Snow melting early warning and intelligent snow removing scheduling method for ice and snow field

Abstract

The invention provides an ice and snow field snow melting early warning and intelligent snow removal scheduling method, which comprises the following steps: acquiring wet snow water content, snow layer density abnormity and field climate difference data through a multi-source sensor; the method comprises the following steps: performing primary processing on original data acquired by radar detection, infrared imaging and a micro weather station to obtain an initial distribution map of wet snow characteristics and local weather conditions; and according to the corrected wet snow characteristic distribution map, in combination with the initial distribution map and temperature influence factors and melting rate fluctuation characteristics obtained by data analysis of the micro meteorological station, aiming at equipment thrust loss and working surface friction loss, correcting acquisition deviation of multi-source sensor data, and obtaining a dynamic update map of the snow accumulation state in the risk area.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method for early warning of snowmelt and intelligent snow removal scheduling on an ice and snow field. Background Art

[0002] The safety and operational efficiency of snow and ice venues are of vital strategic importance in cold regions, directly impacting public safety, transportation, and the smooth running of major events. Ensuring that snow accumulation-free venues during extreme weather conditions has become a core research focus. However, existing methods often struggle to fully account for the dynamic changes in multiple data sources when addressing complex weather and site environments. They also lack in-depth analysis of the interactions between weather microscopic characteristics and site physical conditions. This is particularly true in data integration and real-time response, resulting in often delayed or inaccurate early warning and operation scheduling. A key challenge in this area lies in the complex coupling between snowfall characteristics and site conditions. First, as snowfall transitions from dry to wet, the increased water content of snow crystals significantly alters the physical properties of the snow, reducing heat absorption efficiency and increasing the energy consumption and difficulty of snowmelt operations. A further challenge is that the increased viscosity of wet snow directly leads to greater resistance for snow removal equipment during operation, potentially causing unstable equipment operation or uneven snow accumulation. This uneven distribution further exacerbates differences in snow density under the influence of wind and snow, exacerbating snow accumulation in certain areas of the site and creating a dual dilemma for early warning and operation scheduling. Therefore, a key issue that needs to be addressed is how to scientifically adjust the fusion method and time calibration mechanism of multi-source data to accurately predict snow accumulation trends and optimize snow removal operation paths, given the dynamic changes in snowfall characteristics and the interaction between site physical conditions. Summary of the Invention

[0003] The present invention provides a method for early warning of snowmelt and intelligent snow removal scheduling in an ice and snow field, which mainly includes: Multi-source sensors were used to collect data on wet snow moisture content, snow density anomalies, and site climate differences. The raw data collected by radar detection, infrared imaging, and micro-meteorological stations were preliminarily processed to identify critical moisture content points and uneven water distribution, resulting in an initial distribution map of wet snow characteristics and local climate conditions. Based on the spatiotemporal variation of wet snow moisture content and snow density anomalies in the initial distribution map, combined with the snow density gradient and accumulation unevenness index, anomalies of low snowmelt efficiency and increased viscosity are detected. If the critical point of moisture content exceeds the preset threshold or the uneven distribution of moisture increases, the initial distribution map is analyzed to identify potential risk areas for snowmelt and snow removal operations. The system continuously monitors the moisture content of wet snow in potential risk areas using radar and infrared sensors, analyzing changes in particle cohesion and surface friction coefficient. If the moisture content of the wet snow exceeds a preset threshold or particle cohesion increases, the contribution ratio of the radar and infrared sensor data is corrected in real time to obtain a corrected wet snow characteristic distribution map. If the threshold is not exceeded, the current weight is maintained and monitoring continues. Based on the corrected wet snow characteristic distribution map, combined with the initial distribution map and the temperature influencing factors and melting rate fluctuation characteristics obtained through analysis of micro-meteorological station data, the acquisition deviation of multi-source sensor data was corrected for equipment thrust loss and working surface friction loss, resulting in a dynamically updated map of snow accumulation status in the risk area. Based on the changes in snow accumulation in risk areas in the dynamically updated atlas, combined with the peak viscosity and local accumulation thickness in the calibrated wet snow characteristic distribution map, the impact of resistance accumulation rate and path obstruction frequency on snow removal operations was analyzed. Constraints for snow removal path planning were constructed based on the impact analysis results, and the resistance distribution of snow removal equipment in different areas was determined based on the constraints. If the resistance distribution shows that the accumulated rate of snow removal resistance in a certain area exceeds the preset threshold or the peak load of the equipment exceeds the preset safety range, the operating route of the snow removal equipment will be optimized in real time to avoid areas with excessive accumulation thickness. An optimized snow removal operation scheduling plan will be generated based on the adjusted operating route. If the resistance does not exceed the safety range, the existing route will be maintained and the friction loss data of the working surface will be recorded. According to the optimized runway snow removal operation scheduling plan, combined with the records of uneven moisture distribution and snow layer compression rate in the historical wet snow characteristics data, the changing trends of path obstruction frequency and uneven accumulation index are analyzed. The trend analysis results are input into the pre-established prediction model, and the trend changes of snow melting efficiency and uneven snow distribution around the runway within the preset time in the future are output to dynamically adjust the optimized snow removal operation scheduling plan.

[0004] The technical solution provided by the embodiment of the present invention may have the following beneficial effects: The present invention discloses a method for early warning of snowmelt and intelligent snow removal scheduling in ice and snow venues. The method collects data such as wet snow moisture content and snow layer density through multi-source sensors, constructs an initial distribution map, performs anomaly detection on low snowmelt efficiency and enhanced viscosity, and determines potential risk areas. The method continuously monitors changes in wet snow characteristics in risk areas, dynamically adjusts data fusion weights, and obtains a corrected distribution map. Combining temperature influencing factors and melting rate fluctuation characteristics, a time series compensation method is used to correct data deviations and update snow accumulation status. The resistance accumulation rate and path obstruction frequency are analyzed to construct snow removal path planning constraints and optimize the operation route in real time. According to the trend changes analyzed by historical data, the snow removal operation scheduling plan is dynamically adjusted to improve the efficiency and safety of runway snow removal operations in wet snow environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] Figure 1 This is a flow chart of a method for early warning of snowmelt and intelligent snow removal scheduling for an ice and snow venue according to the present invention. DETAILED DESCRIPTION

[0006] The following will describe the technical solutions in the embodiments of the present invention in detail with reference to the accompanying drawings. The described embodiments are only a part of the embodiments of the present invention.

[0007] like Figure 1 In this embodiment, a method for early warning of snowmelt and intelligent snow removal scheduling at an ice and snow field may specifically include: In step S101, multi-source sensors are used to collect data on wet snow moisture content, snow layer density anomalies, and site climate differences. Based on the moisture content critical point and uneven moisture distribution characteristics, the raw data collected by radar detection, infrared imaging, and micro-meteorological stations are preliminarily processed to obtain an initial distribution map of wet snow characteristics and local climate conditions.

[0008] By analyzing the differences in the reflection intensity of electromagnetic waves emitted by radar detectors in wet snow layers with different moisture contents, the reflection coefficient distribution data of the vertical profile of the snow layer is obtained. A preset comparison table of reflection coefficient and moisture percentage is used to determine the moisture percentage of each depth layer. Infrared imaging equipment is used to capture temperature radiation images of the surface and shallow layers of the snow layer. For abnormal points where the temperature value exceeds the preset threshold of the surrounding area, the temperature gradient change characteristics are extracted from the infrared image to determine the spatial coordinates of areas with uneven moisture distribution. Based on the wind speed, air temperature, and humidity parameter sequences collected by the micro-meteorological station, combined with the moisture percentage values ​​of each depth layer obtained by radar detection and the location coordinates of areas with uneven moisture distribution determined by infrared imaging, the radar data, infrared data, and meteorological data are subjected to dimensionality reduction and feature fusion using principal component analysis to obtain an initial distribution map of wet snow characteristics and local climate conditions.

[0009] Specifically, the application of radar detectors in wet snow moisture detection relies on the differences in the propagation characteristics of electromagnetic waves in different media. When electromagnetic waves emitted by the radar penetrate the snow layer, the dielectric constant of liquid water is much higher than that of ice crystals and air. Snow layers with different moisture contents produce reflected signals of varying intensities.

[0010] Specifically, the radar system transmits electromagnetic pulses of a specific frequency and records the time delay and amplitude attenuation of the reflected waves, thereby obtaining the reflection coefficient distribution of the vertical profile of the snow layer. A comparison table of reflection coefficient and moisture percentage is empirical data obtained through extensive laboratory calibration. Generally, for every 1% increase in moisture content, the reflection coefficient increases by a specific value. This correlation makes it possible to directly infer moisture content from radar signals. The introduction of infrared imaging technology provides another dimension of information for the spatial identification of moisture content distribution. Due to its large specific heat capacity, the temperature of liquid water in wet snow changes relatively slowly, appearing as obvious temperature anomalies in infrared images.

[0011] In one possible implementation, thermal radiation images captured by an infrared camera are digitized and then identified using a set temperature threshold. Extracting temperature gradient characteristics involves calculating the temperature difference between adjacent pixels. When the temperature of an area exceeds the surrounding average temperature by a preset threshold, the area is marked as a potential high-water content region, and its spatial coordinates are recorded.

[0012] It's important to note that principal component analysis plays a key role in multi-source data fusion. Radar data provides depth-wise moisture content information, infrared data provides surface temperature distribution information, and environmental parameters collected by meteorological stations—these three types of data have different dimensions and scales. Principal component analysis constructs a covariance matrix, calculates eigenvalues ​​and eigenvectors, and projects high-dimensional data into a low-dimensional space, retaining the principal components with the largest variance. This dimensionality reduction not only reduces data redundancy but also highlights the inherent connections between data from different sensors. The fused eigenvectors simultaneously reflect the vertical structural characteristics and horizontal distribution of wet snow. Combined with the temporal variations of local climate conditions, they ultimately form an initial distribution map that encompasses multidimensional information such as spatial location, moisture content, and temperature distribution. This comprehensive map provides a reliable data foundation for subsequent avalanche warning and risk assessment.

[0013] In step S102, based on the spatiotemporal variation characteristics of wet snow moisture content and snow layer density anomalies in the initial distribution map, combined with the snow layer density gradient and the accumulation unevenness index, anomaly detection is performed on the low snowmelt efficiency and increased viscosity. If it is detected that the moisture content critical point exceeds the preset threshold or the moisture unevenness is aggravated, the potential risk areas for snowmelt and snow removal operations are determined through analysis of the initial distribution map.

[0014] The difference in wet snow moisture content between adjacent grid points is extracted from the initial distribution map, and the moisture content gradient is calculated. The difference between the snow layer density at each grid point and the regional average density is also taken as the density deviation. The snowmelt retardation parameter is calculated by multiplying the moisture content gradient and the density deviation. The snowmelt retardation parameter is compared with a preset snowmelt efficiency threshold. When the snowmelt retardation parameter exceeds the threshold, it is considered inefficient. The density gradient is calculated by dividing the density difference between adjacent snow layers by the snow layer thickness. The accumulation inequality index is calculated by combining the ratio of the standard deviation of the snow layer thickness in each region to the average thickness. When the accumulation inequality index exceeds a preset critical value, it is considered to indicate increased viscosity. If inefficient snowmelt or increased viscosity is detected, the snowmelt retardation parameter, density gradient, and accumulation inequality index are extracted for the corresponding grid area. A risk score is calculated through weighted summation. Grid locations with risk scores exceeding the preset risk threshold are used to identify potential risk areas for snowmelting and snow removal operations.

[0015] Specifically, the calculation of the snowmelt retardation parameter is based on the fundamental principles of wet snow's physical properties. The moisture content gradient reflects the uneven distribution of water within the snow layer. Large differences in moisture content between adjacent grid points indicate a region of water accumulation or loss. Density deviation reveals anomalies in the snow layer's compaction. The difference between the density of each grid point and the regional average density can identify areas of overcompaction or looseness.

[0016] Specifically, the snowmelt retardation parameter is obtained by multiplying the moisture gradient and density deviation. This calculation method can comprehensively reflect the hindering effect of moisture distribution and density anomalies on the snowmelt process.

[0017] In one possible implementation, the snowmelt efficiency is determined based on the comparison of the snowmelt retardation parameter with a preset threshold. When the snowmelt retardation parameter exceeds the threshold, it means that the snowmelt process in the area is significantly hindered. This obstruction mainly comes from two aspects: the excessive concentration of local water due to high water content gradients to form ice layers, and the reduced heat conduction efficiency caused by density anomalies. The density gradient is calculated by dividing the density difference of adjacent snow layers by the snow layer thickness, reflecting the rate of change of the snow layer structure in the vertical direction. The accumulation unevenness index is obtained by the ratio of the standard deviation of the snow layer thickness in each area to the average thickness. When the index exceeds the critical value, it indicates that the unevenness of the snow layer accumulation has reached a level that affects the snowmelt operation.

[0018] It should be noted that the determination of increased viscosity is closely related to the uneven accumulation index. Uneven snow accumulation leads to localized stress concentrations, reducing the fluidity of wet snow in these areas and significantly increasing viscosity. In actual snow removal operations, this phenomenon manifests as increased propulsion resistance of mechanical equipment and reduced snow removal efficiency. The risk score is calculated using a weighted summation approach, assigning different weights to the snowmelt retardation parameter, density gradient, and uneven accumulation index. The weighting is based on historical data statistics and expert experience. The snowmelt retardation parameter typically carries the highest weight because it directly reflects the difficulty of snowmelt. The identification of risk areas involves a systematic evaluation of the risk scores of all grid points. When a grid's risk score exceeds a preset risk threshold, it is marked as a potential risk area. Adjacent high-risk grids are merged to form a continuous risk area. This approach avoids excessive fragmentation of risk areas and facilitates subsequent operation planning. This multi-parameter comprehensive assessment method accurately identifies potential risk areas in snowmelt and snow removal operations, providing operators with precise risk distribution information and improving operational safety and efficiency.

[0019] Step S103: Continuously monitor the moisture content of wet snow in the potential risk area through radar and infrared sensors, analyze the changes in particle cohesion and surface friction coefficient, and if the moisture content of wet snow exceeds the preset threshold or the particle cohesion increases, perform real-time correction on the contribution ratio of radar and infrared sensor data to obtain a corrected wet snow characteristic distribution map. If it does not exceed the threshold, maintain the current weight and continue monitoring.

[0020] Radar sensors collect reflected signal strength from potential risk areas at preset intervals. The wet snow moisture content is calculated based on the signal attenuation. Simultaneously, infrared sensors acquire temperature distribution data from the same area. The particle cohesion parameter is determined by multiplying the temperature value by the snow depth and dividing it by a preset coefficient. The cohesion change rate is calculated based on the difference between the particle cohesion parameter and the previous cohesion parameter. Surface roughness is extracted from the scattering characteristics of the radar reflected signal. The surface friction coefficient is determined by the inverse of the roughness, and the friction change rate is calculated based on the difference between the friction coefficient and the previous value. If the wet snow moisture content exceeds a preset threshold or the cohesion change rate exceeds a critical value, the radar data weight is adjusted based on the product of the excess value and a preset proportional coefficient, and the infrared data weight is correspondingly reduced. The adjusted weights are then used to perform a weighted fusion of the two sensor data types to produce a calibrated wet snow characteristic distribution map. If the threshold is within the threshold, the original weights are retained and moisture content monitoring continues.

[0021] Specifically, the particle cohesion parameter is determined based on the physical relationship between temperature distribution and water migration in wet snow. When the temperature of the snow layer rises, the surface of the ice crystals begins to melt, forming a liquid water film. This water film forms capillary bridges between adjacent ice crystals, creating a cohesive effect.

[0022] Specifically, the product of temperature and snow depth reflects the cumulative effect of heat in the snow layer. Deeper snow is exposed to heat for a longer period of time, resulting in a correspondingly stronger adhesion. A preset coefficient is introduced to convert the temperature-depth product into a standardized adhesion value. This coefficient is obtained through extensive experimental calibration and is typically adjusted for different snow conditions.

[0023] In one possible implementation, the calculation of the rate of change of adhesion reveals the dynamic evolution of the snow layer structure. By comparing the difference in the adhesion parameter between the current moment and the previous moment, the acceleration or slowdown trend of the adhesion process can be identified. When the rate of change is positive and continues to increase, it indicates that the snow layer is consolidating rapidly. In this case, the snow removal equipment needs a greater thrust to destroy the snow layer structure. The determination of the surface friction coefficient depends on the analysis of radar scattering characteristics. The rough snow surface will produce more diffuse scattering, resulting in increased irregularities in the echo signal. The inverse of the roughness is proportional to the friction coefficient because the friction resistance of the smooth surface is smaller, which is conducive to the propulsion of the snow removal equipment.

[0024] It should be noted that the weight adjustment mechanism is designed to account for the varying reliability of different sensors under specific conditions. When the moisture content exceeds a threshold, the penetration of the radar signal is affected. In this case, the weight of the radar data needs to be reduced, and the contribution of the infrared data needs to be increased accordingly. The extent of the excess multiplied by a preset scaling factor determines the magnitude of the weight adjustment. For example, if the moisture content exceeds the threshold by 10%, the weight adjustment is 0.1 times the scaling factor. This dynamic adjustment ensures accurate monitoring results under various snow conditions. The calibrated wet snow characteristic distribution map integrates the weighted multi-source sensor information. During the weighted fusion process, the final value at each grid point is equal to the radar measurement multiplied by its weight plus the infrared measurement multiplied by its weight. This fusion approach preserves the advantages of radar in depth detection while leveraging the accuracy of infrared in surface temperature monitoring. The continuous monitoring mechanism ensures real-time tracking of changing snow conditions. When parameters are within the threshold, the original weight configuration is maintained, avoiding data fluctuations caused by frequent adjustments. Meanwhile, new monitoring data continues to be collected at regular intervals, providing continuous data support for subsequent risk assessments.

[0025] In step S104, based on the corrected wet snow characteristic distribution map, combined with the initial distribution map and the temperature influencing factors and melting rate fluctuation characteristics obtained by analyzing the micro-meteorological station data, the collection deviation of the multi-source sensor data is corrected for the equipment thrust loss and the working surface friction loss, and a dynamically updated map of the snow status in the risk area is obtained.

[0026] Based on the difference in moisture content between the corrected wet snow characteristics distribution map and the initial distribution map at the same grid point, combined with the ambient temperature time series collected by the micro-meteorological station, the temperature impact factor is calculated by dividing the difference between the current temperature and the freezing point by a preset temperature constant. The melt rate fluctuation characteristics are determined by dividing the moisture content difference between adjacent moments by the time interval. The snowmelt dynamic index is calculated by multiplying the temperature impact factor and the melt rate fluctuation characteristics. The equipment thrust loss value is calculated by multiplying the snowmelt dynamic index by a preset equipment thrust reference value. The working surface friction loss value is determined by multiplying the snowmelt dynamic index by a surface friction reference coefficient. To account for the acquisition deviations of multiple sensors at different times, the radar, infrared, and meteorological data are time-aligned using linear interpolation based on the time difference between each sensor's acquisition time and the reference time. The corrected sensor data, combined with the equipment thrust loss value and the working surface friction loss value, are used to update the snow density, moisture content, and temperature values ​​at each grid point, resulting in a dynamically updated map of the snow accumulation status within the risk area.

[0027] Specifically, the calculation of the temperature impact factor reflects the driving effect of ambient temperature on the change of wet snow state. When the ambient temperature is above freezing, the greater the temperature difference, the more intense the snowmelt process.

[0028] Specifically, a standardized temperature impact factor is calculated by dividing the temperature difference between the current temperature and freezing point by a preset temperature constant. This constant is typically set at 10 degrees Celsius, so a temperature impact factor of 0.5 for a 5-degree temperature difference indicates a moderate snowmelt driving force. This standardization process makes data comparable across regions and seasons.

[0029] In one possible implementation, determining the melting rate fluctuation characteristics involves comparing the differences between the corrected map and the initial map. The difference in water content at the same grid point at two moments in time reflects the snowmelt progress at that location, and dividing it by the time interval yields the melting rate. Fluctuations are characterized by changes in the melt rate over multiple consecutive time periods. Fluctuations in the rate indicate unstable snowmelt in that area, and this instability often signals potential operational risks.

[0030] It should be noted that the introduction of the snowmelt dynamic index comprehensively considers two key factors: temperature and melt rate. The product of the temperature impact factor and the melt rate fluctuation characteristic comprehensively reflects the complexity of the snowmelt process. When the temperature impact factor is 0.8 and the melt rate fluctuation characteristic is 1.2, the snowmelt dynamic index is 0.96. A value close to 1 indicates an active and unstable snowmelt process. The equipment thrust loss value is obtained by multiplying the snowmelt dynamic index by the thrust reference value. The reference value is pre-set based on the equipment model and rated power. The higher the snowmelt dynamic index, the greater the additional thrust required. The calculation of working surface friction loss uses a similar principle. The surface friction reference coefficient is pre-calibrated based on the snow quality type, typically 0.3 for dry snow and up to 0.7 for wet snow. The product of the snowmelt dynamic index and the reference coefficient directly reflects the frictional resistance under actual operating conditions. The advantage of this calculation method is that it can dynamically reflect the impact of the snowmelt process on operating conditions. Linear interpolation plays a key role in time alignment correction. Different sensors have different collection cycles. Radar may collect data every 30 seconds, infrared every 20 seconds, and weather stations every minute. By recording the time difference between each sensor's collection time and a unified reference time, and performing linear interpolation between the values ​​of the two previous and next collection points, we can derive an estimate of the reference time.

[0031] For example, if the radar measured a moisture content of 15% 10 seconds before the reference time and 18% 20 seconds after the reference time, the interpolated value at the reference time would be 16%. The generation of the dynamically updated map integrates corrected multi-source data and calculated loss values. The update of each grid point not only includes the physical parameters directly measured by the sensor but also incorporates loss information related to equipment operation, forming a more comprehensive description of the snow state. This comprehensive dynamic map provides accurate data support for subsequent operational decisions, significantly improving the efficiency and safety of snow melting and snow removal operations.

[0032] In step S105, based on the changes in the snow accumulation status in the risk area in the dynamically updated map, combined with the viscosity peak and local accumulation thickness in the corrected wet snow characteristic distribution map, the impact of the resistance accumulation rate and path obstruction frequency on the snow removal operation is analyzed. Based on the impact analysis results, constraint bars for snow removal path planning are constructed, and the resistance distribution status of snow removal equipment in different areas is determined by the constraint conditions.

[0033] Based on the dynamically updated map, the time series of snow density and moisture content at each grid point within the risk area are extracted. Combined with the peak viscosity and local accumulation thickness at the corresponding location in the calibrated wet snow characteristic distribution map, the resistance accumulation rate is calculated by dividing the difference in viscosity between adjacent moments by the time interval. The frequency of path obstruction is determined by the number of times the viscosity exceeds a preset threshold per unit time. The difficulty of snow removal operations is assessed by multiplying the resistance accumulation rate and the path obstruction frequency. Areas identified as high-difficulty areas are determined when the product exceeds a preset difficulty threshold. A comprehensive ranking of the difficulty level and thickness ratio is performed, combining the ratio of local accumulation thickness to a preset thickness benchmark. A peak load threshold is set based on the difference between the upper limit of snow removal equipment thrust and the resistance accumulation rate. Using the clearance priority as a path selection constraint and the peak load threshold as an operation intensity constraint, a constraint set is constructed, encompassing both regional access restrictions and operation duration limits. By comparing the resistance accumulation rate and the peak load threshold across each grid area, the resistance distribution of snow removal equipment in different areas is determined.

[0034] Specifically, calculations of the resistance accumulation rate reveal the dynamics of the viscosity of wet snow. Peak viscosity represents the maximum resistance to snow removal equipment within a specific area, typically occurring at a critical point where the snow reaches saturation but has not yet fully liquefied.

[0035] Specifically, by tracking the changes in viscosity at adjacent moments, it's possible to accurately identify the growth trend of drag. For example, if the viscosity in a certain area increases from 300 Newtons to 450 Newtons within 10 minutes, the drag accumulation rate is 15 Newtons per minute. This rapid increase indicates that a difficult-to-remove mixed layer of ice and snow is forming in that area.

[0036] In one possible implementation, statistics on the frequency of path obstructions reflect the continuity risk of snow removal operations. A preset threshold is typically set at 1.5 times the normal propulsion resistance of the equipment. Frequently exceeding this threshold indicates that the equipment needs to repeatedly adjust operating parameters or change its propulsion route. The number of times the threshold is exceeded per unit time is directly correlated to operational efficiency; a higher frequency indicates that the path is less suitable for continuous operation. This frequency statistics provide important risk assessment information for subsequent path planning.

[0037] It should be noted that the difficulty of snow removal operations is assessed using the product of the resistance accumulation rate and the frequency of path obstructions as a comprehensive indicator. This product relationship reflects both the absolute growth and fluctuation characteristics of resistance. When the accumulation rate is 20 Newtons per minute and the obstruction frequency is 6 times per hour, the difficulty assessment value is 120. Areas exceeding the preset difficulty threshold of 100 are classified as high-difficulty areas. The ratio of local accumulation thickness to a preset thickness benchmark further refines priority determination. The thickness benchmark is typically set at 30 cm. When the actual thickness reaches 45 cm, a ratio of 1.5 indicates that the area warrants priority. Clearance priority is determined by combining both difficulty level and thickness ratio. By comprehensively ranking all grid areas, a clear order of operations can be established. Areas of high difficulty but moderate thickness may be given lower priority than areas of medium difficulty but exceeding the thickness standard. This flexible prioritization ensures the rational allocation of snow removal resources. The peak load threshold for the equipment is set based on a safety margin between the equipment's upper thrust limit and actual resistance. When the equipment's maximum thrust is 2000 Newtons and the resistance accumulation rate in a certain area reaches 30 Newtons per minute, the equipment's limit will be reached after 40 minutes of continuous operation. Therefore, the peak load threshold for this area is set to 1800 Newtons. The construction of a set of constraints forms a complete system of operating rules. Regional access restrictions are based on clearing priority values; equipment is only allowed to enter areas with a certain priority level. Operation duration limits are calculated based on the peak load threshold to prevent equipment from overloading due to prolonged operation in high-resistance areas. By comparing the real-time monitored resistance accumulation rate with the set threshold, the operability of each area can be dynamically determined. The resistance distribution map intuitively displays the difficulty distribution across the entire operation area, providing a scientific basis for optimizing snow removal paths and significantly improving snow removal efficiency and equipment safety in adverse weather conditions.

[0038] Step S106: If the resistance distribution shows that the accumulated rate of snow removal resistance in a certain area exceeds the preset threshold or the peak load of the equipment exceeds the preset safety range, the operating route of the snow removal equipment is optimized and adjusted in real time to avoid areas with excessive local accumulation thickness. An optimized snow removal operation scheduling plan is generated based on the adjusted operating route. If the resistance does not exceed the safety range, the existing route is maintained and the friction loss data of the operating surface is recorded.

[0039] If the resistance distribution map indicates that the snow removal resistance accumulation rate in a grid area exceeds a preset threshold or the peak equipment load exceeds a preset safety range, the Dijkstra algorithm is used to calculate an alternative route avoiding the high-resistance area. Grid locations where localized snow accumulation exceeds the clearing capacity limit are identified, generating a set of coordinates for the areas requiring detours. Based on the coordinates of the areas requiring detours, detour nodes are inserted between adjacent nodes of the original route to form an adjusted route sequence consisting of a starting point, waypoints, and an end point. The operation duration is calculated by calculating the ratio of the length of each segment in the adjusted route sequence to the average operating speed. The equipment load distribution scheme is then determined based on the resistance distribution of each segment, generating an optimized snow removal operation scheduling plan. If neither the resistance accumulation rate nor the peak equipment load exceeds the safety range, the current route is maintained. Sensors are used to collect real-time data on the working surface friction coefficient and equipment propulsion resistance. Data on the working surface friction loss, including the acquisition time, location coordinates, friction coefficient, and resistance value, is stored and recorded.

[0040] Specifically, the Dijkstra algorithm is applied to snow removal path optimization based on the shortest path principle in graph theory. The work area is divided into grid nodes, each representing an area to be cleared. The edges between nodes represent feasible work paths, and the weight of the edges is determined by the cumulative rate of resistance on that section of road.

[0041] Specifically, when the resistance in a certain area exceeds a safety threshold, the algorithm sets the weights of all incoming edges of the corresponding nodes in that area to infinity, forcing the path to bypass that area. Starting from the starting node, the algorithm gradually calculates the minimum resistance path to each node, ultimately determining the optimal route that avoids high-resistance areas.

[0042] In one possible implementation, determining the upper limit of clearing capacity involves balancing the physical limitations of the equipment and operational efficiency. When the local accumulation thickness reaches 60 centimeters, even if the equipment can be forced to advance, the speed will drop to less than 20% of the normal speed. Continuing to work at this time is not only extremely inefficient but also accelerates equipment wear. Therefore, such areas are marked as coordinates that need to be detoured, waiting for specialized heavy equipment to handle or adopt a layered clearing method. The establishment of a set of detour area coordinates ensures that conventional snow removal equipment can operate efficiently within its capabilities.

[0043] It's important to note that the adjustment process for the route sequence must ensure path continuity and accessibility. The original route may be a straight-line approach. When a detour is necessary, the algorithm inserts new waypoints between adjacent nodes. For example, if the original route runs directly from node A to node B, and a high-resistance area C is detected in the middle, the algorithm calculates a detour path ADEB, with newly inserted waypoints D and E. This insertion method ensures continuous operation of the snow removal equipment, avoiding frequent starts, stops, and turns. The calculation of operation duration fully accounts for the differences in operating conditions across different road sections. The transit time is calculated by dividing the length of each section by the average operating speed for that section. The average operating speed is closely related to the resistance distribution of that section. On sections with low resistance, the equipment can maintain a high speed; on sections with high resistance but not exceeding a threshold, the speed is reduced accordingly. The equipment load distribution plan is formulated based on the resistance characteristics of each section, reducing the propulsion speed to control the load on high-resistance sections and increasing the speed appropriately on low-resistance sections to improve overall efficiency. Real-time recording of friction loss data on the working surface provides an important basis for equipment maintenance and operation optimization. The sensor collects data at regular intervals, including the current GPS location, friction coefficient measurements, and the equipment's propulsion resistance. The friction coefficient reflects the friction characteristics of the contact surface between the snow layer and the equipment, typically ranging from 0.2 to 0.8. Propulsion resistance is directly related to the equipment's energy consumption and wear. This data is stored in timestamp order, forming a complete record of the operation process. By analyzing the friction loss characteristics in different areas and time periods, high-risk operating conditions prone to equipment failure can be identified, providing data support for the development of preventive maintenance plans and significantly extending equipment life.

[0044] Step S107: Based on the optimized runway snow removal operation scheduling plan, combined with the records of uneven moisture distribution and snow layer compressibility in the historical wet snow characteristic data, the changing trends of the path obstruction frequency and the accumulation unevenness index are analyzed, and the trend analysis results are input into a pre-established prediction model. The trend changes in the snow melting efficiency and uneven snow distribution around the runway within a preset time in the future are output, and the optimized snow removal operation scheduling plan is dynamically adjusted.

[0045] Based on the operation sequence for each region in the optimized snow removal scheduling plan, the seasonal water lateral distribution difference and snow layer vertical compression ratio for the corresponding region are extracted from stored historical wet snow characteristics records. Combined with the currently monitored path obstruction frequency and accumulation unevenness index, the four change rate data are calculated by subtracting the current period value from the previous period value and dividing it by the time interval. The change rates of the water lateral distribution difference, snow layer vertical compression ratio, path obstruction frequency, and accumulation unevenness index are used as four-dimensional input vectors and fed into a pre-trained time series prediction network. The network outputs predicted snowmelt efficiency and snow accumulation unevenness at each moment in the future within a preset time period, forming two trend curves. For periods where the predicted snowmelt efficiency falls below a preset efficiency threshold or the predicted snow accumulation unevenness exceeds a preset uniformity threshold, a pre-processing task is added before the corresponding period in the original scheduling plan. By pre-deploying equipment or shortening the operation interval during these periods, a dynamically adjusted and optimized snow removal scheduling plan is generated that includes pre-processing tasks.

[0046] Specifically, the lateral moisture distribution difference reflects the unevenness of the snow's moisture content in the horizontal direction. When one side of the runway is exposed to direct sunlight while the other side is in the shade, the difference in snowmelt rate on both sides leads to uneven moisture distribution.

[0047] Specifically, moisture content data is collected at monitoring points spaced at regular intervals across the runway. The difference in moisture content between adjacent monitoring points is calculated, and the maximum difference represents the lateral moisture distribution difference at that moment. Historical data shows that in early spring, this difference typically peaks at 10:00 AM, when the angle of sunlight creates the most uneven light distribution.

[0048] In one possible implementation, monitoring the vertical compression ratio of the snow layer reveals the densification process of snow accumulation. The initial density of new snowfall is low. As time passes and the new snow on top accumulates, the lower snow layer gradually compresses under the action of gravity. The compression ratio is obtained by measuring the ratio of the initial thickness of the snow layer to the current thickness. When the compression ratio reaches 1.5, it means that the snow layer thickness is compressed to two-thirds of its original thickness. At this time, the density of the snow layer increases significantly, and the resistance to snow removal also increases accordingly. This compression process is particularly obvious when the temperature is close to freezing point, because partial melting and refreezing accelerate the densification of the snow layer.

[0049] It should be noted that the difference method used to calculate the rate of change accurately captures the dynamic characteristics of parameters. For example, if the equipment was blocked three times in the previous hour and seven times in the current hour, the rate of change increases by four times per hour. This continued increase in the rate of change indicates deteriorating snow conditions, necessitating timely adjustments to operational strategies. The four-dimensional vector formed by the four rate-of-change data comprehensively describes the evolving snow conditions, providing rich feature information for the forecasting model. The application of the time series prediction network leverages the temporal patterns inherent in historical data. By learning from past snow condition evolution patterns under similar weather conditions, the network can predict future trends. The input four-dimensional vector undergoes multi-layer processing within the network, extracting deep temporal features. The network outputs two key predictions: snowmelt efficiency, which represents the amount of snow that can be removed per unit time, typically measured in cubic meters per hour; and snow distribution unevenness, which reflects the standard deviation of snow depth across the operational area. The design of preprocessing tasks embodies the concept of preventive maintenance. When the forecast indicates that snowmelt efficiency will fall below 70% of normal during a specific period in the future, the system identifies that period as a high-risk period. By deploying additional snow removal equipment 30 minutes before the high-risk period for pre-treatment, the system can effectively reduce snow accumulation.

[0050] For example, if the forecast indicates a significant drop in snowmelt efficiency at 3 p.m., the adjusted plan schedules a pre-treatment operation at 2:30 p.m. to remove the soft snow on the surface and prevent it from compacting into a difficult-to-remove ice layer during the period of low efficiency. This dynamic adjustment mechanism ensures the continued availability of the runway in adverse weather conditions, significantly improving the airport's operational efficiency and safety.

[0051] The preferred embodiments of the present invention disclosed above are intended only to help illustrate the present invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the present invention to specific embodiments. Obviously, many modifications and variations are possible based on the contents of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better understand and utilize the present invention. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for early warning of snowmelt and intelligent snow removal scheduling at an ice and snow field, characterized in that: The method comprises: Collect data on wet snow moisture content, snow layer density anomalies and site climate differences to generate an initial distribution map of wet snow characteristics and local climate conditions; based on the spatiotemporal variation characteristics of wet snow moisture content and snow layer density anomalies in the initial distribution map, combined with the snow layer density gradient and accumulation unevenness index, detect low snow melting efficiency and enhanced viscosity, and determine potential risk areas for snow melting and snow removal operations; monitor the wet snow moisture content in the potential risk area, analyze the changes in particle adhesion and surface friction coefficient, adjust the data fusion weights, and generate a corrected wet snow characteristic distribution map; based on the corrected wet snow characteristic distribution map, combined with the analyzed temperature influencing factors and melt Based on the fluctuation characteristics of the wet snow characteristic distribution map, a time series compensation method is used to correct the deviation of multi-source sensor data, and a dynamically updated map of the snow accumulation status in the risk area is generated; according to the changes in the snow accumulation status in the risk area in the dynamically updated map, combined with the viscosity peak and local accumulation thickness in the corrected wet snow characteristic distribution map, the constraint conditions for snow removal path planning are constructed; according to the constraint conditions, the operating route of the snow removal equipment is optimized to generate a snow removal operation scheduling plan; according to the snow removal operation scheduling plan, combined with historical wet snow characteristic data, the changing trends of the path obstruction frequency and the accumulation unevenness index are analyzed, and the snow removal operation scheduling plan is adjusted to obtain an adjusted snow removal operation scheduling plan.

2. The method for early warning of snowmelt and intelligent snow removal scheduling at an ice and snow field according to claim 1 is characterized in that: The data on wet snow moisture content, snow density anomalies, and site climate differences are collected to generate an initial distribution map of wet snow characteristics and local climate conditions, including: Obtain the reflectance distribution data of the wet snow layer, and use a preset comparison table to determine the moisture percentage value of each depth layer; capture the surface temperature radiation image of the snow layer, extract the temperature gradient change characteristics, and determine the spatial position coordinates of the area with uneven moisture content distribution; based on the wind speed, air temperature, and humidity parameters collected by the micro-meteorological station, combine the moisture percentage value and the spatial position coordinates to generate the initial distribution map.

3. The method for early warning and intelligent snow removal scheduling for snow melting on ice and snow fields according to claim 1 is characterized in that: Based on the temporal and spatial variation characteristics of wet snow moisture content and snow density anomalies in the initial distribution map, combined with the snow density gradient and accumulation unevenness index, the low snowmelt efficiency and increased viscosity are detected, and potential risk areas for snowmelt and snow removal operations are identified, including: The wet snow moisture content difference is extracted from the initial distribution map, the moisture content gradient value is calculated, the difference between the snow layer density and the regional average density is obtained as the density deviation, and the snowmelt retardation parameter is generated according to the product of the moisture content gradient value and the density deviation; the density gradient is calculated according to the snowmelt retardation parameter and the snow layer thickness, and the accumulation unevenness index is generated by combining the ratio of the standard deviation of the snow layer thickness to the average thickness; a risk score is generated according to the weighted sum of the snowmelt retardation parameter, the density gradient and the accumulation unevenness index, and the potential risk area is determined.

4. The method for early warning of snowmelt and intelligent snow removal scheduling at an ice and snow field according to claim 1 is characterized in that: The monitoring of the moisture content of wet snow in the potential risk area, analysis of changes in particle cohesion and surface friction coefficient, adjustment of data fusion weights, and generation of a calibrated wet snow characteristic distribution map include: The wet snow moisture content is calculated based on the reflection signal intensity of the potential risk area; the particle cohesion parameter is calculated by multiplying the temperature value and the snow layer depth; the surface friction coefficient is calculated based on the rate of change of the particle cohesion parameter and the surface roughness; the radar and infrared data weights are adjusted based on the wet snow moisture content and the particle cohesion parameter, and the adjusted weights are used to fuse the data to generate the corrected wet snow characteristic distribution map.

5. The method for early warning of snowmelt and intelligent snow removal scheduling at an ice and snow field according to claim 1 is characterized in that: The method uses a time series compensation method to correct the deviation of multi-source sensor data based on the corrected wet snow characteristic distribution map, combined with the analyzed temperature influencing factors and melting rate fluctuation characteristics, to generate a dynamically updated map of the snow state in the risk area, including: Based on the difference in moisture content between the corrected wet snow characteristic distribution map and the initial distribution map, combined with the ambient temperature sequence of the micro-meteorological station, the temperature influence factor is calculated; based on the difference in moisture content at adjacent moments, the melting rate fluctuation characteristics are calculated; based on the temperature influence factor and the melting rate fluctuation characteristics, the snowmelt dynamic index is calculated to determine the equipment thrust loss value and the working surface friction loss value; the multi-source sensor data is time-aligned and corrected using a linear interpolation method, and the snow density, moisture content and temperature values ​​are updated to generate the dynamically updated map.

6. The method for early warning of snowmelt and intelligent snow removal scheduling at an ice and snow field according to claim 1 is characterized in that: According to the changes in the snow state in the risk area in the dynamically updated map, combined with the viscosity peak and local accumulation thickness in the corrected wet snow characteristic distribution map, the constraints for snow removal path planning are constructed, including: The snow density and moisture content time series are extracted based on the dynamically updated map, and the resistance accumulation rate is calculated based on the viscosity peak and local accumulation thickness in the corrected wet snow characteristic distribution map; the path obstruction frequency is determined based on the number of times the viscosity exceeds the threshold; the operation difficulty is evaluated based on the product of the resistance accumulation rate and the path obstruction frequency, and a clearing priority value is generated based on the local accumulation thickness; the equipment load peak threshold is set based on the equipment thrust upper limit, and a constraint condition including the clearing priority and load threshold is constructed.

7. The method for early warning of snowmelt and intelligent snow removal scheduling at an ice and snow field according to claim 1 is characterized in that: Optimizing the snow removal equipment operation route according to the constraint conditions and generating a snow removal operation scheduling plan includes: Based on the cleaning priority and equipment load peak threshold in the constraints, grid areas where the resistance accumulation rate or equipment load exceeds the threshold are identified; a path planning algorithm is used to calculate alternative routes that avoid high-resistance areas, and detour path nodes are inserted to generate an adjusted operation route sequence; the operation duration is calculated based on the section length and operation speed of the operation route sequence, and the snow removal operation scheduling plan is generated in combination with the resistance distribution.

8. The method for early warning of snowmelt and intelligent snow removal scheduling at an ice and snow field according to claim 1 is characterized in that: The snow removal operation scheduling plan is adjusted based on the snow removal operation scheduling plan, combined with historical wet snow characteristic data, analyzing the changing trends of the path obstruction frequency and the accumulation unevenness index, and obtaining the adjusted snow removal operation scheduling plan, including: The moisture distribution difference and the snow layer compression ratio are extracted from the historical wet snow characteristic data, and the change rate data is calculated in combination with the currently monitored path obstruction frequency and the uneven accumulation index. The change rate data is used as an input vector and input into a time series prediction network to output a trend change curve of snow melting efficiency and uneven snow distribution. Based on the time period below the threshold in the trend change curve, a preprocessing task is added to the snow removal operation scheduling plan to generate an adjusted snow removal operation scheduling plan.

Images