Desert photovoltaic sand damage monitoring method, device and equipment based on Internet of Things and medium

Abstract

The invention relates to the technical field of equipment monitoring. The desert photovoltaic sand damage monitoring method, device and equipment based on the Internet of Things and the medium are provided, and the method comprises the following steps: performing dynamic detection processing on microbial pollution characteristics on the surface of a photovoltaic module to generate microbial pollution characteristic data; biofilm identification processing is carried out through an AI classification model, and pollution type labels and bioactivity grade data are generated; performing corrosion risk assessment processing according to the pollution type label and the biological activity grade data, and generating a corrosion risk index and a cleaning priority ranking result; performing directional cleaning and protection treatment on the surface of the photovoltaic module to generate a cleaning completion signal; cloud aggregation and trend prediction processing are carried out on the corrosion risk index and the cleaning effect data, and a corrosion trend report and maintenance suggestion are generated, so that the problems of detection dimension splitting, static evaluation and decision lag are solved, and the technical effects of improving the biofilm recognition precision, dynamically quantifying the microbial corrosion risk and adapting to desert unsteady state environment prediction are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment monitoring, and particularly to a method, device, equipment and medium for monitoring desert photovoltaic sand hazards based on the Internet of Things. Background Art

[0002] With the large-scale construction of global desert photovoltaic power stations, photovoltaic modules are long-term exposed to complex environments such as sandstorms, extreme temperature differences and microbial erosion, resulting in increasingly prominent problems of biofilm pollution and chemical corrosion on the surface of the modules.

[0003] However, the related technologies have the following problems: the single detection dimension, the optical / electrochemical sensors work independently, and it is impossible to fuse multi-source data to accurately identify biofilms; the corrosion risk assessment is static, ignoring the coupling effect of spatial propagation and dynamic environment, resulting in the corrosion index deviating from the actual situation; the data-driven prediction ability lags behind, and the linear algorithm is difficult to adapt to the non-steady state environment of the desert, and the decision-making suggestions lag behind the pollution evolution. Summary of the Invention

[0004] Based on this, it is necessary to provide a method, device, equipment and medium for monitoring desert photovoltaic sand hazards based on the Internet of Things for the above technical problems, so as to solve the problems of fragmented detection dimensions, static assessment and lagging decision-making, and achieve the technical effects of improving the accuracy of biofilm recognition, dynamically quantifying the microbial corrosion risk, and adapting to the prediction of the non-steady state environment of the desert.

[0005] In a first aspect, the present application provides a method for monitoring desert photovoltaic sand hazards based on the Internet of Things, and the method includes:

[0006] Dynamically detect and process the microbial pollution characteristics on the surface of the photovoltaic module to generate microbial pollution characteristic data;

[0007] Based on the microbial pollution characteristic data, perform biofilm recognition processing through an AI classification model to generate pollution type labels and bioactivity level data;

[0008] According to the pollution type labels and bioactivity level data, perform corrosion risk assessment processing to generate a corrosion risk index and a cleaning priority ranking result;

[0009] Based on the cleaning priority ranking result, perform directional cleaning and protection processing on the surface of the photovoltaic module to generate a cleaning completion signal;

[0010] Perform cloud aggregation and trend prediction processing on the corrosion risk index and cleaning effect data to generate a corrosion trend report and maintenance suggestions.

[0011] Further, performing cloud aggregation and trend prediction processing on the corrosion risk index and cleaning effect data to generate a corrosion trend report and maintenance suggestions includes:

[0012] Perform noise filtering and missing value filling on the corrosion risk index and cleaning effect data to generate the cleaned corrosion time series data;

[0013] Based on the preset component numbers and timestamps, perform multi-source alignment on the cleaned corrosion time series data to generate a spatio-temporally correlated corrosion time series dataset;

[0014] Input the corrosion time series dataset into the trained long short-term memory neural network model to extract the periodic fluctuation characteristics of the component surface corrosion rate and generate corrosion trend prediction parameters;

[0015] Based on the corrosion trend prediction parameters, perform dynamic programming on the collaborative maintenance time window of the multi-component area to generate a corrosion trend report and maintenance suggestions including maintenance time, priority, and resource allocation strategy.

[0016] Furthermore, input the corrosion time series dataset into the trained long short-term memory neural network model to extract the periodic fluctuation characteristics of the component surface corrosion rate and generate corrosion trend prediction parameters, including:

[0017] Perform time series segmentation on the corrosion time series dataset, divide the time window based on the sandstorm occurrence cycle and microbial activity cycle, and generate multi-scale time series segments;

[0018] Input the multi-scale time series segments into the bidirectional long short-term memory neural network model to capture the forward and backward time series dependencies and generate initial time series feature vectors;

[0019] Perform weight assignment on the initial time series feature vectors based on the attention mechanism to generate high-weight periodic fluctuation characteristics;

[0020] Perform multi-scale fusion on the high-weight periodic fluctuation characteristics and static environmental characteristics to generate corrosion trend prediction parameters.

[0021] Furthermore, perform corrosion risk assessment based on the pollution type label and biological activity level data to generate the corrosion risk index and the cleaning priority ranking result, including:

[0022] Perform multi-source data fusion on the pollution type label and biological activity level data to generate comprehensive microbial pollution assessment data;

[0023] Input the comprehensive microbial pollution assessment data into the trained corrosion risk assessment model to map the coupling relationship between the component surface corrosion rate and power attenuation and generate initial corrosion risk parameters;

[0024] Based on the spatial adjacency of the photovoltaic array and the environmental wind speed data, perform propagation path analysis on the initial corrosion risk parameters to generate a corrosion risk index with dynamically adjusted weights;

[0025] Based on the corrosion risk index adjusted by dynamic weights, perform multi-objective optimization sorting on the polluted areas to generate a cleaning priority sorting result that takes into account both cleaning efficiency and corrosion inhibition.

[0026] Furthermore, based on the spatial adjacency of the photovoltaic array and the environmental wind speed data, perform propagation path analysis on the initial corrosion risk parameters to generate a corrosion risk index adjusted by dynamic weights, including:

[0027] Based on the topological relationship of the spatial layout of the photovoltaic array and the component spacing data, construct a topological network model of the pollution propagation path to generate the pollution diffusion correlation parameter between components;

[0028] Combined with the environmental wind speed data, perform wind direction correction on the pollution diffusion correlation parameter to generate a wind speed weighted propagation probability matrix;

[0029] Input the initial corrosion risk parameters into the risk weighted propagation probability matrix, perform iterative calculation on the risk superposition effect of adjacent components to generate a dynamic propagation risk weight;

[0030] Based on the dynamic propagation risk weight, perform multi-factor fusion on the initial corrosion risk parameters to generate a corrosion risk index adjusted by dynamic weights.

[0031] Furthermore, based on the cleaning priority sorting result, perform directional cleaning and protection on the surface of the photovoltaic components to generate a cleaning completion signal, including:

[0032] Based on the cleaning priority sorting result, perform pre-wetting treatment on the surface of the photovoltaic components in the high-priority area to generate a surface humidity compliance signal;

[0033] According to the surface humidity compliance signal, start the directional cleaning device to perform high-pressure water mist gradient spraying on the target area to generate a preliminary cleaning signal;

[0034] Based on the preliminary cleaning signal, perform self-adaptive spraying of microbial inhibitors on the surface of the cleaned components to generate a protective layer thickness parameter;

[0035] According to the protective layer thickness parameter, perform optical detection on the uniformity of the inhibitor coverage to generate a protective layer formation signal;

[0036] Perform a logical AND operation on the preliminary cleaning signal and the protective layer formation signal to generate a cleaning completion signal.

[0037] Furthermore, perform a logical AND operation on the preliminary cleaning signal and the protective layer formation signal to generate a cleaning completion signal, including:

[0038] Perform a cleaning effect verification process on the preliminary cleaning signal, compare and analyze the multi-spectral reflection data with the preset cleaning threshold, and generate a cleaning compliance flag;

[0039] Perform a thickness and coverage anomaly detection process on the protective layer formation signal, and generate a protective layer qualification flag;

[0040] Based on the current environmental wind speed and sand dust concentration data, perform a dynamic weight assignment process on the cleaning compliance flag and the protective layer qualification flag, and generate a weighted logical AND condition parameter;

[0041] According to the weighted logical AND condition parameter, perform a multi-modal fusion determination process on the preliminary cleaning signal and the protective layer formation signal, and generate a cleaning completion signal.

[0042] In a second aspect, the present application also provides an Internet of Things-based desert photovoltaic sand hazard monitoring device, which includes:

[0043] A pollution dynamic detection module for performing a dynamic detection process on the surface microbial pollution characteristics of photovoltaic modules and generating microbial pollution characteristic data;

[0044] An intelligent biofilm recognition module for performing a biofilm recognition process based on the microbial pollution characteristic data through an AI classification model and generating a pollution type label and bioactivity level data;

[0045] A corrosion risk assessment module for performing a corrosion risk assessment process according to the pollution type label and the bioactivity level data and generating a corrosion risk index and a cleaning priority ranking result;

[0046] An intelligent cleaning and protection module for performing a directional cleaning and protection process on the surface of photovoltaic modules based on the cleaning priority ranking result and generating a cleaning completion signal;

[0047] A cloud prediction and optimization module for performing cloud aggregation and trend prediction processes on the corrosion risk index and the cleaning effect data and generating a corrosion trend report and maintenance suggestions.

[0048] In a third aspect, the present application also provides a computer device, including a memory and a processor, the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of any method in the first aspect of the present application are implemented.

[0049] In a fourth aspect, the present application also provides a computer-readable storage medium, on which a computer program is stored, and is characterized in that when the computer program is executed by a processor, the steps of any method in the first aspect of the present application are implemented.

[0050] The technical solutions provided by this application include the following technical effects: By providing an Internet of Things-based desert photovoltaic sand hazard monitoring method, device, equipment, and medium, the method includes: dynamically detecting and processing the microbial contamination characteristics on the surface of photovoltaic modules to generate microbial contamination characteristic data; based on the microbial contamination characteristic data, performing biofilm recognition processing through an AI classification model to generate contamination type labels and bioactivity level data; according to the contamination type labels and bioactivity level data, performing corrosion risk assessment processing to generate a corrosion risk index and a cleaning priority ranking result; based on the cleaning priority ranking result, performing targeted cleaning and protection processing on the surface of the photovoltaic modules to generate a cleaning completion signal; aggregating the corrosion risk index and cleaning effect data in the cloud and performing trend prediction processing to generate a corrosion trend report and maintenance suggestions, so as to solve the problems of fragmented detection dimensions, static assessment, and lagging decision-making, and achieve the technical effects of improving the accuracy of biofilm recognition, dynamically quantifying the microbial corrosion risk, and adapting to the prediction of the non-steady-state desert environment. Description of the Drawings

[0051] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the following will briefly introduce the drawings required for use in the description of the embodiments or related technologies. Obviously, the drawings in the following description are only some embodiments of this application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0052] Figure 1 It is a flowchart of the Internet of Things-based desert photovoltaic sand hazard monitoring method in an embodiment of the present invention;

[0053] Figure 2 It is a structural diagram of the Internet of Things-based desert photovoltaic sand hazard monitoring device in an embodiment of the present invention. Detailed Embodiments

[0054] In order to make the above objects, features, and advantages of this application more obvious and understandable, the following will describe the specific implementation manners of this application in detail with reference to the drawings. Many specific details are set forth in the following description in order to fully understand this application. However, this application can be implemented in many other ways different from those described herein. Those skilled in the art can make similar improvements without departing from the connotation of the application. Therefore, this application is not limited by the specific embodiments disclosed below.

[0055] As Figure 1 shown, this application provides an Internet of Things-based desert photovoltaic sand hazard monitoring method, and the method includes:

[0056] S101: Dynamically detect and process the microbial contamination characteristics on the surface of the photovoltaic module to generate microbial contamination characteristic data.

[0057] Specifically, an optical sensor (such as a multispectral or hyperspectral imager) and a chemical sensor (such as a pH sensor or an organic acid detection probe) are used to collect the microbial contamination characteristics on the surface of the photovoltaic module in real time or periodically, obtaining raw data such as spectral reflectance and chemical activity signals. The collected raw data is processed by noise filtering and missing value filling to ensure the integrity and accuracy of the data. At the same time, the data is normalized for subsequent analysis. Through a feature extraction algorithm (such as principal component analysis PCA or a deep learning feature extraction model), features related to microbial contamination are extracted from the preprocessed data, including spectral features, chemical activity features (such as organic acid concentration), microbial metabolite features, etc. The extracted multi-source features are fused to generate comprehensive microbial contamination feature data. For example, by combining spectral features and chemical activity features, a comprehensive feature vector that can characterize the degree and type of microbial contamination is generated. The fused feature data is stored as a structured microbial contamination feature data set for subsequent biofilm identification and corrosion risk assessment. Through the above steps, the dynamic detection of the microbial contamination characteristics on the surface of the photovoltaic module and the generation of feature data are realized, providing a basis for subsequent biofilm identification and corrosion risk assessment.

[0058] S102: Based on the microbial contamination feature data, perform biofilm identification processing through an AI classification model to generate contamination type labels and bioactivity level data.

[0059] Specifically, the collected microbial contamination characteristic data is cleaned, including noise filtering and missing value filling, to ensure the integrity and accuracy of the data. At the same time, the data is normalized for subsequent analysis. Features related to biofilm recognition are extracted from the preprocessed data, including spectral features, chemical activity features (such as organic acid concentration), microbial metabolite features, etc. Principal component analysis (PCA) or deep learning feature extraction models are used to extract key features. The extracted multi-source features are fused to generate comprehensive microbial contamination characteristic data. For example, by combining spectral features and chemical activity features, a comprehensive feature vector that can characterize the degree and type of microbial contamination is generated. A trained AI classification model (such as convolutional neural network CNN, support vector machine SVM, etc.) is used to classify the fused feature data. The model needs to be pre-trained with a large amount of labeled data to ensure the accuracy and robustness of the classification. The fused feature data is input into the trained AI classification model, and the model will output contamination type labels (such as lichen, cyanobacteria, etc.) and bioactivity level data (such as low, medium, high activity). According to the output results of the AI classification model, contamination type labels and bioactivity level data are generated for subsequent corrosion risk assessment and cleaning strategy formulation. Through the above steps, the biofilm recognition process based on microbial contamination characteristic data is realized, providing a basis for subsequent corrosion risk assessment and cleaning strategy.

[0060] S103: According to the contamination type label and bioactivity level data, perform corrosion risk assessment processing to generate a corrosion risk index and a cleaning priority ranking result.

[0061] Specifically, the pollution type tags and bioactivity level data are fused to generate a comprehensive microbial pollution characteristic dataset. This step ensures that data from different sources can work together, providing a comprehensive information basis for subsequent corrosion risk assessment. The fused microbial pollution characteristic data is input into a trained corrosion risk assessment model. This model can be a risk assessment-based model, a mass transfer process control-based model, or a comprehensive model based on the electrochemical corrosion mechanism. The model analyzes the relationship between microbial pollution characteristics and corrosion rate to predict the likelihood and severity of corrosion. Considering the spatial adjacency of the photovoltaic array and environmental wind speed data, a propagation path analysis is performed on the initial corrosion risk parameters. By constructing a pollution propagation path topology network model and combining wind speed data to correct the pollution diffusion correlation parameters, a corrosion risk index with dynamically adjusted weights is generated. This step can more accurately reflect the corrosion risk distribution in the actual environment. Based on the corrosion risk index with dynamically adjusted weights, a multi-objective optimization and sorting process is carried out for the polluted areas. Through an optimization algorithm, a cleaning priority sorting result that takes into account both cleaning efficiency and corrosion inhibition is generated to ensure the optimality of the cleaning strategy. Subsequently, the corrosion risk index and the cleaning priority sorting result are generated, providing a scientific basis for subsequent cleaning and maintenance strategies. Through the above steps, the accurate assessment of the corrosion risk on the surface of photovoltaic modules and the optimization of the cleaning strategy are achieved, effectively solving the problems of static assessment and decision-making lag in traditional methods.

[0062] S104: Based on the cleaning priority sorting result, perform directional cleaning and protection on the surface of the photovoltaic module to generate a cleaning completion signal.

[0063] Specifically, according to the pollution degree and corrosion risk, the photovoltaic modules are sorted according to the cleaning priority. Based on the cleaning priority sorting result, the corresponding cleaning equipment is activated. The surface of the photovoltaic modules in the high-priority areas is pre-wetted to reduce the static electricity and dust flying that may occur during the cleaning process and prepare for subsequent cleaning steps. The high-pressure water mist gradient spraying technology is used to clean the surface of the photovoltaic modules. This technology can effectively remove the dust and biofilm on the surface while reducing water resource consumption. An adaptive spraying process of a microbial inhibitor is carried out on the surface of the cleaned module. This step aims to prevent the reformation of the biofilm and extend the duration of the cleaning effect. The uniformity of the inhibitor coverage and the thickness of the protective layer are verified by optical detection means to ensure that the protective layer can effectively inhibit microbial growth. When the above cleaning and protection steps are completed, a cleaning completion signal is generated. This signal can be a simple status indication or a report containing detailed cleaning effect data for subsequent maintenance and management decisions. Through the above steps, it can be ensured that the surface of the photovoltaic module is effectively cleaned and protected, thereby improving its power generation efficiency and service life.

[0064] S105: Aggregate the corrosion risk index and cleaning effect data in the cloud and perform trend prediction processing to generate a corrosion trend report and maintenance suggestions.

[0065] Specifically, aggregate the corrosion risk index and cleaning effect data from different components in the cloud. This step ensures centralized management of the data, providing a unified data basis for subsequent analysis. Perform noise filtering and missing value filling on the aggregated data to generate cleaned corrosion time series data. This step improves the quality of the data, providing reliable data support for subsequent analysis. Use the kernel principal component analysis (KPCA) algorithm to extract features from the cleaned data, reduce the feature dimension, reduce noise interference, and extract key features at the same time. This step can improve the prediction performance of the model. Input the extracted feature data into a trained long short-term memory (LSTM) neural network model to extract and process the periodic fluctuation characteristics of the corrosion rate on the component surface, generating corrosion trend prediction parameters. This step can capture the non-linear correlation between the periodicity of sandstorms, the fluctuation of microbial activity, and the corrosion rate in the desert environment.

[0066] Based on the corrosion trend prediction parameters, perform dynamic programming on the collaborative maintenance time window of the multi-component area to generate a corrosion trend report and maintenance suggestions including maintenance time, priority, and resource allocation strategy. This step ensures the optimality of the maintenance strategy and reduces the operation and maintenance cost. Integrate the corrosion trend prediction results and maintenance suggestions into a detailed corrosion trend report, providing a scientific basis for subsequent maintenance decisions. Through the above steps, the cloud aggregation and trend prediction processing of the corrosion risk index and cleaning effect data are realized, providing accurate decision support for the maintenance of desert photovoltaic power stations.

[0067] An embodiment of the present application provides a desert photovoltaic sand hazard monitoring method based on the Internet of Things, including: dynamically detecting the microbial contamination characteristics on the surface of photovoltaic components to generate microbial contamination characteristic data; based on the microbial contamination characteristic data, performing biofilm recognition through an AI classification model to generate pollution type labels and bioactivity level data; according to the pollution type labels and bioactivity level data, performing corrosion risk assessment to generate a corrosion risk index and a cleaning priority ranking result; based on the cleaning priority ranking result, performing directional cleaning and protection on the surface of the photovoltaic components to generate a cleaning completion signal; aggregating the corrosion risk index and cleaning effect data in the cloud and performing trend prediction processing to generate a corrosion trend report and maintenance suggestions, so as to solve the problems of fragmented detection dimensions, static assessment, and lagging decision-making, and achieve the technical effects of improving the biofilm recognition accuracy, dynamically quantifying the microbial corrosion risk, and adapting to the prediction of the desert non-steady state environment.

[0068] Furthermore, cloud aggregation and trend prediction processing are performed on the corrosion risk index and cleaning effect data to generate a corrosion trend report and maintenance suggestions, including:

[0069] (1) Perform noise filtering and missing value filling on the corrosion risk index and cleaning effect data to generate cleaned corrosion time series data;

[0070] (2) Based on the preset component numbers and timestamps, perform multi-source alignment on the cleaned corrosion time series data to generate a spatio-temporally correlated corrosion time series dataset;

[0071] (3) Input the corrosion time series dataset into the trained long short-term memory neural network model to extract the periodic fluctuation characteristics of the component surface corrosion rate and generate corrosion trend prediction parameters;

[0072] (4) Based on the corrosion trend prediction parameters, perform dynamic programming on the collaborative maintenance time window of the multi-component area to generate a corrosion trend report and maintenance suggestions including maintenance time, priority, and resource allocation strategy.

[0073] Specifically, aggregate the corrosion risk index and cleaning effect data from different components in the cloud. Perform noise filtering and missing value filling on the aggregated data to generate cleaned corrosion time series data. This step ensures the integrity and quality of the data, providing reliable data support for subsequent analysis. Based on the preset component numbers and timestamps, perform multi-source alignment on the cleaned corrosion time series data to generate a spatio-temporally correlated corrosion time series dataset. This step ensures that data from different sources can be accurately matched in time and space, providing a basis for subsequent trend prediction. Use the kernel principal component analysis (KPCA) algorithm to extract features and reduce the dimensionality of the cleaned data, reducing the feature dimension and noise interference while extracting key features. This step can improve the prediction performance of the model. Input the extracted feature data into the trained long short-term memory (LSTM) neural network model to extract the periodic fluctuation characteristics of the component surface corrosion rate and generate corrosion trend prediction parameters. The LSTM model can capture the long-term dependencies of the data through memory units and gating mechanisms, making up for the deficiencies of traditional models in processing time series data. Based on the corrosion trend prediction parameters, perform dynamic programming on the collaborative maintenance time window of the multi-component area to generate a corrosion trend report and maintenance suggestions including maintenance time, priority, and resource allocation strategy. This step ensures the optimality of the maintenance strategy and reduces the operation and maintenance costs. Integrate the corrosion trend prediction results and maintenance suggestions into a detailed corrosion trend report to provide a scientific basis for subsequent maintenance decisions.

[0074] Further, input the corrosion time series dataset into the trained long short-term memory neural network model to extract and process the periodic fluctuation characteristics of the corrosion rate on the component surface, and generate corrosion trend prediction parameters, including:

[0075] (1) Perform time series segmentation processing on the corrosion time series dataset, divide time windows based on the sandstorm occurrence cycle and the microbial activity cycle, and generate multi-scale time series segments;

[0076] (2) Input the multi-scale time series segments into a bidirectional long short-term memory neural network model to capture and process the forward and reverse time series dependency relationships, and generate initial time series feature vectors;

[0077] (3) Perform weight assignment processing on the initial time series feature vectors based on the attention mechanism to generate high-weight periodic fluctuation characteristics;

[0078] (4) Perform multi-scale fusion processing on the high-weight periodic fluctuation characteristics and static environment characteristics to generate corrosion trend prediction parameters.

[0079] Specifically, perform time series segmentation processing on the corrosion time series dataset, divide time windows based on the sandstorm occurrence cycle and the microbial activity cycle, and generate multi-scale time series segments. This step can capture corrosion characteristics at different time scales and provide a basis for subsequent feature extraction. Input the multi-scale time series segments into a bidirectional long short-term memory (BiLSTM) neural network model. BiLSTM can process both the forward and reverse parts of the input sequence simultaneously, thus better capturing the context relationships in the sequence data. The BiLSTM model captures and processes the forward and reverse time series dependency relationships through its two internal LSTM layers respectively to generate initial time series feature vectors. This step can make full use of the historical and future information in the sequence data and improve the prediction ability of the model. Use the attention mechanism to perform weight assignment processing on the initial time series feature vectors to generate high-weight periodic fluctuation characteristics. The attention mechanism can highlight important features and ignore unimportant features, thereby improving the prediction accuracy of the model. Perform multi-scale fusion processing on the high-weight periodic fluctuation characteristics and static environment characteristics to generate corrosion trend prediction parameters. This step can comprehensively consider dynamic and static factors and further improve the accuracy of the prediction.

[0080] Further, perform corrosion risk assessment processing based on the pollution type label and biological activity level data to generate a corrosion risk index and a cleaning priority ranking result, including:

[0081] (1) Perform multi-source data fusion processing on the pollution type label and biological activity level data to generate comprehensive microbial pollution assessment data;

[0082] (2) Input the comprehensive assessment data of microbial contamination into the trained corrosion risk assessment model, perform mapping processing on the coupling relationship between the corrosion rate and power attenuation of the component surface, and generate initial corrosion risk parameters;

[0083] (3) Based on the spatial adjacency of the photovoltaic array and environmental wind speed data, perform propagation path analysis on the initial corrosion risk parameters to generate a corrosion risk index with dynamically adjusted weights;

[0084] (4) According to the corrosion risk index with dynamically adjusted weights, perform multi-objective optimization and sorting on the contaminated areas to generate a cleaning priority sorting result that takes into account both cleaning efficiency and corrosion inhibition.

[0085] Specifically, perform multi-source data fusion on the pollution type labels and bioactivity level data to generate comprehensive assessment data of microbial contamination. This step ensures that data from different sources can work together, providing a comprehensive information basis for subsequent corrosion risk assessment. Input the comprehensive assessment data of microbial contamination into the trained corrosion risk assessment model. This model analyzes the relationship between microbial contamination characteristics and corrosion rate, predicts the possibility and severity of corrosion, and generates initial corrosion risk parameters. Based on the spatial adjacency of the photovoltaic array and environmental wind speed data, perform propagation path analysis on the initial corrosion risk parameters. By constructing a topological network model of the pollution propagation path and combining wind speed data to correct the pollution diffusion correlation parameters, a corrosion risk index with dynamically adjusted weights is generated. According to the corrosion risk index with dynamically adjusted weights, perform multi-objective optimization and sorting on the contaminated areas. Through an optimization algorithm, a cleaning priority sorting result that takes into account both cleaning efficiency and corrosion inhibition is generated to ensure the optimality of the cleaning strategy.

[0086] Furthermore, based on the spatial adjacency of the photovoltaic array and environmental wind speed data, perform propagation path analysis on the initial corrosion risk parameters to generate a corrosion risk index with dynamically adjusted weights, including:

[0087] (1) Based on the topological relationship of the spatial layout of the photovoltaic array and component spacing data, construct a topological network model of the pollution propagation path to generate pollution diffusion correlation parameters between components;

[0088] (2) Combine environmental wind speed data to perform wind direction correction on the pollution diffusion correlation parameters to generate a wind speed weighted propagation probability matrix;

[0089] (3) Input the initial corrosion risk parameters into the risk weighted propagation probability matrix, perform iterative calculation on the risk superposition effect of adjacent components to generate dynamic propagation risk weights;

[0090] (4) Based on the dynamic propagation risk weights, perform multi-factor fusion processing on the initial corrosion risk parameters to generate a corrosion risk index with adjusted dynamic weights.

[0091] Specifically, based on the spatial layout topological relationship of the photovoltaic array and the component spacing data, construct a topological network model of the pollution propagation path to generate the pollution diffusion correlation parameter between components. This step determines the pollution propagation path between components by analyzing the physical layout of the photovoltaic array and the component spacing. Combine the environmental wind speed data to perform wind direction correction processing on the pollution diffusion correlation parameter to generate a wind speed-weighted propagation probability matrix. The wind speed and wind direction have a significant impact on pollution diffusion. By dynamically adjusting the propagation probability matrix, the actual pollution propagation situation can be more accurately reflected. Input the initial corrosion risk parameters into the wind speed-weighted propagation probability matrix, perform iterative calculation processing on the risk superposition effect of adjacent components to generate dynamic propagation risk weights. This step simulates the pollution diffusion effect between components through iterative calculation and dynamically adjusts the risk weights. Based on the dynamic propagation risk weights, perform multi-factor fusion processing on the initial corrosion risk parameters to generate a corrosion risk index with adjusted dynamic weights. Through an optimization algorithm, comprehensively consider the pollution propagation between components and environmental factors to generate a more accurate corrosion risk index.

[0092] Furthermore, based on the cleaning priority sorting result, perform directional cleaning and protection processing on the surface of the photovoltaic module to generate a cleaning completion signal, including:

[0093] (1) Based on the cleaning priority sorting result, perform pre-wetting processing on the surface of the photovoltaic modules in the high-priority area to generate a surface humidity compliance signal;

[0094] (2) According to the surface humidity compliance signal, start the directional cleaning device to perform high-pressure water mist gradient spraying processing on the target area to generate a preliminary cleaning signal;

[0095] (3) Based on the preliminary cleaning signal, perform self-adaptive spraying processing of the microbial inhibitor on the cleaned component surface to generate a protective layer thickness parameter;

[0096] (4) According to the protective layer thickness parameter, perform optical detection processing on the uniformity of the inhibitor coverage to generate a protective layer formation signal;

[0097] (5) Perform logical AND processing on the preliminary cleaning signal and the protective layer formation signal to generate a cleaning completion signal.

[0098] Specifically, according to the cleaning priority sorting result, the surface of the photovoltaic modules in the high-priority area is pre-wetted to reduce static electricity and dust flying during the cleaning process, and at the same time prepare for the subsequent cleaning steps. When the surface humidity reaches the preset standard, a surface humidity compliance signal is generated. After receiving the surface humidity compliance signal, the directional cleaning device is started to perform high-pressure water mist gradient spraying treatment on the target area to effectively remove the dust and biofilm on the module surface. After the treatment is completed, a preliminary cleaning signal is generated. Based on the preliminary cleaning signal, the microbial inhibitor adaptive spraying system is started to spray the cleaned module surface to form a protective layer to inhibit the growth of microorganisms. At the same time, a protective layer thickness parameter is generated for subsequent detection. According to the protective layer thickness parameter, optical detection technology is used to verify the uniformity of the inhibitor coverage and the thickness of the protective layer to ensure that the protective layer can effectively inhibit the growth of microorganisms. After the detection is qualified, a protective layer formation signal is generated. The preliminary cleaning signal and the protective layer formation signal are subjected to a logical AND operation. Only when both signals indicate that the cleaning and protection steps are successfully completed, a cleaning completion signal is generated, indicating the end of the entire cleaning and protection process.

[0099] Further, performing a logical AND operation on the preliminary cleaning signal and the protective layer formation signal to generate a cleaning completion signal includes:

[0100] (1) Performing a cleaning effect verification process on the preliminary cleaning signal, comparing and analyzing the multispectral reflection data with the preset cleaning threshold, and generating a cleaning compliance flag;

[0101] (2) Performing a thickness and coverage anomaly detection process on the protective layer formation signal to generate a protective layer compliance flag;

[0102] (3) Based on the current environmental wind speed and dust concentration data, performing a dynamic weight assignment process on the cleaning compliance flag and the protective layer compliance flag to generate a weighted logical AND condition parameter;

[0103] (4) According to the weighted logical AND condition parameter, performing a multimodal fusion determination process on the preliminary cleaning signal and the protective layer formation signal to generate a cleaning completion signal.

[0104] Specifically, a cleaning effect verification process is performed on the preliminary cleaning signal. By comparing and analyzing the multispectral reflection data with a preset cleaning threshold, a cleaning compliance flag is generated. This step ensures that the cleaning effect meets the expected standard. An abnormal detection process for the thickness and coverage of the protective layer formation signal is performed to generate a qualified protective layer flag. The uniformity of the inhibitor coverage and the thickness of the protective layer are verified through optical detection techniques to ensure that the protective layer can effectively inhibit microbial growth. Based on the current environmental wind speed and dust concentration data, a dynamic weight assignment process is performed on the cleaning compliance flag and the qualified protective layer flag to generate weighted logical AND conditional parameters. This step takes into account the influence of environmental factors on the cleaning and protection effects and ensures the rationality of the weight assignment. According to the weighted logical AND conditional parameters, a multimodal fusion determination process is performed on the preliminary cleaning signal and the protective layer formation signal to generate a cleaning completion signal. The multimodal fusion technology can comprehensively consider data from different modalities and improve the accuracy and reliability of the determination. Through the above steps, the comprehensive determination of the preliminary cleaning signal and the protective layer formation signal is achieved, a cleaning completion signal is generated, and the integrity and effectiveness of the surface cleaning and protection treatment of the photovoltaic module are ensured.

[0105] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0106] In one embodiment, as Figure 2 shown, the present application also provides an Internet of Things-based desert photovoltaic sand hazard monitoring device 200, which includes:

[0107] A pollution dynamic detection module 201 for dynamically detecting the microbial pollution characteristics on the surface of the photovoltaic module and generating microbial pollution characteristic data;

[0108] An intelligent biofilm recognition module 202 for performing biofilm recognition processing based on the microbial pollution characteristic data through an AI classification model to generate pollution type labels and bioactivity level data;

[0109] The corrosion risk assessment module 203 is used to perform corrosion risk assessment processing based on the pollution type label and bioactivity level data, and generate a corrosion risk index and a cleaning priority ranking result;

[0110] The intelligent cleaning and protection module 204 is used to perform directional cleaning and protection processing on the surface of the photovoltaic module based on the cleaning priority ranking result, and generate a cleaning completion signal;

[0111] The cloud prediction and optimization module 205 is used to perform cloud aggregation and trend prediction processing on the corrosion risk index and cleaning effect data, and generate a corrosion trend report and maintenance suggestions.

[0112] Specifically, the pollution dynamic detection module 201: This module is responsible for dynamically detecting the microbial pollution characteristics on the surface of the photovoltaic module, collecting data such as spectral reflectivity and chemical activity signals through optical sensors and chemical sensors, and generating microbial pollution characteristic data. The above data provides a basis for subsequent biofilm recognition and corrosion risk assessment. The intelligent biofilm recognition module 202: Based on the microbial pollution characteristic data, an AI classification model is used to identify the biofilm, and generate a pollution type label (such as lichen, cyanobacteria, etc.) and bioactivity level data (such as low, medium, and high activity). This step improves the accuracy of biofilm recognition through multi-source data fusion and feature extraction. The corrosion risk assessment module 203: Combining the pollution type label and bioactivity level data, it assesses the corrosion risk on the surface of the component, and generates a corrosion risk index and a cleaning priority ranking result. This module ensures the optimality of the cleaning strategy through dynamic weight adjustment and multi-objective optimization ranking. The intelligent cleaning and protection module 204: According to the cleaning priority ranking result, it performs directional cleaning and protection processing on the surface of the photovoltaic module. This module includes steps such as pre-wetting treatment, high-pressure water mist gradient spraying, microbial inhibitor spraying, and optical detection, and then generates a cleaning completion signal to ensure the reliability of the cleaning and protection effects. The cloud prediction and optimization module 205: Aggregates the corrosion risk index and cleaning effect data in the cloud, performs trend prediction through a long short-term memory neural network model, and generates a corrosion trend report and maintenance suggestions. This module provides a scientific basis for the long-term maintenance of the desert photovoltaic power station through dynamic programming of the maintenance time window and resource allocation strategy. Through the collaborative work of the above modules, the device 200 realizes the full-process automation from pollution detection to cleaning and protection, effectively improving the operation efficiency and service life of the photovoltaic module.

[0113] The cloud prediction and optimization module 205 is also used for:

[0114] Performing noise filtering and missing value filling processing on the corrosion risk index and cleaning effect data, and generating the cleaned corrosion time series data;

[0115] Based on the preset component numbers and timestamps, perform multi-source alignment processing on the cleaned corrosion time-series data to generate a spatio-temporally correlated corrosion time-series dataset;

[0116] Input the corrosion time-series dataset into the trained long short-term memory neural network model to extract the periodic fluctuation characteristics of the corrosion rate on the component surface and generate corrosion trend prediction parameters;

[0117] Based on the corrosion trend prediction parameters, perform dynamic programming processing on the collaborative maintenance time window for multi-component areas to generate a corrosion trend report and maintenance suggestions including maintenance time, priority, and resource allocation strategies.

[0118] The cloud prediction optimization module 205 is also used for:

[0119] Perform time-series segmentation processing on the corrosion time-series dataset, divide time windows based on the sandstorm occurrence cycle and microbial activity cycle, and generate multi-scale time-series segments;

[0120] Input the multi-scale time-series segments into the bidirectional long short-term memory neural network model to capture the forward and reverse time-series dependence relationships and generate initial time-series feature vectors;

[0121] Perform weight assignment processing on the initial time-series feature vectors based on the attention mechanism to generate high-weight periodic fluctuation characteristics;

[0122] Perform multi-scale fusion processing on the high-weight periodic fluctuation characteristics and static environmental characteristics to generate corrosion trend prediction parameters.

[0123] The corrosion risk assessment module 203 is also used for:

[0124] Perform multi-source data fusion processing on the pollution type labels and biological activity level data to generate comprehensive microbial pollution assessment data;

[0125] Input the comprehensive microbial pollution assessment data into the trained corrosion risk assessment model to map the coupling relationship between the corrosion rate and power attenuation on the component surface and generate initial corrosion risk parameters;

[0126] Based on the spatial adjacency of the photovoltaic array and the environmental wind speed data, perform propagation path analysis processing on the initial corrosion risk parameters to generate a corrosion risk index with dynamically adjusted weights;

[0127] According to the corrosion risk index with dynamically adjusted weights, perform multi-objective optimization sorting processing on the pollution areas to generate a cleaning priority sorting result that takes into account both cleaning efficiency and corrosion inhibition.

[0128] The corrosion risk assessment module 203 is also used for:

[0129] Based on the topological relationship of the spatial layout of the photovoltaic array and the component spacing data, construct a topological network model of the pollution propagation path, and generate the pollution diffusion correlation degree parameters between components;

[0130] Combined with the environmental wind speed data, perform wind direction correction processing on the pollution diffusion correlation degree parameters to generate a wind speed weighted propagation probability matrix;

[0131] Input the initial corrosion risk parameters into the risk weighted propagation probability matrix, perform iterative calculation processing on the risk superposition effect of adjacent components, and generate dynamic propagation risk weights;

[0132] Based on the dynamic propagation risk weights, perform multi-factor fusion processing on the initial corrosion risk parameters to generate a corrosion risk index after dynamic weight adjustment.

[0133] The intelligent cleaning and protection module 204 is also used for:

[0134] Based on the cleaning priority sorting result, perform pre-wetting treatment on the surface of the photovoltaic components in the high-priority area to generate a surface humidity compliance signal;

[0135] According to the surface humidity compliance signal, start the directional cleaning device to perform high-pressure water mist gradient spraying treatment on the target area to generate a preliminary cleaning signal;

[0136] Based on the preliminary cleaning signal, perform adaptive spraying treatment of microbial inhibitors on the cleaned component surface to generate a protective layer thickness parameter;

[0137] According to the protective layer thickness parameter, perform optical detection processing on the inhibitor coverage uniformity to generate a protective layer formation signal;

[0138] Perform a logical AND operation on the preliminary cleaning signal and the protective layer formation signal to generate a cleaning completion signal.

[0139] The intelligent cleaning and protection module 204 is also used for:

[0140] Perform cleaning effect verification processing on the preliminary cleaning signal, and perform comparison and analysis based on the multi-spectral reflection data and the preset cleaning threshold to generate a cleaning compliance flag;

[0141] Perform thickness and coverage anomaly detection processing on the protective layer formation signal to generate a protective layer qualification flag;

[0142] Based on the current environmental wind speed and dust concentration data, perform dynamic weight allocation processing on the cleaning compliance flag and the protective layer qualification flag to generate a weighted logical AND condition parameter;

[0143] According to the weighted logical AND condition parameter, perform multi-modal fusion determination processing on the preliminary cleaning signal and the protective layer formation signal to generate a cleaning completion signal.

[0144] In one embodiment, the present application further provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps in the above method embodiments are implemented.

[0145] In one embodiment, the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0146] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial descriptions of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the present disclosure solution. A person of ordinary skill in the art can understand and implement it without creative efforts.

[0147] The above embodiments only represent several implementation manners of the embodiments of the present application. The descriptions are relatively specific and detailed, but they should not be construed as limiting the patent scope of the application embodiments. It should be noted that for those of ordinary skill in the art, without departing from the concept of the embodiments of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the embodiments of the present application.

Claims

1. The desert photovoltaic sand hazard monitoring method based on the Internet of Things is characterized in that, The method includes: Dynamically detecting the microbial contamination characteristics on the surface of the photovoltaic module to generate microbial contamination characteristic data; Based on the microbial contamination characteristic data, performing biofilm recognition processing through an AI classification model to generate contamination type labels and bioactivity level data; According to the contamination type labels and bioactivity level data, performing corrosion risk assessment processing to generate a corrosion risk index and a cleaning priority ranking result; Based on the cleaning priority ranking result, performing directional cleaning and protection processing on the surface of the photovoltaic module to generate a cleaning completion signal; Performing cloud aggregation and trend prediction processing on the corrosion risk index and cleaning effect data to generate a corrosion trend report and maintenance suggestions.

2. The method for monitoring desert photovoltaic sand hazards based on the Internet of Things according to claim 1, characterized in that, The performing cloud aggregation and trend prediction processing on the corrosion risk index and cleaning effect data to generate a corrosion trend report and maintenance suggestions includes: Performing noise filtering and missing value filling processing on the corrosion risk index and cleaning effect data to generate cleaned corrosion time series data; Based on the preset component number and timestamp, performing multi-source alignment processing on the cleaned corrosion time series data to generate a spatio-temporally correlated corrosion time series dataset; Inputting the corrosion time series dataset into a trained long short-term memory neural network model to extract the periodic fluctuation characteristics of the corrosion rate on the component surface and generate corrosion trend prediction parameters; Based on the corrosion trend prediction parameters, performing dynamic planning processing on the collaborative maintenance time window of the multi-component area to generate the corrosion trend report and maintenance suggestions including maintenance time, priority, and resource allocation strategy.

3. The method for monitoring desert photovoltaic sand damage based on the Internet of Things according to claim 2, characterized in that, The inputting the corrosion time series dataset into a trained long short-term memory neural network model to extract the periodic fluctuation characteristics of the corrosion rate on the component surface and generate corrosion trend prediction parameters includes: Performing time series segmentation processing on the corrosion time series dataset, dividing time windows based on the sandstorm occurrence cycle and microbial activity cycle to generate multi-scale time series segments; Inputting the multi-scale time series segments into a bidirectional long short-term memory neural network model to capture the forward and backward time series dependence relationships and generate initial time series feature vectors; Performing weight assignment processing on the initial time series feature vectors based on the attention mechanism to generate high-weight periodic fluctuation characteristics; Performing multi-scale fusion processing on the high-weight periodic fluctuation characteristics and static environmental characteristics to generate the corrosion trend prediction parameters.

4. The method for monitoring desert photovoltaic sand hazards based on the Internet of Things according to claim 1, characterized in that, The according to the contamination type labels and bioactivity level data, performing corrosion risk assessment processing to generate a corrosion risk index and a cleaning priority ranking result includes: Performing multi-source data fusion processing on the contamination type labels and bioactivity level data to generate comprehensive microbial contamination assessment data; Inputting the comprehensive microbial contamination assessment data into a trained corrosion risk assessment model to map the coupling relationship between the corrosion rate and power attenuation on the component surface and generate initial corrosion risk parameters; Based on the spatial adjacency of the photovoltaic array and environmental wind speed data, performing propagation path analysis processing on the initial corrosion risk parameters to generate the corrosion risk index with dynamically adjusted weights; Based on the corrosion risk index after the dynamic weight adjustment, perform multi-objective optimization sorting on the polluted area to generate the cleaning priority sorting result that takes into account both cleaning efficiency and corrosion inhibition.

5. The method for monitoring desert photovoltaic sand damage based on the Internet of Things according to claim 4, wherein, Performing propagation path analysis on the initial corrosion risk parameters based on the spatial adjacency of the photovoltaic array and environmental wind speed data to generate the corrosion risk index after dynamic weight adjustment, including: Based on the spatial layout topological relationship of the photovoltaic array and the component spacing data, construct a pollution propagation path topological network model to generate the pollution diffusion correlation degree parameter between components; Combined with the environmental wind speed data, perform wind direction correction on the pollution diffusion correlation degree parameter to generate a wind speed weighted propagation probability matrix; Input the initial corrosion risk parameters into the risk weighted propagation probability matrix, and perform iterative calculation on the risk superposition effect of adjacent components to generate dynamic propagation risk weights; Based on the dynamic propagation risk weights, perform multi-factor fusion on the initial corrosion risk parameters to generate the corrosion risk index after the dynamic weight adjustment.

6. The method for monitoring desert photovoltaic sand damage based on the Internet of Things according to claim 1, wherein Based on the cleaning priority sorting result, perform directional cleaning and protection on the surface of the photovoltaic module to generate a cleaning completion signal, including: Based on the cleaning priority sorting result, perform pre-wetting on the surface of the photovoltaic module in the high-priority area to generate a surface humidity compliance signal; According to the surface humidity compliance signal, start the directional cleaning device to perform high-pressure water mist gradient spraying on the target area to generate a preliminary cleaning signal; Based on the preliminary cleaning signal, perform adaptive spraying of the microbial inhibitor on the surface of the cleaned component to generate the protective layer thickness parameter; According to the protective layer thickness parameter, perform optical detection on the uniformity of the inhibitor coverage to generate a protective layer formation signal; Perform a logical AND operation on the preliminary cleaning signal and the protective layer formation signal to generate the cleaning completion signal.

7. The method for monitoring desert photovoltaic sand hazards based on the Internet of Things according to claim 6, characterized in that, The performing a logical AND operation on the preliminary cleaning signal and the protective layer formation signal to generate the cleaning completion signal includes: Perform cleaning effect verification on the preliminary cleaning signal, and perform comparison and analysis based on multi-spectral reflection data and a preset cleaning threshold to generate a cleaning compliance flag; Perform abnormal detection on the thickness and coverage of the protective layer formation signal to generate a protective layer qualification flag; Based on the current environmental wind speed and dust concentration data, perform dynamic weight allocation on the cleaning compliance flag and the protective layer qualification flag to generate a weighted logical AND condition parameter; According to the weighted logical AND condition parameter, perform multi-modal fusion determination on the preliminary cleaning signal and the protective layer formation signal to generate the cleaning completion signal.

8. The desert photovoltaic sand hazard monitoring device based on the Internet of Things is characterized in that, The device includes: A pollution dynamic detection module for dynamically detecting the microbial pollution characteristics on the surface of the photovoltaic module to generate microbial pollution characteristic data; An intelligent biofilm recognition module for performing biofilm recognition based on the microbial pollution characteristic data through an AI classification model to generate pollution type labels and bioactivity level data; The corrosion risk assessment module is used to perform corrosion risk assessment processing based on the pollution type label and the bioactivity level data, and generate a corrosion risk index and a cleaning priority ranking result; The intelligent cleaning and protection module is used to perform directional cleaning and protection processing on the surface of the photovoltaic module based on the cleaning priority ranking result, and generate a cleaning completion signal; The cloud prediction and optimization module is used to perform cloud aggregation and trend prediction processing on the corrosion risk index and the cleaning effect data, and generate a corrosion trend report and maintenance suggestions.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the Internet of Things-based desert photovoltaic sand hazard monitoring method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the Internet of Things-based desert photovoltaic sand hazard monitoring method according to any one of claims 1 to 7.

Images