Wind-blown snow motion trail simulation tracking method based on wind-blown snow numerical model

Abstract

The invention relates to the technical field of motion trail simulation tracking, in particular to a wind-blown snow motion trail simulation tracking method based on a wind-blown snow numerical model, which comprises the following steps of: extracting a spatial coupling relationship among local meteorological variables, and constructing a meteorological region division basis with higher discrimination capability, so that region division does not depend on univariate layering any more; therefore, the recognition capability of a local strong disturbance area is improved, a meteorological time sequence is segmented and analyzed through a convolutional neural network, a nonlinear time sequence evolution rule among meteorological variables is captured, an interactive mapping relation between meteorological changes and trajectory changes is constructed, the response sensitivity of a prediction model to sudden weather forms is enhanced, and the prediction accuracy is improved. According to the method, weighted evaluation is carried out on multiple groups of model prediction results through a random forest, a final prediction path is optimized according to historical error distribution, structured tracking of error sources is realized, and adaptability and stability of the prediction results are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of motion trajectory simulation and tracking, and particularly to a method for simulating and tracking the motion trajectory of drifting snow based on a numerical model of drifting snow. Background Art

[0002] The technical field of motion trajectory simulation and tracking is a technology that uses computer simulation and digital algorithms to track, predict, and analyze the motion trajectories of objects or phenomena. It predicts motion trajectories based on physical models and performs real-time tracking and analysis of motion states through sensors and computer systems.

[0003] A method for simulating and tracking the motion trajectory of drifting snow based on a numerical model of drifting snow is used to simulate and predict the motion trajectories of snow grains in the drifting snow phenomenon, and provide tracking and analysis of the trajectories. The purpose is to accurately simulate the dynamic trajectories of snow grains during the drifting snow process through a numerical calculation model, the behavioral changes under conditions such as wind speed, snow volume, and geographical environment, and effectively evaluate the severity, influence range of drifting snow events, and their potential threats to facilities such as transportation and buildings.

[0004] The existing technology calculation logic is centered around solving fixed equations, lacking a continuous response mechanism to environmental state changes and the ability to extract local perturbation characteristics. It is difficult to accurately restore the acceleration change process of snow grains under drastic changes in the wind field, resulting in prediction deviations in spatially heterogeneous regions in the simulation results. Due to the lack of the ability to model the non-linear interaction between meteorological variables, the prediction model is difficult to maintain the temporal coherence of the output results during multiple rounds of meteorological changes, and the computational load is limited by the performance of the central node, making it difficult to handle application scenarios with large regional spans and multiple sources of perturbations. Summary of the Invention

[0005] The purpose of the present invention is to solve the deficiencies existing in the prior art and propose a method for simulating and tracking the motion trajectory of drifting snow based on a numerical model of drifting snow.

[0006] To achieve the above purpose, the present invention adopts the following technical solutions: A method for simulating and tracking the motion trajectory of drifting snow based on a numerical model of drifting snow, including the following steps: S1: Obtain real-time meteorological data through sensors, identify the time periods with greater influence on drifting snow by analyzing meteorological parameters in different regions, and calculate the drifting snow environment impact index; S2: Based on the drifting snow environment impact index, extract local spatial features of meteorological data such as wind speed, temperature, and humidity through convolution operations, layer the meteorological regions according to the influence of each feature, and generate a meteorological region feature layer; S3: Based on the meteorological regional feature layer, use a convolutional neural network to segment the time series data, train a model to compare the changes between time nodes, analyze the correlation between meteorological parameters and trajectory changes, and generate predicted values of the blowing snow movement trajectory; S4: Based on the predicted values of the blowing snow movement trajectory, allocate meteorological data to a distributed computing platform, dynamically adjust the task priorities, allocate resources to the areas with greater impacts, and generate simulated data of the blowing snow trajectories in multiple regions; S5: Based on the simulated data of the blowing snow trajectories in multiple regions, use a random forest to perform weighted fusion on the results of multiple models, analyze the prediction errors, judge the adaptability of the models, select the best model for tracking and optimization, and generate optimized predicted results of the blowing snow trajectory.

[0007] As a further solution of the present invention, the specific steps for generating the blowing snow environmental impact index are as follows: Obtain real-time meteorological data through sensors, convert it into meteorological parameters, perform spatial division by region, calculate the wind speed fluctuations, temperature fluctuations, and humidity changes in each region, and map them to the corresponding time periods to generate meteorological parameter fluctuation data; Based on the meteorological parameter fluctuation data, gradually screen the meteorological parameters in each region, analyze the time periods related to the blowing snow phenomenon, identify the time periods affected by blowing snow in each region, and generate blowing snow time period impact data; Based on the blowing snow time period impact data, calculate the change amplitudes of the meteorological parameters in each region, and combine the meteorological impact intensities in each region with the time nodes to generate the blowing snow environmental impact index.

[0008] As a further solution of the present invention, the meteorological parameters include wind speed, temperature, humidity, air pressure, precipitation, radiation, cloud cover, visibility, dew point, meteorological phenomena, atmospheric stability, wind direction, wind frequency, evaporation, snow depth, snow grain size, and snow water equivalent.

[0009] As a further solution of the present invention, the specific steps for generating the meteorological regional feature layer are as follows: Based on the blowing snow environmental impact index, extract data such as wind speed, temperature, and humidity in each region, set thresholds, perform weighting according to the impact degrees, calculate the meteorological impact weights of each region, and generate regional meteorological data weights; Based on the regional meteorological data weights, perform spatial distribution analysis on the meteorological data within the region, layer by layer extract the key meteorological features of each region, and generate meteorological regional local feature data; Based on the meteorological regional local feature data, fuse the meteorological features of each region, perform multi-level feature mapping, and generate the meteorological regional feature layer.

[0010] As a further solution of the present invention, the specific steps for generating the predicted value of the blowing snow movement trajectory are as follows: Based on the meteorological regional feature layer, using a convolutional neural network, divide the data by time interval, extract the meteorological parameters of each time period and classify them into the corresponding regions to generate a time series meteorological data set; Based on the time series meteorological data set, compare the meteorological parameters of different time nodes, analyze the correlation between the changes in meteorological parameters and the blowing snow trajectory, and generate meteorological change and trajectory correlation data through data difference detection; Based on the meteorological change and trajectory correlation data, compare the change trends of meteorological parameters and the blowing snow trajectory between each time node to generate the predicted value of the blowing snow movement trajectory.

[0011] As a further solution of the present invention, the convolutional neural network follows the formula:

[0012] Where: represents the output feature of the th layer at time step , represents the input feature of the previous layer at time step , represents the weight of the th layer and the th convolutional kernel, represents the bias term of the th layer, represents the non-linear activation function, represents the time window width, represents the adjustment factor at time step , represents the th layer's dynamic adjustment coefficient.

[0013] As a further solution of the present invention, the specific steps for generating the multi-region blowing snow trajectory simulation data are as follows: Based on the predicted value of the blowing snow movement trajectory, perform spatial allocation on the meteorological data, and allocate the data to each computing node according to the regional importance to generate the regional meteorological data allocation result; Based on the regional meteorological data allocation result, adjust the priority of the computing tasks according to the urgency of the influence of the blowing snow trajectory, and allocate more computing resources to important regions to generate regional task priority data; Based on the regional task priority data, perform parallel processing of the optimized computing tasks on the distributed platform to generate multi-region blowing snow trajectory simulation data.

[0014] As a further solution of the present invention, for the spatial allocation, first, according to the regional data in the predicted value of the blowing snow movement trajectory, combined with the meteorological influence range of each region, the importance of each region is evaluated, and according to the evaluation result, the regional data is allocated to the corresponding computing nodes according to the influence, so as to obtain the regional meteorological data allocation result.

[0015] As a further solution of the present invention, the specific steps for generating the optimized prediction result of the blowing snow trajectory are as follows: Based on the multi-region blowing snow trajectory simulation data, a random forest is used to weight the output results of each model, adjust according to the influence degree of each region, integrate the predicted values of each region, and generate the model weighted result; Based on the model weighted result, the prediction error of each region is statistically analyzed, the error value is calculated one by one, and the regions with larger errors are marked. By comparing with the set error threshold, the regional error analysis data is generated; Based on the regional error analysis data, analyze the performance of each model in different regions, select the most suitable model by comparing the error sizes, and perform adjustment and optimization to generate the optimized prediction result of the blowing snow trajectory.

[0016] As a further solution of the present invention, for the random forest, according to the formula:

[0017] Where: represents the final weighted output result, represents the weight coefficient of the represents the predicted value of the represents the total number of regions, represents the meteorological factor weight of the represents the terrain adaptability coefficient of the represents the historical data adaptability coefficient of the

[0018] Compared with the prior art, the advantages and positive effects of the present invention are as follows: 1. In the present invention, by extracting the spatial coupling relationship between local meteorological variables, a more discriminative basis for meteorological region division is constructed, so that the region division no longer depends on single-variable stratification, thereby improving the recognition ability of local strong disturbance regions; 2. In the present invention, through the convolutional neural network to segment and analyze the meteorological time series, capture the non-linear time series evolution law between meteorological variables, and construct the interaction mapping relationship between meteorological changes and trajectory changes, enhancing the response sensitivity of the prediction model to sudden weather patterns; 3. In the present invention, through a distributed computing architecture, multi-region simulation tasks are prioritized and resources are dynamically allocated according to the influence intensity, avoiding waste of computing resources and improving the computing accuracy of key regions; 4. In the present invention, through random forests, weighted evaluation is performed on the prediction results of multiple groups of models, and the final prediction path is optimized based on the historical error distribution, realizing structured tracking of the error sources and improving the adaptability and stability of the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a schematic diagram of the main steps of the present invention; Figure 2 is a detailed schematic diagram of S1 of the present invention; Figure 3 is a detailed schematic diagram of S2 of the present invention; Figure 4 is a detailed schematic diagram of S3 of the present invention; Figure 5 is a detailed schematic diagram of S4 of the present invention; Figure 6 is a detailed schematic diagram of S5 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0021] Embodiment 1 Please refer to Figure 1 , the present invention provides a technical solution: a method for simulating and tracking the movement trajectory of drifting snow based on a numerical model of drifting snow, including the following steps: S1: Obtain real-time meteorological data through sensors, identify the time periods with greater influence on drifting snow by analyzing the meteorological parameters in different regions, and calculate the environmental influence index of drifting snow; S2: Based on the environmental influence index of drifting snow, extract the local spatial features of meteorological data such as wind speed, temperature, and humidity through convolution operations, layer the meteorological regions according to the influence of each feature, and generate a meteorological region feature layer; S3: Based on the meteorological region feature layer, use a convolutional neural network to segment the time series data, train the model to compare the changes between time nodes, analyze the correlation between meteorological parameters and trajectory changes, and generate predicted values of the movement trajectory of drifting snow; S4: Based on the predicted values of the movement trajectory of drifting snow, allocate meteorological data to a distributed computing platform, dynamically adjust the task priorities, and allocate resources to regions with greater influence to generate multi-region drifting snow trajectory simulation data; S5: Based on the simulated data of snow-drifting trajectories in multiple regions, use random forest to perform weighted fusion on the results of multiple models, analyze the prediction errors, judge the adaptability of the models, select the best model for tracking and optimization, and generate the optimized prediction results of snow-drifting trajectories.

[0022] Please refer to Figure 2 , and the specific steps to generate the snow-drifting environment impact index are as follows: Obtain real-time meteorological data through sensors, convert it into meteorological parameters, conduct spatial division by region, calculate the wind speed fluctuations, temperature fluctuations, and humidity changes in each region, and map them to the corresponding time periods to generate meteorological parameter fluctuation data; Based on the meteorological parameter fluctuation data, gradually screen the meteorological parameters in each region, analyze the time periods related to the snow-drifting phenomenon, identify the snow-drifting impact time periods within each region, and generate snow-drifting time period impact data; Based on the snow-drifting time period impact data, calculate the change range of meteorological parameters in each region, and combine the meteorological impact intensity within each region with time nodes to generate the snow-drifting environment impact index; Based on the original meteorological observation data output by the sensors, use the wavelet transform method to perform time series processing on the three meteorological signals of wind speed, temperature, and humidity. The selected wavelet basis is the 4th-order Daubechies wavelet, and the decomposition level is 5. Sequentially perform wavelet multi-scale decomposition operations on each meteorological parameter. Complete the calculation of local volatility by extracting the approximate coefficients and detail coefficients of each layer. When processing, set the analysis window width to 15 minutes and the sliding step size to 1 minute. Reconstruct the decomposed signal along the time axis to construct a sequence of the changing trends of meteorological parameters. Subsequently, use the K-means++ clustering method for spatial division. The input samples are the average values of the reconstructed wind speed, temperature, and humidity time series within the specified regional grid. The initialization method of the clustering center is the default mechanism of the K-means++ algorithm, and the number of clusters is set to 8. Use the Euclidean distance to measure the meteorological differences between grids, complete the attribution classification of meteorological data in the spatial grid, and finally generate meteorological parameter fluctuation data; Based on the fluctuating data of meteorological parameters, the double-threshold mutation detection method is used to perform area-by-area fluctuation identification processing on the time series of wind speed, air temperature, and humidity. Upper and lower thresholds are set for each of the three meteorological variables in each grid. The upper and lower limits of mutation are set by superimposing 2 times the standard deviation on the basis of historical moving average. Meteorological data points outside the range are tagged as abnormal points. Subsequently, the sliding window matching algorithm is used for time period identification. The window size is set to 3 hours, and the sliding step is 15 minutes. The total number of abnormal points in each window is counted. If the number of abnormal points in each of the three consecutive windows exceeds 60% of the window capacity, the time is marked as a possible time period affected by blowing snow, and the windows with adjacent time not exceeding 1 hour are merged to obtain a continuous time period interval. After the processing is completed, the blowing snow time period impact data is generated by combining the area number and the time period information; Based on the blowing snow time period impact data, the Z-Score normalization method is used to perform normalization processing on the change amplitudes of wind speed, air temperature, and humidity during the corresponding time period. The normalization process standardizes the meteorological data based on the historical 30-day moving average and standard deviation within the area, and the normalized amplitude values of the three parameters are output respectively. Subsequently, the exponentially weighted moving average method is used to perform a weighting operation on the normalized data. The weighting factor is set to 0.4 to make the recent changes have a higher weight. The weighted data sequences of wind speed, air temperature, and humidity are output respectively. Then, based on the set linear weights for fusion, the wind speed weight is set to 0.5, the air temperature is set to 0.3, and the humidity is set to 0.2. The fusion value is bound with the area number and the corresponding time node and encoded into a three-element data set, and finally the blowing snow environment impact index is generated.

[0023] Meteorological parameters include wind speed, air temperature, humidity, air pressure, precipitation, radiation, cloud cover, visibility, dew point, meteorological phenomena, atmospheric stability, wind direction, wind frequency, evaporation, snow depth, snow grain size, and snow water equivalent.

[0024] Please refer to Figure 3 , the specific steps to generate the meteorological regional feature layer are as follows: Based on the blowing snow environment impact index, data such as wind speed, air temperature, and humidity of each area are extracted, weighted according to the degree of influence by setting thresholds, and the meteorological influence weights of each area are calculated to generate the regional meteorological data weights; Based on the regional meteorological data weights, spatial distribution analysis of the meteorological data within the area is performed, and the key meteorological features of each area are extracted layer by layer to generate the meteorological regional local feature data; Based on the meteorological regional local feature data, the meteorological features of each area are fused, and multi-level feature mapping is performed to generate the meteorological regional feature layer; Based on the blowing snow environment impact index, the fuzzy comprehensive evaluation method is used to weight the wind speed, temperature, and humidity data in each region. First, the fuzzy grade set is set as a five-level equally spaced division interval, corresponding to five impact levels: extremely low, low, medium, high, and extremely high. The weight vector is set as 0.4 for wind speed, 0.35 for temperature, and 0.25 for humidity. The maximum membership principle is used to construct a fuzzy matrix for each input data. Subsequently, the fuzzy synthesis operation is performed, and the weighted average type operation structure is selected for the synthesis method. The comprehensive evaluation value is output separately for each region during the calculation process as the regional meteorological weight index. Finally, all regional indicators are mapped to the corresponding geographical grid numbers to output the regional meteorological data weights; Based on the regional meteorological data weights, the inverse distance weighted interpolation method is used to analyze the spatial distribution of meteorological data within the region. The interpolation weight function is set as a power function structure, and the power parameter is set as 2. The data of the 10 meteorological stations closest to each grid cell are used as reference points. The interpolation weight coefficient is calculated based on the inverse ratio relationship between the distance from the interpolation center point to the reference points. The interpolation filling operations of wind speed, temperature, and humidity in the regional space are completed. Subsequently, the threshold extraction method based on hierarchical clustering is used to decompose the filled spatial data layer by layer. The similarity measurement method is set as the Manhattan distance, and the merging strategy is the minimum distance principle. The grid cells with a fluctuation amplitude higher than 20% of the median are extracted and classified for each meteorological factor, and the merging results are grouped by region number to output the local characteristic data of the meteorological region; Based on the local characteristic data of the meteorological region, the autoencoder neural network method is used to fuse the meteorological characteristics of each region. The dimension of the input layer is set as the number of meteorological factor types multiplied by the number of regions, the activation function is ReLU, the dimension of the encoding layer is gradually compressed to one-fourth of the input dimension, L2 regularization is used for constraint, the decoding layer reconstructs the compressed representation, the mean square error is used as the loss function, the Adam optimizer is used during the training process, the learning rate is set as 0.001, the batch size is 32, and the number of iterations is set as 100. The principal component transformation is performed on the output feature vector of the intermediate layer after fusion, and the feature dimensions with an accumulated contribution rate of the first 80% are extracted as the high-dimensional representation of the regional meteorology, and finally the meteorological region feature layer is generated.

[0025] Please refer to Figure 4 , the specific steps to generate the predicted values of the blowing snow movement trajectory are as follows: Based on the meteorological region feature layer, a convolutional neural network is used to divide the data by time interval, extract the meteorological parameters of each time period and classify them into the corresponding regions to generate a time series meteorological data set; Based on the time series meteorological data set, the meteorological parameters at different time nodes are compared, the correlation between the changes in meteorological parameters and the blowing snow trajectory is analyzed, and the correlation data between meteorological changes and the trajectory are generated through data difference detection; Based on the meteorological change and trajectory correlation data, compare the change trends of meteorological parameters and drifting snow trajectories between different time nodes to generate predicted values of drifting snow movement trajectories; Based on the meteorological regional feature layer, use the convolutional neural network method to process the meteorological data for time interval division. Set the time interval to one hour per segment, and use a three-dimensional tensor structure as the input format. The input dimension is set as the number of channels multiplied by the number of regions multiplied by the number of time steps. The number of channels corresponds to three parameters: wind speed, temperature, and humidity. The convolutional kernel size is set to 3×3, the stride is set to 1, the padding method is same, and the activation function is ReLU. The output feature map dimension of the first layer of convolution is 64, the second layer of convolution dimension is 128, the pooling kernel size of the max pooling layer is 2×2, and the pooling stride is 2. After the convolution operation, extract the wind speed, temperature, and humidity within each time period as time-dimensional vectors respectively, and classify the vectors into corresponding spatial regions according to the region index. Finally, construct a mapping table of the joint index of the region number and the time period, and the output is a time series meteorological data set; Based on the time series meteorological data set, use the difference detection algorithm based on the Laplacian distance to compare and analyze the meteorological parameters in different time periods. The input is the meteorological vectors of the same region in adjacent time periods. The algorithm judges the change trend of wind speed, temperature, and humidity between two time points by constructing an adjacency graph and calculating the Laplacian eigenvalues of each node in the graph. Use the judgment criterion that the eigenvalue difference is greater than the set threshold to mark the parameter changes. The threshold is set to the median of the 7-day moving standard deviation of each parameter. Then, compare the marked change points with the historical records of drifting snow trajectories, and use the vector cosine similarity method to evaluate the relative consistency between the change points and the trajectory morphology. Retain the associated data when the similarity is greater than 0.85. Finally, output the paired information of the region index and the time period and its similarity result to generate meteorological change and trajectory correlation data; Based on the meteorological change and trajectory correlation data, use the gated recurrent unit network method to model and analyze the meteorological parameters and trajectory change trends at different time nodes. The network structure includes a single-layer GRU unit. The input sequence length is set to 12, and the time step corresponds to the hourly unit. Each input sequence includes three parameters: wind speed, temperature, and humidity, as well as the displacement increments of the drifting snow trajectory in the x and y directions. The hidden layer dimension is set to 256, the activation function is tanh, and the update gate and reset gate are controlled by the sigmoid function respectively. The output sequence and the actual trajectory change perform mean square error loss calculation. During the training process, use the Adam optimizer, set the learning rate to 0.0005, the batch size to 64, and the number of training epochs to 200. Encode the predicted output into a prediction index according to the region and time node, and finally generate the predicted value of the drifting snow movement trajectory.

[0026] Convolutional neural network, according to the formula:

[0027] Wherein: represents the output feature of the -th layer at time step ; represents the input feature of the previous layer at time step ; represents the -th convolutional kernel weight of the -th layer; represents the bias term of the -th layer; represents the non-linear activation function; represents the time window width; represents the adjustment factor at time step ; represents the dynamic adjustment coefficient of the -th layer; Execution process: First, through each time step , calculate the weighted sum of the input feature of the previous layer and the convolutional kernel weight of the -th layer through convolution operation to obtain the convolution result, then add the result to the bias term , and process it through the non-linear activation function . Next, introduce the adjustment factor to dynamically adjust the response at time step so as to adaptively correct the movement trajectory of blowing snow in different time periods. Finally, the dynamic adjustment coefficient is applied to adjust the sensitivity of time step t according to the historical movement pattern of blowing snow, so as to accurately simulate the trajectory changes of blowing snow in each time period, and the movement trajectory of blowing snow can be more accurately simulated and tracked in the numerical model.

[0028] Please refer to Figure 5 for the specific steps to generate the multi-region blowing snow trajectory simulation data: Based on the predicted values of the blowing snow movement trajectory, spatially allocate the meteorological data, and allocate the data to each computing node according to the regional importance to generate the regional meteorological data allocation result; Based on the regional meteorological data allocation result, adjust the priority of the computing tasks according to the urgency of the blowing snow trajectory impact, and allocate more computing resources to important regions to generate the regional task priority data; Based on the regional task priority data, perform parallel processing of the optimized computing tasks on the distributed platform to generate the multi-region blowing snow trajectory simulation data; Based on the predicted values of the drifting snow movement trajectories, a data space partitioning method based on K-D Tree is used to perform a space allocation operation on meteorological data, constructing a three-dimensional space index structure, where the node coordinates use longitude, latitude, and wind speed as the three-axis coordinates. An index sequence is constructed for all samples in the input meteorological dataset in a Z-shaped scan order. The maximum capacity of each node is set to 64 records, and the splitting threshold is the median of the number of samples in the current layer. The tree structure is recursively constructed to a depth of no less than 12 layers. During the allocation process, the region number and spatial coordinates in the trajectory prediction values are used as query parameters, and each meteorological data is assigned to the target region number according to the spatial neighborhood returned by the K-nearest neighbor query, and is hashed and mapped to the specified computing node based on the region number. Finally, the regional meteorological data allocation result is output; Based on the regional meteorological data allocation result, the TOPSIS preference ranking method is used to set the regional task priorities. First, an evaluation matrix is constructed, and the column items include the trajectory displacement amplitude, the trajectory slope change rate, the historical trajectory overlap degree, and the meteorological disturbance frequency. Vector normalization operations are performed on the data in each column of the matrix, and the weight vectors are set to 0.4, 0.3, 0.2, and 0.1 respectively. The Euclidean distances between each region and the ideal optimal solution and the worst solution are calculated, and the ranking is performed according to the relative closeness. The ranking index is mapped to the task queue, and the region number with the highest priority corresponds to the task mark with the highest weight. Subsequently, a task scheduling dictionary is constructed, and the ranking result is encoded into a key-value pair structure and input into the scheduling engine to output the regional task priority data; Based on the regional task priority data, an RDD parallel computing mechanism based on Apache Spark is used to perform optimization computing tasks on a distributed platform. First, each regional task object is encapsulated into a ResilientDistributed Dataset in key-value pair format. Each record uses the region number as the Key and the task execution script as the Value. The SparkContext object is called and the number of task partitions is set to 64. The scheduling policy is set to preferentially schedule the Key marked as high. The mapPartitions function is used to batch process each task, and the maximum number of records processed in each partition is set to 100. During the task execution process, the physical constraint parameters of the drifting snow are synchronized uniformly using broadcast variables. Finally, after each node finishes execution, the output data is merged through the reduceByKey function to generate multi-regional drifting snow trajectory simulation data.

[0029] For space allocation, first, based on the regional data in the predicted values of the drifting snow movement trajectories, combined with the meteorological influence ranges of each region, the importance of each region is evaluated. And according to the evaluation results, the regional data is allocated to the corresponding computing nodes according to the influence, obtaining the regional meteorological data allocation result.

[0030] Please refer to Figure 6 , and the specific steps to generate the optimized prediction result of the drifting snow trajectory are as follows: Based on the simulated data of snow drift trajectories in multiple regions, the output results of each model are weighted using random forest, adjusted according to the influence degree of each region, and the predicted values of each region are integrated to generate the weighted results of the model; Based on the weighted results of the model, the prediction errors of each region are statistically analyzed, the error values are calculated one by one, and the regions with larger errors are marked. By comparing with the set error threshold, the regional error analysis data is generated; Based on the regional error analysis data, the performance of each model in different regions is analyzed. The most suitable model is selected by comparing the error sizes, adjusted and optimized, and the optimized prediction results of snow drift trajectories are generated; Based on the simulated data of snow drift trajectories in multiple regions, the output results of each model are weighted using the random forest regression algorithm. When constructing the model, the number of base learners is set to 100 decision trees, the maximum depth of each tree is set to 15, the minimum number of samples for splitting is 4, the feature selection method is automatic, the Bootstrap sampling strategy is used to resample the input data, the region number is used as the grouping index, and the predicted output values of each tree are recorded after the data of each group is input into the forest model for prediction. Subsequently, the regional influence factor is set according to the standard deviation of wind speed fluctuation in each region, and all the prediction results of each region are weighted according to the regional influence factor. Finally, the prediction results of each region are merged to output the weighted results of the model; Based on the weighted results of the model, the prediction errors of each region are calculated using the point-by-point error statistical method. The input parameters are the comparison pairs composed of the predicted values and the true observed values of the trajectories in each region. First, the predicted trajectories and the observed trajectories in each region are registered along the time axis, and the trajectory data is synchronized using an interpolation step of 5 minutes. Subsequently, the absolute difference between the predicted value and the observed value is calculated at each time point. The error threshold is set to 5 meters, the error values are accumulated by time period and the mean value is calculated as the regional average error. After sorting the errors of all regions, the region numbers with errors higher than the threshold are marked, and a triple data structure containing the region index, error value, and threshold status is constructed to output the regional error analysis data; Based on the regional error analysis data, the performance of multiple candidate models in each region is compared and analyzed using the model accuracy optimization algorithm. The input data is the error matrix between the predicted results and the true values of multiple models in each region. The rows of the matrix represent the model numbers, and the columns represent the region numbers. The cumulative error values of each model in different regions are statistically analyzed, the model number with the smallest error in each region is selected as the optimal model for the region, a mapping table between the region number and the optimal model number is established, the parameters of all optimal models are extracted and reconstructed, the sampling parameters, tree depth parameters, and splitting strategies are inherited and optimized, and finally all region models are retrained based on the optimized parameters and the merged final prediction results are output to generate the optimized prediction results of snow drift trajectories.

[0031] Random forest, according to the formula:

[0032] Where: represents the final weighted output result, represents the weight coefficient of the represents the predicted value of the represents the total number of regions, represents the meteorological factor weight of the represents the terrain adaptability coefficient of the represents the historical data adaptability coefficient of the Execution process: First, the predicted value of each region comes from an independent drifting snow numerical model, which simulates the movement trajectory of drifting snow within each region. Then, by introducing the weight coefficient the predicted value is initially weighted according to the influence degree of each region. Considering the importance of different regions in the simulation of the drifting snow trajectory, then the meteorological factor weight is introduced to adjust the meteorological conditions of each region, so that the influence of meteorological conditions on the simulation results can be accurately reflected. Next, the terrain adaptability coefficient is introduced to adjust the performance of drifting snow under different terrain conditions and affect the drifting snow trajectory. Finally, the historical data adaptability coefficient is introduced to adjust the influence of the stability based on regional historical data on the simulation of the drifting snow trajectory and enhance the adaptability of the model to unstable regions. The final weighted output result is the synthesis of the predicted values of all regions and can more accurately simulate and track the movement trajectory of drifting snow.

[0033] The above is only a preferred embodiment of the present invention and does not limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for simulating and tracking the trajectory of blowing snow based on a numerical model of blowing snow, characterized in that: The following steps are involved: S1: Obtain real-time meteorological data through sensors, analyze meteorological parameters in different regions, identify the time periods with greater impact on wind and snow, and calculate the wind and snow environmental impact index; S2: Based on the wind and snow environmental impact index, local spatial features of meteorological data such as wind speed, temperature, and humidity are extracted through convolution operations, and meteorological regions are stratified according to the influence of each feature to generate a meteorological region feature layer; S3: Based on the meteorological area feature layer, a convolutional neural network is used to segment the time series data, a training model is trained to compare the changes between time nodes, the correlation between meteorological parameters and trajectory changes is analyzed, and a prediction value of the wind and snow movement trajectory is generated; S4: Based on the predicted value of the wind and snow movement trajectory, the meteorological data is distributed to the distributed computing platform, the task priority is dynamically adjusted, the resources are allocated to the areas with greater impact, and the multi-region wind and snow trajectory simulation data is generated; S5: Based on the multi-regional snowstorm trajectory simulation data, random forest is used to perform weighted fusion on multiple model results, analyze the prediction error, judge the model adaptability, select the best model for tracking optimization, and generate the snowstorm trajectory optimization prediction result.

2. The method for simulating and tracking the movement trajectory of snowstorms based on a snowstorm numerical model according to claim 1, characterized in that: The specific steps of generating the wind and snow environmental impact index are as follows: Real-time meteorological data is obtained through sensors, converted into meteorological parameters, and spatially divided by region. Wind speed fluctuations, temperature fluctuations, and humidity changes in each region are calculated and mapped to corresponding time periods to generate meteorological parameter fluctuation data; Based on the meteorological parameter fluctuation data, the meteorological parameters of each region are gradually screened, the time periods related to the wind and snow phenomenon are analyzed, the wind and snow impact time periods in each region are identified, and the wind and snow impact data are generated; Based on the wind and snow period impact data, the variation range of meteorological parameters in each area is calculated, and the meteorological impact intensity in each area is combined with the time node to generate a wind and snow environmental impact index.

3. The method for simulating and tracking the movement trajectory of snowstorms based on a snowstorm numerical model according to claim 1, characterized in that: The meteorological parameters include wind speed, temperature, humidity, air pressure, precipitation, radiation, cloud cover, visibility, dew point, meteorological phenomena, atmospheric stability, wind direction, wind frequency, evaporation, snow depth, snow grain size and snow water equivalent.

4. The method for simulating and tracking the movement trajectory of snowstorms based on a snowstorm numerical model according to claim 1, characterized in that: The specific steps of generating the meteorological region characteristic layer are: Based on the wind and snow environmental impact index, the wind speed, temperature, humidity and other data of each region are extracted, and the meteorological impact weight of each region is calculated by setting a threshold and weighting according to the impact degree to generate the regional meteorological data weight; Based on the regional meteorological data weights, spatial distribution analysis is performed on the meteorological data in the region, key meteorological features of each region are extracted layer by layer, and local feature data of the meteorological region is generated; Based on the local characteristic data of the meteorological region, the meteorological characteristics of each region are integrated, multi-level characteristic mapping is performed, and a meteorological region characteristic layer is generated.

5. The method for simulating and tracking the movement trajectory of snowstorms based on a snowstorm numerical model according to claim 1, characterized in that: The specific steps of generating the predicted value of the wind and snow motion trajectory are as follows: Based on the meteorological regional feature layer, a convolutional neural network is used to divide the data by time interval, and the meteorological parameters of each time period are extracted and classified into the corresponding area to generate a time series meteorological data set; Based on the time series meteorological data set, the meteorological parameters at different time nodes are compared, the correlation between the change of meteorological parameters and the trajectory of wind and snow is analyzed, and the meteorological change and trajectory correlation data are generated through data difference detection; Based on the meteorological change and trajectory correlation data, the meteorological parameters between each time node and the change trend of the wind and snow trajectory are compared to generate a predicted value of the wind and snow movement trajectory.

6. The method for simulating and tracking the movement trajectory of snowstorms based on a snowstorm numerical model according to claim 1, characterized in that: The convolutional neural network, according to the formula: in: Indicates Layer at time step The output features of Indicates that the previous layer is at time step The input features of Indicates Tier convolution kernel weights, Indicates The bias term of the layer, represents a nonlinear activation function, represents the time window width, Indicates that at time step The adjustment factor on Indicates Dynamic adjustment factor of the layer.

7. The method for simulating and tracking the movement trajectory of snowstorms based on a snowstorm numerical model according to claim 1, characterized in that: The specific steps of generating the multi-region wind and snow trajectory simulation data are as follows: Based on the predicted value of the wind and snow movement trajectory, the meteorological data is spatially allocated, and the data is allocated to each computing node according to the regional importance to generate a regional meteorological data allocation result; Based on the regional meteorological data allocation result, the priority of the computing tasks is adjusted according to the urgency of the impact of the wind and snow track, more computing resources are allocated to important areas, and regional task priority data is generated; Based on the regional task priority data, the optimized computing tasks are processed in parallel on a distributed platform to generate multi-region wind and snow trajectory simulation data.

8. The method for simulating and tracking the movement trajectory of snowstorms based on a snowstorm numerical model according to claim 7, characterized in that: The spatial allocation first evaluates the importance of each region based on the regional data in the predicted value of the wind and snow movement trajectory, combined with the meteorological influence range of each region, and based on the evaluation results, allocates the regional data to the corresponding computing nodes according to the influence to obtain the regional meteorological data allocation result.

9. The method for simulating and tracking the movement trajectory of snowstorms based on a snowstorm numerical model according to claim 1, characterized in that: The specific steps of generating the wind and snow trajectory optimization prediction result are as follows: Based on the multi-regional wind and snow trajectory simulation data, the output results of each model are weighted using random forests, adjusted according to the degree of influence of each region, and the predicted values ​​of each region are integrated to generate a model weighted result; Based on the weighted results of the model, the prediction errors of each region are counted, the error values ​​are calculated one by one and the regions with larger errors are marked, and regional error analysis data is generated by comparing with the set error threshold; Based on the regional error analysis data, the performance of each model in different regions is analyzed, and the most suitable model is selected by comparing the error sizes. Adjustments and optimizations are performed to generate optimized prediction results for wind and snow trajectories.

10. The method for simulating and tracking the movement trajectory of snowstorms based on a snowstorm numerical model according to claim 1, characterized in that: The random forest, according to the formula: in: represents the final weighted output result, Indicates The weight coefficient of the region, Indicates The predicted value of the region, represents the total number of regions, Indicates The weight of regional meteorological factors, Indicates The terrain adaptability coefficient of the region, Indicates The historical data adaptation coefficient of the region.

Images