Power transmission line icing thickness prediction method, system and equipment based on digital-analog dual-drive model, and medium

Abstract

The invention belongs to the technical field of power transmission line detection, discloses a power transmission line icing thickness prediction method, system and device based on a digital-analog dual-drive model and a medium, and aims at solving the problem that an existing method is low in prediction precision. The method comprises the following steps: acquiring historical meteorological data of the location of a target line and corresponding power system data, performing missing value supplementation, and performing feature correlation analysis by adopting PCC to screen features so as to construct an input feature set; building a base learning layer, deploying an LSTM model in a meta learning layer, and taking the output of the base learning layer as the input of the meta learning layer to obtain a multi-model fused integrated learning model; inputting the input feature set into an integrated learning model to generate a training prediction result, and evaluating the training prediction result based on a physical rule to obtain a digital-analog dual-drive prediction model; and predicting the target line based on a digital-analog dual-drive prediction model, and evaluating a predicted value by using an error index and a fitting index so as to complete prediction of the icing thickness of the target line.

Description

Technical Field

[0001] The present invention belongs to the technical field of transmission line detection, and specifically relates to a method, system, equipment and medium for predicting ice thickness on a transmission line based on a digital-analog dual-drive model. Background Art

[0002] As a vital component of the power system, the safe and stable operation of transmission lines is crucial. In cold climates, climate factors pose a significant challenge to transmission line safety. Snowy weather can cause ice to form on transmission lines, increasing loads and potentially leading to serious safety incidents such as line breakage, equipment damage, and even power outages. Therefore, accurately predicting the thickness of ice on transmission lines is crucial for improving transmission line safety and ensuring stable power system operation.

[0003] Currently, most research on the prediction of ice thickness on transmission lines is still limited to single machine learning models based on meteorological data or historical experience, with estimates made through mathematical models. However, under complex and changing climatic conditions, the predictive accuracy of these methods is often significantly compromised. In particular, traditional single models suffer from large errors when dealing with variable and complex natural conditions, making it difficult to fully reflect the diversity and complexity of actual icing phenomena. Therefore, existing methods still have shortcomings in accurately predicting ice thickness on transmission lines. In addition, existing prediction methods often focus on the influence of a single factor and lack prediction models that comprehensively consider the interactions of multiple factors. Summary of the Invention

[0004] Based on the above-mentioned shortcomings and deficiencies in the prior art, one of the objects of the present invention is to solve at least one or more of the above-mentioned problems in the prior art. In other words, one of the objects of the present invention is to provide a method, system, equipment and medium for predicting ice thickness of transmission lines based on a digital-analog dual-drive model that meets one or more of the above-mentioned needs, so as to achieve the purpose of improving the accuracy and reliability of ice thickness prediction of transmission lines.

[0005] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical solutions: In the first aspect, the present invention provides a method for predicting ice thickness of transmission lines based on a digital-analog dual-drive model, comprising the following steps: S1, obtaining historical meteorological data and corresponding power system data of the target line location, supplementing missing values of the historical meteorological data and the corresponding power system data, performing feature correlation analysis using the Pearson linear correlation coefficient measurement index PCC, screening features based on the correlation analysis results, and constructing an optimized input feature set; S2, building a base learning layer based on five regression algorithms: random forest model, linear regression model, support vector regression model, decision tree, and K-nearest neighbor regression model, and deploying an LSTM model in the meta-learning layer. The output of the base learning layer is used as the input of the meta-learning layer to obtain an integrated learning model of multi-model fusion; S3, the input feature set constructed in step S1 is input into step S2 to obtain the integrated learning model, and a training prediction result is generated. The training prediction result is evaluated based on physical laws and the integrated learning model is optimized based on the evaluation result, thereby obtaining an optimized digital-analog dual-drive prediction model; S4, based on the optimized digital-analog dual-drive prediction model obtained in step S3, the ice thickness of the target line is predicted to obtain a predicted value, and the predicted value is evaluated using an error index and a fitting index, thereby completing the ice thickness prediction of the target line.

[0006] As a preferred solution, the missing value supplementation in step S1 adopts corresponding compensation strategies based on the data type, including using median filling for discrete numerical fields and linear interpolation for continuous variables; The screening feature in step S1 is to eliminate the features whose PCC value is lower than the preset threshold. The expression of the Pearson linear correlation coefficient measurement index PCC is: , where is the PCC value of each feature, and They have different characteristics, and Characteristics and The standard deviation of .

[0007] As a preferred solution, the evaluation of the training prediction results based on physical laws and the optimization of the integrated learning model based on the evaluation results are specifically as follows: calculating the determination coefficient R2 based on the training prediction results, and triggering model adjustment when the determination coefficient R2 is less than a preset value; The calculation formula of the determination coefficient R2 is: , where is the true value for training, is the training prediction value, is the average of the true training values, is the total number of samples, is the sample serial number.

[0008] As a preferred solution, when triggering model adjustment, it includes: calculating the error between the training prediction result and the training true value, obtaining the data error term through the mean square error, and feeding it back to the LSTM model in the meta-learning layer to adjust its weight.

[0009] As a preferred solution, when triggering the model adjustment, it also includes: modifying the physical parameters of the LSTM model in the meta-learning layer so that the training prediction results meet the residual constraints of the physical equation and are minimized during the training process; the residual constraint expression of the physical equation is , Where, is the physical error term, is the total number of samples, For the training prediction value changes over time, is the actual temperature, is the melting temperature of ice, is the coefficient of wind deflection angle, Indicates the degree of influence of wind deflection angle on ice thickening, is the actual angle of the wind deflection angle, The base diameter.

[0010] As a preferred solution, when triggering model adjustment, the minimum total loss is used as the objective function; The expression of the total loss is , where is the weight of the adjusted LSTM model, To adjust the physical parameters of the LSTM model, To adjust the weights of the previous LSTM model, To adjust the physical parameters of the front LSTM model.

[0011] As a preferred solution, the error indicators include root mean square error and mean absolute error; the fitting indicators are explained variance and determination coefficient R2; The expression of the root mean square error is: , the expression of the mean absolute error is , the expression of the explained variance is , where is the true value, is the predicted value, is the total number of samples, is the sample serial number.

[0012] In a second aspect, the present invention provides a transmission line ice thickness prediction system based on a digital-analog dual-drive model, which is used to implement the transmission line ice thickness prediction method as described in the first aspect.

[0013] In a third aspect, the present invention provides an electronic device, wherein the computer device includes a memory, a processor, and a computer program. When the computer program is executed by the processor, the method for predicting ice thickness of a transmission line as described in the first aspect is implemented.

[0014] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for predicting ice thickness of a transmission line as described in the first aspect.

[0015] Compared with the prior art, the present invention has the following beneficial effects: On the one hand, the present invention constructs a physical constraint model based on thermodynamics and fluid mechanics equations to ensure that the prediction results strictly follow the natural laws of ice formation and melting.

[0016] On the other hand, the present invention uses machine learning algorithms to dynamically mine nonlinear relationships in multi-source data such as historical meteorological data and line loads, effectively correcting parameter deviations of traditional physical models.

[0017] In its implementation, the two models are collaboratively optimized through a residual feedback mechanism. This collaborative optimization approach not only addresses the problem of single-model prediction failure in extreme weather conditions but also reduces reliance on labeled data. This method achieves high-precision predictions even in mountainous areas and at high altitudes, where monitoring data is sparse. Furthermore, through iterative real-time data updates, the model's dynamic responsiveness is further enhanced.

[0018] In summary, the present invention provides reliable support for power grid anti-icing decision-making with both physical interpretability and data adaptability, helping to significantly reduce operation and maintenance costs and accident risks. Further or more detailed beneficial effects will be described in conjunction with specific examples in the detailed implementation. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 It is a schematic diagram of the process of predicting ice thickness of transmission lines according to an embodiment of the present invention.

[0021] Figure 2 It is a structural diagram of the electronic device described in Example 3 of the present invention.

[0022] Figure 3 This is a correlation analysis diagram of the Pearson linear correlation coefficient measurement indicator PCC described in Example 5 of the present invention.

[0023] Figure 4 This is a comparison chart of the predicted value and the true value described in Example 5 of the present invention.

[0024] Figure Number: 200. Electronic equipment; 201. Processor; 202. Communication bus; 203. User interface; 204. Network interface; 205. Memory. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present invention will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present invention.

[0026] In the following description, multiple embodiments of the present invention are provided. Different embodiments may be replaced or combined, and therefore the present invention may be considered to include all possible combinations of the same and / or different embodiments described. Thus, if one embodiment includes features A, B, and C, and another embodiment includes features B and D, then the present invention should also be considered to include embodiments that include one or more of all other possible combinations of A, B, C, and D, even if such embodiments may not be explicitly described in the following text.

[0027] The following description provides examples and does not limit the scope, applicability, or examples set forth in the claims. Changes may be made to the functions and arrangements of the elements described without departing from the scope of the present invention. Various examples may appropriately omit, replace, or add various processes or components. For example, the described method may be performed in an order different from the order described, and various steps may be added, omitted, or combined. Furthermore, features described in some examples may be combined in other examples.

[0028] In order to facilitate a better understanding of the embodiments of the present invention, before explaining the specific implementation methods of the present invention in detail, its application scenarios are first described.

[0029] The transmission line ice thickness prediction method described in the embodiments of this specification is applied to power grid operation and maintenance in high-altitude mountainous areas, extreme weather emergency response, and unmanned monitoring area coverage. In these scenarios, the application of the transmission line ice thickness prediction method is intended to reduce the risk of line disconnection and tower collapse accidents caused by icing, optimize the scheduling of anti-icing and de-icing resources, reduce operation and maintenance costs, and improve the stability and reliability of the power grid under complex climatic conditions.

[0030] The following is a brief explanation of the Pearson linear correlation coefficient metric, random forest model, linear regression model, support vector regression model, decision tree, K-nearest neighbor regression model, LSTM model, and ensemble learning model involved in multiple embodiments of this specification: The Pearson Correlation Coefficient (PCC) is a statistical indicator used to measure the degree of linear correlation between two variables. Its value range is [-1, 1]. A PCC value of 1 indicates a perfect positive correlation between the two variables, a PCC value of -1 indicates a perfect negative correlation between the two variables, and a PCC value of 0 indicates no linear correlation between the two variables. In the proposed transmission line ice thickness prediction method, the PCC is used to measure the degree of linear correlation between various features in meteorological and power system data and ice thickness. This allows the selection of features with the greatest impact on ice thickness prediction to construct an optimized input feature set.

[0031] The Random Forest model (RF) is a machine learning algorithm based on ensemble learning, consisting of multiple decision trees. When constructing a random forest model, multiple training subsets are generated by sampling the training data with replacement (bootstrap sampling), and then a decision tree is constructed for each training subset. During the decision tree construction process, a subset of features is randomly selected from all features for node splitting to increase model diversity. Ultimately, the random forest model obtains the final prediction result by voting (for classification problems) or averaging (for regression problems) the prediction results of multiple decision trees. The random forest model has advantages such as strong resistance to overfitting, ability to handle high-dimensional data, and insensitivity to outliers. It is used in the base learning layer of the transmission line ice thickness prediction method described above to improve the model's prediction accuracy and stability.

[0032] A linear regression model (LR) is a statistical model that predicts by establishing a linear relationship between a dependent variable and one or more independent variables. The goal of a linear regression model is to estimate the regression coefficients by minimizing the sum of squared errors, thereby obtaining the optimal linear regression equation. The linear regression model is simple to understand and computationally efficient. In the proposed method for predicting ice thickness on transmission lines, it serves as a regression algorithm in the base learning layer, capturing the linear relationship between features and ice thickness.

[0033] The Support Vector Regression (SVR) model is the application of support vector machines to regression problems. Its basic idea is to find an optimal hyperplane that minimizes the total deviation of all training sample points from the hyperplane. Unlike traditional regression methods, SVR allows for a certain error range, and loss is only calculated when the deviation between the predicted value and the true value of a sample point exceeds this range. By introducing a kernel function, SVR can map nonlinear problems in low-dimensional space to high-dimensional space, thereby finding linear relationships in high-dimensional space and achieving regression prediction of nonlinear data. SVR has the advantages of strong generalization and the ability to process high-dimensional and nonlinear data. In the above-mentioned transmission line ice thickness prediction method, it is used as a regression algorithm in the base learning layer to improve the model's ability to handle complex data relationships.

[0034] A decision tree (DT) is a machine learning model that makes decisions based on a tree structure. It recursively partitions the data set into multiple subsets, each corresponding to a leaf node in the decision tree. During the decision tree construction process, the optimal features are selected for node splitting, resulting in the highest purity (for classification problems) or lowest variance (for regression problems) in the resulting subsets. Decision trees offer advantages such as intuitive understanding, the ability to process both discrete and continuous data, and a lack of strict requirements for data distribution. However, they are prone to overfitting when processing complex data. In the described transmission line ice thickness prediction method, a decision tree is used as a regression algorithm in the base learning layer, and is combined with other regression algorithms to construct an ensemble learning model to improve the model's predictive performance.

[0035] The K-Nearest Neighbors Regression model is an instance-based learning method. Its basic idea is to find the K nearest sample points in the training dataset for a sample point to be predicted. The target value of the sample point to be predicted is then predicted based on the target values of these K sample points. In regression problems, the average of the target values of these K sample points is typically used as the prediction result. The KNN algorithm does not require model training; it only requires storing the training dataset and performing distance calculations and neighbor searches during prediction. While the KNN algorithm is simple and intuitive, it is sensitive to the local structure of the data and has high computational complexity, especially when the data volume is large. In the described transmission line ice thickness prediction method, KNN is used as a regression algorithm in the base learning layer, enriching the feature extraction methods of the ensemble learning model and improving the model's prediction accuracy.

[0036] The LSTM model (Long Short-Term Memory) is a specialized recurrent neural network (RNN) designed to address the vanishing and exploding gradient problems that traditional RNNs often encounter when processing long sequences of data. LSTMs utilize gating mechanisms (input gate, forget gate, and output gate) to control the flow of information, enabling them to memorize long-term historical information. In the prediction of ice thickness on power transmission lines, the ice accumulation process is a time-varying process. The LSTM model can effectively capture long-term dependencies in time series data and accurately predict ice thickness trends. In the proposed method for predicting ice thickness on power transmission lines, the LSTM model is deployed in the meta-learning layer to further integrate and optimize the prediction results of multiple regression algorithms in the base learning layer, improving the model's predictive performance.

[0037] Ensemble learning is a machine learning method that improves model performance by combining the predictions of multiple base learners. In this method, a base learning layer is constructed based on five regression algorithms: random forest, linear regression, support vector regression, decision tree, and K-nearest neighbor regression. Each base learner learns and models the data from a different perspective, capturing distinct features and patterns within the data. An LSTM model is then deployed in the meta-learning layer, taking the multiple predictions from the base learning layer as input. The LSTM model then fuses and optimizes these predictions to produce the final prediction. By combining the strengths of multiple base learners, the ensemble learning model can reduce model variance, improve generalization, and enhance prediction accuracy, playing a significant role in the proposed method for predicting ice thickness on transmission lines.

[0038] Example 1: like Figure 1 As shown, this embodiment provides a method for predicting ice thickness of transmission lines based on a digital-analog dual-drive model, and the specific steps are as follows: Step S1 is data preprocessing and feature screening: historical meteorological data and corresponding power system data of the target line location are obtained, missing values of the historical meteorological data and corresponding power system data are supplemented, and feature correlation analysis is performed using the Pearson linear correlation coefficient measurement indicator PCC. Features are screened based on the correlation analysis results to construct an optimized input feature set.

[0039] Specifically, data preprocessing involves addressing missing information in the historical database and implementing corresponding compensation strategies based on data type differences. In this embodiment, discrete numeric fields are padded using the median, while continuous variables are smoothed using linear interpolation. In the categorical attribute dimension, high-frequency category substitution effectively preserves data distribution characteristics. For invalid samples with severe noise interference, significant abnormal fluctuations, or no correlation with the target variable (e.g., no icing records), data cleaning is performed to optimize the dataset quality by removing interfering items, thereby improving the accuracy of the prediction model.

[0040] Specifically, feature screening involves using the Pearson Correlation Coefficient (PCC) metric to quantitatively analyze feature associations in attribute correlation studies of historical datasets. By calculating the degree of linear dependency between variables and constructing a feature importance ranking matrix, a scientific variable screening mechanism is established for the prediction model. For feature dimensions with multicollinearity or insufficient statistical significance, dimensionality reduction is performed, and feature importance ranking is constructed using this metric to accurately screen input features for the prediction model. PCC satisfies:

[0041] Where: is the PCC value of each feature, and They have different characteristics, and Characteristics and The standard deviation of .

[0042] Step S2 is to build an integrated learning model: a base learning layer is built based on five regression algorithms, namely, random forest model, linear regression model, support vector regression model, decision tree and K nearest neighbor regression model, and an LSTM model is deployed in the meta-learning layer. The output of the base learning layer is used as the input of the meta-learning layer to obtain an integrated learning model of multi-model fusion.

[0043] Step S3 is physical law correction and model optimization: the input feature set constructed in step S1 is input into step S2 to obtain the integrated learning model, generate training prediction results, evaluate the training prediction results based on physical laws, and optimize the integrated learning model based on the evaluation results, thereby obtaining an optimized digital-analog dual-drive prediction model.

[0044] Specifically, the coefficient of determination (R2) is calculated first. Model adjustments are triggered only when R2 is less than a preset value. This triggers adjustments to the weights and physical parameters of the LSTM model in the meta-learning layer.

[0045] More specifically, the preset value in this embodiment is 0.7, that is, when R2 is less than 0.7, the model adjustment is triggered.

[0046] More specifically, the calculation of the determination coefficient R2 satisfies:

[0047] In the above formula: is the true value for training, is the training prediction value, is the average of the true training values, is the total number of samples, is the sample serial number.

[0048] More specifically, the weights of the LSTM model in the meta-learning layer are adjusted based on the data error term. The data error term is calculated based on the observed data to calculate the error between the training prediction value and the training true value. The data error term of the data-driven model is calculated by the mean square error (MSE), and its calculation satisfies

[0049] In the above formula: is the data error term, is the true value for training, is the training prediction value, is the total number of samples, is the sample serial number.

[0050] More specifically, the physical parameters of the LSTM model in the meta-learning layer are adjusted based on the physical error term, that is, through the residual constraint of the physical equation, to ensure that the prediction solution of the neural network strictly satisfies the known physical laws, measure the degree to which the prediction results deviate from the physical laws, and force the model to correct the parameters during training to minimize the residual. Its calculation satisfies

[0051] In the above formula: is the physical error term, is the total number of samples, For the training prediction value changes over time, is the actual temperature, is the melting temperature of ice, is the coefficient of wind deflection angle, Indicates the degree of influence of wind deflection angle on ice thickening, is the actual angle of the wind deflection angle, The base diameter.

[0052] More specifically, when the model adjustment is triggered, the weights and physical parameters of the LSTM model are updated through the optimizer to minimize the total loss, and the total loss satisfies

[0053] In the above formula: is the weight of the adjusted LSTM model, To adjust the physical parameters of the LSTM model, To adjust the weights of the previous LSTM model, To adjust the physical parameters of the front LSTM model.

[0054] Step S4 is ice thickness prediction and evaluation: based on the optimized digital-analog dual-drive prediction model obtained in step S3, the ice thickness of the target line is predicted to obtain a predicted value, and the predicted value is evaluated using the error index and the fitting index, thereby completing the ice thickness prediction of the target line.

[0055] Specifically, this embodiment uses error indicators and fit indicators to evaluate the performance of the prediction model. The error indicators are root mean square error (RMSE) and mean absolute error (MAE), and the fit indicators are explained variance (EV) and coefficient of determination R2. The specific expressions are as follows:

[0056]

[0057]

[0058] Where: is the true value, is the predicted value, is the total number of samples, is the sample serial number.

[0059] Example 2: This embodiment provides a transmission line ice thickness prediction system based on a digital-analog dual-drive model, which is used to implement the transmission line ice thickness prediction method as described in the first embodiment.

[0060] Example 3: like Figure 2 As shown, this embodiment provides an electronic device, which may include: at least one processor, at least one network interface, a user interface, a memory, and at least one communication bus.

[0061] The communication bus can be used to realize the connection and communication among the above components.

[0062] The user interface may include buttons, and the optional user interface may also include a standard wired interface or a wireless interface.

[0063] The network interface may include, but is not limited to, a Bluetooth module, an NFC module, a Wi-Fi module, and the like.

[0064] Among them, the processor may include one or more processing cores. The processor uses various interfaces and lines to connect the various parts of the entire electronic device, and performs various functions of the electronic device and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory, and calling data stored in the memory. Optionally, the processor can be implemented in at least one hardware form of DSP, FPGA, PLA. The processor can integrate one or a combination of CPU, GPU and modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to handle wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor, but may be implemented separately through a chip.

[0065] The memory may include either RAM or ROM. Optionally, the memory may include non-transitory computer-readable media. The memory may be used to store instructions, programs, codes, code sets, or instruction sets. The memory may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, sound playback function, image playback function, etc.), instructions for implementing each of the aforementioned method embodiments, etc.; the data storage area may store data related to each of the aforementioned method embodiments, etc. The memory may optionally be at least one storage device located remotely from the aforementioned processor. The memory, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and a prediction application. The processor may be configured to invoke the prediction application stored in the memory and execute the steps of the transmission line ice thickness prediction method described in the aforementioned embodiment.

[0066] Example 4: This embodiment provides a computer-readable storage medium having instructions stored therein, which, when executed on a computer or processor, causes the computer or processor to execute the above-mentioned Figure 1 If the components of the electronic device are implemented in the form of software functional units and sold or used as independent products, they can be stored in the computer-readable storage medium.

[0067] In the above embodiments, all or part of the embodiments can be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this specification are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted via the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer, or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital versatile disc (DVD)), or a semiconductor medium (eg, a solid state drive (SSD)).

[0068] Those skilled in the art will appreciate that all or part of the process steps in the method of the first embodiment described above can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When executed, the program can include the processes of the embodiments of the above methods. The aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks. The technical features of this embodiment and the implementation scheme can be combined in any manner unless they conflict.

[0069] Embodiment 5: In order to verify the effectiveness of the method for predicting ice thickness of transmission lines based on the digital-analog dual-drive model described in this specification, this embodiment is applied based on actual scenarios. The correlation analysis of the Pearson linear correlation coefficient measurement index PCC during the application process is as follows: Figure 3 As shown, the predicted value is compared with the true value. Figure 4 The evaluation results of the predicted values are shown in Table 1.

[0070] Table 1:

[0071] Based on the above, this embodiment verifies the effectiveness of the method for predicting ice thickness on transmission lines based on the digital-analog dual-drive model described in this specification.

[0072] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations. However, those skilled in the art should be aware that the present invention is not limited by the order of the actions described, because according to the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0073] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0074] The foregoing is merely an exemplary embodiment of the present invention and is not intended to limit the scope of the present invention. That is, any equivalent changes and modifications made in accordance with the teachings of the present invention are still within the scope of the present invention. A person skilled in the art will readily come up with the embodiments of the present invention after considering the specification and practicing the disclosure herein. The present invention is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary technical means in the art that are not described in the present invention. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present invention are defined by the claims.

Claims

1. A method for predicting ice thickness on transmission lines based on a digital-analog dual-drive model, characterized in that: Including steps: S1. Obtain historical meteorological data and corresponding power system data for the target line location, supplement missing values in the historical meteorological data and corresponding power system data, perform feature correlation analysis using the Pearson linear correlation coefficient (PCC) metric, and select features based on the correlation analysis results to construct an optimized input feature set. S2. Build a base learning layer based on five regression algorithms: random forest model, linear regression model, support vector regression model, decision tree, and K-nearest neighbor regression model. Deploy an LSTM model in the meta-learning layer, and use the output of the base learning layer as the input of the meta-learning layer to obtain an integrated learning model of multi-model fusion. S3. Inputting the input feature set constructed in step S1 into step S2 to obtain the integrated learning model, generating a training prediction result, evaluating the training prediction result based on physical laws, and optimizing the integrated learning model based on the evaluation result, thereby obtaining an optimized digital-analog dual-drive prediction model; S4. Based on the optimized digital-analog dual-drive prediction model obtained in step S3, the ice thickness of the target line is predicted to obtain a predicted value, and the predicted value is evaluated using an error index and a fitting index, thereby completing the ice thickness prediction of the target line.

2. The method for predicting ice thickness on power transmission lines based on a digital-analog dual-drive model according to claim 1 is characterized by: The missing value supplementation in step S1 adopts corresponding compensation strategies based on the data type, including median filling for discrete numerical fields and linear interpolation for continuous variables; The screening feature in step S1 is to eliminate the features whose PCC value is lower than the preset threshold. The expression of the Pearson linear correlation coefficient measurement index PCC is: , where is the PCC value of each feature, and They have different characteristics, and Characteristics and The standard deviation of .

3. The method for predicting ice thickness of transmission lines based on a digital-analog dual-drive model according to claim 1 is characterized in that: The evaluating the training prediction results based on physical laws and optimizing the integrated learning model based on the evaluation results are specifically: Calculating a determination coefficient R2 based on the training prediction result, and triggering model adjustment when the determination coefficient R2 is less than a preset value; The calculation formula of the determination coefficient R2 is: , where is the true value for training, is the training prediction value, is the average of the true training values, is the total number of samples, is the sample serial number.

4. The method for predicting ice thickness of transmission lines based on a digital-analog dual-drive model according to claim 3 is characterized in that: When model adjustments are triggered, they include: The error between the training prediction result and the training true value is calculated, and the data error term is obtained by the mean square error, which is fed back to the LSTM model in the meta-learning layer to adjust the weight.

5. The method for predicting ice thickness of transmission lines based on a digital-analog dual-drive model according to claim 4 is characterized in that: When model adjustments are triggered, also include: Modifying the physical parameters of the LSTM model in the meta-learning layer so that the training prediction results satisfy the residual constraints of the physical equations and are minimized during the training process; The expression of the residual constraint of the physical equation is ， Where, is the physical error term, is the total number of samples, For the training prediction value changes over time, is the actual temperature, is the melting temperature of ice, is the coefficient of wind deflection angle, Indicates the degree of influence of wind deflection angle on ice thickening, is the actual angle of the wind deflection angle, The base diameter.

6. The method for predicting ice thickness on power transmission lines based on a digital-analog dual-drive model according to claim 5 is characterized by: When triggering model adjustment, the minimum total loss is used as the objective function; The expression of the total loss is , where is the weight of the adjusted LSTM model, To adjust the physical parameters of the LSTM model, To adjust the weights of the previous LSTM model, To adjust the physical parameters of the front LSTM model.

7. The method for predicting ice thickness on power transmission lines based on a digital-analog dual-drive model according to claim 5 is characterized by: The error indicators include root mean square error and mean absolute error; The fitting indices mentioned are explained variance and coefficient of determination R2; The expression of the root mean square error is: , the expression of the mean absolute error is , the expression of the explained variance is , where is the true value, is the predicted value, is the total number of samples, is the sample serial number.

8. A transmission line ice thickness prediction system based on a digital-analog dual-drive model, characterized in that: Used to implement the method for predicting ice thickness on transmission lines as described in any one of claims 1 to 7.

9. A computer device comprising a memory, a processor, and a computer program, wherein: When the computer program is executed by a processor, the method for predicting ice thickness on a transmission line according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for predicting ice thickness on a transmission line according to any one of claims 1 to 7 is implemented.

Images