as follows:
dimensionless poisson equation:
dimensionless nernst-planck equation:
dimensionless continuity equation:
dimensionless navier-stokes equation:
dimensionless energy equation:
dimensionless array:
wherein the content of the first and second substances,
is a partial differential operator,. Epsilon.is the dielectric constant,. Phi.is the potential,. F is the Faraday constant, c i Is the ion concentration of the i-th ion, z i Is the valence charge number, D, of the i-th ion i Is diffusion coefficient of i-th ion, alpha i Simplified Soret coefficient for the ith ion (where i =1 represents a cation and i =2 represents an anion), σ is the nanochannel surface charge density, L and R represent the nanochannel length and radius, respectively, u is the velocity, R is the ion density g Is the general gas constant, T is the temperature, p is the pressure, μ is the dynamic viscosity, ρ is the density, C p Is specific heat, k f And k s Thermal conductivity coefficient of fluid and solid respectively, sigma f Is the electrical conductivity,. DELTA.T is the temperature difference, C h Is high side concentration, C l The concentration is low concentration side concentration. With superscripted parameters being dimensionless parameters, e * ，/>
φ * ，c i * ，u * ，D i * ，p * ，μ * ，σ f * Theta is respectively a dimensionless dielectric constant, a dimensionless partial differential operator, a dimensionless potential, a dimensionless concentration, a dimensionless speed, a dimensionless diffusion coefficient, a dimensionless pressure, a dimensionless viscosity, a dimensionless conductivity, a dimensionless excess temperature, a parameter with a subscript of m is a characteristic parameter, D is i，m C is a characteristic diffusion coefficient (where i =1 represents a cation, i =2 represents an anion), C m ，μ m ，ε m ，T m ，σ f，m Respectively the characteristic concentration,Characteristic viscosity, characteristic dielectric constant, characteristic temperature, characteristic conductivity. In particular, the characteristic concentration C m Taking as a low concentration value, a characteristic temperature T m Taken as the mean temperature, characteristic diffusion coefficient D i，m Intrinsic viscosity [ mu ] m Characteristic dielectric constant ε m Characteristic conductivity σ f，m The physical property values at the characteristic temperature were obtained.
In the prediction method, the dimensionless array is simplified as follows:
where P is power, η is efficiency, P is * Is a dimensionless power.
In the prediction method, a dimensionless array is ensured to be unchanged by using a similarity principle, parameter values in the dimensionless array are changed to obtain corresponding output performance under different working conditions, the sample is expanded by a limited number of experimental samples, and a large amount of requirements of machine learning on the number of samples in an original data set are met, wherein the method specifically comprises the following steps:
η j ＝η 0
wherein, the subscript j represents the sample to be extended, the subscript 0 represents the known sample, the input and output parameters of the known sample are known, and the input parameter of the extended sample is, for example, the diffusion coefficient D j Concentration of C j Surface charge density σ j Nano channel radius R j Dielectric constant ε j It is known to obtain the power and efficiency of the samples to be extended according to the above formula, and to implement the sample extension.
In the prediction method, the dimensional variables of the input parameters are normalized to be (0,1) intervals, and the formula is as follows:
wherein x represents a sampleOriginal value of data, x min Represents the minimum value of the sample data, x max Representing the maximum value of the sample data.
In the prediction method, a graphic processor is adopted for parallel acceleration, a ReLU function is adopted for an activation function to realize nonlinearity, updating of weight w and bias b is realized through a self-adaptive moment estimation and back propagation algorithm, structural parameter optimization is realized, mean Square Error (MSE) is adopted for a loss function, and average relative error (MRE) is adopted for an error function, wherein the calculation formulas of the ReLU, the MSE and the MRE are respectively expressed as:
wherein x is the value of the input variable, n is the number of samples per batch, y target Is a sample label value, y output And predicting the value of the network model.
In the prediction method, the hyper-parameter optimization comprises the optimization of the number of network layers, the number of neurons in each layer, the learning rate of an optimizer and the size of a mini-batch.
Has the advantages that:
according to the method, through dimensionless formulation of the ion permeation power generation control equation, the dimensionless array of the ion permeation power generation physical process is provided, secondary problems are ignored, physical parameters in the control equation are reduced, and the simplified dimensionless array is obtained. The simplified dimensionless array also takes into account the effect of temperature on ion permeation power generation. By utilizing a similarity principle, under the condition that dimensionless control parameters are not changed, one physical condition is expanded to other different conditions by multiplying corresponding factors, and a group of dimensionless samples are converted into a large number of groups of experimental samples, so that the expansion of a physical information database is realized, and the sample requirement of neural network training on a data set is met. The dimensional variable of the physical parameter of the leading factor is used as input through the ANN, the output power and the energy conversion efficiency are used as output parameters, the nonlinear relation among the geometric structure, the concentration difference, the ion physical property and the energy conversion performance of the nano channel is established, and finally the machine learning-based ion permeation power generation prediction model with high precision and high acceleration ratio is obtained.
The description is only an outline of the technical solution of the present invention, and in order to make the technical means of the present invention more clearly understood, the present invention is implemented by those skilled in the art according to the content of the description, and in order to make the description and other objects, features and advantages of the present invention more obvious, the following is exemplified by the specific embodiments of the present invention.
Drawings
Various other advantages and benefits of the present invention will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. It is obvious that the drawings described below are only some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive effort. Also, like parts are designated by like reference numerals throughout the drawings.
In the drawings:
fig. 1 is a schematic connection diagram of a machine learning-based ion-permeable power generation prediction system according to an embodiment of the present disclosure;
FIG. 2 is a schematic flow chart diagram illustrating a prediction method of a machine learning-based ion-permeable power generation prediction system according to another embodiment of the present disclosure;
FIG. 3 is a schematic diagram of an ANN model according to another embodiment of the present disclosure;
fig. 4 is a graph comparing performance of prediction and simulation results of a model test module of a machine learning-based ion permeation power generation prediction system according to another embodiment of the present disclosure.
The invention is further explained below with reference to the figures and examples.
Detailed Description
Specific embodiments of the present invention will be described in more detail below with reference to fig. 1 to 4. While specific embodiments of the invention are shown in the drawings, it should be understood that the invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
It should be noted that certain terms are used throughout the description and following claims to refer to particular components. As one skilled in the art will appreciate, various names may be used to refer to a component. The description and claims do not intend to distinguish between components that differ in name but not function. In the following description and in the claims, the terms "include" and "comprise" are used in an open-ended fashion, and thus should be interpreted to mean "include, but not limited to. The description which follows is a preferred embodiment of the invention, but is made for the purpose of illustrating the general principles of the invention and not for the purpose of limiting the scope of the invention. The scope of the present invention is defined by the appended claims.
For the purpose of facilitating understanding of the embodiments of the present invention, the following description will be made in detail by taking several specific embodiments as examples with reference to the accompanying drawings, and the drawings are not to be construed as limiting the embodiments of the present invention.
In one embodiment, as shown in fig. 1 to 3, an ion permeation power generation prediction system based on machine learning provided by one embodiment is disclosed, which includes a data acquisition module, a neural network module and a model test module;
the data acquisition module is used for establishing an input and output database of the ion permeation power generation and carrying out certain pretreatment so as to be used by the neural network module.
The neural network module is used for constructing a prediction neural network of the ion permeation power generation and performing neural network training by using the database in the data acquisition module.
The model testing module is used for adjusting the hyper-parameters in the neural network prediction model and obtaining the ion permeation power generation prediction model with high precision and high acceleration ratio.
The prediction system is characterized in that the data acquisition module comprises an ion permeation power generation dimensional equation, a dimensionless array, a simplified dimensionless array, a sample expansion database and normalization preprocessing.
Further, the dimensional governing equation is based on the prediction system.
Poisson equation:
nernst-planck equation:
/>
continuity equation:
the Navier-Stokes equation:
energy equation:
wherein, the first and the second end of the pipe are connected with each other,
is a partial differential operator,. Epsilon.is the dielectric constant,. Phi.is the potential,. F is the Faraday constant, c i Is the ion concentration of the i-th ion, z i Is the valence charge number, D, of the i-th ion i Is diffusion coefficient of i-th ion, alpha i Simplified Soret coefficient for the ith ion (where i =1 represents a cation and i =2 represents an anion). u is the velocity, R g Is the general gas constant, T is the temperature, pIs pressure, mu is dynamic viscosity, rho is density, c p Is specific heat, k f Is the thermal conductivity of the fluid, σ f E is the electric field strength, which is the conductivity of the fluid.
Furthermore, according to the prediction system, dimensionless analysis is carried out on the ion permeation energy conversion so as to unify multiple physical parameters of the ion permeation energy conversion. Obtaining a dimensionless control equation and a dimensionless array corresponding to the dimensionless control equation by introducing dimensionless variables:
dimensionless poisson equation:
dimensionless nernst-planck equation:
dimensionless continuity equation:
dimensionless navier-stokes equation:
dimensionless energy equation:
wherein, sigma is the surface charge density of the nano-channel, and delta T is the temperature difference. L and R represent nanochannel length and radius, respectively. k is a radical of f And k s Fluid and solid thermal conductivities, respectively. The parameters marked with the indices are dimensionless parameters. Epsilon * ，
φ * ，c i * ，u * ，D i * ，p * ，μ * ，σ f * Dimensionless dielectric constant, dimensionless partial differential operator, dimensionless potential, dimensionless concentration of the ith ion, dimensionless velocity, dimensionless diffusion coefficient, dimensionless pressure, dimensionless viscosity, dimensionless conductivity of the fluid, respectively. Dimensionless parameters are each defined as the ratio of a dimensional parameter to a characteristic parameter. Theta is the dimensionless passing temperature and is defined as (T-T) m ) and/Delta T. The parameters with m subscripts are characteristic parameters. C m ，D i，m ，μ m ，ε m ，T m ，σ fm Respectively, the characteristic concentration, the characteristic diffusion coefficient of the ith ion, the characteristic viscosity, the characteristic dielectric constant, the characteristic temperature and the characteristic conductivity of the fluid. Characteristic concentration C m Can be taken as a low concentration value, characteristic temperature T m The average temperature may be taken. Characteristic diffusion coefficient D i，m Intrinsic viscosity [ mu ] m Characteristic dielectric constant ε m Characteristic conductivity σ f，m The values are physical properties at a specific temperature.
According to a similar principle, a dimensionless number of the physical process of ion permeation power generation is provided, and an initial dimensionless array consisting of 12 dimensionless parameters is obtained:
wherein, C h And C l Representing high and low ion concentrations, respectively.
Further, based on the basic assumptions of the physical problem, there are:
therefore, the influence of convection and Joule heat can be ignored, dimensionless physical parameters in a control equation are reduced, a univalent electrolyte solution is used as a reference, and a simplified dimensionless array is obtained, so that the unification of the physical parameters in the ion permeation power generation process is realized, and the method can be used for guiding a modeling experiment. After simplification, a dimensionless array consisting of 6 dimensionless parameters is obtained:
in the specific embodiment, according to the similarity principle, when the same dimensionless arrays are provided, the dimensionless arrays are kept unchanged, and under different physical working conditions, the dimensionless output power P and the dimensionless efficiency η can be obtained through the following reverse derivation:
η j ＝η 0
wherein, the subscript j represents the sample to be extended, the subscript 0 represents the known sample, and the input and output parameters of the known sample are known. Input parameters for the extended samples, e.g. diffusion coefficient D j Concentration C of j Surface charge density σ j Radius of nanochannel R j Dielectric constant ε j Are known. On the premise of no dimensional number change, the power and efficiency of the sample to be expanded can be obtained according to the formula, and the sample expansion is realized so as to meet the large demand of machine learning on the number of samples in the original data set.
In this embodiment, according to the prediction system, the dimensional variables of the input parameters are normalized to be (0,1) intervals, so as to eliminate adverse effects caused by singular sample data, and the specific formula is as follows:
wherein x represents the original value of the sample data, x min Represents the minimum value of the sample data, x max Representing the maximum value of the sample data.
After normalization processing, the speed of gradient descent optimal solution can be increased, and the convergence of the training network is improved.
The prediction system according to, wherein the neural network module comprises training set input, full connection layer (hidden layer), structure parameter optimization.
Further, according to the prediction system, an artificial neural network ANN model is adopted to construct a training model, the training model is composed of an input layer, an output layer and a hidden layer, wherein the output power P and the energy conversion efficiency eta are output parameters, and the rest physical parameters, such as concentration C and high-temperature end temperature T h Low temperature end temperature T 1 The ion diffusion coefficient D, the dielectric constant epsilon, the nano-channel length L, the nano-channel radius R and the dimensional variables of the surface charge density sigma are taken as input so as to realize the nonlinear regression from the input to the output.
Further, according to the prediction system, the activation function selects the ReLU to realize the non-linearity and avoid the phenomena of gradient disappearance and the like by adopting the GPU for parallel acceleration. And updating the weight w and the bias b through an Adam optimizer and a back propagation BP algorithm, and realizing structural parameter optimization. The loss function adopts mean square error MSE, and the error function adopts average relative error MRE. The calculation formulas of MSE and MRE are respectively expressed as:
where x is the value of the input variable, n is the number of samples per batch, y target Is a sample tag value, y output And predicting the value of the network model. And (4) gradually iterating and updating the parameter weight of each layer through back propagation of the error to finally obtain the minimum Loss, and training until convergence to represent that the initial training of the model is finished.
Further, the prediction system according to, wherein the model testing module comprises a hyper-parametric optimization and ion-permeation power generation prediction model.
Further, according to the prediction system, the hyper-parameter optimization mainly comprises optimization of the number of network layers, the number of neurons in each layer, the learning rate of an optimizer and the size of a mini-batch.
And testing the trained network by using the test set, and modifying the model according to the test result. And if the overfitting phenomenon occurs, adjusting the network structure parameters and the model hyperparameters conveniently until the precision meets the requirement. If the test effect is good, the parameter data of the neural network and the output result thereof are stored.
Further, according to the prediction system, the ion permeation power generation prediction model can achieve high-precision and high-acceleration-ratio results and guide the actual process of ion permeation power generation.
In another example, as shown in fig. 2, the present disclosure provides a machine learning-based ion-permeable power generation prediction system and a construction method thereof, including the following steps:
s100: the measuring unit measures a plurality of physical parameters of the ion permeation power generation, an ion permeation power generation dimensional equation is constructed based on the physical parameters, and the dimensionless conversion unit generates an ion permeation power generation dimensionless equation and a corresponding dimensionless array thereof based on the ion permeation power generation dimensional equation;
s200: the sample expansion unit is used for expanding based on the dimensionless array by adopting a similarity principle to obtain an expansion sample, the expansion sample and the dimensionless array form sample data, and the normalization processing unit normalizes the sample data;
s300: the sample data is divided into training data and test data according to a preset proportion, the neural network module is configured to construct a prediction neural network of the ion permeation power generation based on the artificial neural network, the neural network training is carried out based on the sample data to construct the prediction neural network of the ion permeation power generation, wherein the output power P and the energy conversion efficiency eta are output parameters, and the rest physical parameters, such as concentration C and high-temperature end temperature T, are used as output parameters h Low temperature end temperature T 1 Of the ion diffusion systemD, dielectric constant epsilon, nano-channel length L, nano-channel radius R and dimension variables of surface charge density sigma are taken as input;
s400: the super-parameter optimization unit adopts GPU acceleration, activates function nonlinearity and combined optimization of a back propagation algorithm, updates weight w and bias b to optimize a prediction neural network of ion permeation power generation, and evaluates a data processing result through a loss function MSE;
s500: and the test unit tests the optimized prediction neural network of the ion permeation power generation based on the test data, adjusts the hyper-parameters in the network according to the training result, repeats iteration until the error reaches the expectation, and saves the model as the prediction model of the ion permeation power generation.
In another example, as shown in fig. 3, an ANN model structure diagram provided by an embodiment of the present disclosure is shown.
The calculation example of the invention can directly read the concentration C and the temperature T at the high temperature end h Low temperature end temperature T l The ion diffusion coefficient D, the dielectric constant epsilon, the nano-channel length L, the nano-channel radius R and the surface charge density sigma eight characteristic parameters which are dominant factors are used as input, and two output parameters of the output power P and the energy conversion efficiency eta which correspond to the characteristic parameters are used as label values of the training sample.
Before model training, input and output parameters of a data set are normalized to be (0,1) intervals, so that the numerical problem can be avoided, and the network can be rapidly converged.
After the network successfully reads the data set, training a neural network under an ANN deep learning framework to realize a nonlinear regression task. The network layer number is 6, and the sub-structure comprises an input layer, an output layer and four hidden layers. The dimension of the input parameters is 8, the dimension of the output parameters is 2, and the intermediate layer parameters are 1000, 500, 200 and 20 respectively. The input layer neuron receives an input signal, and each neuron of the hidden layer and the output layer is fully connected with all neurons of the adjacent layer.
The input of each neuron in the hidden layer and the output layer is the weighted sum of the output values of all neurons in the previous layer. Let x be m n Is the input value, y, of the mth neuron of the nth layer in the neural network m n And b m n Respectively the output value and the offset of the neuron, w im n-1 The connection weight of the neuron and the ith neuron of the n-1 layer is as follows:
where σ () represents an activation function.
The activation function adopts ReLU to avoid gradient disappearance and realize a nonlinear network, and the formula of ReLU is expressed as follows:
where x represents the input variable value. ReLU will make the output of some neurons 0, making the network sparse and reducing the interdependence of parameters, alleviating the over-fitting problem. After the feature extraction layer by layer, the regression result of the neural network is finally output as 2 parameters.
The loss function adopts Mean Square Error (MSE), the error function adopts average relative error (MRE), and convergence errors of the samples on the training data set and the testing data set are respectively calculated. The network is trained by the Adam optimizer, and the learning rate is set to 0.001/(1 + epoch × β), where epoch is the training number and β =0.001. And (4) error back propagation, gradually iterating and updating the parameter weights of all layers, finally obtaining the minimum loss, training until convergence, and representing that the initial training of the model is finished.
In another embodiment, as shown in fig. 4, an embodiment of the present disclosure provides a performance comparison graph of model test module prediction and simulation results of a machine learning-based ion-permeation power generation prediction system. In order to quantitatively measure the prediction effect of the ion permeation power generation prediction model, so that high precision and high acceleration ratio can be realized, and the actual process of ion permeation power generation can be effectively guided, taking output power P as an example, five experimental samples (Case 1-5) are randomly selected from a data set to perform performance comparison of the prediction result of the ion permeation power generation prediction system based on machine learning and the experimental simulation result, and the results are shown in FIG. 4 and Table 1:
TABLE 1 Performance comparison of model prediction results and Experimental simulation results
As can be seen from fig. 4 and table 1, for five randomly chosen experimental samples, the total time consumption of the prediction model for the output power results is much less than that of the experimental simulation, 2047 seconds is saved, the acceleration ratio is up to 683, and the maximum error of the prediction model is only 2.39%. The result fully proves that the prediction model has a good prediction effect and certain practicability, and effectively guides the actual process of ion permeation power generation.
The prediction method describes physical mathematical characteristics of ion permeation power generation energy conversion, and provides a simplified dimensionless array of the ion permeation power generation physical process through dimensionless control equation. Under the guidance of dimensionless numbers, a group of dimensionless samples are expanded into a plurality of groups of samples, so that the expansion of a physical information database is realized, and the sample requirement of neural network training on a data set is met. On the basis, an ANN prediction model is constructed based on machine learning by taking dimension variables of the physical parameters of the main factors subjected to normalization preprocessing as input and taking the efficiency of output power and energy conversion efficiency as output parameters, the accuracy of the prediction model reaches 97.6% through combined optimization of GPU acceleration, an activation function, an optimizer and a back propagation algorithm, the hyperparameter in the network is adjusted according to a training result, and finally the ion osmotic power generation prediction model based on machine learning with high precision and high acceleration ratio is obtained and can guide judgment of the main parameters of the osmotic power generation system and estimation of the efficiency.
Although embodiments of the present invention have been described above with reference to the accompanying drawings, the present invention is not limited to the specific embodiments and the application fields, and the specific embodiments are illustrative, instructive and not restrictive. Those skilled in the art, having the benefit of this disclosure, may effect numerous modifications to the disclosed embodiments without departing from the scope of the invention as defined by the appended claims.
Claims (9)
1. A prediction system of ion permeation power generation based on machine learning is characterized in that: which comprises the steps of preparing a mixture of a plurality of raw materials,
a data acquisition module configured to acquire sample data of ion permeation power generation, the data acquisition module comprising;
a measurement unit that measures a plurality of physical parameters of the ion permeation power generation;
a dimensionless transformation unit connected to the measurement unit and generating a plurality of dimensionless parameters based on the plurality of physical parameters and forming a dimensionless array based on the dimensionless parameters;
the sample expansion unit is connected with the dimensionless conversion unit and is expanded by adopting a similarity principle based on the dimensionless array to obtain an expansion sample, the expansion sample and the dimensionless array form sample data, and the sample data is divided into training data and test data according to a preset proportion;
a normalization processing unit connected to the sample expansion unit and normalizing the sample data;
the neural network module is configured to construct a prediction neural network and neural network training of ion permeation power generation based on an artificial neural network ANN, the neural network module connected with the data acquisition module comprises a plurality of network layers, the substructure of the neural network module comprises an input layer, an output layer and four full-connection layers, neurons of the input layer receive training data, each neuron of the hidden layer and the output layer is in full connection with all neurons of the adjacent layer, and the neural network training is carried out based on sample data to form the prediction neural network of ion permeation power generation;
a model testing module configured to adjust hyper-parameters in a predictive neural network of ion osmotic power generation to obtain an ion osmotic power generation predictive model, the model testing module comprising;
the hyper-parameter optimization unit is connected with the neural network module, adopts a Graphic Processing Unit (GPU) to perform parallel acceleration, activates the combination optimization of function nonlinearity and back propagation algorithm, and updates the weight w and the bias b to optimize the prediction neural network of the ion permeation power generation;
and the testing unit is connected with the data acquisition module and the neural network module so as to test the prediction neural network of the ion permeation power generation based on the test data until the error reaches the expectation, and the model is saved as the ion permeation power generation prediction model.
2. The machine learning-based prediction system for ion-osmotic power generation of claim 1, wherein preferably: the measuring unit includes a first measuring portion that obtains a dielectric constant, a concentration, an ion diffusion coefficient, a nanochannel length, a radius, and a surface charge density, a temperature sensor that measures temperature data, a power level that measures output power P, and a second measuring portion that measures energy conversion efficiency η.
3. A prediction method using the prediction system for machine learning-based ion osmotic power generation according to claim 1 or 2, characterized in that it comprises,
s100: the measuring unit measures a plurality of physical parameters of the ion permeation power generation, an ion permeation power generation dimensional equation is constructed based on the physical parameters, and the dimensionless conversion unit generates an ion permeation power generation dimensionless equation and a corresponding dimensionless array thereof based on the ion permeation power generation dimensional equation;
s200: the sample expansion unit is used for expanding based on the dimensionless array by adopting a similarity principle to obtain an expansion sample, the expansion sample and the dimensionless array form sample data, and the normalization processing unit normalizes the sample data;
S300: the sample data is divided into training data and test data according to a preset proportion, the neural network module is configured into a prediction neural network for constructing the ion permeation power generation based on the artificial neural network, the neural network training is carried out based on the sample data to form the prediction neural network for the ion permeation power generation, wherein the output power P and the energy conversion efficiency eta are output parameters, the concentration C and the temperature T at the high-temperature end are used as output parameters, and the temperature T at the high-temperature end is used as a temperature parameter h Low temperature end temperature T l Dimensional variables of an ion diffusion coefficient D, a dielectric constant epsilon, a nano-channel length L, a nano-channel radius R and a surface charge density sigma are taken as input;
s400: the hyper-parameter optimization unit adopts a Graphic Processing Unit (GPU) for acceleration, activates the combination optimization of function nonlinearity and back propagation algorithm, and updates the weight w and the bias b to optimize the prediction neural network of the ion permeation power generation;
s500: and the test unit tests the optimized prediction neural network of the ion permeation power generation based on the test data, adjusts the hyper-parameters in the network according to the training result, repeats iteration until the error reaches the expectation, and saves the model as the prediction model of the ion permeation power generation.
4. The prediction method according to claim 3, characterized in that: the dimensionless control equation and the dimensionless array corresponding to it are:
dimensionless poisson equation:
dimensionless nernst-planck equation:
dimensionless continuity equation:
dimensionless navier-stokes equation:
dimensionless energy equation:
dimensionless array:
wherein the content of the first and second substances,
is a partial differential operator,. Epsilon.is the dielectric constant,. Phi.is the potential,. F is the Faraday constant, c i Is the ion concentration of the i-th ion, z i Is the number of valence charges of the i-th ion, D i Is diffusion coefficient of i-th ion, alpha i Simplified Soret coefficient for the ith ion, where i =1 represents a cation, i =2 represents an anion, σ is the nanochannel surface charge density, L and R represent the nanochannel length and radius, respectively, u is the velocity, R g Is the general gas constant, T is the temperature, p is the pressure, μ is the dynamic viscosity, ρ is the density, C p Is specific heat, k f And k s Thermal conductivity coefficient of fluid and solid respectively, sigma f Is the electrical conductivity,. DELTA.T is the temperature difference, C h Is high side concentration, C l For low concentration side concentrations, the parameters marked with a letter are dimensionless parameters, epsilon * ，/>
φ * ，c i * ，u * ，D i * ，p * ，μ * ，σ f * Theta is the dimensionless dielectric constant, the dimensionless partial differential operator, the dimensionless potential, the dimensionless concentration, the dimensionless speed, and the dimensionless spread respectivelyScattering coefficient, dimensionless pressure, dimensionless viscosity, dimensionless conductivity, dimensionless excess temperature, parameter with m index as characteristic parameter, D i，m Is a characteristic diffusion coefficient, C m ，μ m ，ε m ，T m ，σ f，m Respectively, characteristic concentration, characteristic viscosity, characteristic dielectric constant, characteristic temperature and characteristic conductivity.
5. The prediction method according to claim 4, wherein: the dimensionless array is simplified as:
where P is power, η is efficiency, P * Is a dimensionless power.
6. The prediction method of claim 5, wherein a similarity principle is used to ensure that dimensionless arrays are unchanged, parameter values therein are changed to obtain corresponding output performance under different working conditions, and the samples are expanded by a limited number of experimental samples, so as to satisfy the requirement of machine learning on the number of samples in the original data set, as follows:
η j ＝η 0
wherein, the subscript j represents the sample to be extended, the subscript 0 represents the known sample, the input and output parameters of the known sample are known, and the input parameter of the extended sample is, for example, the diffusion coefficient D j Concentration C of j Surface charge density σ j Nano channel radius R j Dielectric constant ε j It is known to obtain the power and efficiency of the samples to be extended according to the above formula, and to implement the sample extension.
7. The prediction method according to claim 3, characterized in that: normalizing the dimension variables of the input parameters, wherein the normalization is an interval (0,1), and the formula is as follows:
wherein x represents the original value of sample data, x min Represents the minimum value of the sample data, x max Representing the maximum value of the sample data.
8. The prediction method according to claim 3, characterized in that: the image processor is adopted for parallel acceleration, the ReLU function is adopted as an activation function to realize nonlinearity, updating of weight w and bias b is realized through adaptive moment estimation and a back propagation algorithm, structural parameter optimization is realized, mean Square Error (MSE) is adopted as a loss function, and average relative error (MRE) is adopted as an error function, wherein the calculation formulas of ReLU, MSE and MRE are respectively expressed as follows:
wherein x is the value of the input variable, n is the number of samples per batch, y target Is a sample tag value, y output And predicting the value of the network model.
9. The prediction method according to claim 3, characterized in that: the hyper-parameter optimization comprises optimization of the number of network layers, the number of neurons in each layer, the learning rate of an optimizer and the size of the mini-batch.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211618751.0A CN115936227A (en) | 2022-12-15 | 2022-12-15 | Machine learning-based ion permeation power generation prediction system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211618751.0A CN115936227A (en) | 2022-12-15 | 2022-12-15 | Machine learning-based ion permeation power generation prediction system and method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115936227A true CN115936227A (en) | 2023-04-07 |
Family
ID=86655624
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211618751.0A Pending CN115936227A (en) | 2022-12-15 | 2022-12-15 | Machine learning-based ion permeation power generation prediction system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115936227A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117393069A (en) * | 2023-11-06 | 2024-01-12 | 上海赫耳墨锶科技有限公司 | Method for determining electrolysis control data of target metal based on neural network |
| CN117476125A (en) * | 2023-12-27 | 2024-01-30 | 豆黄金食品有限公司 | Dried beancurd stick raffinate recovery data processing system based on data analysis |
-
2022
- 2022-12-15 CN CN202211618751.0A patent/CN115936227A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117393069A (en) * | 2023-11-06 | 2024-01-12 | 上海赫耳墨锶科技有限公司 | Method for determining electrolysis control data of target metal based on neural network |
| CN117476125A (en) * | 2023-12-27 | 2024-01-30 | 豆黄金食品有限公司 | Dried beancurd stick raffinate recovery data processing system based on data analysis |
| CN117476125B (en) * | 2023-12-27 | 2024-04-05 | 豆黄金食品有限公司 | Dried beancurd stick raffinate recovery data processing system based on data analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115936227A (en) | Machine learning-based ion permeation power generation prediction system and method | |
| CN101630376B (en) | Soft-sensing modeling method and soft meter of multi-model neural network in biological fermentation process | |
| CN109828532A (en) | A kind of Prediction of Surface Roughness method and process parameter optimizing method based on GA-GBRT | |
| CN110807554B (en) | Generation method and system based on wind power/photovoltaic classical scene set | |
| CN108596369B (en) | Space load prediction method based on multivariate model and blind number theory | |
| CN104462850A (en) | Multi-stage batch process soft measurement method based on fuzzy gauss hybrid model | |
| CN107528824B (en) | Deep belief network intrusion detection method based on two-dimensional sparsification | |
| CN112836884A (en) | Copula-DBiLSTM-based method for accurately predicting multi-element load of comprehensive energy system | |
| CN112541571A (en) | Injection-production connectivity determination method based on machine learning of double parallel neural networks | |
| CN108520310A (en) | Wind speed forecasting method based on G-L mixed noise characteristic v- support vector regressions | |
| CN107729988B (en) | Blue algae bloom prediction method based on dynamic deep belief network | |
| CN107045659A (en) | Predict the method and device of photovoltaic generation power | |
| CN115841076A (en) | Shallow sea layered seabed ground sound parameter inversion method based on BP neural network model | |
| CN114242156A (en) | Real-time prediction method and system for relative abundance of pathogenic vibrios on marine micro-plastic | |
| CN114447378A (en) | Parameter optimization method of proton exchange membrane fuel cell | |
| CN110245398B (en) | Soft measurement deep learning method for thermal deformation of air preheater rotor | |
| CN106405683A (en) | Wind speed forecasting method and device based on G-L mixed noise characteristic kernel ridge regression technology | |
| Zhu et al. | Structural safety monitoring of high arch dam using improved ABC‐BP model | |
| CN114202060A (en) | Method for predicting methylene blue adsorption performance of biomass activated carbon based on deep neural network | |
| Vaněk et al. | On-line estimation of biomass concentration using a neural network and information about metabolic state | |
| CN105787507B (en) | LS SVMs on-line study methods based on budget supporting vector collection | |
| CN111061151A (en) | Distributed energy state monitoring method based on multivariate convolutional neural network | |
| Yang et al. | Prediction and optimization model of activated carbon double layer capacitors based on improved heuristic approach genetic algorithm neural network | |
| CN116401913A (en) | Design and optimization method of hydrogel-based negative hydration swelling metamaterial | |
| Pramanik | Use of artificial neural networks for prediction of cell mass and ethanol concentration in batch fermentation using saccharomyces cerevisiae yeast |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |