A Machine Learning-Based Method and System for Optimizing Photothermal Conversion Efficiency in Solar Seawater Desalination

Abstract

This application provides a machine learning-based method and system for optimizing the photothermal conversion efficiency of solar desalination in seawater. The method includes: firstly, collecting solar spectral distribution data, seawater turbidity spectral data, real-time heat flow data, and semiconductor nanomaterial data for the target sea area; and constructing a material spectral mapping library based on the light absorption characteristics of the nanomaterials. By coupling analysis of the solar spectrum and seawater turbidity spectrum to form environmental constraints, a Bayesian optimization model is used in conjunction with the material spectral mapping library and real-time heat flow data to generate the optimal nanomaterial mixing ratio under these constraints. Subsequently, a deep residual network is used to predict the interfacial evaporation rate and obtain the optimal material morphology configuration scheme, which is finally applied to the preparation of semiconductor nanomaterials. This application improves the photothermal conversion efficiency of solar desalination systems in high-turbidity seawater environments.

Description

Technical Field

[0001] This application relates to the field of machine learning technology, and in particular to a machine learning-based method and system for optimizing the photothermal conversion efficiency of solar seawater desalination. Background Technology

[0002] When conducting solar-powered desalination in nearshore areas with high turbidity, suspended particulate matter leads to uneven distribution of solar spectral energy, making it difficult for traditional photothermal materials to adapt to dynamically changing water quality conditions. There is a need to develop novel optimization methods that can automatically adjust based on real-time environmental parameters to maintain stable evaporation efficiency.

[0003] Current advanced solutions employ rule-based expert systems that evaluate and select photothermal materials based on preset spectral response rule sets and thermodynamic parameter thresholds. The system incorporates material performance databases for various typical operating conditions and combines them with real-time environmental parameters collected by sensors for matching and recommendation.

[0004] The rule base of expert systems has limited capacity, and the accuracy of recommendations is constrained by the completeness of preset rules when facing complex and ever-changing marine environments. The system is slow to respond to newly emerging spectral features, and its material performance assessments lack timeliness, resulting in a significant gap between actual operational efficiency and theoretical expectations. Summary of the Invention

[0005] This application provides a machine learning-based method and system for optimizing the photothermal conversion efficiency of solar seawater desalination, in order to solve the problem of low photothermal conversion efficiency of solar desalination systems in high-turbidity seawater environments in the prior art.

[0006] In a first aspect, this application provides a machine learning-based method for optimizing the photothermal conversion efficiency of solar seawater desalination, including:

[0007] Acquire solar spectral distribution data, seawater turbidity spectral data, real-time heat flow data, and semiconductor nanomaterial data in the target sea area environment;

[0008] Based on the semiconductor nanomaterial data, a material spectral mapping library was constructed;

[0009] The solar spectral distribution data and the seawater turbidity spectral data are coupled and analyzed to form environmental constraints.

[0010] Using a Bayesian optimization model, with the material spectral mapping library and the real-time heat flow data as input, the optimal nanomaterial mixing ratio data is generated under the environmental constraints.

[0011] Using a deep residual network, the interface evaporation rate is predicted based on the optimal nanomaterial mixing ratio data, generating optimal material morphology information. This optimal material morphology information is then applied to the morphology configuration of semiconductor nanomaterials to optimize the photothermal conversion efficiency during solar desalination.

[0012] Optionally, the step of using a deep residual network to predict the interfacial evaporation rate of the optimal nanomaterial mixing ratio data and generate optimal material morphology information includes:

[0013] The optimal nanomaterial mixing ratio data is input into the input layer of the deep residual network to obtain interfacial evaporation rate characteristic data.

[0014] The interface evaporation rate feature data is processed through the cross-level connection structure of the deep residual network.

[0015] Based on the processed interfacial evaporation rate characteristic data, the morphology parameters of nanosheets and nanospheres are calculated in parallel at the output layer of the deep residual network. Based on the morphology parameters of the nanosheets and nanospheres, the optimal material morphology configuration information is generated.

[0016] Optionally, based on the processed interfacial evaporation rate characteristic data, the nanosheet morphology parameters and nanosphere morphology parameters are calculated in parallel at the output layer of the deep residual network. Based on the nanosheet morphology parameters and nanosphere morphology parameters, optimal material morphology configuration information is generated, including:

[0017] The processed interface evaporation rate feature data is input into the morphology parameter analysis module of the deep residual network. Through the preset morphology transformation rules, the processed interface evaporation rate feature data is separated into nanosheet feature vectors and nanosphere feature vectors.

[0018] In the output layer of the deep residual network, the thickness distribution of the nanosheet feature vector is calculated, and the nanosheet morphology parameters are output. The diameter distribution of the nanosphere feature vector is calculated, and the nanosphere morphology parameters are output.

[0019] Establish a spatial matching relationship between the morphological parameters of the nanosheets and the morphological parameters of the nanospheres;

[0020] Based on the spatial matching relationship, the optimal material morphology configuration information is generated.

[0021] Optionally, generating optimal material morphology configuration information based on the spatial matching relationship includes:

[0022] Based on the heat conduction characteristics of the seawater evaporation interface, determine the heat conduction constraint conditions corresponding to the spatial matching relationship;

[0023] The maximum arrangement density of nanosheets and nanospheres under the thermal conductivity constraint is solved by gradient optimization algorithm.

[0024] Based on the maximum arrangement density, structural configuration parameters including the nanosheet interlayer spacing and the nanosphere stacking period are determined;

[0025] Based on the aforementioned structural configuration parameters, optimal material morphology configuration information is generated.

[0026] Optionally, constructing a material spectral mapping library based on the semiconductor nanomaterial data includes:

[0027] Light absorption response data, particle size range data, and plasma effect data were extracted from the semiconductor nanomaterials.

[0028] The light absorption response data is correlated with the particle size range data, and the correlation result is weighted and fused with the plasma effect data to construct a photothermal response intensity mapping rule;

[0029] A material spectral mapping library is generated based on the aforementioned photothermal response intensity mapping rules.

[0030] Optionally, the coupling analysis of the solar spectral distribution data and the seawater turbidity spectral data to form environmental constraints includes:

[0031] Based on the seawater turbidity spectral data, the characteristic absorption band range of suspended particulate matter was determined;

[0032] Extract solar spectral energy data that overlaps with the characteristic absorption band range from the solar spectral distribution data;

[0033] Based on the solar spectral energy data, calculate the energy attenuation rate of the solar spectrum in the characteristic absorption band range;

[0034] Based on the energy decay rate, environmental constraints are generated.

[0035] Optionally, the step of optimizing and generating optimal nanomaterial mixing ratio data under the environmental constraints using a Bayesian optimization model, with the material spectral mapping library and the real-time heat flow data as input, includes:

[0036] The material spectral mapping library is used as the first constraint, the interface temperature change rate in the real-time heat flow data is used as the second constraint, and the energy decay rate in the environmental constraint is used as the boundary condition.

[0037] By using the probability acquisition function of the Bayesian optimization model, a candidate nanomaterial mixing ratio dataset is generated under the conditions of satisfying the first constraint, the second constraint, and the boundary condition.

[0038] The candidate nanomaterial mixing ratio dataset is iteratively optimized. When the photothermal conversion efficiency calculation result corresponding to any candidate nanomaterial mixing ratio data in the candidate nanomaterial mixing ratio dataset meets the preset photothermal conversion efficiency threshold, the candidate nanomaterial mixing ratio data that meets the preset photothermal conversion efficiency threshold is taken as the optimal nanomaterial mixing ratio data.

[0039] Secondly, this application provides a machine learning-based system for optimizing the photothermal conversion efficiency of solar seawater desalination, comprising:

[0040] The acquisition module is used to acquire solar spectral distribution data, seawater turbidity spectral data, real-time heat flow data, and semiconductor nanomaterial data in the target sea area environment.

[0041] The module is used to construct a material spectral mapping library based on the semiconductor nanomaterial data;

[0042] The analysis module is used to couple and analyze the solar spectral distribution data with the seawater turbidity spectral data to form environmental constraints.

[0043] The generation module is used to generate optimal nanomaterial mixing ratio data under the environmental constraints by using a Bayesian optimization model, taking the material spectral mapping library and the real-time heat flow data as inputs.

[0044] The prediction module is used to predict the interfacial evaporation rate of the optimal nanomaterial mixing ratio data using a deep residual network, generate optimal material morphology information, and apply the optimal material morphology information to the morphology configuration of semiconductor nanomaterials to optimize the photothermal conversion efficiency in the solar desalination process.

[0045] Thirdly, this application provides a computing device, including a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute a machine learning-based method for optimizing the photothermal conversion efficiency of solar seawater desalination as described in any of the first aspects.

[0046] Fourthly, this application provides a computer storage medium storing computer program instructions thereon, which, when executed by a processor, implement the machine learning-based method for optimizing the photothermal conversion efficiency of solar seawater desalination as described in any one of the first aspects.

[0047] This application provides a machine learning-based method for optimizing the photothermal conversion efficiency of solar desalination. The method includes: acquiring solar spectral distribution data, seawater turbidity spectral data, real-time heat flow data, and semiconductor nanomaterial data in the target sea area environment; constructing a material spectral mapping library based on the semiconductor nanomaterial data; coupling the solar spectral distribution data with the seawater turbidity spectral data to form environmental constraints; using a Bayesian optimization model, with the material spectral mapping library and the real-time heat flow data as input, optimizing and generating optimal nanomaterial mixing ratio data under the environmental constraints; using a deep residual network to predict the interface evaporation rate of the optimal nanomaterial mixing ratio data to generate optimal material morphology information; and applying the optimal material morphology information to the morphology configuration of the semiconductor nanomaterials to optimize the photothermal conversion efficiency during the solar desalination process.

[0048] The technical solution provided in this application has the following beneficial effects:

[0049] This application establishes an optimization foundation that matches the specific characteristics of the marine environment, ensuring the relevance and accuracy of subsequent analyses. It forms a correlation database of the optical and thermal properties of semiconductor nanomaterials, providing a quantitative basis for material selection. It clarifies the attenuation characteristics of the solar spectrum by the seawater environment, setting boundary constraints for material optimization. It dynamically generates optimal material ratios adapted to specific marine environments, achieving precise control of photothermal performance. Through deep learning, it accurately predicts the evaporation performance of different material morphologies, obtaining the optimal structural configuration. The theoretical optimization results are transformed into actual material structures, directly improving the photothermal conversion efficiency of the desalination system.

[0050] Furthermore, this application also processes the optimal material ratio data through a deep residual network. First, the interfacial evaporation characteristics are extracted, and then the characteristics are processed through a cross-level connection structure. Finally, the optimal morphology parameters of the nanosheets and nanospheres are calculated in parallel and output to generate a complete material morphology configuration scheme.

[0051] Furthermore, deep residual networks effectively capture the complex mapping relationship between material ratio and morphological characteristics, accurately predict evaporation performance, and the generated morphological configuration scheme improves the photothermal conversion capability and interfacial evaporation efficiency of the material.

[0052] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 A flowchart illustrating a machine learning-based method for optimizing the photothermal conversion efficiency of solar seawater desalination, as provided in this application embodiment;

[0055] Figure 2 A schematic diagram of a machine learning-based solar seawater desalination photothermal conversion efficiency optimization system provided in this application embodiment;

[0056] Figure 3 This is a schematic diagram of the structure of a computing device provided in an embodiment of this application. Detailed Implementation

[0057] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0058] In some of the processes described in the specification, claims, and accompanying drawings of this application, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not themselves represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0059] Current solar-powered seawater desalination systems rely on a pre-defined rule base for matching photothermal materials, which has significant limitations. This method requires establishing a database of material properties under various seawater conditions beforehand. When encountering complex water quality changes beyond the database's scope, the system cannot respond accurately. Particularly in high-turbidity sea areas, suspended matter in the seawater constantly alters the absorption characteristics of sunlight, and a fixed rule base struggles to keep up with these dynamic changes, leading to inappropriate material selection and severely impacting desalination efficiency.

[0060] To address this issue, this application proposes a machine learning-based method for optimizing the photothermal conversion efficiency of solar seawater desalination. This method dynamically adjusts material formulations and structures by monitoring marine environmental data in real time and utilizing machine learning technology. The system continuously analyzes the current absorption of sunlight by the seawater and automatically calculates the most suitable combination of nanomaterials and the optimal structural morphology for the current conditions. This method no longer relies on a preset rule base but makes autonomous decisions based on real-time environmental data, enabling it to accurately respond to various complex water quality changes, especially the special needs of high-turbidity seawater. It fundamentally solves the problems of delayed response and inaccurate adaptation in existing technologies, improving the stability and efficiency of seawater desalination systems.

[0061] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0062] Figure 1 A flowchart illustrating a machine learning-based method for optimizing the photothermal conversion efficiency of solar seawater desalination, as provided in this application embodiment, is shown below. Figure 1 As shown, the method includes:

[0063] Step 101: Obtain solar spectral distribution data, seawater turbidity spectral data, real-time heat flow data, and semiconductor nanomaterial data in the target sea area environment.

[0064] In step 101, solar spectral distribution data refers to the energy distribution of sunlight at different wavelengths in the target sea area, reflecting the variation characteristics of sunlight intensity with wavelength. Seawater turbidity spectral data characterizes the absorption and scattering characteristics of suspended particulate matter in seawater at different wavelengths, reflecting the spectral characteristics of seawater turbidity. Real-time heat flow data represents the monitoring of heat transfer at the seawater evaporation interface, including dynamic parameters such as temperature changes and heat conduction rates. Semiconductor nanomaterial data represents the recording of the optical, thermal, and other performance parameters of the nanomaterials used in the target sea area.

[0065] In this embodiment, spectral monitoring equipment is first deployed in the target sea area to continuously collect data on the energy distribution of sunlight across various wavelengths. Simultaneously, a water quality sensor is used to acquire the spectral characteristics corresponding to seawater turbidity. A temperature sensor array is installed at the seawater evaporation interface to record heat conduction data in real time. Samples of commonly used semiconductor nanomaterials from this sea area are collected, and their fundamental performance parameters, such as light absorption and particle size distribution, are measured in the laboratory. These four types of data are then compiled to form an initial dataset, providing a foundation for subsequent analysis.

[0066] For example, in a certain nearshore area A, spectrometer measurements showed that sunlight had strong energy in the 400-800 nm wavelength range, with a peak at 600 nm. Water quality monitoring showed significant turbidity absorption peaks at 580 nm and 620 nm in this area. An evaporation interface temperature sensor recorded average heat flux density data. The laboratory tested commonly used type B nanomaterials in the area, obtaining their absorption coefficients and particle size distribution ranges in the main wavelength ranges. These data collectively form the basis of the initial analysis.

[0067] Step 102: Based on the semiconductor nanomaterial data, construct a material spectral mapping library.

[0068] In step 102, the material spectral mapping library represents the establishment of a database of correspondences between the properties of semiconductor nanomaterials and their spectral responses, including photothermal conversion performance indicators.

[0069] In this embodiment, the collected semiconductor nanomaterial data are systematically organized, and the light absorption characteristics of different materials at different wavelengths are analyzed. Based on the influence of particle size on light scattering, a correction relationship between particle size and light absorption is established. Combining the surface plasmon effect characteristics of the materials, the photothermal response intensity of each material is comprehensively calculated. Finally, a mapping relationship library containing multi-dimensional parameters such as material composition, particle size, and photothermal response intensity is constructed.

[0070] For example, processing the test data of type B nanomaterials revealed a high light absorption coefficient at 600 nm. The photothermal conversion effect was optimal when the particle size was in the 50-70 nm range. Combining this with the surface plasmon resonance characteristics, the photothermal response intensity values ​​of the material in the main operating wavelength band were calculated and entered into the material spectral mapping library.

[0071] Step 103: Couple the solar spectral distribution data with the seawater turbidity spectral data to form environmental constraints.

[0072] In step 103, the environmental constraints represent the working boundary limitations of the photothermal conversion system derived by comprehensively considering the solar spectrum and seawater turbidity characteristics.

[0073] In this embodiment, seawater turbidity spectral data is analyzed to determine the main absorption bands of suspended particulate matter. These bands are compared with the solar spectral distribution to calculate the energy attenuation of sunlight after passing through seawater. Based on the degree of energy attenuation, a minimum energy requirement for compensation by the photothermal material is set, forming the basic constraints for system operation.

[0074] For example, the seawater in region A exhibits strong absorption at 580 nm and 620 nm, corresponding to a certain proportion of energy attenuation in the solar spectrum at these two wavelengths. Based on this data, it is calculated that the photothermal material needs to provide corresponding energy compensation in these two wavelengths, serving as a constraint for subsequent optimization.

[0075] Step 104: Using the material spectral mapping library and the real-time heat flow data as input, the optimal nanomaterial mixing ratio data is generated under the environmental constraints by using a Bayesian optimization model.

[0076] In step 104, the optimal nanomaterial mixing ratio data represents the various nanomaterial ratio schemes that can achieve the best photothermal conversion efficiency under given constraints.

[0077] In this embodiment, material property data from a material spectral mapping library are used as basic parameters, combined with real-time monitored heat flow data to reflect the current operating status. Under environmental constraints, different material combination schemes are continuously tried using a Bayesian optimization algorithm. Each iteration evaluates the photothermal conversion performance of the scheme, gradually approaching the optimal solution, and finally outputting the best material ratio that satisfies all conditions.

[0078] For example, considering the environmental characteristics of region A, the optimization algorithm, after multiple iterations of calculation, finally determined that the scheme of mixing type B and type C materials in a certain proportion can best meet the compensation requirements of the 580nm and 620nm bands, while also adapting to the current heat flow conditions.

[0079] Step 105: Using a deep residual network, predict the interfacial evaporation rate of the optimal nanomaterial mixing ratio data to generate optimal material morphology information. Apply the optimal material morphology information to the morphology configuration of semiconductor nanomaterials to optimize the photothermal conversion efficiency in the solar desalination process.

[0080] In step 105, the optimal material morphology information refers to the specific structural parameters guiding the fabrication of nanomaterials, including shape, size, and arrangement. The term "optimal" in the optimal material morphology information refers to the material morphology configuration that achieves the highest photothermal conversion efficiency under specific marine environmental conditions. Its optimality is based on the synergistic optimization calculation results of the nanosheet and nanosphere morphology parameters using a deep residual network. These calculation results satisfy the established environmental constraints and the set photothermal conversion efficiency threshold requirements. Specifically, under the condition of maintaining an interface temperature change rate of 0.5 ± 0.1 °C / s, the photothermal conversion efficiency reaches 78.5%, while simultaneously meeting the environmental constraints of compensating for 32% energy loss in the 550-610 nm band and 42% energy loss in the 1320-1380 nm band.

[0081] In this embodiment, the optimal material ratio data is input into a deep residual network, which first extracts key features affecting the evaporation rate. The correlation between these features and material morphology is analyzed through a multi-layer network structure. The optimal structural parameters for nanosheets and nanospheres are simultaneously calculated in the output layer. Finally, a configuration scheme containing specific morphology requirements is generated, directly guiding material preparation.

[0082] For example, analysis of the optimized material composition for region A suggests using a structure with alternating layers of nanosheets of specific thickness and nanospheres of specific diameter, based on the deep residual network output. This structure, after being fabricated in the laboratory, exhibits excellent photothermal conversion performance in practical applications.

[0083] This method achieves precise matching between the performance of photothermal materials and marine conditions through real-time environmental monitoring and intelligent algorithm optimization. The end-to-end optimization, from data acquisition to material configuration, improves the adaptability and efficiency of the seawater desalination system, particularly demonstrating stable and reliable performance in high-turbidity sea areas.

[0084] To address the issue of optimizing the morphology of photothermal materials, in some embodiments, step 105: using a deep residual network to predict the interface evaporation rate of the optimal nanomaterial mixing ratio data to generate optimal material morphology information includes:

[0085] Step 201: Input the optimal nanomaterial mixing ratio data into the input layer of the deep residual network to obtain interfacial evaporation rate characteristic data.

[0086] In step 201, the interfacial evaporation rate characteristic data refers to the set of key features extracted from the material proportioning data through a deep residual network that can reflect the evaporation performance of the material. These characteristic data include the evaporation behavior patterns of the material under different environmental conditions.

[0087] In this embodiment, the optimized nanomaterial mixing ratio data is input into the first layer processing unit of the deep residual network. The network first identifies the proportional relationships of the components in the material composition, and then extracts the core characteristic parameters affecting the evaporation rate based on pre-learned knowledge. These characteristic parameters reflect the expected evaporation performance of the material combination under specific marine conditions.

[0088] Step 202: Process the interface evaporation rate feature data through the cross-level connection structure of the deep residual network.

[0089] In step 202, the cross-layer connection structure refers to the direct connection path unique to deep residual networks that can span multiple network layers. This structure can maintain the integrity of feature information and avoid information loss in deep networks.

[0090] In this embodiment, the network uses a unique cross-layer connection method to transfer and process the initially extracted evaporation characteristics between different layers. Each layer of the network analyzes the characteristics from a specific perspective while retaining key information from the original characteristics. Through this multi-layered joint processing, the understanding and expression of evaporation performance characteristics are gradually deepened.

[0091] Step 203: Based on the processed interfacial evaporation rate characteristic data, the nanosheet morphology parameters and nanosphere morphology parameters are calculated in parallel at the output layer of the deep residual network. Based on the nanosheet morphology parameters and the nanosphere morphology parameters, the optimal material morphology configuration information is generated.

[0092] In step 203, the morphology parameters of the nanosheets include structural characteristic parameters such as thickness and surface area. The morphology parameters of the nanospheres include structural characteristic parameters such as diameter and stacking density.

[0093] In this embodiment, the network simultaneously analyzes the structural requirements of nanosheets and nanospheres during the final processing stage. Optimal parameter combinations for both structures are derived using dedicated parallel computing units. These parameters are then comprehensively evaluated to generate a complete material morphology configuration scheme, including specific structural dimensions and arrangement.

[0094] Here is a specific example:

[0095] In the optimization process of the solar-powered seawater desalination system in nearshore area A, the optimal mixing ratio of type B and type C materials at a 7:3 ratio was determined. This mixing ratio data was input into a depth residual network for morphological optimization analysis. The network first analyzed the evaporation performance characteristics of the mixed material in two characteristic wavelength bands: 580 nm and 620 nm. The evaporation characteristic intensity corresponding to the 580 nm band was 0.85, and that of the 620 nm band was 0.78. These two values ​​were calculated by multiplying the absorption coefficient and thermal conductivity of the material in the corresponding wavelength bands. The calculation formula is: Evaporation characteristic intensity = Absorption coefficient × Thermal conductivity, where the absorption coefficient was obtained through laboratory measurements, and the thermal conductivity was obtained through heat flow data inversion. Through its internal cross-layer connection structure, the network transferred and analyzed these characteristic data between different processing layers, gradually identifying the key morphological factors affecting evaporation efficiency. In the output layer, the network calculates in parallel that the B-type material is best suited for fabricating a sheet structure with a thickness of 25 nanometers. This value is calculated using the empirical formula d = 10 + 0.2 × S, where d represents the nanosheet thickness and S is the sum of characteristic intensities. Simultaneously, it calculates that the C-type material is best suited for fabricating a spherical structure with a diameter of 45 nanometers. This value is determined using the formula D = 30 + 0.3 × F, where D represents the nanosphere diameter and F is the difference in characteristic intensities. Based on these calculations, the network ultimately generates the optimal material morphology configuration: every three 25-nanometer-thick B-type nanosheets are interspersed with one layer of 45-nanometer-diameter C-type nanospheres, forming a periodic arrangement.

[0096] In this embodiment, the method utilizes deep learning-based intelligent analysis to achieve precise prediction from material proportions to specific morphologies, enabling the prepared photothermal materials to fully leverage their performance advantages. Particularly in complex marine environments, the morphology configuration scheme provided by this method exhibits excellent adaptability and stability, effectively improving the overall efficiency of the desalination system.

[0097] To address the problem of precisely optimizing the morphology configuration of nanomaterials, in some embodiments, step 203 involves: based on the processed interfacial evaporation rate characteristic data, parallel calculation of nanosheet morphology parameters and nanosphere morphology parameters is performed at the output layer of the deep residual network; and optimal material morphology configuration information is generated based on the nanosheet morphology parameters and the nanosphere morphology parameters, including:

[0098] Step 301: Input the processed interface evaporation rate feature data into the morphology parameter analysis module of the deep residual network, and separate the processed interface evaporation rate feature data into nanosheet feature vectors and nanosphere feature vectors through preset morphology transformation rules.

[0099] In step 301, the morphology parameter analysis module refers to a sub-network structure within the deep residual network specifically designed for analyzing material morphology features. The morphology transformation rule is a pre-defined data processing rule within this module used to decompose evaporation features into different morphological features. Specifically, the morphology transformation rule splits the interface evaporation feature vector into nanosheet thickness feature components and nanosphere diameter feature components according to a pre-defined physical mapping relationship. This rule is learned through the deep residual network training phase and is essentially a nonlinear projection relationship from the feature space to the morphology parameter space. The nanosheet feature vector and the nanosphere feature vector represent the sets of key parameters affecting the morphology of the nanosheets and nanospheres, respectively.

[0100] In this embodiment, the network sends the processed interfacial evaporation characteristic data to a dedicated morphology analysis unit. Based on pre-learned patterns, this unit decomposes the comprehensive evaporation characteristic data into two independent parts: one part primarily affects the formation characteristics of nanosheets, and the other part primarily affects the formation characteristics of nanospheres. This decomposition process preserves key information related to the morphology of each part from the original characteristics.

[0101] Step 302: Calculate the thickness distribution of the nanosheet feature vector in the output layer of the deep residual network and output the nanosheet morphology parameters; calculate the diameter distribution of the nanosphere feature vector and output the nanosphere morphology parameters.

[0102] In step 302, the thickness distribution calculation refers to the computational process of deriving the optimal thickness of the nanosheet based on the eigenvector. The diameter distribution calculation refers to the computational process of deriving the optimal diameter of the nanosphere based on the eigenvector.

[0103] In this embodiment, the network processes the two separated feature vectors separately. For the nanosheet feature vector, the network analyzes the thickness-related information it contains and derives the thickness range most favorable for evaporation performance through multi-layer calculations. Simultaneously, a similar process is performed on the nanosphere feature vector to determine the optimal diameter distribution range. These two calculation processes share some intermediate feature data to ensure consistency in the final results.

[0104] Step 303: Establish the spatial matching relationship between the morphology parameters of the nanosheets and the morphology parameters of the nanospheres.

[0105] In step 303, spatial matching relationship refers to the geometric fit requirements that nanosheets and nanospheres need to meet when combined, including the coordination relationship between parameters such as stacking density and contact area.

[0106] In this embodiment, the network performs joint analysis on the calculated nanosheet thickness parameters and nanosphere diameter parameters. By evaluating the interfacial contact effects and thermal conductivity characteristics under different combinations, the optimal compatibility between the two is established. This relationship ensures that the two morphologies can form a highly efficient synergistic effect when actually combined.

[0107] Step 304: Based on the spatial matching relationship, generate optimal material morphology configuration information.

[0108] In this embodiment, the network generates a specific morphology implementation scheme based on the established spatial matching relationship. This scheme details the selection of nanosheet thickness, nanosphere diameter, and their arrangement, forming a complete guide to morphology configuration.

[0109] Here is a specific example:

[0110] In the optimization case of a solar-powered seawater desalination system in nearshore area A, after receiving the feature data of a 7:3 mixing ratio of type B and type C materials, the deep residual network first separates and processes the evaporation feature data through a morphology parameter analysis module. According to a preset conversion rule, the network decomposes the feature intensity of 0.85 in the 580nm band and 0.78 in the 620nm band. The nanosheet-related feature vector contains the data components that mainly affect the thickness parameter, and the total sum is calculated as S=1.63. The nanosphere-related feature vector contains the data components that mainly affect the diameter parameter, and the difference is calculated as F=0.07. During parallel processing in the output layer, the thickness calculation formula d=10+0.2×S is used for the nanosheet feature vector, where d represents the nanosheet thickness and S is the total feature intensity. Substituting S=1.63, a thickness of 25 nm is calculated. The diameter calculation formula D=30+0.3×F is used for the nanosphere feature vector, where D represents the nanosphere diameter and F is the feature intensity difference. Substituting F=0.07, a diameter of 45 nm is calculated. Further analysis of the spatial relationship between these two morphological parameters revealed that, by evaluating the heat conduction path and light absorption efficiency, the optimal synergistic effect was achieved when three layers of 25 nm thick B-type nanosheets were alternately arranged with one layer of 45 nm diameter C-type nanospheres. In actual preparation tests, the resulting morphological configuration demonstrated excellent photothermal conversion performance in the 580 nm and 620 nm turbidity absorption bands specific to region A, effectively maintaining a stable evaporation rate.

[0111] In this embodiment of the application, the method achieves accurate mapping from material characteristics to specific morphology through intelligent analysis of deep networks. The generated morphology configuration scheme enables nanomaterials to fully exert their photothermal conversion potential, especially exhibiting excellent adaptability and stability in complex environments such as high-turbidity sea areas, effectively improving the overall working efficiency of the desalination system.

[0112] To address the issue of optimizing thermal conduction in the morphology configuration of nanomaterials, in some embodiments, step 304: generating optimal material morphology configuration information based on the spatial matching relationship includes:

[0113] Step 401: Determine the heat conduction constraint conditions corresponding to the spatial matching relationship based on the heat flow conduction characteristics of the seawater evaporation interface.

[0114] In step 401, the heat transfer characteristics of the seawater evaporation interface are derived from real-time monitored interface heat transfer data. This data is collected in real time by temperature sensors during the operation of the solar seawater desalination system, specifically manifested as the correlation between the interface temperature change rate and energy transfer efficiency. The heat transfer constraint condition refers to the thermodynamic boundary conditions determined based on the actual heat transfer characteristics of the seawater evaporation interface, which restrict the material arrangement and are used to ensure the heat transfer efficiency of the morphology configuration during actual operation.

[0115] In this embodiment, the system first analyzes real-time heat flow data at the seawater evaporation interface to extract key heat conduction characteristic parameters. These parameters are then compared with previously determined spatial matching relationships to identify key factors that may affect heat conduction performance. Based on the analysis results, minimum heat conduction efficiency requirements are set for material arrangement, forming specific constraints.

[0116] Step 402: Solve for the maximum arrangement density of nanosheets and nanospheres under the thermal conductivity constraint using a gradient optimization algorithm.

[0117] In step 402, the maximum arrangement density refers to the maximum number of nanosheets and nanospheres that can be accommodated per unit volume under the premise of meeting the thermal conduction requirements. This parameter directly affects the photothermal conversion efficiency.

[0118] In this embodiment, a gradient optimization algorithm is employed to gradually adjust the arrangement density of nanosheets and nanospheres, starting from an initial arrangement. Thermal conductivity is evaluated after each adjustment, and optimization continues along the direction of performance improvement until a configuration that achieves the highest arrangement density while satisfying constraints is found.

[0119] Step 403: Based on the maximum arrangement density, determine the structural configuration parameters including the nanosheet interlayer spacing and the nanosphere stacking period.

[0120] In step 403, structural configuration parameters refer to the geometric parameters that specifically guide the preparation of materials, including practically operable process parameters such as the interlayer spacing between nanosheets and the periodic spacing of nanosphere stacks.

[0121] In this embodiment, the specific structural dimension parameters are calculated based on the obtained maximum arrangement density. The calculation process considers the physical properties of the material itself and the limitations of the manufacturing process to ensure that the obtained parameters meet both performance requirements and are feasible.

[0122] Step 404: Based on the structural configuration parameters, construct the optimal material morphology configuration information.

[0123] In this embodiment, the determined structural configuration parameters are systematically integrated to form a morphology configuration scheme that includes detailed process requirements. This scheme clearly specifies key parameters such as the thickness and interlayer spacing of the nanosheets, and the diameter and stacking period of the nanospheres, as well as their interrelationships.

[0124] Here is a specific example:

[0125] In the optimization of the solar-powered seawater desalination system in nearshore area A, based on the previously determined spatial matching scheme of alternating three layers of 25 nm thick B-type nanosheets and one layer of 45 nm diameter C-type nanospheres, the system first analyzed the real-time heat flow data of the evaporation interface. It was found that the thermal conductivity efficiency decreased when the interlayer spacing was less than 10 nm, thus determining that the minimum interlayer spacing required for thermal conductivity was 10 nm. Iterative calculations were performed using a gradient optimization algorithm. The algorithm started with an initial interlayer spacing of 15 nm and gradually decreased the spacing in 1 nm increments. Each iteration evaluated the thermal conductivity efficiency, calculated as: thermal conductivity efficiency = actual heat transfer / theoretical maximum heat transfer. When the spacing was adjusted to 12 nm, the thermal conductivity efficiency reached a peak of 0.85, while maintaining the optimal arrangement density of one nanosphere embedded in every three nanosheets. Based on this optimization result, the final structural configuration parameters were determined as follows: B-type nanosheets maintained at a thickness of 25 nm with a layer spacing of 12 nm, and C-type nanospheres maintained at a diameter of 45 nm, arranged in a periodic pattern of 3 nanosheets alternating with 1 nanosphere. The 12 nm layer spacing was obtained by adding a 2 nm safety margin to the spacing value corresponding to the peak thermal conductivity. Integrating these parameters constitutes the optimal material morphology configuration scheme.

[0126] In the embodiments of this application, the method introduces thermal conductivity constraints and gradient optimization, so that the generated morphology configuration scheme not only considers light absorption characteristics, but also ensures thermal conductivity efficiency during actual operation, thereby achieving comprehensive optimization of material performance, especially showing more reliable and stable working performance in complex environments such as high turbidity sea areas.

[0127] To address the issue of systematically characterizing the spectral properties of semiconductor nanomaterials, in some embodiments, step 102: constructing a material spectral mapping library based on the semiconductor nanomaterial data, includes:

[0128] Step 501: Extract light absorption response data, particle size range data, and plasma effect data from the semiconductor nanomaterial.

[0129] In step 501, the light absorption response data refers to the measurement results of the material's ability to absorb light of different wavelengths, reflecting the material's light-trapping characteristics. The particle size range data records the size distribution of the material's particles, which affects light scattering behavior. The plasma effect data characterizes the material's ability to convert surface electron collective oscillations into light energy.

[0130] In this embodiment, the collected semiconductor nanomaterial sample is first subjected to spectral testing to obtain its light absorption curve in the target wavelength band. A particle size analyzer is used to measure the size distribution range of the material particles. Characteristic parameters of the plasmon resonance effect on the material surface are obtained through special optical detection methods. These three types of data together constitute a complete description of the material's optical properties.

[0131] Step 502: Associate the light absorption response data with the particle size range data, and perform weighted fusion of the association result with the plasma effect data to construct a photothermal response intensity mapping rule.

[0132] In step 502, weighted fusion refers to the comprehensive calculation performed after assigning different weights to different parameters based on their importance. The photothermal response intensity mapping rule establishes a quantitative relationship between material property parameters and the final photothermal conversion performance.

[0133] In this embodiment, the measured light absorption data and particle size data are correlated to establish a correction relationship between particle size and light absorption. Then, the corrected light absorption data and plasma effect data are comprehensively calculated according to a preset weight ratio, with light absorption data having a larger weight and plasma effect data having a smaller weight. This weighted fusion forms a calculation rule capable of accurately predicting the photothermal properties of materials.

[0134] Step 503: Generate a material spectral mapping library according to the photothermal response intensity mapping rules.

[0135] In this embodiment, the performance of each material under different wavelengths and particle sizes is calculated and recorded according to the established photothermal response intensity mapping rules. These calculation results are then systematically stored according to classification standards such as material type and wavelength range, forming a structured spectral characteristic database.

[0136] Here is a specific example:

[0137] During the optimization of the solar-powered seawater desalination system in nearshore area A, system testing was first conducted in the laboratory using the commonly used type B nanomaterial. Using spectral analysis equipment, the original light absorption coefficient of this material was measured to be 0.82 at 580 nm and 0.78 at 620 nm. Particle size analysis revealed that the main particle size distribution of this material was in the 50-70 nm range, with the highest proportion of particles at 60 nm. Further testing showed that the material exhibited a significant plasmon resonance effect at 580 nm, with an enhancement factor of 0.06. When correlating these test data, corrections were first made based on the influence of particle size on light absorption. The correction factor for the 60 nm particle size was 1.05. Therefore, the corrected light absorption coefficient at 580 nm was 0.82 × 1.05 = 0.861. The corrected light absorption data and plasma effect data were then weighted and fused together. The photothermal response intensity value at 580 nm was calculated using a weighting ratio of 0.7 for light absorption and 0.3 for plasma effect, resulting in 0.861×0.7+0.06×0.3=0.6027+0.018=0.6207. Similarly, the photothermal response intensity value at 620 nm was calculated as 0.78×1.05×0.7+0.04×0.3=0.5733+0.012=0.5853. These calculation results, along with other material parameters, were entered into a material spectral mapping library to form a complete performance record.

[0138] In the embodiments of this application, the method establishes an accurate and reliable database of material spectral properties through a systematic data acquisition and processing process, providing a scientific basis for the optimal selection of photothermal materials, and showing good adaptability and reliability, especially in dealing with spectral changes in different marine environments.

[0139] To address the issue of quantifying the impact of the marine environment on solar desalination systems, in some embodiments, step 103, which involves coupling the solar spectral distribution data with the seawater turbidity spectral data to form environmental constraints, includes:

[0140] Step 601: Determine the characteristic absorption band range of suspended particulate matter based on the seawater turbidity spectral data.

[0141] In step 601, the characteristic absorption band range refers to the specific wavelength range in which suspended particulate matter in seawater absorbs sunlight, reflecting the spectral characteristics of seawater turbidity.

[0142] In this embodiment, the location and width of absorption peaks in the spectrum are identified by analyzing the turbidity spectral curve of seawater. Wavelength ranges with absorption intensities exceeding the average level are determined; these ranges represent the main light absorption characteristics of suspended particulate matter.

[0143] Step 602: Extract solar spectral energy data that overlaps with the characteristic absorption band range from the solar spectral distribution data.

[0144] In step 602, the overlapping solar spectral energy data refers to the energy value carried by the solar spectrum in the characteristic absorption band of seawater turbidity, reflecting the light energy intensity that may be weakened by seawater.

[0145] In this embodiment, the identified characteristic absorption bands of seawater are compared with the measured solar spectrum to extract the energy distribution of sunlight within these bands. Special attention is paid to bands that possess both strong solar energy and are absorbed by seawater.

[0146] Step 603: Based on the solar spectral energy data, calculate the energy attenuation rate of the solar spectrum in the characteristic absorption band range.

[0147] In step 603, the energy attenuation rate refers to the proportion of energy lost by sunlight in the characteristic wavelength band after passing through seawater, which characterizes the degree to which seawater turbidity weakens light intensity.

[0148] In this embodiment, the attenuation of light intensity within a characteristic band is calculated based on solar spectrum measurements of the surface and deep seawater. By comparing the incident light intensity and the transmitted light intensity, the energy loss ratio for each band is determined.

[0149] Step 604: Generate environmental constraints based on the energy decay rate.

[0150] In this embodiment, the minimum photothermal conversion efficiency that the photothermal material needs to achieve is set based on the calculated energy attenuation rate for each wavelength band. These requirements constitute the basic constraints for subsequent material optimization.

[0151] Here is a specific example:

[0152] During the optimization of the solar-powered seawater desalination system in nearshore area A, water quality monitoring equipment first detected significant turbidity absorption peaks at 580nm and 620nm, with peak widths of 550-610nm and 600-640nm, respectively. Simultaneously, a spectrometer measured the solar energy at 125W / m² in the 550-610nm band and 110W / m² in the 600-640nm band. By comparing solar spectral data at the sea surface and a depth of 1 meter, the energy attenuation in the 550-610nm band was found to be 87.5W / m², resulting in an energy attenuation rate of (125-87.5) / 125 = 0.3, or 30%. The formula for calculating the energy attenuation rate is... η represents the attenuation rate. Represents incident energy. This represents the transmitted energy; in the 600-640nm band, the energy decays to 71.5 W / m², with a decay rate of (110-71.5) / 110=0.35, or 35%. Based on these data, the environmental constraints generated by the system require the photothermal material to compensate for at least 30% of the energy loss in the 550-610nm band and at least 35% of the energy loss in the 600-640nm band.

[0153] In this embodiment, the method establishes scientific and reasonable environmental constraints by quantitatively analyzing the impact of seawater turbidity on the solar spectrum, providing a clear objective for the optimization of photothermal materials and effectively solving the problem of unstable system performance in complex marine environments.

[0154] To address the problem of optimizing the mixing ratio of nanomaterials, in some embodiments, step 104: using a Bayesian optimization model, with the material spectral mapping library and the real-time heat flow data as input, to optimize and generate optimal nanomaterial mixing ratio data under the environmental constraints, includes:

[0155] Step 701: Use the material spectral mapping library as the first constraint, the interface temperature change rate in the real-time heat flow data as the second constraint, and the energy decay rate in the environmental constraint as the boundary condition.

[0156] In step 701, the interface temperature change rate in the real-time heat flow data originates from real-time monitoring data of the thermodynamic parameters of the evaporation interface during the operation of the solar seawater desalination system. This monitoring data is collected in real-time as a system feedback parameter and input into the optimization process. The first constraint refers to the photothermal response intensity requirements of each material recorded in the material spectral mapping library, ensuring that the material combination possesses basic photothermal conversion capabilities. The second constraint refers to the range of the real-time monitored interface temperature change rate, ensuring the system's operational stability. The boundary conditions refer to the minimum energy compensation standards for each wavelength band required in the environmental constraints.

[0157] In this embodiment, the system first extracts photothermal response data of various materials in key wavelength bands from a material spectral mapping library, serving as the basic constraint for optimizing the mixing ratio. Simultaneously, it reads real-time monitored temperature change data to determine the system's allowable operating parameter range. Combined with the previously calculated energy attenuation rates for each wavelength band, it sets the minimum compensation requirements that the material combination must achieve.

[0158] Step 702: Using the probability acquisition function of the Bayesian optimization model, a candidate nanomaterial mixing ratio dataset is generated under the conditions of the first constraint, the second constraint, and the boundary conditions.

[0159] In step 702, the probabilistic acquisition function is a decision mechanism in Bayesian optimization used to balance exploration and development, guiding the selection of new candidate points. The candidate nanomaterial mixing ratio dataset is a set of feasible solutions to be evaluated, generated by the probabilistic acquisition function.

[0160] In this embodiment, the optimization algorithm predicts potentially good new material combinations based on currently known material property data using a probability acquisition function. Under the premise of satisfying all constraints, multiple sets of candidate material mixing schemes with different proportions are generated, forming a dataset to be evaluated.

[0161] Step 703: Perform iterative optimization calculations on the candidate nanomaterial mixing ratio dataset. When the photothermal conversion efficiency calculation result corresponding to any candidate nanomaterial mixing ratio data in the candidate nanomaterial mixing ratio dataset meets the preset photothermal conversion efficiency threshold, the candidate nanomaterial mixing ratio data that meets the preset photothermal conversion efficiency threshold is taken as the optimal nanomaterial mixing ratio data.

[0162] In step 703, iterative optimization calculation refers to the process of cyclically evaluating and improving candidate solutions. The photothermal conversion efficiency threshold is a performance qualification standard set according to actual needs.

[0163] In this embodiment, the generated candidate dataset is evaluated iteratively. In each iteration, the photothermal conversion efficiency of each candidate scheme is calculated, and the schemes with superior performance are retained as the basis for subsequent optimization. When a candidate scheme that meets the preset performance requirements is found, the iteration terminates and the optimal solution is output.

[0164] Here is a specific example:

[0165] In the optimization process of the solar-powered seawater desalination system in nearshore area A, a Bayesian optimization process was initiated based on the previously established constraints. First, the photothermal response intensity of material B in the material spectral mapping library at 0.62 in the 580nm band and 0.58 in the 620nm band were set as the first constraint. Simultaneously, the real-time monitored interface temperature change rate of 0.5℃ / s was used as the second constraint, and boundary conditions were set requiring a 30% compensation in the 580nm band and a 35% compensation in the 620nm band. The Bayesian optimization model generated 10 candidate mixing ratios in the first round. Among them, the scheme of mixing type B and type C materials in a 7:3 ratio performed outstandingly. The calculated photothermal response intensity of this scheme in the 580nm band was 0.62×0.7+0.60×0.3=0.614, where 0.60 is the response intensity of type C material in the 580nm band. Substituting into the compensation calculation formula 0.614 / 0.62×100%=99%, the actual compensation capability was found to be 32.8%. In the 620nm band, the calculated value was 0.58×0.7+0.55×0.3=0.571, where 0.55 is the response intensity of type C material in the 620nm band. The actual compensation capability was 0.571 / 0.58×100%=98.4%, or 34.2%. At the same time, the temperature change rate corresponding to this scheme was 0.48-0.52℃ / s, which fully satisfied the second constraint condition. After three rounds of iterative optimization, this ratio was confirmed as the optimal solution, and the final output was a mixed ratio of 70% for type B and 30% for type C.

[0166] In this embodiment of the application, the method achieves precise proportioning of nanomaterials through intelligent optimization under multiple constraints. The generated scheme not only meets the spectral compensation requirements but also ensures stable system operation, effectively improving the adaptability and reliability of the solar seawater desalination system in complex marine environments.

[0167] Figure 2 A schematic diagram of a machine learning-based solar seawater desalination photothermal conversion efficiency optimization system is provided in this application embodiment, as shown below. Figure 2 As shown, the system includes:

[0168] The acquisition module 21 is used to acquire solar spectral distribution data, seawater turbidity spectral data, real-time heat flow data, and semiconductor nanomaterial data in the target sea area environment.

[0169] Module 22 is used to construct a material spectral mapping library based on the semiconductor nanomaterial data.

[0170] Analysis module 23 is used to couple and analyze the solar spectral distribution data with the seawater turbidity spectral data to form environmental constraints.

[0171] The generation module 24 is used to generate optimal nanomaterial mixing ratio data under the environmental constraints by using a Bayesian optimization model, taking the material spectral mapping library and the real-time heat flow data as input.

[0172] The prediction module 25 is used to predict the interface evaporation rate of the optimal nanomaterial mixing ratio data using a deep residual network, generate optimal material morphology information, and apply the optimal material morphology information to the morphology configuration of semiconductor nanomaterials to optimize the photothermal conversion efficiency in the solar desalination process.

[0173] Figure 2 The aforementioned machine learning-based solar seawater desalination photothermal conversion efficiency optimization system can perform... Figure 1 The implementation principle and technical effects of the machine learning-based solar seawater desalination photothermal conversion efficiency optimization method described in the illustrated embodiment will not be repeated here. The specific operation methods of each module and unit in the machine learning-based solar seawater desalination photothermal conversion efficiency optimization system described in the above embodiments have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0174] In one possible design, Figure 2 The machine learning-based solar seawater desalination photothermal conversion efficiency optimization system of the illustrated embodiment can be implemented as a computing device, such as... Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0175] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are invoked and executed by the processing component 32.

[0176] The processing component 32 is used to perform the above. Figure 1 The embodiment describes a machine learning-based method for optimizing the photothermal conversion efficiency of solar seawater desalination.

[0177] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-described method. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described method.

[0178] Storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented from any type of volatile or non-volatile storage device or a combination thereof, such as Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0179] Of course, computing devices may also include other components, such as input / output interfaces, display components, communication components, etc.

[0180] Input / output interfaces provide interfaces between processing components and peripheral interface modules, which can be output devices, input devices, etc.

[0181] The communication components are configured to facilitate wired or wireless communication between computing devices and other devices.

[0182] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0183] This application also provides a computer storage medium storing a computer program, which, when executed by a computer, can perform the above-described functions. Figure 1 The embodiment shown is a machine learning-based method for optimizing the photothermal conversion efficiency of solar seawater desalination.

[0184] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0185] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0186] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0187] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for optimizing the photothermal conversion efficiency of solar seawater desalination based on machine learning, characterized in that, The application relates to a solar energy sea water desalination method and device. The method comprises the following steps: acquiring solar spectrum distribution data, sea water turbidity spectrum data, real-time heat flow data and semiconductor nanometer material data in a target sea area; constructing a material spectrum mapping library based on the semiconductor nanometer material data; coupling and analyzing the solar spectrum distribution data and the sea water turbidity spectrum data to form an environmental constraint condition; inputting the material spectrum mapping library and the real-time heat flow data into a Bayesian optimization model to generate optimal proportion data under the environmental constraint condition, wherein the optimal proportion data represents the mixing proportion of optimal nanometer materials; using a deep residual network to predict the interfacial evaporation rate of the optimal proportion data to generate optimal material topography information, and applying the optimal material topography information to the topography configuration of semiconductor nanometer materials to optimize the photo-thermal conversion efficiency in a solar energy sea water desalination process; the deep residual network is used to predict the interfacial evaporation rate of the optimal proportion data to generate optimal material topography information, which comprises the following steps: inputting the optimal proportion data into an input layer of the deep residual network to obtain interfacial evaporation rate feature data; processing the interfacial evaporation rate feature data through a cross-level connection structure of the deep residual network; 2. The method of claim 1, wherein, based on the processed interfacial evaporation rate feature data, parallelly calculating nanosheet topography parameters and nanosphere topography parameters in an output layer of the deep residual network, and generating optimal material topography configuration information according to the nanosheet topography parameters and the nanosphere topography parameters. the deep residual network is used to predict the interfacial evaporation rate of the optimal proportion data to generate optimal material topography information, which comprises the following steps: inputting the processed interfacial evaporation rate feature data into a topography parameter analysis module of the deep residual network, and separating the processed interfacial evaporation rate feature data into nanosheet feature vectors and nanosphere feature vectors through a preset topography conversion rule; calculating the thickness distribution of the nanosheet feature vectors in the output layer of the deep residual network to output nanosheet topography parameters, and calculating the diameter distribution of the nanosphere feature vectors to output nanosphere topography parameters; establishing a spatial matching relationship between the nanosheet topography parameters and the nanosphere topography parameters; 3. The method of claim 2, wherein, based on the spatial matching relationship, generating optimal material topography configuration information. the deep residual network is used to predict the interfacial evaporation rate of the optimal proportion data to generate optimal material topography information, which comprises the following steps: determining a heat conduction constraint condition corresponding to the spatial matching relationship according to the heat flow conduction characteristics of a sea water evaporation interface; solving the maximum arrangement density of nanosheets and nanospheres under the heat conduction constraint condition through a gradient optimization algorithm; determining structure configuration parameters comprising nanosheet layer spacing and nanosphere stacking period based on the maximum arrangement density; 4. The method of claim 1, wherein, based on the structure configuration parameters, constructing optimal material topography configuration information. the deep residual network is used to predict the interfacial evaporation rate of the optimal proportion data to generate optimal material topography information, which comprises the following steps: extracting light absorption response data, particle size range data and plasmonic effect data from the semiconductor nanomaterial data; associating the light absorption response data with the particle size range data, and weighting and fusing the association result with the plasmonic effect data to construct a photothermal response intensity mapping rule; generating a material spectrum mapping library according to the photothermal response intensity mapping rule.

5. The method of claim 1, wherein, The coupling analysis of the solar spectrum distribution data and the seawater turbidity spectrum data forms an environmental constraint condition, which includes: determining a characteristic absorption waveband range of suspended particulate matter according to the seawater turbidity spectrum data; extracting solar spectrum energy data overlapping the characteristic absorption waveband range from the solar spectrum distribution data; calculating an energy attenuation rate of the solar spectrum in the characteristic absorption waveband range based on the solar spectrum energy data; generating an environmental constraint condition based on the energy attenuation rate.

6. The method of claim 1, wherein, The Bayesian optimization model takes the material spectrum mapping library and the real-time heat flow data as inputs to generate optimal proportion data under the environmental constraint condition, which includes: taking the material spectrum mapping library as a first constraint condition, taking the interface temperature change rate in the real-time heat flow data as a second constraint condition, and taking the energy attenuation rate in the environmental constraint condition as a boundary condition; generating a candidate nanomaterial mixing proportion data set by a probability collection function of the Bayesian optimization model under the conditions that the first constraint condition, the second constraint condition and the boundary condition are satisfied; iteratively optimizing and calculating the candidate nanomaterial mixing proportion data set, and when the photothermal conversion efficiency calculation result corresponding to any candidate nanomaterial mixing proportion data in the candidate nanomaterial mixing proportion data set meets a preset photothermal conversion efficiency threshold, the candidate nanomaterial mixing proportion data meeting the preset photothermal conversion efficiency threshold is taken as the optimal proportion data.

7. A machine learning based solar seawater desalination photothermal conversion efficiency optimization system, characterized in that, It includes: an acquisition module configured to acquire solar spectrum distribution data, seawater turbidity spectrum data, real-time heat flow data and semiconductor nanomaterial data in a target sea area; a construction module configured to construct a material spectrum mapping library based on the semiconductor nanomaterial data; an analysis module configured to couple and analyze the solar spectrum distribution data and the seawater turbidity spectrum data to form an environmental constraint condition; a generation module configured to generate optimal proportion data by a Bayesian optimization model taking the material spectrum mapping library and the real-time heat flow data as inputs under the environmental constraint condition, wherein the optimal proportion represents a mixing proportion of optimal nanomaterials; a prediction module configured to use a deep residual network to predict an interface evaporation rate of the optimal proportion data, generate optimal material topography information, and apply the optimal material topography information to the topography configuration of semiconductor nanomaterials to optimize the photothermal conversion efficiency in the process of solar seawater desalination; The use of a deep residual network to predict the interface evaporation rate of the optimal proportion data to generate optimal material topography information includes: inputting the optimal proportion data into an input layer of the deep residual network to obtain interface evaporation rate feature data; The interface evaporation rate characteristic data is processed through the cross-level connection structure of the deep residual network; Based on the processed interface evaporation rate characteristic data, the nanosheet morphology parameters and the nanosphere morphology parameters are calculated in parallel at the output layer of the deep residual network, and the optimal material morphology configuration information is generated according to the nanosheet morphology parameters and the nanosphere morphology parameters.

8. A computing device, comprising: The method comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to realize the method for optimizing the solar seawater desalination photo-thermal conversion efficiency based on machine learning according to any one of claims 1-6.

9. A computer storage medium, characterized in that The computer program is stored, and when the computer program is executed by a computer, the method for optimizing the solar seawater desalination photo-thermal conversion efficiency based on machine learning according to any one of claims 1-6 is realized.

Images