A method, system, device and medium for predicting ice thickness based on grey correlation

Abstract

The present invention relates to the field of meteorological forecasting and artificial intelligence technology, and in particular to a grey correlation-based ice thickness prediction method, system, device and medium, comprising collecting multi-dimensional ice cover data, dynamically screening key meteorological features through grey correlation; constructing and dividing a cleaned and normalized data set; integrating a bidirectional gated recurrent unit with a multi-head self-attention mechanism to design a hybrid model; performing parameter optimization and testing verification based on a training set; eliminating redundant interference through dynamic feature screening, bidirectionally analyzing the long-term and short-term ice cover evolution laws, strengthening the focus on extreme meteorological event features with an attention mechanism, and realizing cross-climate zone migration deployment in combination with a parameter adaptive compensation module, thereby significantly improving prediction accuracy and early warning capabilities for sudden ice and snow conditions, and providing intelligent decision-making support for power grid disaster prevention.

Description

Technical Field

[0001] The present invention relates to the field of meteorological forecasting and artificial intelligence technology, and in particular to a method, system, device and medium for predicting ice thickness based on grey correlation. Background Art

[0002] Transmission line icing, a significant natural disaster threatening power grid security, exhibits distinct spatial and temporal distribution patterns under specific climatic conditions. This phenomenon is particularly prone to occur in low-temperature, high-humidity environments. The ice accumulation process involves complex interactions between thermodynamics and fluid dynamics, and its destructive effects span mechanical, electrical, and aerodynamic domains. International case studies have shown that severe icing incidents not only damage power grid facilities but can also trigger regional power supply crises, highlighting the need for accurate prediction technologies.

[0003] The development of this research field has undergone three stages of evolution: although the early physical models based on thermodynamic equilibrium equations had clear mechanisms, they were extremely sensitive to micrometeorological dynamic parameters, and their computational efficiency could not meet the needs of real-time monitoring; traditional statistical methods were limited by the nonlinear coupling relationship of meteorological parameters, and their prediction accuracy and generalization capabilities were insufficient; although machine learning technology has opened up new paths for nonlinear modeling, it faces bottlenecks such as complex feature engineering and insufficient utilization of time series information.

[0004] The introduction of deep learning technology marks a methodological revolution. Time series models, such as LSTM (Long Short-Term Memory) networks, effectively capture the temporal dependencies of icing processes through gating mechanisms. Hybrid architecture models further optimize forecasting performance by fusing spatial and temporal features. However, the current technology landscape still faces three limitations: inefficient fusion of heterogeneous multi-source data, delayed response to extreme conditions, and insufficient cross-regional model adaptability. These challenges stem from the inherent scientific challenges of icing forecasting: the nonlinear nature of multi-parameter interactions, the cumulative effects of historical meteorological conditions, the second-order effects of sudden meteorological changes, and geographic climate variability.

[0005] In response to the shortcomings of existing deep models, this study proposes an innovative architecture that integrates grey correlation analysis and a bidirectional gating mechanism. By dynamically screening key meteorological factors through grey correlation, a streamlined and efficient input feature set is constructed; a bidirectional gating unit is used to capture the forward and backward correlation characteristics of time series data to avoid the long-range dependency loss problem of traditional recurrent neural networks; and an attention mechanism is introduced to enhance the model's real-time perception of the sudden change characteristics of ice cover rate. This method achieves a breakthrough in the coordinated characterization of dynamic coupling modeling of meteorological parameters and time evolution laws. Compared with mainstream deep learning solutions, the new model has demonstrated more robust prediction performance in multi-climate zone tests, especially in key working conditions such as rain-snow transitions and sudden temperature changes, providing key technical support for the construction of an active defense-type power grid disaster prevention system. Summary of the Invention

[0006] In view of the above problems in the prior art, the present invention is proposed.

[0007] Therefore, the technical problem addressed by this invention is to filter key features from meteorological data using grey correlation analysis and construct a prediction model by combining differential processing, bidirectional gated recurrent units (BiGRUs), and a multi-head self-attention mechanism. First, grey correlation is used to quantify the correlation between features and the target variable, optimizing the input feature dimensions. Second, first-order differencing is used to eliminate data non-stationarity, improving model training efficiency. Finally, a bidirectional GRU (gated recurrent unit) is used to capture the bidirectional dependencies of time series data, and a multi-head self-attention mechanism dynamically assigns weights to different time steps. Finally, a fully connected layer is used to output the prediction results.

[0008] To solve the above technical problems, the present invention provides the following technical solutions: a method for predicting ice thickness based on grey correlation, comprising: using meteorological stations and sensors to collect ice-related data in real time in different regions and time periods; using grey correlation analysis to filter out important features with a high correlation with ice thickness from features using a threshold; preprocessing the collected data to construct an ice thickness prediction data set and divide it into a training set and a test set; combining a bidirectional gated recurrent unit and a multi-head self-attention mechanism to construct a hybrid prediction model; using the training set to train the combined model, and using the test set data to perform predictions, completing the model test and obtaining the prediction results.

[0009] As a preferred solution of the grey correlation-based ice thickness prediction method described in the present invention, the collecting of ice-related data includes using a weather station and a sensor network to conduct real-time online detection of ice-covered areas of transmission lines, including temperature, humidity, wind speed, ice grade, precipitation, and ice coverage data;

[0010] Among them, ice coverage data includes records of ice thickness on transmission lines over a period of time in the past.

[0011] As a preferred solution of the gray correlation-based ice thickness prediction method described in the present invention, the gray correlation analysis method includes calculating the comprehensive correlation between the input meteorological characteristics and the ice thickness, establishing a comprehensive correlation model according to the relationship between the ice thickness and the meteorological parameters, and obtaining the gray comprehensive correlation.

[0012] As a preferred solution of the gray correlation-based ice thickness prediction method described in the present invention, the screening of important features with a high correlation with ice thickness includes using a gray correlation analysis method to quantify the correlation between features and ice thickness from multi-source meteorological parameters, arranging the obtained gray comprehensive correlation degrees from large to small, and screening features with gray comprehensive correlation degrees higher than a set threshold.

[0013] As a preferred solution of the grey correlation-based ice thickness prediction method described in the present invention, the method comprises: constructing an ice thickness prediction data set and dividing it into a training set and a test set, including taking historical observed ice accumulation data and screened meteorological data with high comprehensive correlation as input time series, taking the predicted value of future ice thickness as the target time series, and constructing an ice thickness time series data set; performing preprocessing, including extracting sample point data, data cleaning, and data normalization; and dividing the data set into a training set and a test set in proportion.

[0014] As a preferred solution of the grey correlation-based ice thickness prediction method described in the present invention, the construction of a hybrid prediction model includes combining a bidirectional gated recurrent unit and a multi-head self-attention mechanism, using a gated recurrent unit to capture the bidirectional dependency of time series data, and assigning weights to time steps through a multi-head self-attention mechanism; after combining the bidirectional gated recurrent unit and the multi-head self-attention mechanism, a feature representation is obtained, which is passed through a fully connected layer to obtain a hybrid prediction model result.

[0015] As a preferred solution of the grey correlation based ice thickness prediction method described in the present invention, the combined model is trained by training the hybrid prediction model with training set data, the data is input into the model for prediction through a sliding window mechanism, and the parameters are adjusted through the optimizer, the learning rate and the number of iterations are set to complete the training and testing.

[0016] Another object of the present invention is to provide an ice thickness prediction system based on grey correlation, which obtains ice-related meteorological and conductor status data in real time through a data acquisition module, and uses a correlation screening module to quantify the grey comprehensive correlation between features and ice thickness and screen key parameters; with the help of a data adjustment module, multi-source heterogeneous data is cleaned, normalized and time-series processed to construct a prediction data set; a hybrid modeling module integrates a bidirectional gated recurrent unit and a multi-head self-attention mechanism to capture the bidirectional dependency and dynamic weight distribution capabilities of time series data respectively; finally, an intelligent optimization module drives model training and testing, and combines differential processing to eliminate non-stationary interference, thereby achieving accurate short-term prediction of ice thickness and providing real-time and reliable decision support for the prevention and control of power grid icing disasters.

[0017] In order to solve the above technical problems, the present invention provides the following technical solutions: a grey correlation-based ice thickness prediction system, comprising: a data acquisition module, a correlation screening module, a data adjustment module, a hybrid modeling module, and an intelligent optimization module;

[0018] The data acquisition module collects ice cover related data in real time using weather stations and sensors in different regions and time periods;

[0019] The correlation screening module uses a grey correlation analysis method and a threshold value to screen out important features with a high correlation with ice thickness from the features;

[0020] The data adjustment module pre-processes the collected data, constructs an ice thickness prediction data set and divides it into a training set and a test set;

[0021] The hybrid modeling module combines the bidirectional gated recurrent unit and the multi-head self-attention mechanism to construct a hybrid prediction model;

[0022] The intelligent optimization module uses the training set to train the combined model, and uses the test set data to make predictions, complete the model test, and obtain the prediction results.

[0023] A computer device includes a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, it implements the steps of the above-mentioned method for predicting ice thickness based on grey correlation.

[0024] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the above-mentioned method for predicting ice thickness based on grey correlation are implemented.

[0025] Beneficial effects of the present invention: The present invention significantly improves the prediction efficiency and adaptability of ice thickness of transmission lines by innovatively integrating grey correlation dynamic screening mechanism and bidirectional gated time series modeling technology; the grey correlation algorithm constructs a multi-dimensional meteorological parameter dynamic evaluation system, intelligently identifies the key driving factors of ice growth, breaks through the static limitations of traditional feature selection, and effectively reduces the interference of redundant data on model performance; the bidirectional gated network synchronously analyzes the forward and backward evolution logic of the ice process through a two-layer recursive architecture, accurately captures the long-term and short-term dependence of ice thickness with meteorological evolution, and overcomes the memory decay defect of conventional recurrent networks in complex time series modeling; the introduction of the self-attention mechanism forms a multi-dimensional feature focus Channel, strengthen the model's ability to capture abnormal waveforms of sudden changes in working conditions such as rain-snow conversion and temperature fluctuations, and significantly improve the warning sensitivity of extreme meteorological events; by constructing a parameter adaptive compensation module, the model can autonomously learn the differences in ice formation mechanisms in different typical climate zones, eliminate the rigid constraints of regional characteristics on the prediction model, and achieve reliable migration across complex terrain environments; through the triple iterative improvement of feature selection, time series modeling and generalization optimization, ice prediction is promoted from traditional single physical indicator analysis to the intelligent decision-making stage of multi-source integration, providing the power grid with all-weather, multi-scenario dynamic assessment support for ice disaster risks, and effectively enhancing the power system's forward-looking defense capabilities against natural disasters. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. 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.

[0027] Figure 1 A flowchart of a method for predicting ice thickness based on grey correlation is provided in accordance with an embodiment of the present invention.

[0028] Figure 2 A grey correlation flow chart of a grey correlation ice thickness prediction method provided in one embodiment of the present invention.

[0029] Figure 3 A grey correlation diagram of ice thickness based on a grey correlation ice thickness prediction method is provided in accordance with an embodiment of the present invention.

[0030] Figure 4 A model building flowchart of a grey correlation ice thickness prediction method provided in one embodiment of the present invention.

[0031] Figure 5 A graph of prediction results of an LSTM model based on a grey correlation ice thickness prediction method provided in one embodiment of the present invention.

[0032] Figure 6 A graph of prediction results of a CNN model based on a grey correlation ice thickness prediction method provided in one embodiment of the present invention.

[0033] Figure 7 A graph showing the prediction results of a BiGRU-MSA model based on a grey correlation ice thickness prediction method provided in one embodiment of the present invention. DETAILED DESCRIPTION

[0034] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0035] Example 1, with reference to Figures 1-4 , which is the first embodiment of the present invention, provides a method for predicting ice thickness based on grey correlation, comprising:

[0036] S1: Use weather stations and sensors to collect ice cover data in real time at different regions and time periods.

[0037] It should be noted that if Figure 1 As shown in step S1, collecting ice-related data includes using meteorological stations and sensor networks to conduct real-time online detection of ice-covered areas on transmission lines, including temperature, humidity, wind speed, ice level, precipitation, and ice coverage data; wherein the ice coverage data includes records of ice thickness on transmission lines over a period of time.

[0038] Furthermore, the collection equipment uses a high-precision weather station and sensor network to record meteorological data related to ice formation and evolution in real time, including temperature sensors, humidity sensors, wind speed and direction sensors, precipitation sensors, and air pressure sensors. Meteorological data is recorded at hourly intervals to ensure that real-time changes in meteorological conditions can be captured.

[0039] Specifically, the temperature sensor is used to record the ambient temperature with an accuracy of ±0.1°C; the humidity sensor is used to measure the relative humidity of the air with an accuracy of ±2%; the wind speed and direction sensor records the wind speed and direction with an accuracy of ±0.2m / s and a wind direction accuracy of ±3°; the precipitation sensor is used to measure precipitation with an accuracy of ±0.1mm; and the pressure sensor records the atmospheric pressure with an accuracy of ±0.3hPa.

[0040] S2: Using the grey correlation analysis method, the threshold value is used to filter out the important features with high correlation with ice thickness from the features.

[0041] It should be noted that if Figure 1 As shown in step S2, screening ice thickness related features includes using the grey correlation analysis method to quantify the correlation between features and ice thickness from multi-source meteorological parameters, arranging the obtained grey comprehensive correlation degrees from large to small, and screening features with grey comprehensive correlation degrees higher than a set threshold.

[0042] Furthermore, the collected meteorological data and temperature are recorded as ,humidity , wind speed , ice class , precipitation , ice thickness ; The grey correlation analysis method is used to calculate the comprehensive correlation between input meteorological characteristics and ice thickness. and meteorological parameters (temperature ,humidity , wind speed , ice class , precipitation ) to establish a comprehensive association model, which can be expressed as:

[0043] ,

[0044] in, Indicates ice thickness With the Meteorological parameters The comprehensive correlation between 、 Represents ice thickness and meteorological parameters The grey absolute correlation degree and grey relative correlation degree, Used to balance the role of absolute correlation and relative correlation;

[0045] Specifically, the grey comprehensive correlation reflects the research sequence. and research The degree of geometric similarity also characterizes the study sequence and research sequence The closeness of the rate of change relative to the initial point;

[0046] For example, if the predicted target sequence is , the sequence of relevant influencing factors is ; If the lengths of the two sequences are the same, the grey absolute correlation degree of the two sequences is expressed as:

[0047] ,

[0048] in:

[0049] ,

[0050] ,

[0051] ,

[0052] = ,

[0053] ,

[0054] in, Represents the target sequence element, Represents the sequence elements of related influencing factors, Represents the predicted target sequence The cumulative absolute value of Represents the sequence of related influencing factors The cumulative absolute value of Represents the predicted target sequence and related influencing factor sequences The absolute value of the difference between the cumulative absolute values of Represents the predicted target sequence The The original value of the elements, Represents the sequence of related influencing factors The The original value of the elements, Indicates the index of the sum, which means traversing from the second element to the second to last element of the sequence; Indicates the length of the sequence, indicating the total number of elements in the sequence. When the target sequence is , the sequence of relevant influencing factors is , if the two sequences have the same length and their initial values are not zero, then and The initial value term can be calculated as and ;

[0055] in:

[0056] ,

[0057] ,

[0058] say and Grey absolute correlation degree for and The grey relative correlation degree is recorded as ;

[0059] ,

[0060] in:

[0061] ,

[0062] ,

[0063] ,

[0064] ,

[0065] = ,

[0066] in, Indicates the first The ratio of the first element to the Indicates the first The ratio of the first element to the Represents the initial value term of the predicted target sequence The cumulative absolute value of Represents the initial value of the sequence of related influencing factors The cumulative absolute value of Represents the initial value term of the predicted target sequence and the initial value of the sequence of related influencing factors The absolute value of the difference between the cumulative absolute values of Represents the original data sequence of the predicted target sequence, that is itself, The original data sequence representing the sequence of relevant influencing factors, that is, itself;

[0067] In addition, the grey relative correlation degree is a representation of the correlation between the change rates of the two sequences relative to the initial point. It is only related to the change rate of the sequence relative to the initial point, but has nothing to do with the size of the values of each sequence.

[0068] Furthermore, the specific comprehensive correlation values of meteorological characteristics and ice thickness are calculated according to the formula, and the values are arranged from large to small. The larger the value of the important feature, the more important it is. The meteorological characteristics with a grey correlation greater than 0.7 are selected. The grey correlation flow chart is as follows: Figure 2 As shown in the figure, the grey relational degree of ice cover is as follows: Figure 3 As shown, Figure 2 The thirteen characteristics are average wind speed, slope, daily precipitation, altitude, minimum temperature, average temperature, ice level, maximum temperature, slope, longitude, latitude, average humidity, and slope aspect.

[0069] S3: Preprocess the collected data, construct an ice thickness prediction dataset and divide it into a training set and a test set.

[0070] It should be noted that if Figure 1 As shown in step S3, constructing an ice thickness prediction dataset and dividing the dataset includes taking historical observed ice accumulation data and screened meteorological data with high comprehensive correlation as input time series, taking the predicted value of future ice thickness as the target time series, and constructing an ice thickness time series data set; performing preprocessing, including extracting sample point data, data cleaning, and data normalization; and dividing the dataset into a training set and a test set in proportion.

[0071] Furthermore, historical ice accumulation data and selected meteorological data with high comprehensive correlation, such as wind speed, daily precipitation, altitude, and microtopography, were used as input time series, and the predicted value of future ice thickness was used as the target time series to construct an ice thickness time series data set. Preprocessing was then performed, including extracting sample point data, data cleaning, and data normalization.

[0072] Specifically, the sample points of the dataset are extracted, that is, the original data generated by climate change during the ice covering process; the real-time data is collected once a day as the original sample points of the dataset, as shown in Table 1;

[0073] Table 1 Dataset samples

[0074] ,

[0075] Among them, data cleaning uses the average method to clean the missing or abnormal sample point data in the ice cover dataset through its previous The average value of the sample points in the adjacent time periods is filled in at each moment. The average value calculation formula is:

[0076] ,

[0077] in represents the calculated average value, represents the moment when missing or abnormal data occurs, represents the number of samples selected, express Moment ago The sample point data at each moment, express Moment ago The sample point data at each moment, express The sample point data at the first moment before the moment;

[0078] Data normalization, transform the original data and map the data to the default The calculation formula for data normalization is expressed as:

[0079] ,

[0080] in, represents the sample point data, represents the minimum value of the sample data, represents the maximum value of the sample data, Represents the data after normalization.

[0081] Furthermore, the dataset is divided into training set and test set in a ratio of 8:2.

[0082] S4: Combine the bidirectional gated recurrent unit and the multi-head self-attention mechanism to build a hybrid prediction model.

[0083] It should be noted that if Figure 1As shown in step S4, constructing a hybrid prediction model includes combining a bidirectional gated recurrent unit and a multi-head self-attention mechanism, using a gated recurrent unit to capture the bidirectional dependency of time series data, and assigning weights to time steps through a multi-head self-attention mechanism; after combining the bidirectional gated recurrent unit and the multi-head self-attention mechanism, a feature representation is obtained, which is passed through a fully connected layer to obtain the result of the hybrid prediction model.

[0084] Furthermore, to improve the training efficiency of the model, the first-order difference method is used to eliminate the non-stationarity of the data. The first-order difference operation makes the original non-stationary series stable, which helps the subsequent model learning process. Training the model on the differenced data can improve the prediction accuracy and stability.

[0085] A bidirectional GRU (Gated Recurrent Unit) is used to capture the bidirectional dependencies of time series data. The bidirectional GRU structure refers to a neural network structure composed of two GRU units at the same time: one processes the input sequence in time order (called forward GRU), and the other processes the input sequence in time reverse order (called backward GRU); for each time step t, the forward GRU generates a hidden state , the backward GRU generates the hidden state , the final output state It is composed of these two states:

[0086] ,

[0087] The symbol “[ ; ]” indicates concatenating two vectors in dimension.

[0088] The structure and calculation method of the forward GRU and backward GRU are the same as those of the unidirectional GRU, except that the processing order is reversed; the update gate, reset gate, and candidate state calculation formula of each GRU are expressed as:

[0089] ,

[0090] ,

[0091] ,

[0092] ,

[0093] Among them, the update gate and reset gate Control the updating and forgetting of information, and is the weight matrix of the update gate, and is the weight matrix of the reset gate, is the Sigmoid activation function, the candidate hidden state The calculation combines the current input and the previous hidden state after reset, represents bit-wise multiplication of elements, and is the weight matrix used to generate candidate hidden states, Represents the hyperbolic tangent activation function; generates the final hidden state of the current time step , is the previous time step The hidden state of contains the information of the previous time step.

[0094] Correspondingly output the hidden state sequence:

[0095] The forward GRU processes the input sequence in chronological order and outputs the corresponding hidden state sequence:

[0096] ,

[0097] The forward GRU processes the input sequence in reverse time order and outputs a corresponding hidden state sequence:

[0098] .

[0099] Furthermore, the attention mechanism calculates the similarity between different time steps and assigns a weight to each time step, which is expressed as:

[0100] ,

[0101] in, , , is the weight matrix of query, key and value, , , are the matrix representations of query, key, and value, respectively, used to calculate the attention weights. The similarity between the query vector and the key vector is measured by calculating the dot product between them, and the similarity is normalized into weights through the softmax function. These weights are then used to perform weighted summation on the value vector.

[0102] ,

[0103] in, It is a mechanism used to enhance the model's ability to focus on specific parts of the input data. is the similarity matrix between query and key, is the dimension of the key, The operation ensures the normalization of the weights, so that the output of each position can be updated based on the weighted information of other positions;

[0104] After passing through the bidirectional GRU and multi-head self-attention mechanism, a feature representation will be obtained, which will be passed to the fully connected layer and finally output the prediction result , expressed as:

[0105] ,

[0106] in, It is a feature representation processed by the bidirectional GRU and multi-head self-attention mechanism, which contains the information of the sequence data at each time step; is the weight matrix of the fully connected layer, which is used to Mapping to the final prediction space; is the bias term of the fully connected layer, which is used to adjust the baseline of the prediction output; the model prediction process is as follows Figure 5 shown.

[0107] S5: Use the training set to train the combined model, and use the test set data to make predictions, complete the model test, and obtain the prediction results.

[0108] It should be noted that if Figure 1 As shown in step S5, the combined model is trained by training the hybrid prediction model with the training set data, inputting the data into the model for prediction through the sliding window mechanism, and adjusting the parameters through the optimizer, setting the learning rate and number of iterations, and completing the training and testing.

[0109] Furthermore, the operating data of multiple days of ice cover period collected by multiple monitoring devices were selected, and a short-term ice thickness prediction model based on grey correlation and BiGRU-MSA (a hybrid model of bidirectional gated recurrent unit and multi-head self-attention mechanism) was trained using the training set data; the model inputs gas data, and the goal is to predict the gas content; through the sliding window mechanism, historical data is used each time to predict the change in gas concentration for the next day, and future gas prediction data is gradually obtained; the entire model is implemented using the PyTorch (deep learning) framework, the number of hidden layer neurons is set, the optimizer uses the Adam (adaptive moment estimation) algorithm, the learning rate is set, and multiple iterations are performed during the training process.

[0110] Example 2, reference Figure 5-Figure 7 The second embodiment of the present invention provides a method for predicting ice thickness based on grey correlation. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0111] We selected operational data collected by multiple monitoring devices over a 625-day ice-cover period, as shown in Table 2. The model's input features consisted of gas data from the previous 10 days, with the goal of predicting gas concentrations on the 11th day. Using a sliding window mechanism, we used 10 days of historical data to predict gas concentrations for the next day, gradually generating future gas forecasts. The entire model was implemented using the PyTorch framework. The model's hidden layer had 64 neurons, and the Adam algorithm was used as the optimizer with a learning rate of 0.02. Training was performed for 150 iterations.

[0112] Table 2 Operational data collected by monitoring equipment during the 625-day ice-cover period

[0113] ,

[0114] The prediction results obtained using the classic model LSTM are as follows Figure 5 As shown in the figure, the prediction accuracy is 33.00%. In the figure, TrueValue is the true value, Predicted Value is the predicted value, Value is the ice thickness value, Best Epoch is the round with the best prediction effect, True vs Predicted Values is the true value relative to the predicted value, and Time is the time.

[0115] Prediction results obtained using the classic model BiGRU Figure 6 As shown in the figure, the prediction accuracy is 36.23%. In the figure, TrueValue is the true value, Predicted Value is the predicted value, Value is the ice thickness value, Best Epoch is the round with the best prediction effect, True vs Predicted Values is the true value relative to the predicted value, and Time is the time.

[0116] After completing the model training and performing the prediction, the comparison between the actual data and the prediction results is shown in the figure below. Figure 7 As shown in the figure, TrueValue is the true value, Predicted Value is the predicted value, Value is the ice thickness value, Best Epoch is the round with the best prediction effect, True vs Predicted Values is the true value relative to the predicted value, and Time is the time.

[0117] From the prediction results, it can be seen that the predicted values have a high degree of overlap with the actual values, with a prediction accuracy of 84.16%, a MAE of 5.1769, an MSE of 49.1601, an RMSE of 7.0114, and a MAPE of 11.99%. This shows that the prediction accuracy of the short-term ice thickness prediction model based on grey correlation and BiGRU-MSA is much higher than that of the classical model, and it can be used in ice thickness monitoring devices and systems to improve measurement accuracy and guide ice disposal decisions.

[0118] Example 3 is the third embodiment of the present invention, which differs from the first two embodiments in that:

[0119] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0120] Logic and / or steps otherwise described herein, which may be considered, for example, as a sequenced list of executable instructions for implementing logical functions, may be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, "computer-readable medium" may be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0121] It should be understood that various components of the present invention may be implemented using hardware, software, firmware, or a combination thereof. In the aforementioned embodiments, multiple steps or methods may be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one or a combination of the following technologies known in the art may be used: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field-programmable gate array (FPGA), etc.

[0122] Example 4 is the fourth embodiment of the present invention, which provides a grey correlation-based ice thickness prediction system, comprising: a data acquisition module, a correlation screening module, a data adjustment module, a hybrid modeling module, and an intelligent optimization module;

[0123] The data acquisition module collects ice cover related data in real time using weather stations and sensors in different regions and time periods;

[0124] The correlation screening module uses a grey correlation analysis method and a threshold value to screen out important features with a high correlation with ice thickness from the features;

[0125] The data adjustment module pre-processes the collected data, constructs an ice thickness prediction data set and divides it into a training set and a test set;

[0126] The hybrid modeling module combines the bidirectional gated recurrent unit and the multi-head self-attention mechanism to construct a hybrid prediction model;

[0127] The intelligent optimization module uses the training set to train the combined model, and uses the test set data to make predictions, complete the model test, and obtain the prediction results.

[0128] This embodiment further provides a computing device applicable to a gray correlation-based ice thickness prediction method, including:

[0129] Memory and processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement a gray correlation-based ice thickness prediction method proposed in the above embodiment.

[0130] This embodiment further provides a storage medium storing a computer program, which, when executed by a processor, implements a method for predicting ice thickness based on grey correlation as proposed in the above embodiment.

[0131] The storage medium proposed in this embodiment and the ice thickness prediction method based on grey correlation proposed in the above embodiment belong to the same inventive concept. For technical details not fully described in this embodiment, please refer to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0132] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A grey correlation-based ice thickness prediction method, characterized by: include: Use weather stations and sensors to collect real-time ice cover data in different regions and time periods; Grey correlation analysis method is used to select important features with high correlation with ice thickness from the features using thresholds; Preprocess the collected data, construct an ice thickness prediction dataset and divide it into a training set and a test set; Combine bidirectional gated recurrent units and multi-head self-attention mechanisms to build a hybrid prediction model; Use the training set to train the combined model, and use the test set data to make predictions, complete the model test, and obtain the prediction results.

2. The method for predicting ice thickness based on grey correlation as claimed in claim 1, characterized in that: The collection of ice-related data includes using a weather station and sensor network to conduct real-time online detection of ice-covered areas on transmission lines, including temperature, humidity, wind speed, ice level, precipitation, and ice coverage data; Among them, ice coverage data includes records of ice thickness on transmission lines over a period of time in the past.

3. The method for predicting ice thickness based on grey correlation as claimed in claim 2, characterized in that: The grey correlation analysis method includes calculating the comprehensive correlation between input meteorological characteristics and ice thickness, establishing a comprehensive correlation model according to the relationship between ice thickness and meteorological parameters, and obtaining the grey comprehensive correlation.

4. The method for predicting ice thickness based on grey correlation as claimed in claim 3, characterized in that: The method of screening out important features with a high correlation with ice thickness includes using a grey correlation analysis method to quantify the correlation between features and ice thickness from multi-source meteorological parameters, arranging the obtained grey comprehensive correlation degrees from large to small, and screening features with a grey comprehensive correlation degree higher than a set threshold.

5. The method for predicting ice thickness based on grey correlation as claimed in claim 4, characterized in that: The method of constructing an ice thickness prediction data set and dividing it into a training set and a test set includes taking historical observed ice accumulation data and screened meteorological data with high comprehensive correlation as input time series, taking the predicted value of future ice thickness as the target time series, and constructing an ice thickness time series data set; Perform preprocessing, including extracting sample point data, data cleaning, and data normalization; and divide the data set into training set and test set in proportion.

6. The method for predicting ice thickness based on grey correlation as claimed in claim 5, characterized in that: The construction of the hybrid prediction model includes combining a bidirectional gated recurrent unit and a multi-head self-attention mechanism, using a gated recurrent unit to capture the bidirectional dependency of time series data, and assigning weights to time steps through a multi-head self-attention mechanism; after combining the bidirectional gated recurrent unit and the multi-head self-attention mechanism, a feature representation is obtained, which is passed through a fully connected layer to obtain a hybrid prediction model result.

7. The method for predicting ice thickness based on grey correlation as claimed in claim 6, characterized in that: The combined model is trained by training the hybrid prediction model with training set data, inputting data into the model for prediction through a sliding window mechanism, and adjusting parameters through an optimizer, setting a learning rate and number of iterations, to complete training and testing.

8. A system for predicting ice thickness based on grey correlation, using the method for predicting ice thickness based on grey correlation according to any one of claims 1 to 7, characterized in that: include: Data acquisition module, association screening module, data adjustment module, hybrid modeling module and intelligent optimization module; The data acquisition module collects ice cover related data in real time using weather stations and sensors in different regions and time periods; The correlation screening module uses a grey correlation analysis method and a threshold value to screen out important features with a high correlation with ice thickness from the features; The data adjustment module pre-processes the collected data, constructs an ice thickness prediction data set and divides it into a training set and a test set; The hybrid modeling module combines the bidirectional gated recurrent unit and the multi-head self-attention mechanism to construct a hybrid prediction model; The intelligent optimization module uses the training set to train the combined model, and uses the test set data to make predictions, complete the model test, and obtain the prediction results.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the grey correlation-based ice thickness prediction method according to any one of claims 1 to 7 are 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 steps of a grey correlation-based ice thickness prediction method according to any one of claims 1 to 7 are implemented.

Images