A road de-icing system and method

Abstract

This invention relates to the technical field of road de-icing, and discloses a road de-icing implementation system and method. The system includes a meteorological monitoring, de-icing agent spraying, and control platform. First, meteorological data is collected and road surface conditions are monitored to obtain an initial road de-icing data set; then, anomaly detection and missing data filling are performed. Next, a road de-icing prediction model is established based on a neural network to predict whether the road will freeze, outputting the road icing result. Then, de-icing work is carried out, and an enhanced Tianying optimization algorithm is introduced to optimize the spraying parameters during the de-icing process, generating a road de-icing implementation strategy. Finally, the road de-icing implementation strategy is adjusted, and a de-icing environment coupling model and alarm mechanism are established. This invention achieves the purpose of road de-icing implementation through the analysis and processing of road de-icing data, and the method is objective and accurate.

Description

Technical Field

[0001] This invention relates to the technical field of road de-icing, specifically to a road de-icing implementation system and method. Background Technology

[0002] Winter road snow and ice accumulation is a global problem threatening traffic safety and impacting traffic efficiency. Statistics show that the number of traffic accidents caused by icy roads increases by 30%-50% compared to normal weather, and can lead to regional traffic paralysis and significant economic losses. In recent years, with the increasing frequency of extreme weather events, traditional passive de-icing methods have become insufficient to meet the demands of modern transportation networks for efficient, real-time, and sustainable de-icing. Although new technologies such as carbon nanotube heating and conductive concrete are gradually emerging, their high cost, complex construction, and insufficient technological maturity limit their practical application. There is an urgent need to explore innovative de-icing solutions that balance environmental friendliness, economy, and reliability.

[0003] Currently, traditional road de-icing methods can be divided into three categories: the first is mechanical de-icing, which physically removes ice and snow using equipment such as snowplows and de-icing rollers; the second is chemical de-icing, which mainly involves spreading chloride-based de-icing agents (such as sodium chloride and calcium chloride) to accelerate the melting of ice and snow by utilizing the freezing point lowering effect; and the third is thermal de-icing, such as laying electric heating cables, ground source heat pump circulation systems, or electrically heating asphalt roads to raise the road temperature through external energy input.

[0004] Traditional road de-icing methods generally rely on repeated mechanical de-icing and timed spraying of de-icing agents. They lack intelligent control capabilities and cannot dynamically adjust de-icing strategies based on weather conditions and the slippery condition of the road surface, often resulting in energy waste and incomplete de-icing. Summary of the Invention

[0005] In view of the problems in the related technologies, the present invention provides a road de-icing system and method to overcome the technical problems existing in the prior art.

[0006] To solve the aforementioned technical problem, the present invention is achieved through the following technical solution:

[0007] This invention is a road de-icing implementation system, specifically including: meteorological monitoring, de-icing agent spraying, and a control platform;

[0008] Meteorological monitoring, including a meteorological environment monitoring module and a road surface condition monitoring module, is used to collect meteorological data and monitor road surface conditions to obtain an initial road icing data set. An adaptive window-based data detection method is used to detect anomalies in the initial road icing data set and fill in missing data to generate a processed road icing data set.

[0009] Snow melting agent spraying, including snow melting agent pumping stations, snow melting agent spraying facilities and snow melting agent auxiliary equipment;

[0010] The control platform is used for snow and ice monitoring and early warning, equipment condition monitoring, and solution spraying control.

[0011] The ice and snow monitoring and early warning system includes establishing a road melting prediction model, predicting whether the road will freeze based on the processed road melting data set, outputting the road freezing result, and segmenting the road to obtain a set of segmented melting road sections.

[0012] The equipment condition monitoring includes monitoring the operation process of the snow melting agent spraying;

[0013] The solution spraying control includes carrying out ice melting work on the ice melting sections in the segmented ice melting section set, introducing an enhanced Tianying optimization algorithm to optimize the spraying parameters in the ice melting work, generating a road ice melting implementation strategy, adjusting the road ice melting implementation strategy according to real-time meteorological data and the use of snow melting agent materials, and establishing a snow melting environment coupling model and alarm mechanism.

[0014] Preferably, the process of collecting meteorological data and monitoring road surface conditions to obtain an initial road de-icing data set includes:

[0015] Meteorological data is collected to obtain a meteorological data set, road surface conditions are monitored to obtain a road surface condition data set, and the road icing time series is recorded to obtain an initial road icing data set.

[0016] Preferably, the anomaly detection method for the initial road icing data set using an adaptive window-based data detection method includes:

[0017] Set a sliding window, dynamically and adaptively adjust the sliding window to generate an adaptive window, calculate the abnormal score of the road ice melting data in the adaptive window, and identify abnormal data;

[0018] Count the number of outliers in the current adaptive window, then calculate the ratio of the number of outliers to the current adaptive window size, and calculate the next adaptive window size;

[0019] Update the current adaptive window size according to the next adaptive window size, then calculate the abnormal score of the road icing data in the next adaptive window, until the initial road icing data set has been traversed, delete all abnormal data, and obtain the processed road icing data set.

[0020] Preferably, the process of filling in missing data includes:

[0021] When the missing data is the first or last road icing data in the adaptive window, the mean of the road icing data in the adaptive window is used to fill in the missing data; otherwise, piecewise linear interpolation is used to fill in the missing data to generate a processed road icing data set.

[0022] Preferably, the establishment of the road icing prediction model, based on the processed road icing data set, predicts whether the road will freeze, and outputs the road icing result, including:

[0023] Obtain the current road icing data set from the processed road icing data set, then obtain the road icing data sample set, set the window step size, divide according to the window step size, and standardize to obtain several road icing data sample subsets;

[0024] The road ice melting prediction model is set to include a neural network model and a support vector machine. The neural network model adopts an encoder-decoder structure. The encoder includes two LSTM layers, and the decoder includes one LSTM layer and a fully connected layer. The activation function is the Tanh function.

[0025] The road ice melting data sample subset is divided into a sample training set and a sample test set, which are then input into a neural network model for training. The prediction results are output to obtain the final road ice melting prediction model.

[0026] Set road icing rules, and mark the road melting data in the prediction results that meet the road icing rules as 1, otherwise mark it as 0;

[0027] The support vector machine is set to use a radial basis function kernel. The prediction results are divided according to the window step size and standardized. They are then input into the support vector machine for training. A maximum number of training rounds is set. Training is stopped when the maximum number of training rounds is reached, and the final support vector machine is obtained.

[0028] The current road icing data set is divided according to the window step size and standardized. It is then input into the final road icing prediction model, and the predicted values ​​of the road icing data are output. The predicted values ​​of the road icing data are marked according to the road icing rules and input into the final support vector machine. The output 1 indicates that the road is icy and the output 0 indicates that the road is not icy, thus obtaining the road icing result.

[0029] Preferably, the segmentation of the road includes:

[0030] Based on the road icing results, the road is segmented, the road segments corresponding to icing are identified and marked as melting segments, and the road segments not icing are identified and marked as non-melting segments, thus obtaining a set of segmented melting segments.

[0031] Preferably, the enhanced Skyhawk optimization algorithm for optimizing spraying parameters during ice melting includes:

[0032] Ice-melting operations are carried out on selected ice-melting sections from the set of segmented ice-melting sections. The spraying parameters during the ice-melting operation are controlled, including the pipeline pressure, ice-melting agent concentration, and spraying amount per unit area of ​​the de-icing agent solution spraying equipment. The pipeline pressure range, ice-melting agent concentration range, and minimum spraying coverage are set. The de-icing agent solution is sprayed, and the energy consumption of the de-icing agent solution spraying equipment, the ice melting time of the ice-melting section, and the uniformity of the de-icing agent solution coverage are recorded.

[0033] Establish an objective function based on minimum energy consumption and maximum ice melting efficiency;

[0034] The sky eagle optimization algorithm is further improved by combining the hill climbing algorithm and the hunting behavior of Harris eagles, resulting in an enhanced sky eagle optimization algorithm. The objective function is used as the fitness function, and a search space is set. In the search space, there exists a sky eagle population. The sky eagle position in the sky eagle population represents a set of spraying parameters. The iterative process of sky eagle position is regarded as the process of optimizing spraying parameters. The optimal fitness function value is used to simulate the optimal objective function value.

[0035] A hill-climbing algorithm is introduced to update the position of the eagle;

[0036] The eagle population has entered the high-altitude hunting and gliding hunting phase. Levy flight simulation eagle hunting was introduced to update the eagle's location.

[0037] Replace the low-flying and gliding hunting phases of the eagle population with the hunting behavior of Harris eagles, and update the eagle positions again until the current iteration count reaches the maximum iteration count, then stop iterating to obtain the final eagle population. Find the eagle position corresponding to the best fitness function value in the final eagle population to obtain the optimized spraying parameters.

[0038] Preferably, the road de-icing implementation strategy includes:

[0039] The optimized spraying parameters include the optimized pipeline pressure, de-icing agent concentration, and spraying rate per unit area of ​​the de-icing agent solution spraying equipment. The de-icing sections in the segmented de-icing section set are selected, and the optimized spraying parameters are used to drive the de-icing agent solution spraying equipment to carry out de-icing work, thus obtaining the road de-icing implementation strategy.

[0040] Preferably, the step of adjusting the road de-icing strategy based on real-time meteorological data and the usage of de-icing agent materials, and establishing a de-icing environment coupling model and alarm mechanism, includes:

[0041] Collect real-time meteorological data, set change threshold and thickness threshold. When the change rate of real-time meteorological data is greater than the change threshold, find the ice melting section corresponding to the real-time meteorological data, measure the ice thickness. When the ice thickness is greater than the ice thickness threshold, adjust the road ice melting strategy and prioritize the ice melting work of the corresponding ice melting section.

[0042] Monitor the usage of de-icing agent materials, then establish a de-icing environment coupling model to calculate the remaining amount in the storage tank; set a remaining amount threshold and a rated threshold. When the remaining amount in the storage tank is less than the remaining amount threshold, replenish the de-icing agent solution in time. When the remaining amount in the storage tank is greater than the rated threshold, trigger the alarm mechanism to complete the road de-icing.

[0043] This invention also discloses a method for road de-icing, specifically including:

[0044] S1. Collect meteorological data and monitor road surface conditions to obtain an initial road icing data set. Use an adaptive window-based data detection method to detect anomalies in the initial road icing data set and fill in missing data to generate a processed road icing data set.

[0045] S2. Establish a road icing prediction model based on neural networks. Based on the processed road icing data set, predict whether the road will freeze, output the road icing result, and segment the road to obtain a set of segmented icing road sections.

[0046] S3. Carry out de-icing work on the de-icing sections in the segmented de-icing section set, introduce the enhanced Tianying optimization algorithm to optimize the spraying parameters in the de-icing work, and generate a road de-icing implementation strategy.

[0047] S4. Adjust the road de-icing implementation strategy based on real-time meteorological data and the use of de-icing agent materials, establish a de-icing environment coupling model and alarm mechanism, and complete the road de-icing implementation.

[0048] The present invention has the following beneficial effects:

[0049] 1. This invention collects meteorological data and monitors road surface conditions, uses an adaptive window-based data detection method to detect anomalies and fill in missing data. By dynamically adjusting the window size, the window is reduced to improve detection sensitivity when the proportion of abnormal data is high, and expanded to reduce computational overhead when the data is relatively normal, thus avoiding the waste of resources caused by frequent adjustments. It achieves a balance between detection accuracy and computational efficiency, and is applicable to anomaly cleaning and filling tasks for various meteorological indicators, providing reliable input for subsequent icing prediction.

[0050] 2. This invention establishes a road icing prediction model that integrates LSTM neural networks and support vector machines, effectively combining physical rules and data-driven approaches to predict whether roads will freeze and output the road icing results. This reduces the cost of manual verification and, compared to a single neural network model, has higher accuracy and reliability as well as stronger adaptability to extreme environments, providing a flexible solution for subsequent road segmentation and icing work.

[0051] 3. This invention introduces an enhanced Eagle Optimization Algorithm to optimize spraying parameters during ice melting operations. It integrates the climbing algorithm and Harris Eagle hunting behavior, which solves the problems of easily getting trapped in local optima, slow convergence speed and poor accuracy compared with traditional algorithms. By simulating the hunting behavior of eagles, it achieves the search for the global optimal solution of spraying parameters, obtains spraying parameters suitable for ice melting sections, significantly reduces energy waste and improves the thoroughness of ice melting.

[0052] 4. This invention improves the speed of de-icing response and the level of intelligence in emergency management by adjusting the road de-icing strategy, establishing a snow melting environment coupling model and alarm mechanism, integrating dynamic adaptive adjustment and alarm mechanism, and constructing a closed-loop response system of resources, environment and equipment.

[0053] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0054] Figure 1 This invention provides an overall schematic diagram of a road de-icing system.

[0055] Figure 2 This invention provides a schematic flowchart of a road de-icing method.

[0056] Figure 3 This invention provides a schematic diagram of the modular structure of a road de-icing system. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Traditional road de-icing methods generally rely on repeated mechanical de-icing and timed spraying of de-icing agents. They lack intelligent control capabilities and cannot dynamically adjust de-icing strategies based on weather conditions and the slippery condition of the road surface, often resulting in energy waste and incomplete de-icing.

[0059] To solve the above technical problems, such as Figure 1 and Figure 3As shown, this embodiment of the invention provides a road de-icing implementation system, specifically including: a meteorological monitoring system, a de-icing agent spraying system, and a management and control platform; the meteorological monitoring system includes a meteorological environment monitoring module and a road surface condition monitoring module, used to collect meteorological data and monitor road surface conditions to obtain an initial road de-icing data set, using an adaptive window-based data detection method to perform anomaly detection on the initial road de-icing data set, and filling in missing data to generate a processed road de-icing data set; the de-icing agent spraying system includes a de-icing agent pumping station, de-icing agent spraying facilities, and de-icing agent auxiliary equipment; the management and control platform is used for ice and snow monitoring and early warning, equipment condition monitoring, and solution spraying management and control; the ice... Snow monitoring and early warning includes establishing a road icing prediction model, predicting whether roads will freeze based on processed road icing data, outputting road icing results, and segmenting roads to obtain a set of segmented icing-melting road sections. Equipment condition monitoring includes monitoring the operation of the de-icing agent spraying process. Solution spraying control includes carrying out icing work on the icing sections in the segmented icing-melting road section set, introducing an enhanced Tianying optimization algorithm to optimize spraying parameters during icing work, generating a road icing implementation strategy, adjusting the road icing implementation strategy based on real-time meteorological data and de-icing agent material usage, and establishing a snow melting environment coupling model and alarm mechanism.

[0060] In a specific embodiment, a meteorological environmental sensor is used to automatically monitor comprehensive meteorological environmental information such as temperature, humidity, pressure, and wind speed. A road surface condition sensor is set up to collect road data such as road slipperiness, road humidity, and road water accumulation. The de-icing agent solution spraying equipment consists of an integrated intelligent device, a liquid storage unit, an intelligent spraying unit, and an electrical system. It controls the automatic spraying of the de-icing agent solution, providing real-time meteorological and road surface condition monitoring data for the embodiments of the present invention, which helps in statistical analysis and decision support.

[0061] In the specific implementation process of the above embodiments, firstly, meteorological data is collected and road surface conditions are monitored to obtain an initial road icing data set. An anomaly detection method based on an adaptive window is used to detect anomalies and fill in missing data, resulting in a processed road icing data set. This method dynamically adjusts the window size; when the proportion of abnormal data is high, the window is reduced to improve detection sensitivity, and when the data is relatively normal, the window is expanded to reduce computational overhead, avoiding resource waste caused by frequent adjustments. This achieves a balance between detection accuracy and computational efficiency, and is suitable for anomaly cleaning and filling tasks for various meteorological indicators, providing reliable input for subsequent icing prediction. Secondly, a road icing prediction model is established. Based on the processed road icing data set, it predicts whether the road will ic up, outputs the road icing result, and segments the road. This method integrates LSTM neural networks and support vector machines, effectively combining physical rules and data-driven approaches, reducing the cost of manual review. Compared to a single neural network model, this method offers higher accuracy and reliability, as well as stronger adaptability to extreme environments. Segmented processing provides a flexible solution for subsequent ice-melting operations. Ice-melting is then carried out on segmented roads, and an enhanced eagle optimization algorithm is introduced to optimize spraying parameters, generating a road ice-melting implementation strategy. This algorithm integrates hill-climbing algorithms and Harris eagle hunting behavior, solving the problems of getting trapped in local optima, slow convergence speed, and poor accuracy compared to traditional algorithms. By simulating eagle hunting behavior, it achieves a global optimal solution search for spraying parameters, obtaining suitable spraying parameters for ice-melting sections, significantly reducing energy waste and improving the thoroughness of ice melting. Finally, the road ice-melting implementation strategy is adjusted, establishing a snow-melting environment coupling model and alarm mechanism. By integrating dynamic adaptive adjustment and alarm mechanisms, a closed-loop response system of resources, environment, and equipment is constructed, improving the ice-melting response speed and the intelligence level of emergency management under extreme weather conditions, thus completing the road ice-melting implementation.

[0062] Furthermore, to better illustrate the technical solution of the embodiments of the present invention, based on the above-mentioned road de-icing system, such as... Figure 2 As shown in the figure, this embodiment of the invention provides a method for road de-icing, which specifically includes the following:

[0063] S1. Collect meteorological data and monitor road surface conditions to obtain an initial road icing data set. Use an adaptive window-based data detection method to detect anomalies in the initial road icing data set and fill in missing data to generate a processed road icing data set.

[0064] S1 includes the following steps:

[0065] S11. Collect meteorological data using multi-element meteorological environmental sensors, including temperature, humidity, pressure, wind speed, etc., to obtain a meteorological data set; then monitor road surface conditions, including road slipperiness, road humidity, road water accumulation, etc., to obtain a road surface condition data set; and record the time points for collecting meteorological data and road surface conditions to generate a road icing time series. Combine the road icing time series, the meteorological data set, and the road surface condition data set to form an initial road icing data set. ,in This represents the m-th time point in the road ice melting time series. This represents the meteorological data collected at the m-th time point. This represents the road surface condition collected at the m-th time point;

[0066] S12. Set a sliding window, dynamically and adaptively adjust the sliding window to generate an adaptive window, use a data detection method based on the adaptive window to perform anomaly detection on the initial road icing data set, and delete abnormal data to obtain the processed road icing data set. The specific steps are as follows:

[0067] S121. Place the adaptive window at the beginning of the initial road icing data set. Calculate the mean and variance of the road icing data within the adaptive window, and then calculate the anomaly score for the nth road icing data point. ,in This represents the ice melting data for the nth road. This represents the average value of road ice melt data within the adaptive window. This represents the variance of road icing data within the adaptive window; an anomaly threshold is set, and when the anomaly score of the nth road icing data exceeds the anomaly threshold, the nth road icing data is considered an anomaly.

[0068] S122. Set the current adaptive window size to w, count the number of outliers within the current adaptive window, and then calculate the ratio of the number of outliers to the current adaptive window size to obtain the current outlier ratio; set the target outlier ratio, and then calculate the next adaptive window size. ,in This represents the window adjustment factor, set to 0.1. Indicates the target abnormality ratio. Indicates the current abnormality rate;

[0069] S123. Move the current adaptive window and update the current adaptive window size according to the next adaptive window size. Then calculate the abnormal score of the road melting data in the next adaptive window and mark the location of the abnormal data. Continue to update the adaptive window size until the initial road melting data set has been traversed. Delete all abnormal data to obtain the processed road melting data set.

[0070] S13. When the missing data is the first or wth road icing data in the adaptive window, the mean of the road icing data in the adaptive window is used to fill in the missing data; otherwise, piecewise linear interpolation is used to fill in the missing data, selecting the road icing data before and after the missing data. The missing data filling value is then... ,in This indicates the road icing data prior to the missing data. This indicates road icing data after data loss. This indicates the time point corresponding to the road icing data before the missing data. This indicates the time point corresponding to the missing road icing data. The missing data points are indicated by the time points corresponding to the missing data; the missing data in the processed road icing data set are filled in sequentially to generate a processed road icing data set.

[0071] In this embodiment, an initial road icing data set is obtained by collecting meteorological data and monitoring road surface conditions. An anomaly detection method based on an adaptive window is used to detect anomalies and fill in missing data to obtain a processed road icing data set. This method dynamically adjusts the window size: when the proportion of data anomalies is high, the window is reduced to improve detection sensitivity; when the data is relatively normal, the window is expanded to reduce computational overhead, avoiding the waste of resources caused by frequent adjustments. It achieves a balance between detection accuracy and computational efficiency and is suitable for anomaly cleaning and filling tasks for various meteorological indicators, providing reliable input for subsequent icing prediction. Specifically, for example, the data set can be simplified by recording only time points. Temperature and road slipperiness are considered, {(t1, (5℃, 60%)), (t2, (20℃, 20%)), (t3, (4℃, 65%)}. The window size is set to 3. The abnormal score for t1 is 0, the abnormal score for t2 is 1.16, and the abnormal score for t3 is -1.22. The abnormal threshold is 1.0, the abnormal ratio is 1 / 3, the target abnormal ratio is 10%, and the window adjustment coefficient is 0.1. At this point, the window size is adjusted to 3, and the abnormal data t2 is deleted. Missing data is filled using piecewise linear interpolation. The time interval between the preceding data of t2 is 5 minutes, and the time interval between the following data of t2 is 5 minutes. The filling temperature is 5℃. The filling is performed sequentially.

[0072] S2. Establish a road icing prediction model based on neural networks. Based on the processed road icing data set, predict whether the road will freeze, output the road icing result, and segment the road to obtain a set of segmented icing road sections.

[0073] S2 includes the following steps:

[0074] S21. Obtain temperature, humidity, pressure, wind speed, road humidity, and road water accumulation from the processed road de-icing data set to form the current road de-icing data set; then obtain meteorological data, road surface conditions, and collection time points from previous years to form a road de-icing data sample set; set a window step size, divide the collection time points in the road de-icing data sample set according to the window step size, and standardize the data to obtain several road de-icing data sample subsets;

[0075] S22. Establish a road icing prediction model based on LSTM neural network and support vector machine to predict the probability of road icing, output the road icing probability, and determine whether the road is icy. The specific steps are as follows:

[0076] S221. The road ice melting prediction model is set to include a neural network model and a support vector machine. The neural network model adopts an encoder-decoder structure. The encoder includes two LSTM layers, and the decoder includes one LSTM layer and a fully connected layer. The activation function is the Tanh function.

[0077] The road icing data sample subset is divided into a sample training set and a sample test set. The sample training set is input into the input layer of the neural network model for training until the neural network model converges, resulting in a trained neural network model. The sample test set is then input into the trained neural network model to output the prediction result. An accuracy threshold is set. When the accuracy of the prediction result is greater than the accuracy threshold, the final road icing prediction model is obtained; otherwise, the weights are adjusted until the accuracy of the prediction result is greater than the accuracy threshold.

[0078] S323. Set road humidity threshold and road water accumulation threshold, and set road icing rules; when the temperature in the prediction result is less than 0 degrees Celsius and the road humidity is greater than the road humidity threshold, it is determined that the road is icing; when the road water accumulation in the prediction result is greater than the road water accumulation threshold and the temperature is less than 0 degrees Celsius, it is determined that the road is icing; mark the road melting data in the prediction result that meets the road icing rules as 1, otherwise mark it as 0;

[0079] S33. Set the support vector machine to use the radial basis kernel function. The prediction results are divided according to the window step size and standardized. The results are then input into the support vector machine for training. Set the maximum number of training rounds. When the maximum number of training rounds is reached, training is stopped to obtain the final support vector machine.

[0080] The current road icing data set is divided according to the window step size and standardized. It is then input into the final road icing prediction model to output the predicted values ​​of road icing data. The predicted values ​​of road icing data are then marked according to the road icing rules and input into the final support vector machine to determine whether the road is icy. The output 1 indicates that the road is icy and the output 0 indicates that the road is not icy, thus obtaining the road icing result.

[0081] S23. Based on the road icing results, the road is segmented, the road segments corresponding to the icing are found and marked as melting road segments, and the road segments not icing are found and marked as non-melting road segments, thus obtaining a set of segmented melting road segments.

[0082] In this embodiment, a road icing prediction model is established. Based on the processed road icing data set, it predicts whether roads will freeze, outputs the road icing result, and segments the roads. This method integrates LSTM neural networks and support vector machines, effectively combining physical rules and data-driven approaches, reducing manual verification costs. Compared to a single neural network model, it has higher accuracy and reliability, and stronger adaptability to extreme environments. Segmented processing provides a flexible solution for subsequent icing work. Specifically, for example, with a window size of 2 hours, data from 6 consecutive hours is used as a sample subset. Features such as temperature (range -25℃ to 10℃) and humidity (0-100%) are Z-score standardized to make the mean 0. The standard deviation is 1; the encoder is a 2-layer LSTM with 64 neurons per layer, processing the input sequence (6 hours × 6 features); the decoder is a 1-layer LSTM (32 neurons) + a fully connected layer (outputting 6 feature predictions), with the activation function being Tanh; the icing rule is set as follows: condition 1: temperature ≤ 0℃ and road humidity ≥ 85%; condition 2: water accumulation ≥ 3mm and temperature ≤ 0℃. If either condition is met, the road is marked as icy (label 1); otherwise, the label is 0; the SVM classifier uses the RBF kernel function, with the predicted values ​​output by the LSTM as features; combined with the GIS map, the icy road sections (label 1) and normal road sections (label 0) are divided at 500-meter intervals, and the prediction results are updated. The segment set is dynamically adjusted to improve the de-icing efficiency;

[0083] S3. Carry out de-icing work on the de-icing sections in the segmented de-icing section set, introduce the enhanced Tianying optimization algorithm to optimize the spraying parameters in the de-icing work, and generate a road de-icing implementation strategy.

[0084] S3 includes the following steps:

[0085] S31. Select the ice-melting sections from the set of segmented ice-melting sections for ice-melting work, control the spraying parameters during the ice-melting work, including the pipeline pressure, ice-melting agent concentration, and spraying amount per unit area of ​​the snow-melting agent solution spraying equipment; and set the pipeline pressure range, ice-melting agent concentration range, and minimum spraying coverage rate, spray the snow-melting agent solution, and then record the energy consumption of the snow-melting agent solution spraying equipment, the ice melting time of the ice-melting section, and the uniformity of the snow-melting agent solution coverage;

[0086] S32. Establish an objective function based on minimum energy consumption and maximum ice melting efficiency. ,in , and Indicates the weighting coefficient. This indicates the energy consumption of the de-icing agent solution spraying equipment. Indicates the time it takes for the ice to melt on the road section. Indicates the uniformity of the snow-melting agent solution's coverage;

[0087] The Skyhawk Optimization Algorithm is further improved by integrating the hill-climbing algorithm and the hunting behavior of Harris eagles, resulting in an enhanced Skyhawk Optimization Algorithm. This enhanced algorithm is then used to optimize the spraying parameters, yielding the optimized spraying parameters. The specific steps are as follows:

[0088] S321. Using the objective function as the fitness function, a search space is defined. A population of eagles exists within this search space, and the eagle positions within the population represent a set of spraying parameters. The iterative process of eagle positions is considered as an optimization process for the spraying parameters. The optimal fitness function value is used to simulate the optimal objective function value. The number of eagles in the population is set to p, and the dimension of the eagle population is set to q. The eagle population is initialized, the current iteration number is t, and the maximum iteration number is T. A hill-climbing algorithm is introduced, and the position of the i-th eagle in the t-th iteration is set to... Distance position Nearest location ,Location Arrive at the location The probability is , Represents a random number within the interval [0, 1]; when hour, Otherwise, position constant;

[0089] S322. The Eagle population enters the high-altitude hunting and gliding hunting phase. The average position of the Eagle population at iteration t is set as follows: The optimal position is At this point, calculate the position of the i-th eagle in the (t+1)-th iteration. Introducing the Levi flight simulation for Eagle Hunting, setting the Levi flight coefficient to [value missing]. The Gaussian distribution coefficient is and , regarding position Update and get ;

[0090] S323, Replace the low-flying and gliding hunting phases of the eagle population with the hunting behavior of the Harris Eagle, and set... This represents a random number within the interval [0, 1]. , , and Represents a random number between the interval [0, q], when At that time, the position of the j-th eagle is randomly selected from the eagle population. Again, regarding the position Update and get ;when At that time, the upper and lower bounds of the search space are set as follows: and Again, regarding the position Update and get At this point, all stages of the t-th iteration are completed, and the next iteration begins. This continues until the current iteration count reaches the maximum iteration count, at which point the iteration stops, and the final eagle population is obtained. The location of the eagle corresponding to the best fitness function value in the final eagle population is then found, and the optimized spraying parameters are obtained.

[0091] S33. The optimized spraying parameters include the optimized pipeline pressure, de-icing agent concentration, and spraying amount per unit area of ​​the de-icing agent solution spraying equipment. The de-icing sections in the segmented de-icing section set are selected, and the optimized spraying parameters are used to drive the de-icing agent solution spraying equipment to carry out de-icing work, thereby obtaining the road de-icing implementation strategy.

[0092] In this embodiment, ice melting is carried out on segmented roads. An enhanced Hawk optimization algorithm is introduced to optimize the spraying parameters during the ice melting process, generating a road ice melting implementation strategy. This algorithm integrates the hill-climbing algorithm and Harris eagle hunting behavior, solving the problems of easily getting trapped in local optima, slow convergence speed, and poor accuracy compared to traditional algorithms. By simulating the hunting behavior of eagles, the algorithm achieves the search for the globally optimal solution of spraying parameters, obtaining spraying parameters suitable for the ice melting sections. Specifically, for example, the pipeline pressure range is set to [0.5, 2.5] MPa, the de-icing agent concentration range is [20%, 40%], the minimum spraying coverage is 90%, and the Hawk algorithm parameters are initialized as follows: population size 50, dimension 3 (corresponding to 3 optimization parameters), maximum number of iterations is 100, and the objective function weight coefficients are 0.4 (energy consumption) and 0.3 (melting time). 0.3 (Coverage); 50 initial parameter combinations are randomly generated, and a hill-climbing algorithm is used for local search. The position is updated during the high-altitude gliding phase of the Skyhawk, and then the Levy flight is introduced to enhance the global search capability, resulting in jump-type parameter adjustments. The position is updated again through the Harris Eagle hunting mechanism. After 100 iterations, the algorithm converges to obtain the optimal parameter combination, resulting in pipeline pressure: 1.8MPa (20% reduction in energy consumption compared to the initial value), de-icing agent concentration: 30% (the minimum effective concentration to meet antifreeze requirements, reducing costs), and unit area spraying rate: 0.12L / m² (coverage increased to 95%, improving ice melting efficiency). By optimizing the algorithm to search for the global optimal solution of spraying parameters, energy waste is significantly reduced and the thoroughness of ice melting is improved. The optimized parameters are input into the automated spraying control system to achieve intelligent and precise ice melting operations.

[0093] S4. Adjust the road de-icing implementation strategy based on real-time meteorological data and the use of de-icing agent materials, establish a de-icing environment coupling model and alarm mechanism, and complete the road de-icing implementation;

[0094] S4 includes the following steps:

[0095] S41. Collect real-time meteorological data, set change threshold and thickness threshold. When the change rate of real-time meteorological data is greater than the change threshold, find the ice melting section corresponding to the real-time meteorological data, measure the ice thickness. When the ice thickness is greater than the ice threshold, adjust the road ice melting strategy and prioritize the ice melting work of the ice melting section corresponding to the real-time meteorological data.

[0096] S42. Monitor the usage of snow-melting agent materials, including snow-melting agent concentration and consumption per unit area, and establish a snow-melting environment coupling model to calculate the remaining amount in the storage tank. Where U represents the initial volume of the storage tank, and k represents the concentration correction factor. This represents the area of ​​the e-th icing segment in the set of segmented icing segments. This represents the unit area consumption of the e-th ice-melting section, and s represents the number of ice-melting sections in the set. Set a reserve threshold. When the reserve in the storage tank is less than the reserve threshold, replenish the de-icing agent solution in time. When the reserve in the storage tank is greater than the rated threshold, trigger the alarm mechanism to complete the road de-icing.

[0097] In this embodiment, the road de-icing strategy is adjusted, a snow melting environment coupling model and alarm mechanism are established, and dynamic adaptive adjustment and alarm mechanisms are integrated to construct a closed-loop response system of resources, environment and equipment. Specifically, for example, the meteorological change threshold is set to 20% / hour, the ice thickness threshold is 5mm, and the system monitors that the temperature in a certain area drops sharply from -2℃ to -8℃ in 1 hour, the total span of the reference temperature range is 10℃, and the change rate is 6℃ / 10℃ = 60% / hour. The corresponding road segment is located: Highway K12-K15 (the road segment areas are 800m², 1000m² and 120m² respectively, and the unit consumption is... The concentrations were 0.15 L / m², 0.18 L / m², and 0.20 L / m², respectively. The ice thickness was measured to be 6.2 mm (exceeding the 5 mm threshold). The system automatically upgraded this road section to a higher priority. The concentration correction coefficient k=1.1 (due to the low temperature, the concentration needs to be increased based on standard 1). The initial storage volume was 5000 L. The calculated remaining volume of the storage tank was 440 L (less than the remaining volume threshold of 500 L). The de-icing agent material was replenished in time. After replenishment, the remaining de-icing agent volume was 4000 L, which was higher than the rated threshold of 2000 L. This improved the de-icing response speed and the level of intelligence in emergency management under extreme weather conditions, and completed the road de-icing implementation.

[0098] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0099] The preferred embodiments of the invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A road de-icing system, characterized in that, include: Meteorological monitoring, including a meteorological environment monitoring module and a road surface condition monitoring module, is used to collect meteorological data and monitor road surface conditions to obtain an initial road icing data set. An adaptive window-based data detection method is used to detect anomalies in the initial road icing data set and fill in missing data to generate a processed road icing data set. Snow melting agent spraying, including snow melting agent pumping stations, snow melting agent spraying facilities and snow melting agent auxiliary equipment; The control platform is used for snow and ice monitoring and early warning, equipment condition monitoring, and solution spraying control. The ice and snow monitoring and early warning system includes establishing a road melting prediction model, predicting whether the road will freeze based on the processed road melting data set, outputting the road freezing result, and segmenting the road to obtain a set of segmented melting road sections. The equipment condition monitoring includes monitoring the operation process of the snow melting agent spraying; The solution spraying control includes carrying out ice melting work on the ice melting sections in the segmented ice melting section set, introducing an enhanced Tianying optimization algorithm to optimize the spraying parameters in the ice melting work, generating a road ice melting implementation strategy, adjusting the road ice melting implementation strategy according to real-time meteorological data and the use of snow melting agent materials, and establishing a snow melting environment coupling model and alarm mechanism. The enhanced Skyhawk optimization algorithm introduced to optimize spraying parameters in ice melting operations includes: Ice-melting operations are carried out on selected ice-melting sections from the set of segmented ice-melting sections. The spraying parameters during the ice-melting operation are controlled, including the pipeline pressure, ice-melting agent concentration, and spraying amount per unit area of ​​the de-icing agent solution spraying equipment. The pipeline pressure range, ice-melting agent concentration range, and minimum spraying coverage are set. The de-icing agent solution is sprayed, and the energy consumption of the de-icing agent solution spraying equipment, the ice melting time of the ice-melting section, and the uniformity of the de-icing agent solution coverage are recorded. Establish an objective function based on minimum energy consumption and maximum ice melting efficiency; The sky eagle optimization algorithm is further improved by combining the hill climbing algorithm and the hunting behavior of Harris eagles, resulting in an enhanced sky eagle optimization algorithm. The objective function is used as the fitness function, and a search space is set. In the search space, there exists a sky eagle population. The sky eagle position in the sky eagle population represents a set of spraying parameters. The iterative process of sky eagle position is regarded as the process of optimizing spraying parameters. The optimal fitness function value is used to simulate the optimal objective function value. A hill-climbing algorithm is introduced to update the position of the eagle; The eagle population has entered the high-altitude hunting and gliding hunting phase. Levy flight simulation eagle hunting was introduced to update the eagle's location. Replace the low-flying and gliding hunting phases of the eagle population with the hunting behavior of Harris eagles, update the eagle positions again until the current iteration count reaches the maximum iteration count, stop the iteration, obtain the final eagle population, find the eagle position corresponding to the best fitness function value in the final eagle population, and obtain the optimized spraying parameters. The road de-icing implementation strategy includes: The optimized spraying parameters include the optimized pipeline pressure, de-icing agent concentration, and spraying rate per unit area of ​​the de-icing agent solution spraying equipment. The de-icing sections in the segmented de-icing section set are selected, and the optimized spraying parameters are used to drive the de-icing agent solution spraying equipment to carry out de-icing work, thus obtaining the road de-icing implementation strategy.

2. The road de-icing system according to claim 1, characterized in that, The process of collecting meteorological data and monitoring road surface conditions to obtain the initial road de-icing data set includes: Meteorological data is collected to obtain a meteorological data set, road surface conditions are monitored to obtain a road surface condition data set, and the road icing time series is recorded to obtain an initial road icing data set.

3. The road de-icing system according to claim 2, characterized in that, The anomaly detection method based on adaptive windows for the initial road icing melt data set includes: Set a sliding window, dynamically and adaptively adjust the sliding window to generate an adaptive window, calculate the abnormal score of the road ice melting data in the adaptive window, and identify abnormal data; Count the number of outliers in the current adaptive window, then calculate the ratio of the number of outliers to the current adaptive window size, and calculate the next adaptive window size; Update the current adaptive window size according to the next adaptive window size, then calculate the abnormal score of the road icing data in the next adaptive window, until the initial road icing data set has been traversed, delete all abnormal data, and obtain the processed road icing data set.

4. The road de-icing system according to claim 3, characterized in that, The filling of missing data includes: When the missing data is the first or last road icing data in the adaptive window, the mean of the road icing data in the adaptive window is used to fill in the missing data; otherwise, piecewise linear interpolation is used to fill in the missing data to generate a processed road icing data set.

5. A road de-icing system according to claim 4, characterized in that, The road icing prediction model is established by predicting whether roads will freeze based on the processed road icing data set, and the output road icing results include: Obtain the current road icing data set from the processed road icing data set, then obtain the road icing data sample set, set the window step size, divide according to the window step size, and standardize to obtain several road icing data sample subsets; The road ice melting prediction model is set to include a neural network model and a support vector machine. The neural network model adopts an encoder-decoder structure. The encoder includes two LSTM layers, and the decoder includes one LSTM layer and a fully connected layer. The activation function is the Tanh function. The road ice melting data sample subset is divided into a sample training set and a sample test set, which are then input into a neural network model for training. The prediction results are output to obtain the final road ice melting prediction model. Set road icing rules, and mark the road melting data in the prediction results that meet the road icing rules as 1, otherwise mark it as 0; The support vector machine is set to use a radial basis function kernel. The prediction results are divided according to the window step size and standardized. They are then input into the support vector machine for training. A maximum number of training rounds is set. Training is stopped when the maximum number of training rounds is reached, and the final support vector machine is obtained. The current road icing data set is divided according to the window step size and standardized. It is then input into the final road icing prediction model, and the predicted values ​​of the road icing data are output. The predicted values ​​of the road icing data are marked according to the road icing rules and input into the final support vector machine. The output 1 indicates that the road is icy and the output 0 indicates that the road is not icy, thus obtaining the road icing result.

6. A road de-icing system according to claim 5, characterized in that, The segmentation of the road includes: Based on the road icing results, the road is segmented, the road segments corresponding to icing are identified and marked as melting segments, and the road segments not icing are identified and marked as non-melting segments, thus obtaining a set of segmented melting segments.

7. A road de-icing system according to claim 6, characterized in that, The adjustment of the road de-icing strategy based on real-time meteorological data and the usage of de-icing agent materials, and the establishment of a de-icing environment coupling model and alarm mechanism, include: Collect real-time meteorological data, set change threshold and thickness threshold. When the change rate of real-time meteorological data is greater than the change threshold, find the ice melting section corresponding to the real-time meteorological data, measure the ice thickness. When the ice thickness is greater than the ice thickness threshold, adjust the road ice melting strategy and prioritize the ice melting work of the corresponding ice melting section. Monitor the usage of de-icing agent materials, then establish a de-icing environment coupling model to calculate the remaining amount in the storage tank; set a remaining amount threshold and a rated threshold. When the remaining amount in the storage tank is less than the remaining amount threshold, replenish the de-icing agent solution in time. When the remaining amount in the storage tank is greater than the rated threshold, trigger the alarm mechanism to complete the road de-icing.

8. A method for road de-icing, characterized in that, The system applied to the road de-icing implementation system as described in any one of claims 1-7 specifically includes: S1. Collect meteorological data and monitor road surface conditions to obtain an initial road icing data set. Use an adaptive window-based data detection method to detect anomalies in the initial road icing data set and fill in missing data to generate a processed road icing data set. S2. Establish a road icing prediction model based on neural networks. Based on the processed road icing data set, predict whether the road will freeze, output the road icing result, and segment the road to obtain a set of segmented icing road sections. S3. Carry out de-icing work on the de-icing sections in the segmented de-icing section set, introduce the enhanced Tianying optimization algorithm to optimize the spraying parameters in the de-icing work, and generate a road de-icing implementation strategy. S4. Adjust the road de-icing implementation strategy based on real-time meteorological data and the use of de-icing agent materials, establish a de-icing environment coupling model and alarm mechanism, and complete the road de-icing implementation.

Images