Icing thickness prediction method and device, equipment and storage medium

Abstract

The invention discloses an icing thickness prediction method, apparatus and device, and a storage medium. The method comprises the steps of obtaining various meteorological data and corresponding real icing thicknesses in a target line area in real time; constructing an icing thickness prediction model and completing model optimization; taking the various meteorological data as the input of a prediction model, and outputting the predicted icing thickness; the target of the model optimization is to enable the error between the predicted icing thickness output by the model and the real icing thickness to be within a certain error range; and inputting various current meteorological data into the optimized prediction model to obtain a prediction value of the current icing thickness. According to the technical scheme of the embodiment of the invention, the icing thickness prediction precision can be improved.

Description

Technical Field

[0001] This invention relates to the field of power transmission line safety monitoring technology, specifically to a method, device, equipment, and storage medium for predicting icing thickness. Background Technology

[0002] Ice accumulation on power transmission lines is a significant hidden danger threatening the safe operation of the power system. Its hazards extend to multiple levels, including mechanical, electrical, equipment, and public safety, and can easily trigger systemic failures and cascading reactions. As a critical transmission link in the power grid, when lines are damaged by ice accumulation, not only is the transmission of electrical energy impossible, but it also causes enormous economic losses and may even lead to safety accidents.

[0003] The occurrence of icing is closely related to a variety of meteorological factors, such as temperature, humidity, wind speed, and wind direction. These factors exhibit a high degree of randomness and uncertainty, especially in areas with complex terrain and variable climates, making meteorological forecasting inherently challenging. While existing technologies have proposed monitoring using imaging or radar, various interference factors still arise during implementation.

[0004] Therefore, there is an urgent need to propose a more accurate method for predicting icing thickness in order to monitor, analyze, and predict the icing situation of power lines. Summary of the Invention

[0005] This invention provides a method, apparatus, device, and storage medium for predicting icing thickness, thereby improving the accuracy of icing thickness prediction.

[0006] In a first aspect, the present invention provides a method for predicting icing thickness, comprising:

[0007] Real-time acquisition of various meteorological data and corresponding actual icing thickness in the target route area;

[0008] Construct an ice thickness prediction model and optimize the model; use the various meteorological data as input to the prediction model and output the predicted ice thickness; the goal of the model optimization is to ensure that the error between the predicted ice thickness output by the model and the actual ice thickness is within a certain error range.

[0009] By inputting various current meteorological data into the optimized prediction model, the predicted value of the current ice thickness is obtained.

[0010] Furthermore, after acquiring various meteorological data for the target route area, the process also includes: removing abnormal data from various meteorological data based on real-time data trends and historical data for the same period; standardizing the various meteorological data after removing abnormal data; and constructing an initial dataset from the standardized meteorological data.

[0011] The process of constructing the initial dataset is as follows: based on the Laida criterion, outliers are initially identified; combined with real-time data trends and historical data of the same period, it is determined whether the extreme values ​​in various meteorological data are true extreme values. If they are true extreme values, they are retained; if they are false values, the corresponding outlier data is removed; the various meteorological data after removing outlier data are standardized using the Z-Score standardization method; and an initial dataset is constructed from the standardized meteorological data.

[0012] Furthermore, before constructing the ice thickness prediction model, the method further includes: dividing various meteorological data into training and testing sets according to a certain ratio;

[0013] The data partitioning process is as follows: meteorological data of various types are stratified according to the order of months, and training set and test set are divided into various types of meteorological data according to a certain proportion; the proportion of various types of meteorological data of different months or different seasons in the training set is the same as the proportion in the test set; the proportion of various types of meteorological data corresponding to different actual icing thicknesses in the training set is the same as the proportion in the test set.

[0014] Furthermore, an attention mechanism is introduced into the ice thickness prediction model to assign weights to various types of meteorological data, and the various types of meteorological data are merged into a feature vector based on the weights assigned to them.

[0015] The process of merging the various meteorological data into a single feature vector is as follows: After inputting the various meteorological data into the prediction model, the hyperbolic tangent activation function values ​​corresponding to each type of meteorological data are calculated separately, and the intermediate scalar of each type of meteorological data is calculated based on the hyperbolic tangent activation function values ​​corresponding to each type of meteorological data; the intermediate scalar of each type of meteorological data is normalized, and the attention score of each type of meteorological data is calculated based on the normalized intermediate scalar of each type of meteorological data; the feature vector of each type of meteorological data after attention weighting is calculated, and then merged into a single feature vector.

[0016] Furthermore, a temporal convolutional network is introduced into the ice thickness prediction model. The temporal sequence of the feature vector is input into the convolutional kernel of the temporal convolutional network, and the predicted ice thickness is output.

[0017] The process of outputting the predicted ice thickness is as follows: the temporal sequence of the feature vector is input into the dilated convolutional network and the radial network for parallel processing; the output of the dilated convolutional network and the output of the radial network are used as the input of the residual network, and the sum of the outputs of the dilated convolutional network and the radial network is used as the output of the residual network; the output of the residual network is used as the input of the fully connected network, and the fully connected network outputs the predicted ice thickness.

[0018] Furthermore, before inputting the temporal sequence of the feature vector into the convolution kernel of the temporal convolutional network, the method further includes: setting the optimal convolution kernel width and convolution kernel dilation coefficient;

[0019] The process of setting the optimal convolutional kernel width and convolutional kernel dilation coefficient is as follows: based on the adaptation range of the convolutional kernel width and convolutional kernel dilation coefficient, select the initial adaptation values ​​of the convolutional kernel width and convolutional kernel dilation coefficient; iteratively optimize the initial adaptation values ​​of the convolutional kernel width and convolutional kernel dilation coefficient based on the Savannah avoidance algorithm; evaluate the prediction performance of the model under the current parameters by setting a fitness function, and select the convolutional kernel width and convolutional kernel dilation coefficient corresponding to the highest fitness as the optimal solution.

[0020] Furthermore, the fitness function is the reciprocal of the mean square error between the predicted icing thickness and the actual icing thickness output by the model under the current parameters.

[0021] In a second aspect, the present invention provides an ice thickness prediction device, comprising:

[0022] The data acquisition module is used to acquire various meteorological data and corresponding actual icing thickness in the target route area in real time;

[0023] The model building module is used to build an ice thickness prediction model and optimize the model; it takes the various meteorological data as input to the prediction model and outputs the predicted ice thickness; the goal of the model optimization is to make the error between the predicted ice thickness output by the model and the actual ice thickness within a certain error range.

[0024] The thickness prediction module is used to input various current meteorological data into the optimized prediction model to obtain the predicted value of the current icing thickness.

[0025] Thirdly, embodiments of the present invention provide an electronic device, the electronic device comprising:

[0026] At least one processor; and a memory communicatively connected to the at least one processor; wherein,

[0027] The memory stores a computer program that can be executed by at least one processor, such that the at least one processor is able to perform the steps of the icing thickness prediction method according to any embodiment of the present invention.

[0028] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing computer instructions that cause a processor to execute steps of the ice thickness prediction method of any embodiment of the present invention.

[0029] Compared with the prior art, the present invention has the following advantages:

[0030] The technical solution of this invention first acquires various meteorological data and corresponding actual icing thicknesses in the target line area in real time; then, it constructs and optimizes an icing thickness prediction model; it uses various meteorological data as input to the prediction model and outputs the predicted icing thickness; the goal of model optimization is to ensure that the error between the predicted icing thickness and the actual icing thickness is within a certain error range; finally, it inputs the current meteorological data into the optimized prediction model to obtain the predicted value of the current icing thickness. The technical solution of this invention can improve the accuracy of icing thickness prediction and provide technical support for power system fault prediction and maintenance optimization. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only preferred embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 A flowchart illustrating an ice thickness prediction method provided in an embodiment of the present invention;

[0033] Figure 2 A schematic diagram illustrating the implementation process of the prediction model algorithm provided in this embodiment of the invention;

[0034] Figure 3 A comparative diagram of prediction results of some models provided in the embodiments of the present invention;

[0035] Figure 4 A comparison chart of MAPE error and MAX error for some models provided in embodiments of the present invention;

[0036] Figure 5 A comparison chart of RMSE errors for some models provided in embodiments of the present invention;

[0037] Figure 6 This is a schematic diagram of the structure of an ice thickness prediction device provided in an embodiment of the present invention;

[0038] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0039] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0040] Figure 1 This is a flowchart illustrating an ice thickness prediction method provided in an embodiment of the present invention. This embodiment is applicable to monitoring the ice thickness of transmission lines in different geographical environments. The method can be executed by an ice thickness prediction device, which can be implemented in software and / or hardware and can be configured in an electronic device.

[0041] like Figure 1 As shown, the method specifically includes:

[0042] S1 acquires various meteorological data and corresponding actual icing thickness in the target route area in real time.

[0043] The target line area refers to the monitoring area where the transmission line to be monitored long-term is located. Meteorological data includes at least the following categories: daily minimum temperature, relative humidity, saturation humidity, absolute humidity, and wind speed. These categories of meteorological data can be recorded in real time through manual inspection or sensor monitoring.

[0044] In one embodiment, after acquiring various meteorological data of the target route area, the method further includes: removing abnormal data from various meteorological data based on real-time data trends and historical data of the same period; standardizing the various meteorological data after removing abnormal data; and constructing an initial dataset from the standardized meteorological data.

[0045] The process of constructing the initial dataset is as follows: First, outliers are initially identified based on the Raida criterion. Combining real-time data trends with historical data from the same period, it is determined whether the extreme values ​​in the meteorological data are true extreme values. If they are true extreme values, they are retained; if they are false values, the corresponding outlier data is removed. Then, the meteorological data after removing outlier data is standardized using the Z-Score standardization method. Finally, an initial dataset is constructed from the standardized meteorological data of various types.

[0046] In this embodiment of the invention, by combining the trend changes of real-time monitoring data with the distribution of historical data, the loss of effective data caused by static elimination is avoided.

[0047] It should be noted that the Laida criterion is an outlier detection method based on the characteristics of normal distribution. It assumes that a set of detection data contains only random errors, calculates the standard deviation, and determines an interval based on a certain probability. Errors exceeding this interval should be discarded. Z-score standardization is a data preprocessing technique that standardizes data based on the mean and standard deviation of the original data. The mean is adjusted to 0, and the standard deviation is adjusted to 1, thus transforming the data to a uniform scale and eliminating the influence of dimensions.

[0048] For any type of meteorological data, Z-score normalization can be performed as follows:

[0049]

[0050] In the formula, This represents the initial data for any type of meteorological data; This represents the mean of any meteorological data. The standard deviation of any type of meteorological data; Standardized data representing any type of meteorological data.

[0051] For example, when meteorological data includes: daily minimum temperature ( ), relative humidity ( ), saturated humidity ( ), absolute humidity ( When the initial dataset contains (V) and wind speed (V), it can be represented as:

[0052]

[0053] S2, Construct an ice thickness prediction model and complete model optimization; use various meteorological data as input to the prediction model and output the predicted ice thickness; the goal of model optimization is to ensure that the error between the predicted ice thickness output by the model and the actual ice thickness is within a certain error range.

[0054] Figure 2 This is a schematic diagram illustrating the implementation process of the prediction model algorithm provided in the embodiments of the present invention, as shown below. Figure 2 As shown, after acquiring various meteorological data in the target area, the data is preprocessed to remove outliers and standardize the units. These data are then divided into training and testing sets. The model is trained using the training set and its accuracy is verified using the testing set. Next, the data is input into the prediction model, where initial parameter settings and adaptive parameter optimization are performed. Weights are assigned to various input data using an attention mechanism, and feature vectors are reconstructed. Finally, a functional relationship between the feature vectors and the actual ice thickness is established, and the accuracy of this relationship is verified.

[0055] In one embodiment, before constructing the icing thickness prediction model, various meteorological data are divided into training and test sets according to a certain ratio.

[0056] The training set is used to build the model, and the test set is used to verify the model's accuracy.

[0057] The data partitioning process is as follows: meteorological data of various types are stratified according to the order of months, and training set and test set are divided into various types of meteorological data according to a certain proportion; the proportion of various types of meteorological data of different months or different seasons in the training set is the same as the proportion in the test set; the proportion of various types of meteorological data corresponding to different actual icing thicknesses in the training set is the same as the proportion in the test set.

[0058] Preferably, the ratio of meteorological data in the training set to that in the test set is set to 8:2.

[0059] In this embodiment of the invention, by maintaining the consistency of the data ratio between the training set and the test set, and ensuring that both the training set and the test set contain data from typical icy and non-icy periods, the model is prevented from fitting only the features of a specific time period.

[0060] Furthermore, an attention mechanism is introduced into the icing thickness prediction model to assign weights to various types of meteorological data, and the various types of meteorological data are merged into a feature vector based on the weights assigned to them.

[0061] The process of merging various meteorological data into a single feature vector is as follows: After inputting various meteorological data into the prediction model, the hyperbolic tangent activation function values ​​corresponding to each type of meteorological data are calculated separately, and the intermediate scalar of each type of meteorological data is calculated based on the hyperbolic tangent activation function values; the intermediate scalar of each type of meteorological data is normalized, and the attention score of each type of meteorological data is calculated based on the normalized intermediate scalar of each type of meteorological data; the feature vector of each type of meteorological data after attention weighting is calculated, and then merged into a single feature vector.

[0062] The calculation process of the intermediate scalar can be represented as:

[0063]

[0064] in, This represents an intermediate scalar value for the i-th type of meteorological data. This represents the transpose of the attention parameter vector. This represents the hyperbolic tangent activation function. Represents the weight matrix. This represents the original feature vector of the i-th type of meteorological data. This represents the paranoia vector.

[0065] It should be noted that the intermediate scalar is used for subsequent calculation of the attention score, the attention parameter vector is a learnable parameter, the hyperbolic tangent activation function is used to introduce non-linear relationships, the weight matrix is ​​used to map the input features to the dimension of the attention score, and the bias vector is also a learnable parameter.

[0066] After obtaining the intermediate scalars of various meteorological data, the softmax function can be used to normalize each intermediate scalar, and the attention score of each meteorological data can be calculated based on the normalized intermediate scalars of various meteorological data.

[0067] The calculation process for attention scores of various meteorological data can be expressed as follows:

[0068]

[0069] in, This represents the attention score for the i-th type of meteorological data. This represents an intermediate scalar value for the i-th type of meteorological data. This indicates that an intermediate scalar is exponentially operated on, which converts the intermediate scalar into a positive number.

[0070] After obtaining the attention scores for various types of meteorological data, the feature vectors of each type of meteorological data after being weighted by attention weights can be calculated, which can be expressed as:

[0071]

[0072] in, This represents the weighted feature vector of the i-th type of meteorological data. This represents the attention score for the i-th type of meteorological data. This represents the original feature vector of the i-th type of meteorological data.

[0073] After obtaining the feature vectors of various meteorological data after attention weighting, the various meteorological data can be merged into a single feature vector, which can be represented as:

[0074]

[0075] in, This represents the merged feature vector. Let represent the weighted feature vector of the i-th type of meteorological data.

[0076] In this embodiment of the invention, by introducing an attention mechanism, different weights can be assigned to different categories of meteorological data features, thereby highlighting key features and helping to make icing predictions more accurate.

[0077] Furthermore, a temporal convolutional network is introduced into the ice thickness prediction model. The temporal sequence of feature vectors is input into the convolutional kernel of the temporal convolutional network, and the predicted ice thickness is output.

[0078] The process of outputting the predicted ice thickness is as follows: the temporal sequence of the feature vector is input into the dilated convolutional network and the radial network for parallel processing; the outputs of the dilated convolutional network and the radial network are used as inputs to the residual network, and the sum of the outputs of the dilated convolutional network and the radial network is used as the output of the residual network; the output of the residual network is used as input to the fully connected network, and the fully connected network outputs the predicted ice thickness.

[0079] It should be noted that the model construction process needs to consider the icing characteristics of the area where the railway line is located, and take into account the icing lag effect. The icing lag effect is essentially due to the asynchronous nature of real-time weather changes and icing thickness changes. Therefore, a temporal convolutional network can be used to extract the temporal features of the input.

[0080] The convolution kernel in a temporal convolutional network is Then the causal convolutional network at time t can be represented as:

[0081]

[0082] Where K represents the kernel width, This represents the output of the convolution kernel. Represents the weighting coefficients of the convolution kernel. This represents the temporal value of the feature vector. The temporal features at time t are obtained by weighted summation of local temporal windows in the input sequence using a convolutional kernel F. The causal convolutional network ensures that the output at time t depends only on the input at time t and earlier.

[0083] The lag effect of icing can involve the cumulative effects of weather over multiple days (such as continuous low temperatures and rain). Dilated convolutions, by inserting "holes" between elements of the convolution kernel, exponentially expand the receptive field without increasing the parameters. By stacking dilated convolutional layers, the receptive field of a 1D convolution kernel (causal convolutional network) can be expanded exponentially, thereby capturing more distant temporal information. A dilated convolutional network can be represented as:

[0084]

[0085] Where p represents the convolution kernel dilation coefficient, This represents the output of the dilated convolutional layer. Represents the weighting coefficients of the convolution kernel. This represents the expanded temporal value of the eigenvector. The causal convolution kernel is... Dilated convolution at time.

[0086] The input and output of a radial network are the same, namely the temporal sequence of the feature vectors themselves.

[0087] Next, a residual network is used to connect the outputs of the dilated convolutional network and the radial network to address the network layer degradation problem. In the residual network, the output of each layer is reused as input to both the dilated convolutional and radial networks. The output of the residual network can be represented as:

[0088]

[0089] in, This represents the output of the residual network. This represents the output of the dilated convolutional layer. This represents the input to the dilated convolutional layer, i.e., the temporal sequence of feature vectors.

[0090] Finally, the fully connected layer is used to predict the current ice thickness:

[0091]

[0092] in, This represents the predicted icing thickness output by the model. This represents the activation function paired with the fully connected layer. This represents the global timing characteristics output by the last residual module.

[0093] In this embodiment of the invention, an initial prediction model is established between the temporal sequence of feature vectors and the predicted ice thickness using a temporal convolutional network.

[0094] Furthermore, before inputting the temporal sequence of feature vectors into the convolution kernel of the temporal convolutional network, the following steps are also included: setting the optimal convolution kernel width and convolution kernel dilation coefficient.

[0095] The process of setting the optimal kernel width and kernel dilation coefficient is as follows: Based on the fit range of the kernel width and kernel dilation coefficient, select the initial fit values ​​of the kernel width and kernel dilation coefficient; iteratively optimize the initial fit values ​​of the kernel width and kernel dilation coefficient based on the Savannah avoidance algorithm; evaluate the predictive performance of the model under the current parameters by setting a fitness function, and select the kernel width and kernel dilation coefficient corresponding to the highest fitness as the optimal solution.

[0096] Select initialization adaptation parameters. Initialize adaptation parameters, including:

[0097] The kernel size range is adapted. In high-altitude areas (>3000m), the icing accumulation period is long, so the optimization range is set to [3,9]; in low-altitude plain areas, it is set to [1,7], and the initial value of the kernel size is set.

[0098] Kernel dilation coefficient range adaptation. The optimization range of the kernel dilation coefficient is set to [1,9], and the initial value of the kernel dilation coefficient is set.

[0099] Other parameter ranges are adapted. Regularization parameter range. The number of network layers is adjusted according to the line length.

[0100] Based on the Savannah avoidance algorithm, iteratively optimize the parameters and set the maximum number of iterations. For 1000, from the first iteration to the... The next iteration. The icing prediction model based on the improved temporal convolutional network uses the Savannah avoidance algorithm for parameter optimization, the formula of which is as follows:

[0101]

[0102] in, This will be the parameter value for the next iteration. This is the current parameter value. This represents the current iteration number. The maximum number of iterations is r, which is a random number between 0 and 1.

[0103] Define a fitness function to evaluate the predictive performance of the model under the current parameters.

[0104] Preferably, the fitness function is the reciprocal of the mean squared error (MSE), as shown in the following formula:

[0105]

[0106] in, This represents the actual icing thickness. The fitness value represents the model's predicted value, and n is the number of samples in the test set. A higher fitness value indicates higher model prediction accuracy.

[0107] Finally, an optimization iteration is performed. The initialized kernel size K and kernel dilation coefficient p are substituted into the temporal convolutional network model, the model is trained using the training set, and the initial fitness is calculated using the test set. For each newly generated candidate parameter solution... The model is then used to calculate the fitness (Fit). If Fit is greater than the current optimal fitness, the optimal parameters are updated. And the corresponding fitness Fit. Substitute the optimal parameters. The optimal icing prediction model is obtained by applying it to the icing thickness prediction model.

[0108] S3 inputs the current meteorological data into the optimized prediction model to obtain the predicted value of the current ice thickness.

[0109] After obtaining the predicted value of the current icing thickness, corresponding line processing decisions can be made based on the relationship between the predicted value and the icing thickness safety threshold.

[0110] Figure 3 This is a comparative diagram of the prediction results of some models provided in the embodiments of the present invention, such as... Figure 3 As shown, the prediction results include those from the NGO neural network model, the WOA neural network model, the prediction model constructed in this invention, and the actual monitored ice thickness. Except for month number 1 of the ice-covered period, where the predicted value is lower than the monitored value, the NGO neural network model consistently overestimates the actual ice thickness in all other months, indicating that the NGO model is overly sensitive to thickness increases and exhibits a single pattern of prediction deviation. The WOA neural network model's predicted value largely coincides with the monitored value in month number 3 of the ice-covered period, but its error fluctuates significantly in other months, lacking stability and failing to consistently reflect actual changes. The ice thickness prediction model proposed in this invention exhibits the smallest error between the predicted and monitored values ​​for all months, with smooth fluctuations. Its overall trend is highly synchronized with the actual monitored values, avoiding the systematic overestimation seen in the NGO model and overcoming the fluctuation problem of the WOA model, demonstrating significantly superior prediction accuracy and stability.

[0111] Figure 4 This is a comparison chart of the MAPE error and MAX error of some models provided in the embodiments of the present invention, as shown in the figure. Figure 4 As shown, the mean absolute percentage error (MAPE) and maximum error of the ice thickness prediction model proposed in this invention are significantly smaller than those of the NGO neural network model and the WOA neural network model, indicating that the prediction accuracy of the ice thickness prediction model proposed in this invention is significantly better.

[0112] Figure 5 This is a comparison chart of RMSE errors for some models provided in embodiments of the present invention, such as... Figure 5 As shown, the root mean square error (RMSE) of the ice thickness prediction model proposed in this invention is significantly smaller than that of the NGO neural network model and the WOA neural network model, which also indicates that the prediction accuracy of the ice thickness prediction model proposed in this invention is significantly better.

[0113] The technical solution in this embodiment of the invention considers that the relationship between meteorological data and different types of icing is often highly nonlinear and complex. By introducing an attention mechanism into the neural network, based on its adaptive learning capability, it can dynamically capture the nonlinear relationship between input features and target variables, thereby more accurately quantifying the differences in the impact of different meteorological data on icing. Simultaneously, the Savannah avoidance algorithm is used to globally optimize the kernel size and kernel dilation coefficient of the temporal convolutional network model, accurately matching the time scale of the data, optimizing the receptive field, and enhancing the model's adaptability to different data.

[0114] In summary, this invention collects various meteorological data, preprocesses it, and then obtains weighted feature variables through a feature attention mechanism. Combined with the Savannah avoidance algorithm, it optimizes the kernel size and kernel dilation coefficient of the temporal convolutional network-based icing prediction model, resulting in the optimal temporal convolutional network icing prediction model. Finally, it predicts icing thickness. This improves the accuracy of icing thickness prediction and addresses the shortcomings of existing technologies, such as insufficient feature extraction, large icing thickness prediction errors, and significant overfitting.

[0115] Figure 6 This is a schematic diagram of the structure of an ice thickness prediction device provided in an embodiment of the present invention, as shown below. Figure 6 As shown, the device specifically includes:

[0116] The data acquisition module 100 is used to acquire various meteorological data and corresponding actual icing thickness in the target line area in real time.

[0117] The model building module 200 is used to build an ice thickness prediction model and perform model optimization; it takes various meteorological data as input to the prediction model and outputs the predicted ice thickness; the goal of model optimization is to make the error between the predicted ice thickness output by the model and the actual ice thickness within a certain error range.

[0118] The thickness prediction module 300 is used to input various current meteorological data into the optimized prediction model to obtain the predicted value of the current icing thickness.

[0119] The technical solutions in the embodiments of the present invention can accurately obtain the predicted value of icing thickness, providing technical support for power system fault prediction, maintenance optimization, etc.

[0120] Figure 7 This is a schematic diagram of an electronic device implementing the icing thickness prediction method of this invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0121] like Figure 7As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0122] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0123] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as icing thickness prediction methods.

[0124] In some embodiments, the icing thickness prediction method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the icing thickness prediction method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the icing thickness prediction method by any other suitable means (e.g., by means of firmware).

[0125] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0126] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0127] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0128] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0129] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0130] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0131] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0132] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting ice thickness, characterized in that, include: Real-time acquisition of various meteorological data and corresponding actual icing thickness in the target route area; Construct an ice thickness prediction model and optimize the model. The various meteorological data are used as input to the prediction model, and the predicted icing thickness is output. The goal of the model optimization is to ensure that the error between the predicted icing thickness output by the model and the actual icing thickness is within a certain error range. By inputting various current meteorological data into the optimized prediction model, the predicted value of the current ice thickness is obtained.

2. The method according to claim 1, characterized in that, After acquiring various meteorological data for the target route area, the process further includes: removing abnormal data from various meteorological data based on real-time data trends and historical data for the same period; standardizing the various meteorological data after removing abnormal data; and constructing an initial dataset from the standardized meteorological data. The process of constructing the initial dataset is as follows: based on the Laida criterion, outliers are initially identified; combined with real-time data trends and historical data of the same period, it is determined whether the extreme values ​​in various meteorological data are true extreme values. If they are true extreme values, they are retained; if they are false values, the corresponding outlier data is removed; the various meteorological data after removing outlier data are standardized using the Z-Score standardization method; and an initial dataset is constructed from the standardized meteorological data.

3. The method according to claim 1, characterized in that, Before constructing the ice thickness prediction model, the method also includes: dividing various meteorological data into training and testing sets according to a certain ratio; The data partitioning process is as follows: meteorological data of various types are stratified according to the order of months, and training set and test set are divided into various types of meteorological data according to a certain proportion; the proportion of various types of meteorological data of different months or different seasons in the training set is the same as the proportion in the test set; the proportion of various types of meteorological data corresponding to different actual icing thicknesses in the training set is the same as the proportion in the test set.

4. The method according to claim 1, characterized in that, An attention mechanism is introduced into the ice thickness prediction model to assign weights to various types of meteorological data, and the various types of meteorological data are merged into a feature vector based on the weights assigned to them. The process of merging the various meteorological data into a single feature vector is as follows: after inputting the various meteorological data into the prediction model, the hyperbolic tangent activation function values ​​corresponding to each type of meteorological data are solved respectively, and the intermediate scalars of each type of meteorological data are calculated based on the hyperbolic tangent activation function values ​​corresponding to each type of meteorological data. The intermediate scalars of various meteorological data are normalized, and the attention scores of various meteorological data are calculated based on the normalized intermediate scalars of various meteorological data. Based on the attention scores of various meteorological data, the feature vectors of various meteorological data after attention weighting are calculated and merged into a single feature vector.

5. The method according to claim 4, characterized in that, The ice thickness prediction model also introduces a temporal convolutional network, which inputs the temporal sequence of feature vectors into the convolutional kernel of the temporal convolutional network and outputs the predicted ice thickness. The process of outputting the predicted ice thickness is as follows: the temporal sequence of the feature vector is input into the dilated convolutional network and the radial network for parallel processing; the output of the dilated convolutional network and the output of the radial network are used as the input of the residual network, and the sum of the outputs of the dilated convolutional network and the radial network is used as the output of the residual network; the output of the residual network is used as the input of the fully connected network, and the fully connected network outputs the predicted ice thickness.

6. The method according to claim 5, characterized in that, Before inputting the temporal sequence of the feature vector into the convolution kernel of the temporal convolutional network, the method further includes: setting the optimal convolution kernel width and convolution kernel dilation coefficient; The process of setting the optimal convolutional kernel width and convolutional kernel dilation coefficient is as follows: based on the adaptation range of the convolutional kernel width and convolutional kernel dilation coefficient, select the initial adaptation values ​​of the convolutional kernel width and convolutional kernel dilation coefficient; iteratively optimize the initial adaptation values ​​of the convolutional kernel width and convolutional kernel dilation coefficient based on the Savannah avoidance algorithm; evaluate the prediction performance of the model under the current parameters by setting a fitness function, and select the convolutional kernel width and convolutional kernel dilation coefficient corresponding to the highest fitness as the optimal solution.

7. The method according to claim 6, characterized in that, The fitness function is the reciprocal of the mean square error between the predicted icing thickness and the actual icing thickness output by the model under the current parameters.

8. An ice thickness prediction device, characterized in that, The apparatus is configured to implement the method according to any one of claims 1-7, the apparatus comprising: The data acquisition module is used to acquire various meteorological data and corresponding actual icing thickness in the target route area in real time; The model building module is used to build an ice thickness prediction model and optimize the model; it takes the various meteorological data as input to the prediction model and outputs the predicted ice thickness; the goal of the model optimization is to make the error between the predicted ice thickness output by the model and the actual ice thickness within a certain error range. The thickness prediction module is used to input various current meteorological data into the optimized prediction model to obtain the predicted value of the current icing thickness.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to said at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor to enable the at least one processor to perform the steps of the icing thickness prediction method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the steps of the ice thickness prediction method according to any one of claims 1-7.

Images