Line icing thickness prediction method and device and electronic equipment

Abstract

The invention discloses a line icing thickness prediction method and device and electronic equipment. The method comprises the following steps: receiving an icing thickness prediction request; in response to the icing thickness prediction request, determining a plurality of line parameters corresponding to the target line according to the line identifier; inputting the plurality of line parameters into a first network layer of the prediction network model, obtaining corresponding output parameters according to a target weight matrix corresponding to the first network layer, and inputting the corresponding output parameters and the plurality of line parameters into a next network layer until a plurality of output parameters are obtained; according to the plurality of output parameters, determining a target icing stage and an icing change rate corresponding to the target line in the preset time; and according to the initial icing thickness, the icing change rate and the target icing stage of the target line, determining the target icing thickness of the target line at the preset time. According to the invention, the technical problems of complex icing generation and disaggregation change process and low accuracy of the predicted line icing thickness in the prior art are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing, in particular to a line icing thickness prediction method and device and electronic equipment. BACKGROUND

[0002] Prediction of line icing thickness is crucial for preventing power failure, ensuring stable power supply and ensuring safety. At present, the method of establishing a line icing model is mainly used to predict the line icing thickness. However, due to the complex process of line icing growth and disappearance, the accuracy of the predicted icing thickness by the line icing model is low, which cannot meet the risk assessment requirements.

[0003] At present, no effective solution has been proposed for the above problems. SUMMARY

[0004] The embodiments of the present application provide a line icing thickness prediction method and device and electronic equipment to at least solve the technical problem of low accuracy of predicted line icing thickness due to the complex process of line icing growth and disappearance in the related art.

[0005] According to an aspect of an embodiment of the present application, a line icing thickness prediction method is provided, comprising: receiving an icing thickness prediction request, wherein the icing thickness prediction request carries a line identifier of a target line; in response to the icing thickness prediction request, determining a plurality of line parameters corresponding to the target line according to the line identifier; inputting the plurality of line parameters into a first network layer of a prediction network model, obtaining a corresponding output parameter according to a target weight matrix corresponding to the first network layer, and inputting the corresponding output parameter and the plurality of line parameters into a next network layer until a plurality of output parameters are obtained, wherein the prediction network model comprises a plurality of network layers connected in series, the plurality of network layers are respectively provided with a corresponding target weight matrix, the plurality of output parameters correspond one-to-one to the plurality of network layers, and the corresponding target weight matrix is used to represent the influence degree of a plurality of input parameters of the corresponding network layer on the line icing thickness; determining a target icing stage and an icing change rate of the target line at a predetermined time according to the plurality of output parameters; and determining a target icing thickness of the target line at the predetermined time according to an initial icing thickness of the target line, the icing change rate and the target icing stage.

[0006] Optionally, before the determining the target ice thickness of the target line at the predetermined time according to the initial ice thickness of the target line, the ice change rate and the target ice phase, the method further comprises: obtaining historical data of the target line, wherein the historical data comprises a plurality of historical line data and initial actual thicknesses corresponding to the plurality of historical line data, the initial actual thicknesses being used to represent the initial ice thickness of the target line corresponding to the historical line data; determining initial predicted thicknesses corresponding to the plurality of historical line data according to an average value of the plurality of initial actual thicknesses; determining a plurality of first differences according to the plurality of initial actual thicknesses and the initial predicted thicknesses corresponding to the plurality of initial actual thicknesses; and determining the initial ice thickness of the target line according to the plurality of first differences and a plurality of line parameters corresponding to the target line.

[0007] Optionally, the determining the initial ice thickness of the target line according to the plurality of first differences and the plurality of line parameters corresponding to the target line comprises: determining an initial prediction model, wherein the initial prediction model is used to determine the initial ice thickness of the target line; updating the initial prediction model according to the plurality of first differences until a target prediction model is obtained, wherein a second difference corresponding to the target prediction model is less than a difference threshold, the second difference being a difference between a corresponding target predicted thickness and a corresponding initial actual thickness, the corresponding target predicted thickness being the initial ice thickness of the target line determined by the target prediction model according to the corresponding historical line data; and determining the initial ice thickness of the target line according to the target prediction model and the plurality of line parameters corresponding to the target line.

[0008] Optionally, before the inputting the plurality of line parameters into a first network layer of a prediction network model, obtaining corresponding output parameters according to a weight matrix corresponding to the first network layer, and inputting the corresponding output parameters and the plurality of line parameters into a next network layer until a plurality of output parameters are obtained, the method further comprises: determining an initial network model, wherein a plurality of initial network layers of the initial network model are respectively provided with initial weight matrices; determining a target error function with a minimum total error as an objective according to the initial network model, wherein the target error function comprises a first error term and a second error term, the first error term being used to represent an error between a predicted ice phase determined by the initial network model and a corresponding actual ice phase, and the second error term being used to represent an error between a predicted change rate determined by the initial network model and a corresponding actual change rate; updating the plurality of initial weight matrices according to the target error function to obtain a plurality of target weight matrices, wherein a total error value corresponding to the plurality of target weight matrices is less than an error threshold; and obtaining the prediction network model according to the plurality of target weight matrices and the initial network model.

[0009] Optionally, the determining the target icing stage and the icing change rate of the target line at the predetermined time according to the plurality of output parameters comprises: determining a first weight value corresponding to each of the plurality of output parameters, wherein the first weight value corresponding to each of the plurality of output parameters is used to represent the influence degree of the corresponding output parameter on the target icing stage of the target line; determining a second weight value corresponding to each of the plurality of output parameters, wherein the second weight value corresponding to each of the plurality of output parameters is used to represent the influence degree of the corresponding output parameter on the icing change rate of the target line; and determining the target icing stage and the icing change rate of the target line at the predetermined time according to the plurality of first weight values, the plurality of second weight values and the plurality of output parameters.

[0010] Optionally, the inputting the plurality of line parameters into the first network layer of the prediction network model, obtaining the corresponding output parameter according to the target weight matrix corresponding to the first network layer, and inputting the corresponding output parameter and the plurality of line parameters into the next network layer until a plurality of output parameters are obtained comprises: determining an influence index corresponding to each of the plurality of input parameters of the corresponding network layer according to the target weight matrix corresponding to the first network layer, wherein the influence index corresponding to each of the plurality of input parameters is used to represent the influence degree of the corresponding input parameter on the line icing thickness; and determining the corresponding output parameter from the plurality of input parameters according to the plurality of influence indexes, wherein the influence index corresponding to the corresponding output parameter is greater than an influence index threshold value.

[0011] Optionally, after the determining the target icing thickness of the target line at the predetermined time according to the initial icing thickness of the target line, the icing change rate and the target icing stage, the method further comprises: obtaining an actual icing thickness of the target line at the predetermined time; determining a thickness difference value according to the actual icing thickness and the target icing thickness; and in a case where the thickness difference value is greater than a predetermined threshold value, updating the prediction network model according to the actual icing thickness to obtain an updated network model, so as to predict the icing thickness of the target line according to the updated network model.

[0012] According to an aspect of an embodiment of the present application, there is provided a line icing thickness prediction apparatus, comprising: a receiving module configured to receive an icing thickness prediction request, wherein the icing thickness prediction request carries a line identification of a target line; a responding module configured to, in response to the icing thickness prediction request, determine, according to the line identification, a plurality of line parameters corresponding to the target line; a first determining module configured to input the plurality of line parameters into a first network layer of a prediction network model, obtain a corresponding output parameter according to a target weight matrix corresponding to the first network layer, and input the corresponding output parameter and the plurality of line parameters into a next network layer until a plurality of output parameters are obtained, wherein the prediction network model comprises a plurality of network layers connected in series, the plurality of network layers are respectively provided with corresponding target weight matrices, the plurality of output parameters correspond to the plurality of network layers one by one, and the corresponding target weight matrix is used to represent the influence degree of a plurality of input parameters of a corresponding network layer on line icing thickness; a second determining module configured to determine, according to the plurality of output parameters, a target icing stage and an icing change rate corresponding to the target line at a predetermined time; and a third determining module configured to determine, according to an initial icing thickness of the target line, the icing change rate and the target icing stage, a target icing thickness of the target line at the predetermined time.

[0013] According to an aspect of an embodiment of the present application, there is provided an electronic device, comprising: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to execute the instructions to implement any of the above line icing thickness prediction methods.

[0014] According to an aspect of an embodiment of the present application, there is provided a computer-readable storage medium, when instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform any of the above line icing thickness prediction methods.

[0015] In the embodiment of the present application, the ice thickness prediction request is received, wherein the ice thickness prediction request carries the line identification of the target line; in response to the ice thickness prediction request, the plurality of line parameters corresponding to the target line are determined according to the line identification; the plurality of line parameters are input into the first network layer of the prediction network model, the corresponding output parameters are obtained according to the target weight matrix corresponding to the first network layer, and the corresponding output parameters and the plurality of line parameters are input into the next network layer until the plurality of output parameters are obtained, wherein the prediction network model comprises a plurality of network layers, the plurality of network layers are connected in series, the plurality of network layers are respectively provided with the corresponding target weight matrix, the plurality of output parameters correspond to the plurality of network layers one by one, and the corresponding target weight matrix is used to represent the influence degree of the plurality of input parameters of the corresponding network layer on the line ice thickness; the target ice stage and the ice change rate of the target line at the predetermined time are determined according to the plurality of output parameters; and the target ice thickness of the target line at the predetermined time is determined according to the initial ice thickness of the target line, the ice change rate and the target ice stage. By inputting the plurality of line parameters into the first network layer of the prediction network model, obtaining the corresponding output parameters according to the target weight matrix corresponding to the first network layer, and inputting the corresponding output parameters and the plurality of line parameters into the next network layer until the plurality of output parameters are obtained, the target ice stage and the ice change rate of the target line at the predetermined time are determined according to the plurality of output parameters, so as to realize the technical effect of determining the target ice thickness of the target line at the predetermined time according to the initial ice thickness of the target line, the ice change rate and the target ice stage. Since the single network layer of the prediction network model can consider the influence degree of the plurality of line parameters on the line ice thickness from a short-term perspective to determine the corresponding output parameters, and then input the corresponding output parameters and the plurality of line parameters into the next network layer to consider the influence degree of the plurality of line parameters on the line ice thickness from a long-term perspective, the target ice stage and the ice change rate determined according to the plurality of output parameters can improve the accuracy of the predicted ice thickness, thereby solving the technical problems of complex ice growth and disappearance change process and low accuracy of the predicted line ice thickness in the related art. BRIEF DESCRIPTION OF DRAWINGS

[0016] The accompanying drawings, which are included to provide a further understanding of the present application and are incorporated in and constitute a part of this application, illustrate embodiments of the present application and serve to explain the present application. In the drawings:

[0017] Figure 1 is a flowchart of the line ice thickness prediction method of the embodiment of the present application;

[0018] Figure 2is a flowchart of the ice growth and disappearance change dynamic prediction and risk assessment method provided by the optional embodiment of the present application.

[0019] Figure 3 is a structural block diagram of the line ice thickness prediction device of the embodiment of the present application. DETAILED DESCRIPTION

[0020] In order for those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the scope of protection of the present application.

[0021] It should be noted that the terms "first", "second", and the like in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0022] Embodiment 1

[0023] According to the embodiments of the present application, an embodiment of a line ice thickness prediction method is provided. It should be noted that the steps shown in the flowchart of the drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical sequence is shown in the flowchart, in some cases, the steps shown or described herein can be executed in an order different from that described herein.

[0024] Figure 1 is a flowchart of the line ice thickness prediction method according to the embodiments of the present application, as shown in Figure 1 The method comprises the following steps:

[0025] Step S102, receiving an ice thickness prediction request, wherein the ice thickness prediction request carries a line identification of a target line.

[0026] In step S102 provided in the present application, the ice thickness prediction request is received.

[0027] The icing thickness prediction request refers to an instruction for determining a target line icing thickness change prediction value.

[0028] The target line refers to a power transmission line that needs to be predicted for icing thickness.

[0029] The line identification refers to a code or label used to identify the target line, such as a line number.

[0030] In this step, the icing thickness prediction request for determining the target line icing thickness prediction value is received, and the line identification of the target line is carried in the icing thickness prediction request. Through the line identification, the target line that needs to be predicted can be directly located, which is the basic step for predicting the target line icing thickness.

[0031] Step S104, in response to the icing thickness prediction request, determines a plurality of line parameters corresponding to the target line according to the line identification.

[0032] In step S104 provided in the present application, a plurality of line parameters corresponding to the target line are determined.

[0033] The line parameters refer to various environmental, topographical, and technical parameters related to the target line, which are the basic data for icing thickness prediction. For example, line temperature, relative humidity, wind speed, wind direction, slope, terrain, span, line diameter, voltage level, geographic location coordinates, etc. These parameters can reflect the structural characteristics of the line itself and the characteristics of the environment in which it is located, and are extremely critical for predicting the icing process.

[0034] In this step, after receiving the icing thickness prediction request, the target line that needs to be predicted for icing thickness is first identified according to the provided line identification in the icing thickness prediction request. Then, the line parameters related to the target line are retrieved and extracted, which will be used as input for the prediction model to determine the future change of icing thickness. Through this step, the plurality of line parameters of the target line are accurately determined, which can take into account the unique properties of each line and the influence of the environment, thereby improving the accuracy and reliability of the icing thickness prediction.

[0035] In step S106, the plurality of line parameters are input to a first network layer of the prediction network model, and corresponding output parameters are obtained according to a target weight matrix corresponding to the first network layer, and the corresponding output parameters and the plurality of line parameters are input to a next network layer until a plurality of output parameters are obtained. The prediction network model includes a plurality of network layers connected in series, and the plurality of network layers are respectively provided with corresponding target weight matrices. The plurality of output parameters correspond to the plurality of network layers one by one, and the corresponding target weight matrix is used to represent the influence degree of the plurality of input parameters of the corresponding network layer on the line icing thickness.

[0036] In step S106 provided in the present application, a plurality of output parameters are obtained.

[0037] The prediction network model refers to a model capable of predicting the change of the icing thickness according to the input line parameters, for example, a long short-term memory network model (LSTM). The single time state node of the hidden layer can consider the influence degree of the plurality of line parameters on the line icing thickness from a short-term perspective, and the plurality of time state nodes can consider the influence degree of the plurality of line parameters on the line icing thickness from a long-term perspective according to the parameters determined by the previous time state node and the input parameters, so as to determine the change of the icing thickness.

[0038] The plurality of network layers refer to network layers for processing the input plurality of line parameters, which can determine the plurality of output parameters with high influence degree on the prediction of the icing thickness from the plurality of line parameters.

[0039] The target weight matrix refers to a matrix used to represent the importance of each layer in the network in the prediction process of a specific input feature. The input parameter with a large corresponding weight value in the input parameter will be determined as the output parameter and will be paid more attention to in predicting the icing thickness.

[0040] The output parameter refers to a parameter with high influence degree on the prediction of the icing thickness determined by the input data after the weight matrix processing in each network layer.

[0041] In this step, the plurality of line parameters are input to the prediction network model for processing. First, the first network layer receives the plurality of line parameters, and determines the corresponding output parameters according to the target weight matrix of the first network layer and the plurality of line parameters. The output parameters are then transmitted to the next network layer together with the original plurality of line parameters for further processing. This process is repeated in each network layer in the model until all network layers are traversed, and finally a plurality of output parameters are obtained. The plurality of output parameters contain multi-level and multi-angle analysis results of the icing thickness.

[0042] Through this step, through the multi-layer network design, the predicted network model can extract and convert the features in the input data from different aspects of short-term and long-term layer by layer, capture the most critical information from the multi-source heterogeneous data, reveal how different parameters interactively affect the change of the ice thickness, help the model to understand the formation mechanism of the ice in a deeper way, and greatly improve the accuracy of the ice thickness prediction.

[0043] In step S108, the target icing stage and the icing change rate of the target line at the predetermined time are determined according to the plurality of output parameters.

[0044] In step S108 provided in the present application, the target icing stage and the icing change rate of the target line at the predetermined time are determined.

[0045] The predetermined time is determined based on the needs of power system operation and maintenance, and the ice thickness of the target line needs to be predicted.

[0046] The target icing stage refers to the icing growth and disappearance stage of the target line at the predetermined time, for example, the non-icing stage, the icing occurrence stage, the icing growth stage, the icing maintenance stage and the icing ablation stage.

[0047] The icing change rate refers to the predicted change speed of the ice thickness of the target line at the predetermined time.

[0048] In this step, the target icing stage of the target line at a specific future time point (i.e. the predetermined time) and the icing thickness change rate at the predetermined time are determined according to the plurality of output parameters determined by the prediction network model. Through the prediction of the icing stage and the icing change rate, real-time and accurate data support can be provided for the operation and decision of the power system, and the efficiency of early warning of icing is effectively improved.

[0049] In step S110, the target ice thickness of the target line at the predetermined time is determined according to the initial ice thickness of the target line, the icing change rate and the target icing stage.

[0050] In step S110 provided in the present application, the target ice thickness of the target line at the predetermined time is determined.

[0051] The initial ice thickness refers to the initial thickness of the ice layer on the target line when the ice thickness prediction request is received and the ice thickness of the target line is predicted. This value is the basis for dynamic ice prediction, and it affects the accumulation and change of the subsequent ice layer thickness.

[0052] The target icing thickness refers to the icing thickness of the target line at the predetermined time determined according to the initial icing thickness, the icing change rate, and the target icing stage.

[0053] After the initial icing thickness of the target line, the icing change rate within the predetermined time, and the target icing stage are obtained in this step, the icing thickness of the target line within the predetermined time can be accurately predicted.

[0054] Through the steps S102-S110, the icing thickness prediction request can be received, where the icing thickness prediction request carries the line identifier of the target line; in response to the icing thickness prediction request, the plurality of line parameters corresponding to the target line are determined according to the line identifier; the plurality of line parameters are input into the first network layer of the prediction network model, the corresponding output parameters are obtained according to the target weight matrix corresponding to the first network layer, and the corresponding output parameters and the plurality of line parameters are input into the next network layer until the plurality of output parameters are obtained, where the prediction network model includes a plurality of network layers, the plurality of network layers are connected in series, the plurality of network layers are respectively provided with corresponding target weight matrices, the plurality of output parameters correspond to the plurality of network layers one by one, and the corresponding target weight matrix is used to represent the influence degree of the plurality of input parameters of the corresponding network layer on the line icing thickness; the target icing stage and the icing change rate of the target line at the predetermined time are determined according to the plurality of output parameters; and the target icing thickness of the target line at the predetermined time is determined according to the initial icing thickness of the target line, the icing change rate, and the target icing stage. By inputting the plurality of line parameters into the first network layer of the prediction network model, obtaining the corresponding output parameters according to the target weight matrix corresponding to the first network layer, and inputting the corresponding output parameters and the plurality of line parameters into the next network layer until the plurality of output parameters are obtained, the target icing stage and the icing change rate of the target line at the predetermined time are determined according to the plurality of output parameters, thereby achieving the technical effect of determining the target icing thickness of the target line at the predetermined time according to the initial icing thickness of the target line, the icing change rate, and the target icing stage. Since the single network layer of the prediction network model can consider the influence degree of the plurality of line parameters on the line icing thickness from a short-term perspective to determine the corresponding output parameters, and then input the corresponding output parameters and the plurality of line parameters into the next network layer to consider the influence degree of the plurality of line parameters on the line icing thickness from a long-term perspective, the target icing thickness of the target line determined according to the target icing stage and the icing change rate determined according to the plurality of output parameters can improve the accuracy of the predicted icing thickness, thereby solving the technical problems of complex icing growth and disappearance change process and low accuracy of the predicted line icing thickness in the related art.

[0055] As an optional embodiment, before determining the target ice thickness of the target line at the predetermined time according to the initial ice thickness of the target line, the ice change rate and the target ice stage, the method further comprises: obtaining historical data of the target line, wherein the historical data comprises a plurality of historical line data and initial actual thicknesses corresponding to the plurality of historical line data, the initial actual thicknesses being used to represent initial ice thicknesses of the target line corresponding to the historical line data; determining initial predicted thicknesses corresponding to the plurality of historical line data according to average thicknesses of the initial actual thicknesses; determining a plurality of first differences according to the initial actual thicknesses and the initial predicted thicknesses corresponding to the initial actual thicknesses; and determining the initial ice thickness of the target line according to the plurality of first differences and a plurality of line parameters corresponding to the target line.

[0056] In this embodiment, specific steps of determining the initial ice thickness of the target line are illustrated.

[0057] Here, the historical data refers to ice condition data of the target line collected in a plurality of historical time periods.

[0058] Here, the historical line data refers to line parameters corresponding to different ice stages and different ice thicknesses of the target line in a plurality of historical time periods.

[0059] Here, the initial actual thickness refers to a real initial ice thickness of the target line in a time period in the historical data.

[0060] Here, the average thickness refers to an average value of initial ice thicknesses of the target line in a plurality of historical time periods, and can be used as a basic reference for the initial predicted thickness.

[0061] Here, the initial predicted thickness refers to a preliminary estimated value of the ice thickness of the target line in an environment corresponding to the historical line data based on statistical analysis of the historical data.

[0062] Here, the first difference refers to a difference between the initial actual thickness of the ice of the target line in the historical data and the initial predicted thickness of the ice thickness of the target line predicted based on the historical line data.

[0063] In this step, first, the historical icing data of the target line is collected, including the line data under different historical conditions and the corresponding icing thickness, i.e. the initial actual thickness. Then, based on the initial icing thickness in the historical data, the average value of multiple initial actual thicknesses is determined as a preliminary estimate. This estimate constitutes the initial predicted thickness, which serves as the starting point for the subsequent iterative optimization process. By comparing the initial actual thickness in each historical record with the initial predicted thickness based on the average value, the difference between the two is calculated to obtain the first difference. The first difference reflects the difference between the model prediction and the actual situation, which can be used to adjust the model parameters to improve the prediction accuracy. Finally, using the first difference obtained from the historical data and the current multiple line parameters of the target line, the current initial icing thickness value of the target line is determined, laying a solid foundation for subsequent dynamic prediction of icing thickness and risk assessment.

[0064] Through this step, the correlation between icing thickness and environmental factors under historical conditions is determined through in-depth analysis of historical data, significantly improving the accuracy and reliability of the prediction model in handling future icing events, determining accurate initial icing thickness, reducing uncertainty in subsequent icing growth and decay prediction, and helping to reduce false positives and false negatives, avoiding unnecessary resource waste and safety accidents.

[0065] As an optional embodiment, determining the initial icing thickness of the target line according to the plurality of first differences and the plurality of line parameters corresponding to the target line comprises: determining an initial prediction model, wherein the initial prediction model is used to determine the line initial icing thickness of the target line; updating the initial prediction model according to the plurality of first differences until a target prediction model is obtained, wherein the second difference corresponding to the target prediction model is less than the difference threshold, the second difference corresponding to the target prediction model is the difference between the corresponding target prediction thickness and the corresponding initial actual thickness, the corresponding target prediction thickness is the line initial icing thickness determined by the target prediction model according to the corresponding historical line data; determining the initial icing thickness of the target line according to the target prediction model and the plurality of line parameters corresponding to the target line.

[0066] In this embodiment, the specific steps of determining the initial icing thickness of the target line according to the plurality of first differences and the plurality of line parameters corresponding to the target line are described.

[0067] Here, the initial prediction model is involved, which refers to the preliminary model for predicting icing thickness, determined based on historical data, used for preliminary estimation of icing thickness, and its purpose is to provide a starting point for subsequent training and optimization of the model.

[0068] Here, the target prediction model is involved, which refers to the final model determined by iterative updating of the initial prediction model through training and optimization process, which can more accurately predict the initial thickness of icing.

[0069] wherein the second difference refers to the error value between the ice thickness prediction value output by the target prediction model and the actual ice thickness.

[0070] wherein the difference threshold refers to a set error standard for judging whether the model reaches the expected prediction accuracy. When the second difference is less than the difference threshold, it indicates that the prediction error of the model is within an acceptable range, and the model training is considered successful, with the ability to accurately predict the ice thickness.

[0071] In this step, first, an initial prediction model is determined for preliminary estimation of the ice thickness. Then, by analyzing multiple first differences (i.e., the difference between the model prediction value and the actual ice thickness), the initial prediction model is updated and optimized until a target prediction model is found whose prediction error (second difference) is lower than the set difference threshold, i.e., the model's ability to predict ice thickness has reached a high degree of accuracy. Finally, based on the optimized target prediction model and the line parameters of the target line, the initial ice thickness of the target line under specific meteorological conditions can be accurately predicted. This process embodies the iterative optimization strategy of model training, ensuring the accuracy and reliability of the prediction results.

[0072] Through this step, iterative optimization is performed to continuously reduce the error value of the predicted initial ice thickness, obtain the target prediction model, and improve the accuracy of the prediction. At the same time, using multiple historical data for model training helps the model learn more extensive data patterns and improves the model's generalization ability, enabling accurate prediction in various complex environments in practical applications.

[0073] As an optional embodiment, the plurality of line parameters are input into a first network layer of the prediction network model, and the corresponding output parameters are obtained based on the weight matrix corresponding to the first network layer. The corresponding output parameters and the plurality of line parameters are input into the next network layer until the plurality of output parameters are obtained. Further comprising: determining an initial network model, wherein the plurality of initial network layers of the initial network model are respectively provided with initial weight matrices; determining a target error function with the objective of minimizing the total error based on the initial network model, wherein the target error function includes a first error term and a second error term, the first error term is used to represent the error between the predicted ice accumulation stage determined by the initial network model and the corresponding actual ice accumulation stage, and the second error term is used to represent the error between the predicted change rate determined by the initial network model and the corresponding actual change rate; updating the plurality of initial weight matrices based on the target error function to obtain a plurality of target weight matrices, wherein the total error value corresponding to the plurality of target weight matrices is less than the error threshold; obtaining the prediction network model based on the plurality of target weight matrices and the initial network model.

[0074] In this embodiment, the specific steps of determining the prediction network model are illustrated.

[0075] In this step, the initial network model is determined, and the initial network model refers to the initial model at the beginning of training for determining the relationship between the line parameter and the ice thickness variation process, which is adjusted through the training data and the optimization algorithm later.

[0076] In this step, the target error function is involved, which refers to a function for measuring the difference between the prediction results of the network model and the true results, guiding the adjustment of the network model parameters.

[0077] In this step, the initial weight matrix is involved, which refers to the initial value of the weight in each layer of the neural network at the beginning of model training, which is usually initialized with random numbers.

[0078] In this step, the error threshold is involved, which refers to a preset value for determining whether the model training has reached an acceptable accuracy. When the value of the target error function is lower than the error threshold, the model is considered to be trained or to achieve the expected performance.

[0079] In this step, first, the initial network model is determined, and the multiple network layers of the initial network model are provided with the initial weight matrix. The prediction ability of the initial network model is relatively weak. The first error term and the second error term are determined, which measure the errors of the initial network model in the ice stage prediction and the ice thickness variation rate prediction, respectively. By combining these two error terms by weighting, a comprehensive target error function is formed. According to the total error value of the target error function, the initial weight matrix is adjusted to gradually improve the prediction accuracy of the model. When the total error value of the model falls below the set error threshold, the training process ends. At this time, the weight matrix is the final target weight matrix, and based on these target weight matrices and the entire network structure, the final prediction network model is obtained.

[0080] Through this step, the multi-task learning framework is used to optimize the ice stage classification and the ice thickness variation rate prediction simultaneously. This training strategy helps the model to learn more comprehensive and complex data patterns, improves the overall prediction accuracy and generalization ability, and at the same time, reduces the necessity of repeated learning between different tasks, saves the computing resources, and improves the efficiency of model training.

[0081] As an optional embodiment, the target icing stage and the icing change rate of the target line at the predetermined time are determined according to the plurality of output parameters, including: determining a first weight value corresponding to each of the plurality of output parameters, wherein the corresponding first weight value is used to represent the influence degree of the corresponding output parameter on the icing stage of the target line; determining a second weight value corresponding to each of the plurality of output parameters, wherein the corresponding second weight value is used to represent the influence degree of the corresponding output parameter on the icing change rate of the target line; and determining the target icing stage and the icing change rate of the target line at the predetermined time according to the plurality of first weight values, the plurality of second weight values and the plurality of output parameters.

[0082] In this embodiment, the specific steps of determining the target icing stage and the icing change rate of the target line at the predetermined time according to the plurality of output parameters are described.

[0083] The first weight value refers to the importance or influence of each output parameter in determining the icing stage. The first weight value reflects the proportion of a certain output parameter in determining the icing stage of the target line.

[0084] The second weight value refers to the relative importance of each output parameter in predicting the icing thickness change rate.

[0085] In this step, first, the plurality of output parameters related to the icing thickness change of the target line are determined by the prediction network model, and then the parameters are weighted according to the first weight value and the second weight value. The weighted result reflects the contribution of each parameter in predicting the icing stage and the change rate, and finally the prediction result of the icing condition of the target line (i.e. the icing stage classification probability and the icing change rate) is determined according to the weighted result. Through this step, the first weight value and the second weight value are dynamically adjusted to realize the fine prediction of the icing stage and the icing change rate, and improve the reliability of the prediction.

[0086] As an optional embodiment, the plurality of line parameters are input into the first network layer of the prediction network model, the corresponding output parameters are obtained according to the target weight matrix corresponding to the first network layer, and the corresponding output parameters and the plurality of line parameters are input into the next network layer until the plurality of output parameters are obtained, including: determining an influence index corresponding to each of the plurality of input parameters of the corresponding network layer according to the target weight matrix, wherein the corresponding influence index is the influence degree of the corresponding input parameter on the icing thickness of the line; and determining the corresponding output parameter from the plurality of input parameters according to the plurality of influence indexes, wherein the corresponding influence index of the corresponding output parameter is greater than an influence index threshold.

[0087] In this embodiment, the specific steps of determining the corresponding output parameter are described.

[0088] wherein the influence index is involved, the influence index refers to an indicator measuring the degree of influence of each line parameter on the ice thickness prediction result. The influence index is calculated based on the target weight matrix, reflecting the relative importance of different parameters in the ice formation process.

[0089] wherein the influence index threshold is involved, the influence index threshold refers to a pre-set numerical standard for screening line parameters that have a significant impact on ice thickness prediction. Only when the influence index of the input parameter exceeds the threshold, the parameter will be further considered and processed, which helps the model focus on key information and improve prediction efficiency.

[0090] In this step, first, all line parameters are input into the first network layer of the prediction network model, and the influence degree of each parameter is preliminarily evaluated through the target weight matrix of the layer to calculate the corresponding influence index. The influence index reflects the importance of the line parameter in the ice thickness prediction, i.e. the contribution of the parameter to the ice process. Next, the prediction network model will screen out those parameters whose influence index exceeds the influence index threshold based on the size of the influence index, i.e. the parameters that have a significant impact on the ice thickness prediction. These screened parameters will be input into the subsequent network layer together with the output parameters of the first network layer. In this process, the prediction network model focuses on the information that is really important to the prediction result through layer-by-layer processing and screening, gradually constructs the prediction result of the ice thickness, and improves the accuracy of the prediction.

[0091] As an optional embodiment, after determining the target ice thickness of the target line at the predetermined time according to the initial ice thickness of the target line, the ice change rate and the target ice stage, the method further comprises: obtaining the actual ice thickness of the target line at the predetermined time; determining a thickness difference according to the actual ice thickness and the target ice thickness; in a case where the thickness difference is greater than a predetermined threshold, updating the prediction network model according to the actual ice thickness to obtain an updated network model, so as to predict the ice thickness of the target line according to the updated network model.

[0092] In this embodiment, the specific steps of determining the updated network model are described.

[0093] wherein the actual ice thickness is involved, the actual ice thickness refers to the actual observed ice thickness on the target line at the predetermined time.

[0094] wherein the updated network model is involved, the updated network model refers to the optimization and update of the prediction network model by adjusting the parameters (such as weight matrix, bias term, etc.) of the model when the prediction accuracy of the model does not reach the predetermined threshold, and the updated network model can provide more accurate results in subsequent prediction.

[0095] In this step, after the target ice thickness of the target line at the predetermined time is determined, the actual ice thickness of the line at the same time point is obtained. Then, the difference between the target ice thickness and the actual ice thickness, i.e., the thickness difference, is calculated. If the thickness difference exceeds a predetermined threshold, it indicates that there is a significant difference between the prediction result of the model and the actual ice condition, and at this time, the prediction network model needs to be updated according to the actual ice thickness, so as to obtain an updated network model, ensuring that the prediction ability of the model can be closer to the actual situation and improving the prediction accuracy.

[0096] Based on the above embodiments and optional embodiments, an optional implementation is provided, which is specifically described as follows.

[0097] In the related art, the monitoring and prediction method for ice coating can monitor and warn the ice coating condition of the power transmission line to a certain extent, but they generally have problems of single monitoring means, insufficient prediction accuracy, poor real-time performance and the like. Especially in the dynamic change monitoring and risk assessment of the ice coating process, the existing technology cannot meet the demand of safe operation of the power system.

[0098] In view of this, the optional implementation of the present application provides an overhead power transmission line ice coating growth and disappearance change dynamic prediction and risk assessment method for joint monitoring, which combines multi-source heterogeneous data including meteorological elements, geographical elements, line data and the like, and constructs a dynamic ice coating process prediction system.

[0099] The optional implementation of the present application solves the problem of insufficient utilization of multiple data sources by the existing model by introducing a multi-task learning framework based on a long short-term memory network combined with an attention mechanism (MTL-LSTM-Attention) multi-task learning framework, which can predict the ice thickness change rate and the ice coating process type every 3 hours within 72 hours in the future in real time, and dynamically assess the ice coating risk of the power transmission line at different time points through a risk assessment model, guiding the power system operation and maintenance personnel to take timely measures.

[0100] Figure 2 is a flowchart of the ice coating growth and disappearance change dynamic prediction and risk assessment method provided by the optional implementation of the present application, as shown in Figure 2As shown, the optional embodiment of the present application can obtain and preprocess meteorological elements, geographical elements, line data and other multi-source heterogeneous data, construct an icing feature sample library, combine icing monitoring data and time series microclimate prediction data, apply gradient boosting tree method (GBDT) to predict the initial thickness of the icing occurrence stage. Introduce the MTL-LSTM-Attention multi-task learning framework to jointly predict the icing thickness change rate and icing process type within 72 hours with a time interval of 3 hours. A joint loss function is constructed, and the model parameters are optimized by gradient descent method to minimize the joint loss function and improve the model performance. Through the multi-task learning model, the model can capture the dynamic change of the icing thickness and the icing process type at the same time, and finally realize the dynamic prediction of the icing growth and disappearance change. Based on the dynamic change prediction result of the icing thickness, the icing thickness at each future time point is calculated and the icing risk level is evaluated. The present application can provide real-time and accurate risk assessment for the operation safety of the power transmission line and improve the early warning and response ability to icing disasters.

[0101] The detailed steps of the optional embodiment of the present application will be introduced below.

[0102] S1, receiving an icing thickness prediction request.

[0103] S2, in response to the icing thickness prediction request, determining a plurality of line parameters corresponding to the target line according to the line identification.

[0104] Obtain and preprocess multi-source heterogeneous data, including meteorological elements, geographical elements, ground object types, terrain types, line data, icing process types and icing change rates, and the preprocessing process includes normalization, dimensionality reduction and feature extraction and other existing technologies to form a sample data set (Same as the above line parameters):

[0105]

[0106] Wherein, represents the number of data, represents the i-th data; represents temperature, unit: ℃; represents relative humidity, unit: %; represents wind speed, unit: m / s; represents wind direction, unit: °; represents slope, unit: °; represents aspect, unit: °; represents elevation, unit: m; represents surface roughness, dimensionless; represents the ground object type, , the set From small to large, grassland, water, building land, transportation land, woodland, farmland; Indicates the terrain type, , from small to large, as plain, hilly, mountainous and high mountainous land; Indicates the span, unit: m; Indicates the line diameter, unit: cm; Indicates the voltage level, unit: kV; Indicates the ice growth rate, unit: mm / h; Indicates the ice growth stage type, dimensionless, , the elements in set A from large to small respectively represent the non-icing stage, the occurrence stage, the growth stage, the maintenance stage and the ablation stage.

[0107] Using the transmission line icing monitoring or ice observation equipment, the first observed ice thickness in a single icing process is marked as the initial ice thickness, denoted as , which represents the external conditions reaching the critical value of icing occurrence, and finally forms the sample data set

[0108]

[0109] , wherein is the initial ice thickness, if is not in the icing occurrence stage, then .

[0110] Before predicting the line ice thickness, the initial ice thickness of the target line needs to be determined, and the specific steps are as follows:

[0111] A1, initialize the model: set the initial prediction value ( ) as the average value of all sample labels (i.e. initial actual thickness, ), denoted as:

[0112]

[0113] , wherein represents the number of samples in the sample data set.

[0114] A2, iterative training:

[0115] The residual error of the current model is calculated as:

[0116]

[0117] , wherein represents the residual error of the th sample in the th iteration; ​For the first The sample at the th The predicted value of each iteration; For the first The initial thickness of ice accretion on each sample.

[0118] Determine a new decision tree based on residual training To fit the residuals, for:

[0119]

[0120] in, Indicates the first The new decision tree obtained through training in rounds of iteration; For decision trees to sample The predicted value.

[0121] Update the model predictions to get:

[0122]

[0123] in, For the first The sample at the th Predicted values ​​after rounds of iteration; This represents the learning rate, which controls the size of the step size for each update.

[0124] A3. Repeat the iterative training in step A2, after a predetermined... After iteration, the initial icing thickness prediction model is expressed as:

[0125]

[0126] in, This represents the final predicted result for the initial thickness of the ice layer. Indicates the number of iterations; This refers to the decision tree trained in each iteration.

[0127] S3. Input multiple line parameters into the first network layer of the prediction network model, obtain the corresponding output parameters according to the target weight matrix of the first network layer, and input the corresponding output parameters and multiple line parameters into the next network layer until multiple output parameters are obtained.

[0128] B1. Constructing an ice-covered sample dataset This includes meteorological elements, geographical elements, route data, icing process types, and icing change rates. M1 is:

[0129]

[0130] in, For input features, including meteorological elements (such as temperature , humidity , wind speed , wind direction , etc.), geographical elements (such as elevation , slope , aspect , surface roughness , land cover type , terrain type , etc.), line data (such as span , conductor diameter , voltage level , etc.); For output labels, including ice thickness change rate (regression task) and ice process type (classification task).

[0131] B2, build a long short-term memory network model (LSTM), set the input layer, hidden layer and output layer structure.

[0132] The specific steps of building a long short-term memory network model include:

[0133] a. Set the input layer, generate time series input features as for:

[0134]

[0135] where, denotes the length of the historical time window, denotes the step size.

[0136] b. Set the hidden layer of the LSTM network, at each time step , the update formula of the LSTM unit includes:

[0137] The update formula of the forget gate is:

[0138]

[0139] The update formula of the input gate is:

[0140]

[0141]

[0142] The update formula of the cell state is:

[0143]

[0144] The update formula of the output gate is:

[0145]

[0146]

[0147] where, , , and are weight matrices, , , and are bias vectors, is a nonlinear activation function (sigmoid activation function), is a hyperbolic tangent function (tanh activation function), is a forget gate, is an input gate, is a candidate memory cell state, is a memory cell state, is the output of the hidden layer cell node at time t.

[0148] B3, introduce attention mechanism to weight process the output of LSTM.

[0149] The attention weight is calculated as:

[0150]

[0151]

[0152] where, is the attention score of each time step t, is the step length, here , , are the parameters of the attention mechanism layer;

[0153] The weighted hidden state c of the LSTM is obtained by weighted sum of the hidden state as:

[0154]

[0155] B4, according to the setting of MTL framework, the context vector weighted by attention mechanism is a weighted representation of the input sequence. The context vector is passed into the fully connected network , and each layer of neurons performs linear transformation and nonlinear activation. The formula of the hidden layer output is:

[0156]

[0157] where, and These are the weight matrix and bias vector of a fully connected network, respectively. This is the output of layer l of a fully connected network.

[0158] S4. Based on multiple output parameters, determine the target icing stage and icing change rate of the target line at a predetermined time.

[0159] Predicted output of rate of change of ice thickness ( The rate of icing change mentioned above is:

[0160]

[0161] in, It is the hidden output of the last layer of a fully connected network. It is the weight matrix for the regression task. That is the corresponding bias.

[0162] Icing process type prediction output ( The icing stage (same as the target mentioned above) is as follows:

[0163]

[0164] in, It is the weight matrix for the classification task. It is the bias of the classification task, which is ultimately determined by the classification function ( The function outputs the classification probability.

[0165] Construct a joint loss function and optimize it by minimizing the joint loss function using gradient descent to improve the model's prediction performance. This is achieved through the following steps:

[0166] C1. Define the loss function for the icing thickness change rate model. :

[0167]

[0168] in, The rate of change of ice thickness predicted by the model; This represents the actual rate of change in ice thickness.

[0169] C2. Define the loss function for the icing process type classification model. for:

[0170]

[0171] in, For the sample Belongs to the icing process type The actual label; Samples predicted by the model belong The probability of the type of icing process; This represents the number of categories for the icing process type.

[0172] C3. Construct the joint loss function By weighted merging the two loss functions mentioned above, the performance of the model on regression and classification tasks can be optimized. (Joint loss function) for:

[0173]

[0174] in, and These are the weight coefficients for the regression and classification tasks, respectively, used to balance the contributions of the two tasks in the joint loss function.

[0175] C4. Optimize the joint loss function using gradient descent. This is done to minimize the joint loss function and improve the model's predictive performance.

[0176] S5. Based on the initial icing thickness of the target line, the rate of icing change, and the target icing stage, determine the target icing thickness of the target line at a predetermined time.

[0177] Based on the initial ice thickness prediction and predicting the rate of change in ice thickness By summing up the changes in ice thickness at various times, the ice thickness values ​​for each time point in the next 72 hours can be simulated. for:

[0178]

[0179]

[0180] in, This is a time interval in hours; the value here is 3. For time steps; For the future; For time step The rate of change of ice thickness; This is the correction value for the initial thickness of the icing layer; To predict the ice thickness at any given time.

[0181] The assessment of the icing process end time is achieved through the following steps:

[0182] Based on simulations, the ice thickness values ​​at various times over the next 72 hours are predicted. ,when When the time is right, the icing process is considered to be over.

[0183] Through the above optional implementation, at least the following beneficial effects can be achieved:

[0184] (1) A dynamic icing process prediction system is constructed by combining multi-source heterogeneous data, including meteorological elements, geographical elements, line data, etc.

[0185] (2) The MTL-LSTM-Attention multi-task learning framework is introduced to solve the problem of insufficient utilization of multiple data sources by existing models.

[0186] (3) The icing thickness change rate and icing process type every 3 hours within the next 72 hours can be predicted in real time, and the icing risk of the power transmission line at different time points can be dynamically evaluated through the risk evaluation model, guiding the power system operation and maintenance personnel to take timely measures.

[0187] It should be noted that, for the foregoing method embodiments, in order to simply describe, they are all expressed as a series of action combinations, but those skilled in the art should know that the present application is not limited by the action sequence described, because according to the present application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions and modules involved are not necessarily essential to the present application.

[0188] From the above description of the embodiments, those skilled in the art can clearly understand that the method according to the above embodiments can be realized by means of software and the necessary general hardware platform, of course, it can also be realized by hardware, but in many cases the former is a better implementation. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes a plurality of instructions for causing a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the method of each embodiment of the present application.

[0189] Embodiment 2

[0190] According to the embodiments of the present application, a device for implementing the above line icing thickness prediction method is also provided, Figure 3 The structure block diagram of the line icing thickness prediction device according to the embodiments of the present application is shown in Figure 3 The device includes a receiving module 302, a responding module 304, a first determining module 306, a second determining module 308, and a third determining module 310, which are described in detail below.

[0191] The receiving module 302 is configured to receive an icing thickness prediction request, wherein the icing thickness prediction request carries a line identification of a target line; the response module 304 is connected to the receiving module 302 and configured to, in response to the icing thickness prediction request, determine a plurality of line parameters corresponding to the target line according to the line identification; the first determination module 306 is connected to the response module 304 and configured to input the plurality of line parameters into a first network layer of a prediction network model, obtain corresponding output parameters according to a target weight matrix corresponding to the first network layer, and input the corresponding output parameters and the plurality of line parameters into a next network layer until a plurality of output parameters are obtained, wherein the prediction network model comprises a plurality of network layers connected in series, the plurality of network layers are respectively provided with corresponding target weight matrices, the plurality of output parameters correspond to the plurality of network layers one by one, and the corresponding target weight matrix is used to represent the influence degree of a plurality of input parameters of the corresponding network layer on the line icing thickness; the second determination module 308 is connected to the first determination module 306 and configured to determine a target icing stage and an icing change rate corresponding to the target line at a predetermined time according to the plurality of output parameters; and the third determination module 310 is connected to the second determination module 308 and configured to determine a target icing thickness of the target line at the predetermined time according to an initial icing thickness of the target line, the icing change rate and the target icing stage.

[0192] It should be noted that the receiving module 302, the response module 304, the first determination module 306, the second determination module 308 and the third determination module 310 correspond to steps S102 to S110 in the method for predicting the line icing thickness, and the plurality of modules have the same instances and application scenarios as the corresponding steps, but are not limited to the content disclosed in the above embodiment 1.

[0193] Embodiment 3

[0194] According to another aspect of the embodiments of the present application, an electronic device is also provided, which comprises a processor and a memory for storing processor-executable instructions, wherein the processor is configured to execute the instructions to implement the line icing thickness prediction method of any one of the above.

[0195] Embodiment 4

[0196] According to another aspect of the embodiments of the present application, a computer-readable storage medium is also provided, which, when the instructions in the computer-readable storage medium are executed by a processor of an electronic device, enables the electronic device to perform the line icing thickness prediction method of any one of the above.

[0197] The above-mentioned embodiment numbers of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments.

[0198] In the above-mentioned embodiments of the present application, the description of each embodiment is focused on, and the part not described in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0199] In several embodiments provided in the present application, it should be understood that the disclosed technical content can be implemented by other ways. Among them, the above-mentioned device embodiments are only schematic, for example, the division of the units can be a logical function division, and actual implementation can have another division way, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units or modules shown or discussed can be indirect coupling or communication connection through some interfaces, units or modules, which can be electrical or other forms.

[0200] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place or distributed to multiple units. Part or all of the units can be selected to achieve the purpose of the embodiment scheme according to actual needs.

[0201] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The above integrated unit can be realized in the form of hardware or in the form of software functional unit.

[0202] The integrated unit, if realized in the form of software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part of the prior art or the whole or part of the technical solutions can be embodied in the form of software product, which is stored in a storage medium and includes a plurality of instructions to make a computer device (which can be a personal computer, a server or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present application. The foregoing storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and various program code storage media.

[0203] The above-mentioned is only the preferred embodiment of the present application, and it should be pointed out that for ordinary skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, and these improvements and refinements should be regarded as the protection scope of the present application.

Claims

1. A method for predicting the thickness of icing on power lines, characterized in that, include: Receive an icing thickness prediction request, wherein the icing thickness prediction request carries the line identifier of the target line; In response to the icing thickness prediction request, multiple line parameters corresponding to the target line are determined based on the line identifier; The multiple line parameters are input into the first network layer of the prediction network model. Based on the target weight matrix corresponding to the first network layer, the corresponding output parameters are obtained. The corresponding output parameters and the multiple line parameters are then input into the next network layer until multiple output parameters are obtained. The prediction network model includes multiple network layers connected in series. Each of the multiple network layers is set with a corresponding target weight matrix. The multiple output parameters correspond one-to-one with the multiple network layers. The corresponding target weight matrix is ​​used to represent the degree of influence of the multiple input parameters of the corresponding network layer on the icing thickness of the line. Based on the multiple output parameters, the target icing stage and icing change rate of the target line at a predetermined time are determined; Based on the initial icing thickness of the target line, the icing change rate, and the target icing stage, the target icing thickness of the target line at the predetermined time is determined.

2. The method according to claim 1, characterized in that, Before determining the target icing thickness of the target line at a predetermined time based on the initial icing thickness of the target line, the icing change rate, and the target icing stage, the method further includes: The historical data of the target line is obtained, wherein the historical data includes multiple historical line data and the initial actual thickness corresponding to the multiple historical line data respectively, and the corresponding initial actual thickness is used to represent the initial icing thickness of the target line under the corresponding historical line data. Based on the average thickness corresponding to multiple initial actual thicknesses, the initial predicted thickness corresponding to the multiple historical line data is determined respectively; Based on the plurality of initial actual thicknesses and the initial predicted thicknesses corresponding to the plurality of initial actual thicknesses, a plurality of first differences are determined; The initial icing thickness of the target line is determined based on the multiple first differences and the multiple line parameters corresponding to the target line.

3. The method according to claim 2, characterized in that, The step of determining the initial icing thickness of the target line based on the plurality of first differences and the plurality of line parameters corresponding to the target line includes: An initial prediction model is determined, wherein the initial prediction model is used to determine the initial icing thickness of the target line; Based on the multiple first differences, the initial prediction model is updated until the target prediction model is obtained. The second difference corresponding to the target prediction model is less than the difference threshold. The corresponding second difference is the difference between the corresponding target predicted thickness and the corresponding initial actual thickness. The corresponding target predicted thickness is the initial icing thickness of the line determined by the target prediction model based on the corresponding historical line data. Based on the target prediction model and multiple line parameters corresponding to the target line, the initial icing thickness of the target line is determined.

4. The method according to claim 1, characterized in that, The process of inputting the multiple line parameters into the first network layer of the prediction network model, obtaining the corresponding output parameters based on the weight matrix corresponding to the first network layer, and inputting the corresponding output parameters and the multiple line parameters into the next network layer until multiple output parameters are obtained further includes: An initial network model is determined, wherein each of the multiple initial network layers corresponding to the initial network model is provided with an initial weight matrix; Based on the initial network model, a target error function is determined with the goal of minimizing the total error. The target error function includes a first error term and a second error term. The first error term represents the error between the predicted icing stage determined by the initial network model and the corresponding actual icing stage. The second error term represents the error between the predicted rate of change determined by the initial network model and the corresponding actual rate of change. Based on the target error function, multiple initial weight matrices are updated to obtain multiple target weight matrices, wherein the total error value corresponding to the multiple target weight matrices is less than the error threshold; The prediction network model is obtained based on the multiple target weight matrices and the initial network model.

5. The method according to claim 1, characterized in that, Determining the target icing stage and icing change rate of the target line at a predetermined time based on the multiple output parameters includes: Determine the first weight value corresponding to each of the plurality of output parameters, wherein the corresponding first weight value is used to represent the degree of influence of the corresponding output parameter on the icing stage of the target line; Determine the second weight values ​​corresponding to the plurality of output parameters respectively, wherein the corresponding second weight values ​​are used to represent the degree of influence of the corresponding output parameters on the icing change rate of the target line; Based on multiple first weight values, multiple second weight values, and the multiple output parameters, the target icing stage and icing change rate of the target line at the predetermined time are determined.

6. The method according to claim 1, characterized in that, The process involves inputting the multiple line parameters into the first network layer of the prediction network model, obtaining corresponding output parameters based on the target weight matrix corresponding to the first network layer, and then inputting the corresponding output parameters and the multiple line parameters into the next network layer until multiple output parameters are obtained, including: Based on the corresponding target weight matrix, the influence index corresponding to multiple input parameters of the corresponding network layer is determined, wherein the corresponding influence index is the degree of influence of the corresponding input parameter on the icing thickness of the line; Based on multiple influence indices, the corresponding output parameter is determined from the multiple input parameters, wherein the influence index corresponding to the output parameter is greater than the influence index threshold.

7. The method according to any one of claims 1 to 6, characterized in that, After determining the target icing thickness of the target line at a predetermined time based on the initial icing thickness of the target line, the icing change rate, and the target icing stage, the method further includes: Obtain the actual icing thickness of the target line at the predetermined time; The thickness difference is determined based on the actual icing thickness and the target icing thickness; If the thickness difference is greater than a predetermined threshold, the prediction network model is updated based on the actual icing thickness to obtain an updated network model, which is then used to predict the icing thickness of the target line.

8. A device for predicting the thickness of icing on power lines, characterized in that, include: A receiving module is used to receive an icing thickness prediction request, wherein the icing thickness prediction request carries the line identifier of the target line; A response module is used to respond to the icing thickness prediction request and determine multiple line parameters corresponding to the target line based on the line identifier. The first determining module is used to input the plurality of line parameters into the first network layer of the prediction network model, obtain the corresponding output parameters according to the target weight matrix corresponding to the first network layer, and input the corresponding output parameters and the plurality of line parameters into the next network layer until multiple output parameters are obtained. The prediction network model includes multiple network layers connected in series. Each of the multiple network layers is set with a corresponding target weight matrix. The multiple output parameters correspond one-to-one with the multiple network layers. The corresponding target weight matrix is ​​used to represent the degree of influence of the multiple input parameters of the corresponding network layer on the icing thickness of the line. The second determining module is used to determine the target icing stage and icing change rate of the target line at a predetermined time based on the multiple output parameters. The third determining module is used to determine the target icing thickness of the target line at a predetermined time based on the initial icing thickness of the target line, the icing change rate, and the target icing stage.

9. An electronic device, characterized in that, include: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the line icing thickness prediction method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, When the instructions in the computer-readable storage medium are executed by the processor of the electronic device, the electronic device is able to perform the line icing thickness prediction method as described in any one of claims 1 to 7.

Images