Method for dynamic assessment and real-time early warning of meteorological disaster risk of distributed photovoltaic power station

Abstract

Embodiments of the present disclosure disclose a distributed photovoltaic power station meteorological disaster risk dynamic assessment and real-time early warning method. A specific embodiment of the method comprises: preprocessing photovoltaic power station meteorological disaster related data; inputting the preprocessed photovoltaic power station meteorological disaster related data into a first disaster risk assessment model, a second disaster risk assessment model and a third disaster risk assessment model respectively; integrating the first disaster risk assessment result, the second disaster risk assessment result and the third disaster risk assessment result to generate a meteorological disaster risk value; setting a dynamic threshold according to the meteorological disaster risk value, dividing a risk level and generating an early warning signal; and linking a distributed photovoltaic power station monitoring system for real-time early warning. The embodiment realizes multi-platform data interconnection and intercommunication, is more suitable for the decentralized and multi-scenario operation and maintenance requirements of distributed photovoltaic power stations, and expands the technical application range.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of computer technology, specifically to a method for dynamic assessment and real-time early warning of meteorological disaster risks for distributed photovoltaic power stations. Background Technology

[0002] Against the backdrop of global energy transition and the advancement of the "dual-carbon" strategy, distributed photovoltaic (PV) power stations have become an important support for the popularization of clean energy due to their advantages such as flexible layout and short construction cycle. However, distributed PV power stations are mostly deployed on building rooftops, transportation facilities, or remote areas, making them susceptible to complex and changeable meteorological environments. Natural disasters such as typhoons, rainstorms, lightning strikes, and extreme heat are frequent, threatening the operational stability of the power stations and significantly increasing maintenance costs and operational risks.

[0003] In the field of meteorological disaster risk early warning for photovoltaic power plants, traditional early warning technologies mostly rely on static risk classification and single-factor threshold analysis. In recent years, however, machine learning, multi-model fusion, and other technologies have been gradually combined to explore refined early warning, multi-source data fusion, and dynamic disaster risk assessment. For example, focusing on typhoon risk for coastal photovoltaic power plants, Landsat remote sensing imagery and an improved random forest algorithm were used to extract photovoltaic distribution data from 2000 to 2023. Combined with 215 historical disaster questionnaires, a wind speed-photovoltaic loss rate model was established, clearly showing that the loss rate surges to 5.5% when the wind speed exceeds 20 m / s. Simultaneously, Google Earth was used... High-resolution image correction using Pro and other methods was performed to supplement the latest data from 2023, addressing the lag issue of traditional datasets that only go up to 2022. Marine risk was also categorized. For landslide and collapse disasters at photovoltaic power stations in Guizhou, the impact of rainfall was quantified using the information content method. An artificial neural network was constructed using coordinates, rainfall, and disaster susceptibility level as inputs, reducing the false alarm rate of Level 3 forecasts by 5.04%-9.65% and the false alarm rate of Level 4 warnings by 6.03%-11.17% compared to conventional methods, thus improving the accuracy of regional early warnings. Focusing on landslide risks at plateau photovoltaic power stations, an ANP-FBN model was constructed integrating 10 disaster-causing factors. Combined with the ID rainfall threshold, rainfall probability and risk zoning were superimposed to achieve dynamic assessment. Monitoring showed that 5 days after rainfall, the proportion of "high-to-extremely high" risk areas increased from 15.89% to 40.06%. Although focusing on earthquake early warning, its SVM model construction and 10-fold cross-validation parameter optimization method also provide a reference for rapid early warning of short-term extreme weather at photovoltaic power stations, helping to improve model accuracy.

[0004] However, existing photovoltaic power plant meteorological disaster early warning technologies have significant shortcomings: First, the generalization ability of single models is weak, making it difficult to cope with high-frequency fluctuations in meteorological data and regional heterogeneity; second, data fusion is insufficient, failing to systematically integrate multi-source data such as remote sensing images, ground sensors, and historical operation and maintenance logs, resulting in incomplete identification of disaster-causing factors; third, threshold settings are static, failing to consider the dynamic changes in power plant equipment status and environmental parameters, leading to high false alarm and false alarm rates; fourth, early warning is disconnected from operation and maintenance, lacking visual presentation and actionable response suggestions, failing to form a "early warning-response" closed loop. Summary of the Invention

[0005] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0006] Some embodiments of this disclosure propose methods, devices, and electronic equipment for dynamic assessment and real-time early warning of meteorological disaster risks for distributed photovoltaic power plants, in order to solve the technical problems mentioned in the background section above.

[0007] Firstly, some embodiments of this disclosure provide a method for dynamic assessment and real-time early warning of meteorological disaster risks for distributed photovoltaic power plants. The method includes: collecting meteorological disaster-related data for photovoltaic power plants, wherein the aforementioned meteorological disaster-related data includes: disaster-causing factor data, disaster-bearing body data, and recovery capacity data; preprocessing the aforementioned meteorological disaster-related data to generate preprocessed meteorological disaster-related data for photovoltaic power plants; and constructing a first disaster risk assessment model, a second disaster risk assessment model, and a third disaster risk assessment model, wherein the first disaster risk assessment model is a support vector machine, the second disaster risk assessment model is a convolutional neural network, and the third disaster risk assessment model is a multi-mode neural network. A meta-linear regression model is used. The preprocessed meteorological disaster-related data from the photovoltaic power station are input into the first, second, and third disaster risk assessment models, respectively, to obtain the first, second, and third disaster risk assessment results. These results are then integrated to generate a meteorological disaster risk value. Based on this risk value, a dynamic threshold is set, risk levels are classified, and early warning signals are generated. These early warning signals are displayed through a visual early warning platform and linked to the distributed photovoltaic power station monitoring system for real-time early warning.

[0008] Secondly, some embodiments of this disclosure provide a dynamic assessment and real-time early warning device for meteorological disaster risks of distributed photovoltaic power stations. The device includes: a data acquisition unit configured to acquire meteorological disaster-related data of the photovoltaic power station, wherein the aforementioned meteorological disaster-related data includes: disaster-causing factor data, disaster-bearing body data, and recovery capacity data; a preprocessing unit configured to preprocess the aforementioned meteorological disaster-related data of the photovoltaic power station to generate preprocessed meteorological disaster-related data of the photovoltaic power station; and a construction unit configured to construct a first disaster risk assessment model, a second disaster risk assessment model, and a third disaster risk assessment model, wherein the first disaster risk assessment model is a support vector machine, the second disaster risk assessment model is a convolutional neural network, and the third disaster risk assessment model is a multiple linear regression model. The system comprises three components: an input unit, configured to input the preprocessed meteorological disaster-related data of the photovoltaic power station into the first disaster risk assessment model, the second disaster risk assessment model, and the third disaster risk assessment model, respectively, to obtain the first disaster risk assessment result, the second disaster risk assessment result, and the third disaster risk assessment result; an integration unit, configured to integrate the first disaster risk assessment result, the second disaster risk assessment result, and the third disaster risk assessment result to generate a meteorological disaster risk value; a classification unit, configured to set a dynamic threshold based on the meteorological disaster risk value, classify the risk level, and generate an early warning signal; and an early warning unit, configured to display the early warning signal through a visual early warning platform and link it with the distributed photovoltaic power station monitoring system for real-time early warning.

[0009] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0010] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.

[0011] The above-described embodiments of this disclosure have the following beneficial effects: Through the distributed photovoltaic power station meteorological disaster risk dynamic assessment and real-time early warning method of some embodiments of this disclosure, a photovoltaic power station meteorological disaster original dataset, a power station multi-dimensional feature dataset, and a standardized meteorological disaster feature dataset are constructed. At the dataset construction stage, refined correlation matching of three types of data—disaster-causing factors, disaster-bearing bodies, and recovery capabilities—is achieved. Then, through a multi-model integration architecture centered on random forest and incorporating SVM, MLR, and CNN, the spatiotemporal coupling relationship of multiple data fields is deeply analyzed, effectively solving the problem of traditional technologies' difficulty in handling "meteorological-equipment-operation and maintenance" data collaboration, significantly improving the accuracy and dynamic adaptability of photovoltaic power station meteorological disaster risk early warning; multiple initial data indicators are selected, and 11 core features are obtained through HFS, PCA, and mRMR algorithm calculations and optimizations, greatly enriching the feature dimensions of photovoltaic power station risk assessment, strengthening the correlation between meteorological data and power station operation data, and providing a basis for... Multi-model integrated assessment provides high-quality data support, making it easier to accurately quantify risks in different regions and under different equipment conditions. Based on multi-model integration, a dynamic threshold adjustment mechanism with "cloud-edge-device" collaboration is introduced. The weights of each basic model are optimized by combining Shapley values, and the generalization ability of the model is enhanced by Bagging technology. At the same time, time-series data analysis is embedded to improve the efficiency of real-time risk capture. This not only solves the problem of poor adaptability of traditional static thresholds, but also avoids the prediction bias of a single model in extreme weather scenarios, further ensuring the timeliness of early warning response and the practicality of operation and maintenance linkage. The developed early warning system supports custom thresholds, alarm templates and regional division functions, and reserves API interfaces to connect with meteorological centers and power grid dispatching platforms. Compared with existing closed early warning tools, it not only improves users' customized management capabilities, but also realizes data interconnection and interoperability across multiple platforms, which is more in line with the decentralized and multi-scenario operation and maintenance needs of distributed photovoltaic power stations and expands the scope of technology application. Attached Figure Description

[0012] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0013] Figure 1 This is a flowchart of some embodiments of the method for dynamic assessment and real-time early warning of meteorological disaster risks of distributed photovoltaic power stations according to this disclosure; Figure 2 This is a data type summary diagram of meteorological disaster risk assessment for distributed photovoltaic power plants based on the dynamic assessment and real-time early warning method for meteorological disaster risks of distributed photovoltaic power plants disclosed herein; Figure 3This is a diagram illustrating the overall concept of a multi-model integration based on the dynamic assessment and real-time early warning method for meteorological disaster risks of distributed photovoltaic power stations disclosed herein. Figure 4 This is a schematic diagram of an early warning signal response mechanism based on the dynamic assessment and real-time early warning method for meteorological disaster risks of distributed photovoltaic power stations disclosed herein; Figure 5 This is a schematic diagram of the dynamic adjustment threshold principle based on the dynamic assessment and real-time early warning method for meteorological disaster risks of distributed photovoltaic power stations disclosed herein; Figure 6 This is a structural diagram of a distributed photovoltaic power station monitoring system in the dynamic assessment and real-time early warning method for meteorological disaster risks of distributed photovoltaic power stations according to this disclosure; Figure 7 This is a structural schematic diagram of some embodiments of the distributed photovoltaic power station meteorological disaster risk dynamic assessment and real-time early warning device according to the present disclosure; Figure 8 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0014] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0015] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0016] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0017] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0018] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0019] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] refer to Figure 1 The flowchart 100 illustrates some embodiments of the dynamic assessment and real-time early warning method for meteorological disaster risks of distributed photovoltaic power plants according to this disclosure. The method includes the following steps: Step 101: Collect meteorological disaster-related data for photovoltaic power plants.

[0021] In some embodiments, the implementing entity (e.g., a computing device) of the method for dynamic assessment and real-time early warning of meteorological disaster risks at distributed photovoltaic power stations can collect meteorological disaster-related data for photovoltaic power stations. This meteorological disaster-related data includes: disaster-causing factor data, disaster-bearing body data, and resilience data. The disaster-causing factor data includes: extreme daily precipitation, historical rainfall frequency, temperature, air humidity, total suspended particulate matter, maximum wind speed, and average wind speed; the disaster-bearing body data includes: the power generation capacity of the photovoltaic power station and the number of damaged equipment; and the resilience data includes: personnel deployment data and equipment repair technology data for the photovoltaic power station.

[0022] refer to Figure 2 The data classification framework is based on "disaster-causing factors - disaster-bearing bodies - resilience". Disaster-causing factor data is acquired through the China Meteorological Data Network, the ERA5 reanalysis data platform, and on-site sensors at power plants, covering extreme daily precipitation, historical heavy rainfall frequency, temperature, air humidity, total suspended particulate matter concentration, maximum wind speed, and average wind speed. Data on the bearing bodies is acquired through photovoltaic power plant operation and maintenance logs and equipment parameter manuals, including power generation capacity, number of damaged equipment, component type, and installation tilt angle. Resilience data is collected through on-site surveys of power plants, including the scale of manpower deployment, equipment repair technology level, and historical fault repair cycles. The final result is a multi-source raw dataset covering power plants of different sizes in the target area.

[0023] Step 102: Preprocess the above-mentioned meteorological disaster-related data of photovoltaic power plants to generate preprocessed meteorological disaster-related data of photovoltaic power plants.

[0024] In some embodiments, the aforementioned implementing entity may preprocess the aforementioned meteorological disaster-related data of the photovoltaic power station to generate preprocessed meteorological disaster-related data of the photovoltaic power station.

[0025] For example, a hierarchical feature selection algorithm can be used, with "mutual information and information gain ratio" as joint metrics, through the mutual information formula: ,in, Representative disaster-causing factors Risk indicators mutual information, Represents the disaster-causing factors. Represents risk indicators. For factor information entropy, Let's consider the conditional entropy. We calculate the correlation strength between each hazard-causing factor and the risk indicator, while also introducing the information gain ratio formula: ,in, Representative disaster-causing factors Risk indicators The information gain ratio is optimized to avoid interference from high-entropy, low-correlation factors, thereby eliminating low-impact and redundant factors and retaining core disaster-causing factors such as extreme daily precipitation, peak instantaneous wind speed, extreme temperature, and TSP concentration. Principal component analysis is then used to preprocess the core factors by centering them to eliminate dimensional interference. Eigenvalues ​​are then solved using the covariance matrix, and principal components are selected based on a cumulative variance contribution rate ≥ 90% to compress data dimensions and reduce feature redundancy. Subsequently, Z-score standardization is used to convert indicators with different dimensions, such as temperature, wind speed, and TSP concentration, into a standard normal distribution, unifying the data scale to ensure comparability. Finally, missing values ​​caused by sensor failures or transmission interruptions are filled using linear interpolation, and extreme outliers are handled using three criteria, ultimately forming a standardized dataset (preprocessed photovoltaic power station meteorological disaster-related data).

[0026] Step 103: Construct the first disaster risk assessment model, the second disaster risk assessment model, and the third disaster risk assessment model.

[0027] In some embodiments, the aforementioned implementing entity may construct a first disaster risk assessment model, a second disaster risk assessment model, and a third disaster risk assessment model. The first disaster risk assessment model is a support vector machine (SVM), the second disaster risk assessment model is a convolutional neural network (CNN), and the third disaster risk assessment model is a multiple linear regression (MLR) model.

[0028] Among them, SVM excels at handling nonlinear data classification and regression problems. By using the RBF kernel function, difficult-to-classify sample data in the collected dataset is mapped to a high-dimensional feature space, making the originally unseparable low-dimensional meteorological data linearly separable, thereby constructing an optimal linear separator and finding the optimal solution.

[0029] The kernel function of SVM is: ; in, , The input feature vector is low-dimensional and contains disaster-causing factors such as average wind speed and equipment operating parameters such as power generation. The kernel function parameters are used to control the smoothness of the function.

[0030] The objective function of SVM is: ; The constraints are as follows: . n This represents the number of training samples, i.e., the total number of samples in the dataset used to train the support vector machine model. The sample label is (+1 or -1). Let be the normal vector of the hyperplane. For bias terms, The penalty coefficient is... As slack variables, This is a high-dimensional mapping function for the RBF kernel. In risk assessment scenarios, the SVM is extended to a regression model using the ε-SVR form to output continuous risk values.

[0031] Among them, the MLR model plays a core role in "linearly quantifying the risk contribution of each indicator". By constructing a linear equation, it quantifies the relationship between multiple independent variables and the dependent variable, and can intuitively output the contribution weight of each indicator to the risk value. It has the advantages of strong interpretability and high computational efficiency, and provides a traceable risk attribution basis for operation and maintenance decisions.

[0032] MLR model: ;

[0033] Regression plane equation: .

[0034] in, This represents the predicted risk value for meteorological disasters. As a disaster-causing factor indicator, The weight coefficients for each indicator can be determined using the least squares method. ε Let the random error vector of the regression model satisfy the expected value. =0, covariance matrix = , It is the variance of the error term. It is an identity matrix, representing the assumption that the error terms are homoscedastic.

[0035] Among them, CNN possesses powerful local feature extraction capabilities, automatically uncovering deep coupling features in meteorological data through convolutional and pooling layers. Unlike traditional models that rely on superficial data relationships, it uses a multi-level architecture to mine deep nonlinear relationships between meteorological indicators and power plant risks, achieving accurate risk quantification, especially in extreme weather scenarios. The CNN model expression is as follows: Convolutional layer feature extraction: ; in, The input channel number corresponds to indicators such as temperature and wind speed. The kernel size is the convolution kernel size. For convolution kernel weights, For the input feature map, For bias terms, This is the ReLU activation function.

[0036] MSE loss function: ; in, The number of samples in the test set. This represents the actual risk value. These are predicted values.

[0037] Adam learning rate adjustment: ; in, The initial learning rate, , For momentum parameters, This represents the number of iterations.

[0038] Using a standardized dataset encompassing disaster-causing factors, disaster-bearing bodies, and resilience as input, the system employs three models: The SVM model uses the RBF kernel function to map low-dimensional linearly inseparable data to a high-dimensional space, combining grid search to optimize the penalty coefficient and kernel function parameters, maintaining stable predictive performance under conventional meteorological scenarios; the MLR model solves for regression coefficients using the least squares method, linearly quantifying the contribution weight of each indicator to risk, achieving efficient basic risk value output; and the CNN model, through multi-layer convolution and pooling operations, mines the deep coupling characteristics of meteorological indicators in extreme disasters, outputting highly reliable risk values. These three models complement each other in terms of interpretability, stability, and accuracy, respectively adapting to different needs such as rapid initial assessment, routine monitoring, and extreme weather warnings, providing comprehensive and reliable basic predictive support for subsequent integrated optimization.

[0039] In practice, the aforementioned implementing entities can construct the first disaster risk assessment model, the second disaster risk assessment model, and the third disaster risk assessment model through the following steps: The first step is to train the first, second, and third disaster risk assessment models and set initial thresholds in the cloud. The second step involves fine-tuning the first, second, and third disaster risk assessment models at the edge nodes.

[0040] Step 104: Input the above-mentioned preprocessed meteorological disaster-related data of photovoltaic power station into the first disaster risk assessment model, the second disaster risk assessment model and the third disaster risk assessment model respectively to obtain the first disaster risk assessment result, the second disaster risk assessment result and the third disaster risk assessment result.

[0041] In some embodiments, the executing entity can input the preprocessed meteorological disaster-related data of the photovoltaic power station into the first disaster risk assessment model, the second disaster risk assessment model, and the third disaster risk assessment model, respectively, to obtain the first disaster risk assessment result, the second disaster risk assessment result, and the third disaster risk assessment result. The first disaster risk assessment result is the output of the first disaster risk assessment model (representing the disaster risk value). The second disaster risk assessment result is the output of the second disaster risk assessment model (representing the disaster risk value). The third disaster risk assessment result is the output of the third disaster risk model (representing the disaster risk value). The higher the disaster risk value, the more severe the disaster.

[0042] For example, based on preprocessed historical training data, after completing parameter training for three basic models—SVM, MLR, and CNN—model performance is evaluated from three individual evaluation indicators: “hazard of disaster-causing factors (L), vulnerability of disaster-bearing bodies (S), and degree of impact on recovery capabilities (P),” combined with the weighting results of the XGBoost algorithm.

[0043] The hazard level (L) of a disaster-causing factor is determined by the formula: ; in, The weights of each hazard factor determined by the XGBoost algorithm, To assess the model's ability to quantify disaster-causing factors such as extreme daily precipitation and peak instantaneous wind speed, the model outputs a single disaster-causing factor risk level.

[0044] The vulnerability (S) of a disaster-bearing body is determined by the formula: ; in, Due to the vulnerability of power generation capacity, To assess the vulnerability of equipment to damage, we validate the accuracy of the model in assessing the risk of damage to disaster-bearing structures.

[0045] The degree of impact on recovery ability (P) is determined by the formula: ; in, Data on the impact of manpower deployment The data is used to assess the impact of equipment repair technology and to verify the accuracy of the model in assessing the power plant's ability to resume operation.

[0046] After normalizing and weighting the three evaluation indicators, the combined risk of photovoltaic power plants under meteorological disasters is obtained by integrating and superimposing the individual evaluation indices: ; Step 105: Using the above formula, the first disaster risk assessment result can be obtained.R svm The above-mentioned second disaster risk assessment results R cnn Compared with the above-mentioned third disaster risk assessment results R mlr The results of the first, second, and third disaster risk assessments are integrated to generate a meteorological disaster risk value.

[0047] In some embodiments, the aforementioned implementing entity can integrate the results of the first disaster risk assessment, the second disaster risk assessment, and the third disaster risk assessment to generate a meteorological disaster risk value. To address the differences in the results of the basic models, a weighted voting method is used to balance the bias of individual models. Combined with a stacking ensemble strategy, the outputs of the three models are fused using a random forest as the core. Simultaneously, a diverse range of sub-models are generated through Bagging technology to improve prediction accuracy and generalization ability.

[0048] Random Forest Ensemble Framework Setup: Refer to Figure 3 Based on the performance differences of the base models, a Bagging ensemble strategy is adopted to construct a framework. First, the training set is sampled with replacement to generate multiple Bootstrap sample subsets. Second, each subset is input into SVM, MLR, and CNN models respectively to obtain the prediction results of the three base models. Then, a random forest regressor is constructed, with the base model predictions as input features and the actual risk value as the output target. A preliminary ensemble risk value is generated by voting among multiple decision trees. Finally, Shapley values ​​are introduced to quantify the model contribution and achieve dynamic weight adjustment—emphasizing the advantages of CNN in extreme disaster scenarios, emphasizing the stability of SVM in normal meteorological scenarios, while retaining the reference value of MLR, thereby improving the prediction accuracy of different scenarios.

[0049] In practice, the aforementioned implementing entities can integrate the results of the first, second, and third disaster risk assessments through the following steps to generate a meteorological disaster risk value: The first step involves assigning weights to the aforementioned first, second, and third disaster risk assessment results, respectively, to obtain the first weighted disaster risk assessment result, the second weighted disaster risk assessment result, and the third weighted disaster risk assessment result. For example, the XGBoost algorithm is first used to weight the output results of each model, and historical disaster data is used to obtain predicted risk values ​​through trained SVM, CNN, and MLR models, respectively. As a feature, input the corresponding quantified value of actual disaster loss. To achieve the objective function, an XGBoost regression model is trained. The optimal weights for the predictions of the three base models are automatically learned by optimizing the objective function, resulting in a weighted combination. as close as possible It also outputs the importance of each model in the ensemble, i.e., the optimal weight for the prediction results. In practical applications, by analyzing the decision path of the XGBoost model for the sample, or by using interpretive tools such as SHAP, the approximate contribution of each basic model on the sample can be deduced. For example, in a typhoon warning, the mechanism may automatically assign a higher weight, such as 0.50, to CNN which is good at extreme weather, a medium weight, such as 0.35, to SVM, and a lower weight, such as 0.15, to MLR.

[0050] The second step is to integrate the results of the first, second, and third weighted disaster risk assessments to generate meteorological disaster risk values.

[0051] Step 106: Based on the above meteorological disaster risk values, set dynamic thresholds, classify risk levels, and generate early warning signals.

[0052] In some embodiments, the aforementioned implementing entity may set dynamic thresholds, classify risk levels, and generate early warning signals based on the aforementioned meteorological disaster risk values.

[0053] Reference Figure 4 The initial threshold settings were determined using finite element analysis: 80% of the ultimate strain of the power plant material was set as the red warning threshold. When the actual strain of the photovoltaic power plant material structure reaches or exceeds this threshold, the warning system will immediately trigger a red alarm. 80% of the red threshold was set as the orange warning threshold. When the actual strain exceeds this value, the system will issue an orange alarm. 80% of the orange threshold was set as the yellow warning threshold. When the strain reaches this threshold, the system will push a yellow alarm. When the actual strain is lower than the yellow threshold, the system displays a green status, indicating that the power plant is operating normally and no special action is required.

[0054] Reference Figure 5 The risk threshold setting must be continuously optimized and adjusted based on the actual operating conditions of the photovoltaic power station.

[0055] In practice, the aforementioned implementing entities can set dynamic thresholds, classify risk levels, and generate early warning signals through the following steps: The first step is to use the preset strain level of the material's ultimate strain as the red threshold, and then set orange, yellow and green thresholds step by step.

[0056] The second step is to dynamically calibrate the dynamic threshold based on real-time monitoring data.

[0057] Step 107: Display the aforementioned warning signals through a visual warning platform and link them with the distributed photovoltaic power station monitoring system for real-time warning.

[0058] In some embodiments, the aforementioned implementing entity can display the aforementioned warning signals through a visual warning platform and link the distributed photovoltaic power station monitoring system for real-time warnings.

[0059] In practical applications, the dynamic adjustment of the threshold relies on a comparison mechanism between the theoretical model and actual data. The theoretical model is based on the initial threshold determined by finite element analysis, while the actual data includes historical disaster loss records, equipment failure repair cases, and real-time monitoring data. The direction of threshold deviation is located by analyzing the difference between the two. Simultaneously, multi-dimensional values ​​such as external environment, equipment status, and power plant operating parameters are comprehensively incorporated. By collecting and analyzing this dynamic data in real time, the initial threshold is specifically corrected, reducing the false alarm rate and false negative rate of the early warning system.

[0060] The distributed photovoltaic power station monitoring system uses Python / Java to build the backend service architecture, integrates Echarts and Cesium visualization components on the front end, and connects to the actual sensors of the photovoltaic power station to establish a real-time data stream transmission channel, ensuring low latency and high stability of data transmission.

[0061] Based on the underlying architecture, the existing machine learning model is embedded into the system, enabling it to automatically analyze sensor data and accurately assess the potential impact of meteorological disasters on photovoltaic power plants.

[0062] Based on the system's preset initial thresholds, the risk level is automatically predicted through model calculation. A risk warning function is developed to generate multi-level warning signals, including low-risk alerts, medium-risk notifications, and high-risk alarms, while simultaneously realizing the visualization of risks.

[0063] For different warning levels, a decision support module is configured to enable the system to automatically match and recommend operation and maintenance response measures, providing decision support for operation and maintenance personnel.

[0064] Develop customizable features and optimize user customization capabilities to support users in adjusting warning thresholds, setting alarm templates, and dividing management areas according to actual needs, thereby improving system adaptability.

[0065] Finally, API interfaces are set up, and data exchange interfaces are reserved to enable data interconnection and interoperability between the system and government meteorological centers, dispatch centers, and third-party intelligent monitoring platforms, thereby expanding the system's application scenarios.

[0066] Further reference Figure 7 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of a dynamic assessment and real-time early warning device for meteorological disaster risks in distributed photovoltaic power stations. These device embodiments are similar to... Figure 1 Corresponding to the method embodiments shown, the distributed photovoltaic power station meteorological disaster risk dynamic assessment and real-time early warning device can be specifically applied to various electronic devices.

[0067] like Figure 7 As shown, a distributed photovoltaic power station meteorological disaster risk dynamic assessment and real-time early warning device 700 according to some embodiments includes: a data acquisition unit 701, a preprocessing unit 702, a construction unit 703, an input unit 704, an integration unit 705, a partitioning unit 706, and an early warning unit 707. The data acquisition unit 701 is configured to acquire meteorological disaster-related data for the photovoltaic power station, including disaster-causing factor data, disaster-bearing body data, and recovery capacity data. The preprocessing unit 702 is configured to preprocess the meteorological disaster-related data to generate preprocessed meteorological disaster-related data for the photovoltaic power station. The construction unit 703 is configured to construct a first disaster risk assessment model, a second disaster risk assessment model, and a third disaster risk assessment model, wherein the first disaster risk assessment model is a support vector machine, the second disaster risk assessment model is a... The convolutional neural network and the third disaster risk assessment model are multiple linear regression models. Input unit 704 is configured to input the preprocessed meteorological disaster-related data of the photovoltaic power station into the first, second, and third disaster risk assessment models respectively to obtain the first, second, and third disaster risk assessment results. Integration unit 705 is configured to integrate the first, second, and third disaster risk assessment results to generate a meteorological disaster risk value. Classification unit 706 is configured to set a dynamic threshold based on the meteorological disaster risk value, classify the risk level, and generate an early warning signal. Early warning unit 707 is configured to display the early warning signal through a visual early warning platform and link the distributed photovoltaic power station monitoring system for real-time early warning.

[0068] It is understandable that the various units recorded in the distributed photovoltaic power station meteorological disaster risk dynamic assessment and real-time early warning device 700 are related to the reference. Figure 1 The steps in the described method correspond accordingly. Therefore, the operations, features, and beneficial effects described above for the method are also applicable to the distributed photovoltaic power station meteorological disaster risk dynamic assessment and real-time early warning device 700 and the units contained therein, and will not be repeated here.

[0069] The following is for reference. Figure 8 It shows a schematic diagram of the structure of an electronic device (e.g., a computing device) 800 suitable for implementing some embodiments of the present disclosure. Figure 8The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0070] like Figure 8 As shown, the electronic device 800 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 801, which can perform various appropriate actions and processes according to a program stored in a read-only memory 802 or a program loaded from a storage device 808 into a random access memory 803. The random access memory 803 also stores various programs and data required for the operation of the electronic device 800. The processing unit 801, the read-only memory 802, and the random access memory 803 are interconnected via a bus 804. An input / output interface 805 is also connected to the bus 804.

[0071] Typically, the following devices can be connected to the input / output interface 805: input devices 806 including, for example, a touchscreen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; output devices 807 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 808 including, for example, magnetic tape, hard disk, etc.; and communication devices 809. Communication device 809 allows electronic device 800 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 8 An electronic device 800 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 8 Each box shown can represent a device or multiple devices as needed.

[0072] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 809, or installed from a storage device 808, or installed from a read-only memory 802. When the computer program is executed by the processing device 801, it performs the functions defined above in the methods of some embodiments of this disclosure.

[0073] It should be noted that, in some embodiments of this disclosure, the computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0074] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0075] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to: collect meteorological disaster-related data of photovoltaic power plants, wherein the aforementioned meteorological disaster-related data of photovoltaic power plants includes: disaster-causing factor data, disaster-bearing body data, and recovery capacity data; preprocess the aforementioned meteorological disaster-related data of photovoltaic power plants to generate preprocessed meteorological disaster-related data of photovoltaic power plants; and construct a first disaster risk assessment model, a second disaster risk assessment model, and a third disaster risk assessment model, wherein the first disaster risk assessment model is a support vector machine, the second disaster risk assessment model is a convolutional neural network, and the third disaster risk assessment model is a multi-mode neural network. A meta-linear regression model is used. The preprocessed meteorological disaster-related data from the photovoltaic power station are input into the first, second, and third disaster risk assessment models, respectively, to obtain the first, second, and third disaster risk assessment results. These results are then integrated to generate a meteorological disaster risk value. Based on this risk value, a dynamic threshold is set, risk levels are classified, and early warning signals are generated. These early warning signals are displayed through a visual early warning platform and linked to the distributed photovoltaic power station monitoring system for real-time early warning.

[0076] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0077] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0078] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0079] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A method for dynamic assessment and real-time early warning of meteorological disaster risks for distributed photovoltaic power stations, characterized in that, include: Collect meteorological disaster-related data for photovoltaic power plants, including: disaster-causing factor data, disaster-bearing body data, and recovery capacity data; The meteorological disaster-related data of the photovoltaic power station are preprocessed to generate preprocessed meteorological disaster-related data of the photovoltaic power station; A first disaster risk assessment model, a second disaster risk assessment model, and a third disaster risk assessment model are constructed. The first disaster risk assessment model is a support vector machine, the second disaster risk assessment model is a convolutional neural network, and the third disaster risk assessment model is a multiple linear regression model. The preprocessed meteorological disaster-related data of the photovoltaic power station are input into the first disaster risk assessment model, the second disaster risk assessment model and the third disaster risk assessment model respectively to obtain the first disaster risk assessment result, the second disaster risk assessment result and the third disaster risk assessment result; The results of the first disaster risk assessment, the second disaster risk assessment, and the third disaster risk assessment are integrated to generate a meteorological disaster risk value; Based on the meteorological disaster risk value, a dynamic threshold is set, risk levels are classified, and early warning signals are generated. The warning signals are displayed through a visual warning platform and linked to the distributed photovoltaic power station monitoring system for real-time warnings.

2. The method for dynamic assessment and real-time early warning of meteorological disaster risks for distributed photovoltaic power stations according to claim 1, characterized in that, The disaster-causing factor data includes: extreme daily precipitation, historical number of rainstorms, temperature, air humidity, total suspended particulate matter, maximum wind speed, and average wind speed; the disaster-bearing body data includes: the power generation capacity of the photovoltaic power station and the number of damaged equipment; the recovery capacity data includes: personnel deployment data and equipment repair technology data of the photovoltaic power station.

3. The method for dynamic assessment and real-time early warning of meteorological disaster risks for distributed photovoltaic power stations according to claim 2, characterized in that, The process of integrating the first disaster risk assessment result, the second disaster risk assessment result, and the third disaster risk assessment result to generate a meteorological disaster risk value includes: The first disaster risk assessment result, the second disaster risk assessment result, and the third disaster risk assessment result are respectively weighted to obtain the first weighted disaster risk assessment result, the second weighted disaster risk assessment result, and the third weighted disaster risk assessment result; The first weighted disaster risk assessment result, the second weighted disaster risk assessment result, and the third weighted disaster risk assessment result are integrated to generate a meteorological disaster risk value.

4. The method for dynamic assessment and real-time early warning of meteorological disaster risks for distributed photovoltaic power stations according to claim 1, characterized in that, The step of setting dynamic thresholds, classifying risk levels, and generating early warning signals based on the meteorological disaster risk value includes: The red threshold is set based on the preset strain level of the material's ultimate strain, and orange, yellow and green thresholds are set step by step. The dynamic threshold is dynamically calibrated based on real-time monitoring data.

5. The method for dynamic assessment and real-time early warning of meteorological disaster risks for distributed photovoltaic power stations according to claim 1, characterized in that, The construction of the first disaster risk assessment model, the second disaster risk assessment model, and the third disaster risk assessment model includes: The first, second, and third disaster risk assessment models are trained and initial thresholds are set in the cloud. Fine-tuning of the first, second, and third disaster risk assessment models is performed at the edge nodes.

6. A dynamic assessment and real-time early warning device for meteorological disaster risks of distributed photovoltaic power stations, characterized in that, include: The data acquisition unit is configured to collect meteorological disaster-related data of photovoltaic power plants, wherein the meteorological disaster-related data of photovoltaic power plants includes: disaster-causing factor data, disaster-bearing body data, and recovery capacity data; The preprocessing unit is configured to preprocess the meteorological disaster-related data of the photovoltaic power station to generate preprocessed meteorological disaster-related data of the photovoltaic power station; The building unit is configured to build a first disaster risk assessment model, a second disaster risk assessment model, and a third disaster risk assessment model, wherein the first disaster risk assessment model is a support vector machine, the second disaster risk assessment model is a convolutional neural network, and the third disaster risk assessment model is a multiple linear regression model; The input unit is configured to input the preprocessed meteorological disaster-related data of the photovoltaic power station into the first disaster risk assessment model, the second disaster risk assessment model and the third disaster risk assessment model respectively, to obtain the first disaster risk assessment result, the second disaster risk assessment result and the third disaster risk assessment result; An integration unit is configured to integrate the first disaster risk assessment result, the second disaster risk assessment result, and the third disaster risk assessment result to generate a meteorological disaster risk value; The division unit is configured to set a dynamic threshold based on the meteorological disaster risk value, divide the risk level, and generate an early warning signal; The early warning unit is configured to display the early warning signal through a visual early warning platform and to link with the distributed photovoltaic power station monitoring system for real-time early warning.

7. An electronic device, characterized in that, include: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1 to 5.

8. A computer-readable medium, characterized in that, It stores a computer program thereon, wherein the computer program, when executed by a processor, implements the method as described in any one of claims 1 to 5.

Images