A CNN-based nano-ion electroosmosis energy conversion power prediction method and system

Abstract

The application discloses a kind of based on CNN's nano-ion electrophoresis energy conversion power prediction method and system, collect the physical field parameter of nano-ion electrophoresis energy conversion device under different operating conditions to construct training dataset;Multi-layer multi-dimensional CNN model is constructed, and the training dataset after pre-processing and standardization is used to train CNN model;The physical field parameter of nano-ion electrophoresis energy conversion device and corresponding nano geometry structure parameter are re-collected to construct test sample dataset, and are pre-processed and standardized;Test dataset is input into trained CNN model, and the prediction result of maximum power generation is obtained, realize the accurate prediction of nano-ion electrophoresis energy conversion power.The depth of nano-ion electrophoresis energy conversion power prediction model is reduced, the model is lightened, and the maximum power generation of nano-ion electrophoresis energy conversion device under different conditions has higher prediction accuracy, and superior generalization ability is shown.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of nano-ion electro-osmotic energy conversion, and particularly relates to a nano-ion electro-osmotic energy conversion power prediction method and system based on a CNN. BACKGROUND

[0002] In the current world, energy transformation has entered a full acceleration period from a start-up and accumulation period. Therefore, the non-renewable and high-pollution problems of fossil energy need to be solved by new sustainable development methods. In the field of new energy, it is of great significance to efficiently convert a large amount of solar energy and salt difference energy in nature into electric energy. Since the 1950s, inspired by the principle of electric eel attack and prey hunting, scholars have carried out extensive research on ion electro-osmotic energy conversion. In order to strengthen the directional migration process of ions in the nano-channel, some scholars have experimentally built a nano-channel ion electro-osmotic energy conversion device to qualitatively and quantitatively study the energy conversion process. However, in the research process, the analysis of the power generation of the energy conversion device under different working conditions needs to artificially change the geometric structure of the nano-channel and the parameters of each physical field in the device, which consumes a lot of manpower and material resources and has low efficiency. Therefore, it is of great research significance and application value to directly and effectively predict the maximum power generation of the conversion device through the geometric parameters of the nano-channel and other physical field parameters in a short period of time.

[0003] With the advent of the big data era, artificial intelligence has been applied in various fields due to its high efficiency, intelligence, and cost savings. Among them, the deep learning networks such as convolutional neural network and recurrent neural network have shown strong feature learning ability, which can extract the internal connection characteristics between the nano-pore geometric structure, physical field parameters and the power generation of the device in the ion electro-osmotic energy conversion process, so as to fit the optimal network weight to predict the maximum power generation of the device. Therefore, the prediction of the power generation of the nano-ion electro-osmotic energy conversion by the deep learning technology has important guiding significance for the optimization design of the ion electro-osmotic energy conversion device, and can effectively improve the efficiency of the electro-osmotic energy conversion, which has important scientific significance and application value. SUMMARY

[0004] The technical problem to be solved by the present application is to provide a nano-ion electro-osmotic energy conversion power prediction method and system based on a CNN to solve the technical problems of large consumption of manpower and material resources and low efficiency in the simulation verification process in the field of nano-ion electro-osmotic energy conversion by establishing a multi-layer multi-dimensional CNN to accurately predict the maximum power generation of the nano-ion electro-osmotic energy conversion device under different working conditions, and accurately predicting the power generation under different working conditions of the existing conversion device through the learned characteristics.

[0005] The application adopts the following technical scheme:

[0006] A CNN-based nano-ion electroosmosis energy conversion power prediction method, comprising the following steps:

[0007] S1, collecting the temperature field, concentration field and surface charge density data of the nano-ion electroosmosis energy conversion device under different operating conditions;

[0008] S2, the temperature field, concentration field and surface charge density data obtained in step S1, and the nano-channel and nano-pore geometric structure parameters of the nano-ion electroosmosis energy conversion device are discretely sampled, and after pretreatment and standardization, a training data set is constructed;

[0009] S3, constructing a multi-dimensional deep CNN model as a prediction model for the power generation of the nano-ion electroosmosis energy conversion;

[0010] S4, using the training data set obtained in step S2 to train the multi-dimensional deep CNN model constructed in step S3;

[0011] S5, re-collecting the temperature field, concentration field and surface charge density data of the nano-ion electroosmosis energy conversion device, and after pretreatment and standardization, as a test data set; input the test data set into the multi-dimensional deep CNN model obtained in step S4 to predict the maximum power generation under the test condition, and realize the prediction of the power generation of the ion electroosmosis energy conversion.

[0012] Specifically, in step S1, the collected data includes the radius of the nano-channel large hole / small hole, the length / width of the channel internal nano-pore geometric structure, the channel surface charge density, the temperature and the concentration ratio of the two side liquid tanks.

[0013] Specifically, in step S2, the temperature field and concentration field inside the nano-ion electroosmosis energy conversion device of different types are discretely valued, and they are combined with the corresponding nano-channel and nano-pore geometric structure parameters to construct a training sample data set.

[0014] Further, the standardization process is as follows:

[0015]

[0016] Wherein, is the average value of all physical quantities in the i-th group of data, and i is the standard deviation of all physical quantities in the i-th group of data.

[0017] Specifically, in step S3, the multi-dimensional deep CNN includes an input layer, a multi-dimensional feature extraction layer, a full connection layer and an output layer; all initial network weights and biases in the multi-layer multi-dimensional CNN are randomly generated and subject to normal distribution.

[0018] Further, the input layer: receives the seven-dimensional training data after standardization processing;

[0019] The multi-dimensional feature extraction layer: includes five parallel one-dimensional convolution layers, one feature merging layer, an excitation layer and one two-dimensional convolution layer, the output of each one-dimensional convolution layer is taken as the input of the feature merging layer;

[0020] The full connection layer: takes the features extracted by the feature extraction layer as the input value, and outputs one element representing the maximum power predicted by the network model.

[0021] Further, the parallel one-dimensional convolution layer: the input channel number is 1, the output channel number is 16, and the convolution kernel size is 1, 1, 3, 3 and 5 respectively;

[0022] The feature merging layer: merges the features extracted by the parallel one-dimensional convolution layer in the dimension of the one-dimensional convolution channel;

[0023] The excitation layer: the excitation function is set to ReLU function;

[0024] The two-dimensional convolution layer: the input channel number is 16, the output channel number is 32, and the convolution kernel size is 3;

[0025] Input x in Multi-dimensional feature extraction is performed by multiple one-dimensional convolution layers, and the final output x is determined by the two-dimensional convolution layer out .

[0026] Further, the operation process of the multi-dimensional feature extraction layer is represented as:

[0027] x in =[T,C,Q s ,r l ,r s ,P l ,P w ] T

[0028]

[0029]

[0030] x out =K conv,2D (W conv,2D ,C L )+b conv,2D

[0031] Wherein, T and C are nanometer ion electrophoresis energy conversion temperature and concentration ratio respectively; Q s is the surface charge density of nanochannel; r l , r srespectively are the large and small pore radii of the nanochannel; P l , P w respectively are the height and width of the nanopore geometry; is the output of the i-th one-dimensional convolutional layer; W conv,2D respectively are the weights of the i-th one-dimensional convolutional layer and the two-dimensional convolutional layer; C L is the output of the feature merging layer; b conv,2D respectively are the biases of the i-th one-dimensional convolutional layer and the two-dimensional convolutional layer.

[0032] Specifically, in step S4, the number of training rounds of the multi-dimensional deep CNN model is 100 rounds, and the learning rate decay value is set to 0.5 every 50 iterations during the training process.

[0033] In a second aspect, an embodiment of the present application provides a CNN-based nanometer iontophoresis energy conversion power prediction system, comprising:

[0034] The acquisition module acquires temperature field, concentration field and surface charge density data of the nanometer iontophoresis energy conversion device under different operating conditions;

[0035] The processing module discretely samples the temperature field, concentration field and surface charge density data obtained by the acquisition module, and the nanochannel and nanopore geometry parameters of the nanometer iontophoresis energy conversion device, and constructs a training data set after preprocessing and standardization;

[0036] The construction module constructs a multi-dimensional deep CNN model as a prediction model for the power generation of the nanometer iontophoresis energy conversion device;

[0037] The training module trains the multi-dimensional deep CNN model constructed by the construction module using the training data set obtained by the processing module;

[0038] The prediction module reacquires the temperature field, concentration field and surface charge density data of the nanometer iontophoresis energy conversion device, and inputs the test data set into the multi-dimensional deep CNN model obtained by the training module to predict the maximum power generation under the test condition, thereby realizing the prediction of the power generation of the iontophoresis energy conversion device.

[0039] Compared with the prior art, the present application has at least the following beneficial effects:

[0040] A CNN-based nanometer iontophoresis energy conversion power prediction method is proposed, which uses a multi-dimensional CNN to predict the power of the nanometer iontophoresis energy conversion device, and can accurately predict the maximum power generation of the conversion device under different conditions.

[0041] Further, a simulation model of the nano-iontophoresis energy conversion device is established, and physical field parameters such as a temperature field and a concentration field in the model are set, and physical field parameters under different working conditions in the power generation process of the device and corresponding nano geometric structure parameters are obtained through numerical simulation to construct a training sample data set.

[0042] Further, the sample data is preprocessed and standardized to eliminate dimensional differences between various parameters in the data, which is beneficial to improve the speed of the root mean square error loss function in gradient descent to obtain an optimal solution, and can avoid the gradient explosion problem in model training.

[0043] Further, the multi-layer multi-dimensional CNN architecture can effectively extract the internal features of the nano-iontophoresis energy conversion device parameters, and can greatly improve the accuracy of the prediction model.

[0044] Further, the network model structure is reasonably set, and the optimal hyperparameter is selected, which is beneficial to improve the prediction accuracy; and the full connection layer can integrate and refine the extracted features to realize the accurate prediction of the network model.

[0045] Further, the data obtained by using a shorter sampling time to train the network model can effectively extract the features without wasting too much calculation cost.

[0046] It can be understood that the beneficial effects of the above-mentioned second aspect can be referred to the related description in the above-mentioned first aspect, which will not be repeated here.

[0047] In summary, the present application can accurately predict the maximum power generation of the nano-iontophoresis energy conversion device under different structures and different working conditions.

[0048] The technical solutions of the present application will be further described in detail below with the help of the drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 The logical flowchart of the present application is shown in the figure;

[0050] Figure 2 The nano-iontophoresis energy conversion schematic diagram is shown in the figure;

[0051] Figure 3 The multi-dimensional feature extraction layer architecture diagram is shown in the figure;

[0052] Figure 4 The multi-dimensional CNN model architecture diagram is shown in the figure;

[0053] Figure 5 The error trend diagram of the training set and the validation set in the training process of the network model is shown in the figure;

[0054] Figure 6A plot of the test set true power values versus the model predicted power values. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the present application will be apparently and completely described below with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all the other embodiments obtained by those skilled in the art without creative work belong to the protection scope of the present application.

[0056] In the description of the present application, it should be understood that the terms "comprising" and "including" indicate the existence of described features, integers, steps, operations, elements, and / or components, but do not exclude one or more other features, integers, steps, operations, elements, components, and / or sets thereof.

[0057] It should also be understood that the terms used in the specification of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification and the appended claims of the present application, unless otherwise clearly indicated by the context, the singular forms "a", "an" and "the" are intended to include the plural forms as well.

[0058] It should be further understood that the term "and / or" used in the specification and the appended claims of the present application means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations, for example, A and / or B can represent three cases of A alone, A and B together, and B alone. In addition, the character " / " in the present application generally represents an "or" relationship between the front and rear associated objects.

[0059] It should be understood that although the terms first, second, third, etc. may be used in the embodiments of the present application to describe the preset ranges, etc., these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from each other. For example, the first preset range can also be referred to as the second preset range, and similarly, the second preset range can also be referred to as the first preset range, without departing from the scope of the embodiments of the present application.

[0060] Depending on the context, the word "if" as used herein can be interpreted to mean "when" or "while" or "in response to determining" or "in response to detecting". Similarly, the phrase "if determined" or "if detecting (a stated condition or event)" can be interpreted to mean "when determined" or "in response to determining" or "when detecting (a stated condition or event)" or "in response to detecting (a stated condition or event)", depending on the context.

[0061] The various structural diagrams according to the disclosed embodiments of the present application are shown in the drawings. These diagrams are not drawn to scale, in which certain details are exaggerated for clarity and others omitted. The shapes and relative sizes of the various regions, layers, and their relative positions illustrated in the drawings are merely exemplary and may deviate in actuality due to manufacturing tolerances or technical limitations, and regions / layers with different shapes, sizes, and relative positions can be additionally designed according to actual needs by those skilled in the art.

[0062] The application provides a CNN (Convolutional Neural Network)-based nano-ion electroosmosis energy conversion power prediction method, adopts a simulation model as a verification object, specifically establishes a simulation model of a nano-ion electroosmosis energy conversion device, sets physical field parameters such as a temperature field and a concentration field in the model, obtains physical field parameters under different working conditions in a power generation process of the device through numerical simulation, and constructs a training sample data set in combination with corresponding nano geometric structure parameters; a multi-layer multi-dimensional CNN model is established, the sample data set constructed above is used for training the model, so that the model can accurately predict the maximum power generation of the nano-ion electroosmosis energy conversion device under different working conditions; and for a conversion device with a brand-new nano channel and nano pore structure within a certain range, the accurate prediction of the maximum power generation under different working conditions can be realized through the learned characteristics of the existing device.

[0063] Referring to Figure 1 The application provides a CNN-based nano-ion electroosmosis energy conversion power prediction method, which comprises the following steps:

[0064] S1, a simulation model of a nano-ion electroosmosis energy conversion device is established, physical field parameters such as a temperature field and a concentration field in the model are set, physical field parameters under different working conditions in a power generation process of the device are obtained through numerical simulation, and a training sample data set is constructed in combination with corresponding nano geometric structure parameters;

[0065] Referring to Figure 2 As shown in the figure, the application relies on a nano-ion electroosmosis energy conversion device schematic diagram, which comprises two side NaCl solution storage tanks with a concentration difference, a nano channel with a surface nano structure, can more comprehensively provide a reference for simulation of the nano-ion electroosmosis energy conversion device, and meets the verification needs of the application.

[0066] According to Figure 2Referring to, the simulation model of nano-iontophoresis energy conversion device is established by using COMSOL Multiphysics software, and the power generation of the device under different working conditions is simulated. On this basis, according to the existing research (Ren Q, Zhu H, Chen K, et al. Similarity principle based multi-physical parameter unification and comparison in salinity-gradient osmotic energy conversion [J]. Applied Energy, 2022, 307.) Results, the large hole / small hole radius of the nano channel, the height / width of the nano-pore geometry inside the channel, the temperature field temperature in the device, the concentration ratio of the two sides of the NaCl solution storage tank, and the surface charge density of the channel are reasonably selected for sampling, and the training sample data set is constructed. In order to make the sample amount in the sample data set meet the training needs of the subsequent CNN model, 15 different conversion device geometries are set in the simulation model, and the temperature and concentration ratio of the device under each geometry are simulated and randomly sampled within a certain range; Through the above sampling method, the total amount of parameters can reach more than 14000, which meets the training needs of the CNN model.

[0067] S2, the parameters of the nano-iontophoresis energy conversion device collected in step S1 are pretreated and standardized, the pretreatment is to integrate the data and make it into the format supported by the corresponding training model;

[0068] The standardization method is:

[0069]

[0070] Among them, is the average value of all physical quantities in the ith group of data, σ i is the standard deviation of all physical quantities in the ith group of data.

[0071] After standardization, the influence of the dimensional difference between parameters is eliminated, the numerical value of each parameter obeys the normal distribution, which can speed up the gradient descent to find the optimal solution during network training, which is helpful for training; The structure of the data after standardization does not change.

[0072] S3, construct a multi-dimensional deep CNN as a prediction model for the power of nano-iontophoresis energy conversion;

[0073] Please refer to Figure 4The architecture of a multi-layer, multi-dimensional CNN includes an input layer, a multi-dimensional feature extraction layer, a fully connected layer, and an output layer. All initial network weights and biases in a multi-layer, multi-dimensional CNN are randomly generated, follow a normal distribution, and are continuously optimized during network training.

[0074] The architecture of the multi-dimensional CNN model is as follows:

[0075] Input layer: Receives training data containing seven-dimensional physical parameters after standardization;

[0076] Multi-dimensional feature extraction layer:

[0077] It includes 5 one-dimensional convolutional layers and 1 two-dimensional convolutional layer, with the output of each one-dimensional convolutional layer serving as the input to the feature merging layer.

[0078] This structure increases the network depth of the model, enabling the extraction of hidden features from the multidimensional physical parameters of the input from multiple dimensions.

[0079] Fully connected layer: The features extracted by the feature extraction layer are used as input values. The input value is a single element that represents the maximum output power predicted by the network model.

[0080] Please see Figure 3 The architecture of the multi-dimensional feature extraction layer includes five parallel one-dimensional convolutional layers ( (i = 1, 2, ..., 5), 1 feature merging layer, 1 activation layer, and 1 two-dimensional convolutional layer (K conv,2D The settings are as follows:

[0081] Parallel one-dimensional convolutional layer: The number of input channels is set to 1, the number of output channels is set to 16, and the kernel size is set to 1, 1, 3, 3, and 5 respectively.

[0082] Feature merging layer: The features extracted by the parallel one-dimensional convolutional layer are merged along the dimension of the one-dimensional convolutional channel.

[0083] Excitation layer: The excitation function is set to the ReLU function.

[0084] Two-dimensional convolutional layer: the number of input channels is set to 16, the number of output channels is set to 32, and the kernel size is set to 3.

[0085] Enter x in Multi-dimensional feature extraction is performed by multiple one-dimensional convolutional layers, and the final output x is determined by a two-dimensional convolutional layer. out .

[0086] The operation process of the multi-dimensional feature extraction layer is represented as follows:

[0087] x in =[T,C,Q s ,rl r s P l P w ] T

[0088]

[0089]

[0090] x out = K conv,2D (W conv,2D , C L ) + b conv,2D

[0091] where T, C are nanion electroosmosis energy conversion temperature and concentration ratio, respectively; Q s is nanochannel surface charge density; r l , r s are nanochannel large hole radius and small hole radius, respectively; P l , P w are nanopore geometric structure height and width, respectively; is the output of the i-th one-dimensional convolution layer; W conv,2D are the weights of the i-th one-dimensional convolution layer and the two-dimensional convolution layer, respectively; C L is the output of the feature merging layer; b conv,2D are the biases of the i-th one-dimensional convolution layer and the two-dimensional convolution layer, respectively.

[0092] S4, using the training data preprocessed and standardized in step S2 to train the multi-layer multi-dimensional CNN model constructed in step S3;

[0093] The above training data is input into the multi-layer multi-dimensional CNN for training. Since a deep neural network needs to continuously extract features from a large number of samples and optimize the weight parameters in the network to achieve an ideal prediction effect, it is recommended that the number of training sample parameters collected under each working condition be more than 1500. The sample parameters can be set to different temperature fields, concentration fields, and different nanochannel geometric structures to increase the diversity of the samples and make the model more widely applicable. The number of network model training rounds is set to 100 rounds, and the learning rate decay value is set to 0.5 every 50 iterations during the training process.

[0094] S5, the geometry parameters of the nanochannel and nanopore of the nanion electroosmosis energy conversion device are reset, the corresponding temperature field, concentration field and surface charge density data are collected, preprocessed and standardized as a test sample set, the test sample data set is used to predict the maximum power generation, and the prediction accuracy of the model is tested.

[0095] The geometry parameters of the nanochannel and the nanopore of the nanion electrophoresis energy conversion device are re-collected, the collected parameters are pre-processed and standardized, a test data set is constructed, and the trained multi-layer multi-dimensional CNN in step S4 is input for calculation to realize the prediction of the nanion electrophoresis energy conversion power.

[0096] In another embodiment of the present application, a CNN-based nanion electrophoresis energy conversion power prediction system is provided, which can be used to realize the above-mentioned CNN-based nanion electrophoresis energy conversion power prediction method. Specifically, the CNN-based nanion electrophoresis energy conversion power prediction system comprises a collection module, a processing module, a construction module, a training module and a prediction module.

[0097] The collection module collects the temperature field, concentration field and surface charge density data of the nanion electrophoresis energy conversion device under different operating conditions.

[0098] The processing module discretely samples the temperature field, concentration field and surface charge density data obtained by the collection module, and the geometry parameters of the nanochannel and the nanopore of the nanion electrophoresis energy conversion device, and constructs a training data set after pre-processing and standardization.

[0099] The construction module constructs a multi-dimensional deep CNN model as a prediction model of the nanion electrophoresis energy conversion power.

[0100] The training module trains the multi-dimensional deep CNN model constructed by the construction module using the training data set obtained by the processing module.

[0101] The prediction module re-collects the temperature field, concentration field and surface charge density data of the nanion electrophoresis energy conversion device, and inputs the test data set into the multi-dimensional deep CNN model obtained by the training module to predict the maximum power generation under the test condition, thereby realizing the prediction of the nanion electrophoresis energy conversion power.

[0102] In still another embodiment of the present application, a terminal device is provided, which comprises a processor and a memory, the memory being configured to store a computer program, the computer program comprising program instructions, and the processor being configured to execute the program instructions stored in the computer storage medium. The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic, discrete hardware components, etc., which are the computing core and control core of the terminal, and are suitable for implementing one or more instructions, and are specifically suitable for loading and executing one or more instructions to implement a corresponding method flow or a corresponding function; the processor in the embodiments of the present application can be used for the operation of the CNN-based nano-ion electroosmosis energy conversion power prediction method, including:

[0103] Collecting temperature field, concentration field and surface charge density data of the nano-ion electroosmosis energy conversion device under different operating conditions; discretely sampling the temperature field, concentration field and surface charge density data, and the geometric structure parameters of the nano-channels and nano-pores of the nano-ion electroosmosis energy conversion device, constructing a training data set after preprocessing and standardization; constructing a multi-dimensional deep CNN model as a prediction model of the power generation of the nano-ion electroosmosis energy conversion device; training the multi-dimensional deep CNN model using the training data set; re-collecting the temperature field, concentration field and surface charge density data of the nano-ion electroosmosis energy conversion device, and using the data as a test data set after preprocessing and standardization; inputting the test data set into the multi-dimensional deep CNN model to predict the maximum power generation under the test condition, and realizing the prediction of the power generation of the nano-ion electroosmosis energy conversion device.

[0104] In another embodiment of the present application, the present application also provides a storage medium, specifically a computer readable storage medium (Memory), which is a memory device in a terminal device, used for storing programs and data. It can be understood that the computer readable storage medium herein can include an internal storage medium in the terminal device, and of course can also include an expansion storage medium supported by the terminal device. The computer readable storage medium provides a storage space, which stores an operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer readable storage medium herein can be a high-speed RAM memory, or a non-volatile memory such as at least one disk memory.

[0105] The one or more instructions stored in the computer readable storage medium can be loaded and executed by the processor to implement the corresponding steps of the CNN-based nanometer iontophoresis energy conversion power prediction method in the above embodiments; the one or more instructions in the computer readable storage medium are loaded and executed by the processor to perform the following steps:

[0106] Collecting temperature field, concentration field and surface charge density data of the nanometer iontophoresis energy conversion device under different operating conditions; discretely sampling the temperature field, concentration field and surface charge density data, and the geometric structure parameters of the nanochannel and nanopore of the nanometer iontophoresis energy conversion device, constructing a training data set after preprocessing and standardization; constructing a multi-dimensional deep CNN model as a prediction model of the nanometer iontophoresis energy conversion power generation power; training the multi-dimensional deep CNN model using the training data set; re-collecting the temperature field, concentration field and surface charge density data of the nanometer iontophoresis energy conversion device, and using them as a test data set after preprocessing and standardization; inputting the test data set into the multi-dimensional deep CNN model to predict the maximum power generation power under the test condition, and realizing the prediction of the iontophoresis energy conversion power generation power.

[0107] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0108] The prediction model of the nano-ion electro-osmotic energy conversion power is verified in a cross-validation manner, and the error trend during training is specifically referred to, and the final error is closer to 0, and the effect is better, as shown in the following formula: Figure 5 The loss function is selected as a smooth average absolute error (SmoothL1Loss) during training, so that the process of gradient descent to find the optimal solution during training is smoother, and the convergence speed is faster.

[0109] The average absolute error of the model prediction after training is 0.012834, and the comparison chart of the predicted power value and the real power value of the test case is as shown in the following formula: Figure 6

[0110] In summary, the present application is a kind of nano-ion electro-osmotic energy conversion power prediction method and system based on CNN, taking a simulation model as a verification example, first, a simulation model of the nano-ion electro-osmotic energy conversion device is established, the physical field parameters such as temperature field and concentration field in the model are set, and the physical field parameters under different working conditions during the power generation process of the device and the corresponding nano-channel geometric structure parameters are obtained by numerical simulation to construct a training sample data set;A multi-layer multi-dimensional CNN model is established, and the constructed sample data set is used for training, so that it can predict the maximum power generation of the nano-ion electro-osmotic energy conversion device under different working conditions;And for the conversion device with new nano-channel and nano-pore structure within a certain range, the learned characteristics of the existing device can realize accurate prediction of the maximum power generation under different working conditions.

[0111] ​Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is exemplified, and in actual application, the above-mentioned functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be realized in the form of hardware or software. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction, and do not limit the protection scope of the present application. The specific working process of the units and modules in the above system can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0112] In the above embodiments, the description of each embodiment has its own emphasis, and the parts not described or recorded in detail in a certain embodiment can be referred to the related description of other embodiments.

[0113] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in the present application can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0114] In the embodiments provided by the present application, it should be understood that the disclosed apparatus / terminal and method can be implemented by other ways. For example, the above-mentioned apparatus / terminal embodiments are only schematic, and the division of the modules or units is only a logical function division, and there can be another division way in actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0115] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0116] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0117] The integrated module / unit, if realized in the form of a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer-readable storage medium. The computer program can implement the steps of each method embodiment when executed by a processor. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms. The computer-readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the computer-readable medium can include or exclude contents according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0118] The present application is described with reference to flowcharts and / or block diagrams according to the methods, devices (systems), and computer program products of embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one flow or multiple flows and / or blocks Figure 1 The device that implements the functions specified in one block or multiple blocks.

[0119] These computer program instructions can also be stored in a computer-readable memory that can guide the computer or other programmable data processing devices to work in a specific way, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction devices that implement the functions specified in the flowcharts and / or block diagrams. Figure 1one or more processes and / or blocks Figure 1 the function specified in the one or more blocks.

[0120] These computer program instructions can also be loaded into computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable data processing devices to generate a computer-implemented process, so that the instructions executed on the computer or other programmable data processing devices provide a process for implementing the flow Figure 1 one or more processes and / or blocks Figure 1 the function specified in the one or more blocks.

[0121] The above is only to illustrate the technical idea of the present application, and cannot limit the protection scope of the present application. Any modification made according to the technical idea of the present application on the basis of the technical scheme falls within the protection scope of the claims of the present application.

Claims

1. A method for predicting the energy conversion power of nano-ion electroosmosis based on CNN, characterized in that, Includes the following steps: S1. Collect temperature field, concentration field and surface charge density data of the nano-ion electroosmosis energy conversion device under different operating conditions; S2. Discretely sample the temperature field, concentration field, and surface charge density data obtained in step S1, as well as the geometric parameters of the nanochannels and nanopores of the nano-ion electroosmosis energy conversion device, and construct a training dataset after preprocessing and standardization. S3. Construct a multi-dimensional deep CNN model as a prediction model for the power generation of nano-ion electroosmosis energy conversion. The multi-dimensional deep CNN includes an input layer, a multi-dimensional feature extraction layer, a fully connected layer, and an output layer. All initial network weights and biases in the multi-layer multi-dimensional CNN are randomly generated and follow a normal distribution. Input layer: Receives standardized seven-dimensional training data. Multi-dimensional feature extraction layer: includes 5 parallel one-dimensional convolutional layers, 1 feature merging layer, activation layer and 1 two-dimensional convolutional layer, with the output of each one-dimensional convolutional layer serving as the input to the feature merging layer; Fully connected layer: It takes the features extracted by the feature extraction layer as input value and outputs an element that represents the maximum power generation predicted by the network model. Parallel one-dimensional convolutional layer: 1 input channel, 16 output channels, and kernel sizes of 1, 1, 3, 3, and 5 respectively; Feature merging layer: Merges the features extracted by the parallel one-dimensional convolutional layers along the dimension of the one-dimensional convolutional channel; Excitation layer: The excitation function is set to the ReLU function; Two-dimensional convolutional layer: 16 input channels, 32 output channels, and a kernel size of 3; enter Multi-dimensional feature extraction is performed by multiple one-dimensional convolutional layers, and the final output is determined by two-dimensional convolutional layers. ; S4. Use the training dataset obtained in step S2 to train the multi-dimensional deep CNN model constructed in step S3; S5. Re-collect the temperature field, concentration field, and surface charge density data of the nano-ion electroosmosis energy conversion device, and use the preprocessed and standardized data as the test dataset. The test dataset is input into the multi-dimensional deep CNN model obtained in step S4 to predict the maximum power generation under the test conditions, thereby achieving the prediction of the power generation from ion electroosmosis energy conversion.

2. The CNN-based method for predicting the energy conversion power of nano-ion electroosmosis according to claim 1, characterized in that, In step S1, the collected data includes the macropore / micropore radius of the nanochannel, the length / width of the internal nanopore geometry, the surface charge density of the channel, the temperature, and the concentration ratio of the liquid tanks on both sides.

3. The CNN-based method for predicting the energy conversion power of nano-ion electroosmosis according to claim 1, characterized in that, In step S2, the temperature field and concentration field inside different types of nano-ion electroosmosis energy conversion devices are discretely measured, and then combined with the corresponding nanochannel and nanopore geometric parameters to construct a training sample dataset.

4. The CNN-based method for predicting the energy conversion power of nano-ion electroosmosis according to claim 3, characterized in that, The standardization process specifically involves: in, It is the first i The average value of all physical quantities in the set of data. It is the first i The standard deviation of all physical quantities in the set of data.

5. The CNN-based method for predicting the energy conversion power of nano-ion electroosmosis according to claim 1, characterized in that, The operation process of the multi-dimensional feature extraction layer is represented as follows: in, , These represent the energy conversion temperature and concentration ratio of nano-ion electroosmosis, respectively. The surface charge density of the nanochannel; , These are the macropore radius and micropore radius of the nanochannels, respectively. , These represent the height and width of the nanopore geometry, respectively. For the first i The output of a one-dimensional convolutional layer; , The first i The weights of one-dimensional and two-dimensional convolutional layers; This is the output of the feature merging layer; , The first i The bias of one-dimensional convolutional layers and two-dimensional convolutional layers.

6. The CNN-based method for predicting the energy conversion power of nano-ion electroosmosis according to claim 1, characterized in that, In step S4, the training rounds of the multi-dimensional deep CNN model are 100, and the learning rate decay value is set to 0.5 every 50 iterations during training.

7. A CNN-based nano-ion electroosmosis energy conversion power prediction system, characterized in that, include: The data acquisition module collects temperature field, concentration field, and surface charge density data of the nano-ion electroosmosis energy conversion device under different operating conditions. The processing module discretely samples the temperature field, concentration field, and surface charge density data obtained by the acquisition module, as well as the geometric parameters of the nanochannels and nanopores of the nano-ion electroosmosis energy conversion device, and constructs a training dataset after preprocessing and standardization. A module was built to construct a multi-dimensional deep CNN model as a predictive model for the power generation of nano-ion electroosmosis energy conversion. The training module uses the training dataset obtained from the processing module to train the multi-dimensional deep CNN model constructed by the building module. The multi-dimensional deep CNN includes an input layer, a multi-dimensional feature extraction layer, a fully connected layer, and an output layer. All initial network weights and biases in the multi-layer multi-dimensional CNN are randomly generated and follow a normal distribution. The input layer receives standardized seven-dimensional training data. Multi-dimensional feature extraction layer: includes 5 parallel one-dimensional convolutional layers, 1 feature merging layer, activation layer and 1 two-dimensional convolutional layer, with the output of each one-dimensional convolutional layer serving as the input to the feature merging layer; Fully connected layer: It takes the features extracted by the feature extraction layer as input value and outputs an element that represents the maximum power generation predicted by the network model. Parallel one-dimensional convolutional layer: 1 input channel, 16 output channels, and kernel sizes of 1, 1, 3, 3, and 5 respectively; Feature merging layer: Merges the features extracted by the parallel one-dimensional convolutional layers along the dimension of the one-dimensional convolutional channel; Excitation layer: The excitation function is set to the ReLU function; Two-dimensional convolutional layer: 16 input channels, 32 output channels, and a kernel size of 3; enter Multi-dimensional feature extraction is performed by multiple one-dimensional convolutional layers, and the final output is determined by two-dimensional convolutional layers. ; The prediction module re-collects temperature field, concentration field, and surface charge density data of the nano-ion electroosmosis energy conversion device, and uses the preprocessed and standardized data as a test dataset. The test dataset is then input into the multi-dimensional deep CNN model obtained by the training module to predict the maximum power generation under the test conditions, thereby achieving the prediction of the power generation of the ion electroosmosis energy conversion.

Images