An unmanned mine car on-board environment monitoring and dynamic path planning method and system

Abstract

The application discloses an unmanned mine vehicle-mounted environment monitoring and dynamic path planning method and system, belongs to the technical field of unmanned mine transportation, and aims to solve the problems that the existing unmanned mine vehicle lacks environment sensing capability, path planning does not consider pollution risk, and a purification strategy lags behind, leading to easy erosion of core equipment and energy consumption redundancy. The application collects multi-source environment data through a vehicle-mounted multi-sensor, constructs a dynamic pollution space-time model by fusing mine area fixed monitoring station data, realizes predictive hierarchical purification regulation by using a model predictive control framework, adds an environment exposure cost factor in a path planning cost function, realizes the collaborative decision of purification power regulation and path re-planning, can optimize operation energy consumption on the premise of protecting core equipment from high pollution erosion, and effectively improves the operation reliability and operation efficiency of the unmanned mine vehicle.

Description

Technical Field

[0001] This invention relates to the field of unmanned mining transportation technology, and in particular to a method and system for onboard environmental monitoring and dynamic path planning of unmanned mining vehicles. Background Technology

[0002] Unmanned mining trucks, equipped with various sensors such as LiDAR, cameras, and millimeter-wave radar, combined with high-precision positioning systems and intelligent decision-making algorithms, achieve unmanned operation in transportation. However, the mining environment has unique characteristics: blasting operations generate large amounts of dust, and the operation of heavy equipment emits exhaust gases, resulting in significant spatial and temporal unevenness in air quality. High concentrations of dust and harmful gases not only threaten the health of personnel but also seriously damage the core equipment of the unmanned mining trucks. Dust accumulation on sensor surfaces reduces the perception accuracy of LiDAR and cameras, while corrosive gases can damage computing units and electrical equipment, thus affecting the reliability and safety of the unmanned driving system.

[0003] Chinese patent CN117092994B discloses a method for monitoring the status of unmanned mining trucks. This method plans the driving path of the convoy, marks dangerous areas, monitors the driving status data of mining trucks in dangerous and non-dangerous areas, identifies abnormal mining trucks and controls them in coordination with the main mining truck, and judges and adjusts the possibility of collision between different convoys to ensure the orderly operation of mining trucks in the mine. This patent focuses on driving safety issues caused by mine truck malfunctions and road anomalies. It uses speed monitoring and trajectory analysis to identify and prevent collisions with abnormal mine trucks, but it still has significant technical shortcomings: First, the path planning only aims to "minimize the passage through dangerous areas," where "dangerous areas" are defined as areas with a high incidence of historical traffic accidents. It fails to consider environmental pollution factors such as dust and harmful gases, making it impossible to prevent mine trucks from entering highly polluted areas and causing corrosion and damage to core equipment. Second, it only monitors vehicle operating data such as speed and braking status, without configuring environmental monitoring sensors for gases and dust, lacking the ability to perceive environmental pollutants in the mining area and predicting potential damage to equipment from environmental pollution. Third, it lacks an air purification system and control strategy for core equipment, only addressing driving safety issues through braking and path avoidance, without addressing equipment environmental protection and energy consumption optimization for purification. This makes it difficult to meet the comprehensive requirements of unmanned mine operation for equipment reliability, energy consumption control, and green operation. Therefore, those skilled in the art urgently need to solve the above technical problems. Summary of the Invention

[0004] To address the technical problems in the prior art, this invention deploys vehicle-mounted environmental sensors for gases and dust, integrates fixed monitoring data from the mining area to construct a dynamic pollution model, and compensates for the lack of environmental perception capabilities. It incorporates environmental pollution risks into the path planning cost function, enabling proactive avoidance of highly polluted areas. Simultaneously, it configures a graded and controlled vehicle-mounted purification system, which collaborates with path planning to optimize purification energy consumption while protecting core equipment from pollution, thus meeting the comprehensive needs of unmanned mine operation.

[0005] To achieve the above-mentioned technical objectives, the present invention is implemented through the following technical solution: a method for onboard environmental monitoring and dynamic path planning of unmanned mining trucks, characterized by comprising the following steps:

[0006] S1. Environmental Data Acquisition: The environmental data acquisition module, composed of a gas detection unit, a dust detection unit, and an intelligent online detection unit, collects data on harmful gases, dust concentrations, vehicle location, and environmental images around the mining truck, and uploads them to the cloud database, while simultaneously accessing map database data.

[0007] S2. Dynamic pollution model construction: Based on multi-source data from a cloud database, a dynamic pollution model of the mining area is constructed.

[0008] S3. Pollution Level Classification: Based on the output results of the dynamic pollution model, the air pollution level of each area in the mining area is classified.

[0009] S4. Purification power level control: Determine the pollution level of the mine car at its current location. If it is low pollution, control the air purification system to operate at low power. If it is medium pollution, control the air purification system to operate at medium power. If it is high pollution, control the air purification system to operate at high power.

[0010] S5. High-pollution entry prediction: Summarize the pollution levels of the locations of unmanned mining trucks in the entire mining area, and predict whether the mining trucks are about to enter high-pollution areas through the dynamic pollution model prediction and analysis module.

[0011] S6. Collaborative Strategy Execution: If it is determined that the vehicle will not enter a high-pollution area, the current purification strategy and driving path will be maintained; if it is determined that the vehicle is about to enter a high-pollution area, the air purification system will be controlled to operate at high power first, and then the driving path of the mining truck will be replanned. The path planning results will be synchronously transmitted back to the cloud database to form a closed loop.

[0012] By adopting the above technical solution, and through the complete closed-loop operation logic of environmental perception, pollution modeling, hierarchical control, risk prediction, and collaborative execution, the core innovation of this patented method is comprehensively covered. It breaks through the limitations of existing solutions that only focus on driving safety and lack environmental protection capabilities, and can achieve a balanced optimization of unmanned mining vehicle equipment protection and transportation efficiency, with a comprehensive and complete protection scope.

[0013] Furthermore, the data acquisition and dynamic pollution model construction in steps S1 to S2 are specifically as follows: the gas detection unit collects concentration data of carbon monoxide, sulfur dioxide, nitrogen oxides, and volatile organic compounds; the dust detection unit collects concentration data of PM2.5, PM10, and total suspended particulate matter; and the intelligent online detection unit collects centimeter-level positioning data and environmental image data. After preprocessing, all data are uploaded to the cloud. After fusing data from fixed monitoring stations, a continuous pollutant concentration distribution raster map is generated using a common Kriging interpolation algorithm. This raster map is then updated to the map database with a delay of less than 10 seconds to form a dynamic pollution spatiotemporal distribution model.

[0014] By adopting the above technical solutions, the construction method of the dynamic pollution model clarifies the complete process of multi-source data fusion, spatial interpolation, temporal prediction, and map synchronization, ensuring the spatiotemporal accuracy and real-time updates of the pollution model. This not only makes the technical solutions fully public and reproducible, but also further improves stability by refining the technical features.

[0015] Furthermore, the pollution level classification and power regulation rules in steps S3 to S4 are specifically as follows:

[0016] Criteria for determining low pollution levels: Furthermore, the concentration of harmful gases meets the standards, corresponding to the low-power operation mode of the air purification system;

[0017] Criteria for determining medium pollution level: Or, if the level of a single harmful gas exceeds the standard by ≤50%, the corresponding power operation mode of the air purification system should be selected.

[0018] Criteria for determining high pollution levels: If the concentration of a single harmful gas exceeds the standard by more than 50%, or if multiple gases exceed the standard simultaneously, the corresponding air purification system will operate in high-power mode.

[0019] By adopting the above technical solutions, the corresponding control rules between the pollution level classification standards and the purification power make the operation logic of the purification system clear and implementable, and can adapt to the optimal operation mode according to the actual pollution scenario, avoiding redundant energy consumption.

[0020] Furthermore, the prediction of entry into the high-pollution area in step S5 adopts a probability calculation formula, specifically:

[0021] Discretize the 5-minute prediction time domain as follows: The control step, calculate the... The probability of entering a high-pollution zone at the corresponding position of each control step. ,in, For the first Predicting pollution concentrations The threshold for high pollution concentration;

[0022] Calculate the overall probability of entering a highly polluted area within the predicted time domain: Under normal transportation tasks When the time is right, it is determined that the mining truck is about to enter a highly polluted area; the threshold for high-value tasks is lowered to 0.40-0.50, and the threshold for time-priority tasks is raised to 0.70-0.80.

[0023] By adopting the above technical solution, the implementation logic of the prediction of entry into highly polluted areas realizes the advance prediction of pollution risks rather than the post-event response. It can trigger protective actions in advance to reduce the risk of pollution exposure of core equipment. At the same time, it adapts to the dynamic threshold settings of different task priorities, effectively improving the scenario adaptability.

[0024] Furthermore, the path planning in step S6 adopts a comprehensive cost function formula, specifically:

[0025] Construct the path synthesis cost function: ,in For the cost of time, For the cost of distance, As a result of environmental exposure, the environmental risk coefficients for low-pollution, medium-pollution, and high-pollution areas are 1, 2-3, and 4-6, respectively.

[0026] The improved A* algorithm or Dijkstra's algorithm is used to find the driving path with the lowest overall cost as the planning result. The path planning result is synchronously sent back to the cloud database to update the data link and complete the closed-loop operation.

[0027] By adopting the above technical solutions, the cost calculation logic of route planning incorporates environmental exposure costs into the consideration dimensions of route planning for the first time. This breaks through the limitations of traditional route planning, which only focuses on transportation efficiency and driving safety, and achieves synergistic optimization of route selection and pollution protection, further amplifying the comprehensive benefits.

[0028] Further, in step S6, the path planning specifically includes:

[0029] S61. Define the node cost calculation rules:

[0030] When using the A* algorithm for path search, the node evaluation function satisfies: ;

[0031] in, The path from the starting point to the current node The cumulative comprehensive cost is calculated using the following formula: ;

[0032] in, For the current node The estimated total cost to the destination of the route is calculated using the following formula: ;

[0033] When using the improved Dijkstra's algorithm for path search, the cumulative cost per node is used. As a core indicator for node evaluation;

[0034] in, The path from the starting point to the current node The total number of road segments included in the planned route; , , These are preset time cost weights, distance cost weights, and environmental exposure cost weights, which are dynamically adjusted according to the priority of the unmanned mining truck transportation task. , , Each is the current node Estimated travel time to the end of the route, estimated travel distance, and estimated environmental exposure cost;

[0035] S62. Define the single-segment cost calculation rule: For any segment in the path... For each road segment, the calculation method for the three types of costs is as follows:

[0036] Time cost satisfied: ,in This is the actual length of the road section. The design speed for unmanned mining trucks on this section of road;

[0037] Distance cost satisfies: ;

[0038] Environmental exposure cost satisfied: ; represents the estimated travel time of the unmanned mining truck on this section of road. This is the environmental risk coefficient corresponding to the road section, which corresponds to the real-time pollution level of the road section.

[0039] S63. Path Search Iteration Rule: When performing path search using the improved Dijkstra algorithm, the cumulative comprehensive cost is selected in each iteration. The smallest unvisited node is expanded until the target destination is reached, thus obtaining the initial optimal driving path;

[0040] S64. Dynamic Replanning Triggering Rule: After the dynamic pollution model of the mining area completes its periodic update, the edge weights of all road segments affected by the changes in pollution distribution are automatically recalculated. If the change in the edge weight of any road segment in the planned driving path exceeds the preset threshold, the path replanning process is immediately triggered, and the calculation logic of S61 to S63 is repeated to iteratively obtain the updated optimal driving path that integrates environmental protection requirements and transportation efficiency requirements.

[0041] By adopting the above technical solutions, a complete system solution is constructed at the functional module architecture level, covering the entire chain of technical features from hardware perception to software decision-making. It can effectively cover infringement scenarios on the product side, prohibit infringers from producing and selling unmanned mining truck monitoring and planning systems with corresponding functions, and provide comprehensive protection.

[0042] This invention also discloses an unmanned mining truck onboard environmental monitoring and dynamic path planning system, characterized in that it includes:

[0043] The environmental data acquisition module consists of a gas detection unit, a dust detection unit, and an intelligent online detection unit. It is deployed at key locations on the mine truck body to collect multi-source environmental and vehicle status data around the mine truck.

[0044] A cloud-based database is used to store collected environmental and vehicle data, and synchronously accesses map database data.

[0045] The dynamic pollution model construction module is used to generate a dynamic spatiotemporal distribution model of pollution in mining areas based on multi-source data in the cloud.

[0046] The air pollution level classification module is used to classify the air pollution level of each area in the mining area; the graded control module is used to control the air purification system to operate in low power mode, medium power mode and high power mode according to the pollution level of the current location of the mining truck.

[0047] The dynamic pollution model prediction and analysis module is used to summarize the pollution level data of all mining trucks and predict the probability that a mining truck will soon enter a high-pollution area.

[0048] The collaborative execution module is used to select an execution strategy based on the prediction results: when there is no high pollution risk, the current strategy is maintained; when there is a high pollution risk, the purification system is first upgraded to high power operation, and then the driving path is replanned and the results are sent back to the cloud database.

[0049] By adopting the above technical solution, the hardware configuration of the environmental data acquisition module is further limited, ensuring the accuracy and reliability of the acquisition of various types of environmental data, and providing accurate data source support for subsequent model building and decision-making.

[0050] Furthermore, the environmental data acquisition module specifically comprises:

[0051] The gas detection unit adopts a combination of electrochemical and infrared sensors;

[0052] The dust detection unit uses a sensor based on the principle of laser scattering.

[0053] The intelligent online detection unit integrates a GPS or Beidou dual-mode positioning module with a high-definition camera, and uses environmental image data for dust source identification and auxiliary marking of pollution scenes.

[0054] By adopting the above technical solutions, the sub-unit composition and corresponding functions of the dynamic pollution model construction module ensure the construction accuracy and update efficiency of the dynamic pollution model, providing accurate basic support for subsequent pollution level classification and risk prediction, and further refining the core technical features of the system.

[0055] Furthermore, the dynamic pollution model construction module includes:

[0056] The data fusion unit is used to perform time alignment, spatial registration, and outlier removal on asynchronously collected mobile monitoring data and fixed monitoring station data, and to perform weighted fusion with a weight of 0.55-0.75 for fixed monitoring stations and 0.25-0.45 for mobile monitoring data.

[0057] Spatial interpolation unit, with built-in ordinary Kriging interpolation algorithm and spherical, exponential and Gaussian variogram models, generates continuous pollutant concentration distribution raster map;

[0058] The time-series analysis unit, with a built-in ARIMA model or LSTM neural network model, predicts the pollution diffusion trend in the next 1-2 hours.

[0059] The map update unit is used to synchronize pollution distribution data to the map database in real time.

[0060] The air purification system adopts a distributed architecture, including: a dedicated purification module for the core control cabinet, which provides a 50-100Pa positive pressure clean environment for the computing unit and communication equipment;

[0061] The sensor cluster self-cleaning system provides air curtain isolation and periodic purging for key sensors such as lidar and cameras. The purging interval is dynamically adjusted according to the pollution level.

[0062] The electric drive system protection unit provides cooling and air filtration for the battery pack and motor controller, with the filtration level dynamically adjusted according to the pollution level.

[0063] By adopting the above technical solutions, the architecture and corresponding functions of the distributed air purification system have achieved targeted and graded protection for different core components of the mining truck, avoiding redundant energy consumption caused by overall purification, improving the accuracy of protection and energy consumption optimization, and further improving the environmental protection function of the system.

[0064] Furthermore, the collaborative execution module also includes a health prediction unit, which has a built-in filter cartridge remaining life calculation model. This model calculates the remaining life of the filter cartridge based on three parameters: the pressure difference before and after the filter, the cumulative operating time, and the environmental pollution load.

[0065] Lifetime attrition rate: ,in For current detection of differential pressure, For initial pressure difference, The differential pressure threshold for scrapping;

[0066] The three-level maintenance warning is triggered based on the remaining lifespan percentage: remaining lifespan > 30% is normal, 15%-30% triggers a level one warning, 5%-15% triggers a level two warning, and ≤5% triggers a level three warning to prompt replacement of the filter element.

[0067] By adopting the above technical solutions, the functions of filter element remaining life prediction and maintenance early warning have improved the system's full life cycle operation and maintenance capabilities, effectively reducing equipment operation and maintenance costs and the risk of unplanned downtime, and further enhancing the practical value of the technical solutions.

[0068] The beneficial effects of this invention are:

[0069] 1. This invention achieves a technological paradigm shift from protecting the driver to ensuring the safety of critical equipment by configuring a distributed precision protection architecture for the core control cabinet, sensor cluster, and electric drive system of the mining truck. It also upgrades the air purification system from an auxiliary comfort system to a core mission support and equipment reliability support system, filling the technological gap in equipment environmental protection in unmanned mining scenarios.

[0070] 2. This invention integrates mobile monitoring data from multiple unmanned mining vehicles with data from fixed monitoring stations in the mining area, and uses spatial statistical methods to construct a real-time dynamic pollution model of the mining area, generating a high-precision, low-latency pollutant concentration distribution layer, providing accurate data support for subsequent path planning and equipment protection strategy generation;

[0071] 3. This invention achieves predictive purification control by adopting a model predictive control framework. It can predict the pollution level of the area that the mining truck is about to enter in advance, perform pressurization and high-efficiency purification operations on key equipment compartments in advance, and automatically switch to low-power maintenance mode in clean sections. This realizes the technological upgrade from reactive purification to predictive on-demand purification, and effectively optimizes the energy consumption of the entire vehicle.

[0072] 4. By introducing an environmental exposure cost factor into the traditional path planning cost function, this invention can dynamically plan the travel path that minimizes damage to equipment, reduce the exposure time of core equipment in highly polluted environments, effectively extend the service life of core components, and reduce equipment operation and maintenance costs.

[0073] 5. This invention achieves coordinated control of air purification power adjustment and path replanning by setting a strategy adjustment module, which can optimize the energy consumption of the entire vehicle while ensuring that the core equipment is protected from pollution, and effectively improve the adaptability and operational reliability of the unmanned mining truck in harsh mining environments.

[0074] 6. This invention uses unmanned mining vehicles as mobile environmental monitoring nodes, and utilizes vehicle sensor arrays to collect real-time environmental data of the entire mining area. The data is then uploaded to a cloud-based digital twin platform to generate a real-time air quality heat map. This can simultaneously support multiple application scenarios such as dynamic path planning, environmental compliance report generation, and optimization of mining production processes, thus expanding the application value boundaries of the system. Attached Figure Description

[0075] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0076] Figure 1 This is a schematic diagram of the overall architecture of the present invention;

[0077] Figure 2 This is a system operation flowchart of the present invention;

[0078] Figure 3 This is a comparison diagram of environmental exposure before and after path replanning in an embodiment of the present invention;

[0079] Figure 4 This is a graph showing the filter element pressure difference increase and maintenance early warning results in an embodiment of the present invention. Detailed Implementation

[0080] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. For some well-known technologies, those skilled in the art may also use other alternative methods to implement the invention. Moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0081] like Figure 1 , Figure 2 As shown, the present invention provides a method for onboard environmental monitoring and dynamic path planning of an unmanned mining truck, characterized by comprising the following steps:

[0082] S1. Environmental Data Acquisition: The environmental data acquisition module, composed of a gas detection unit, a dust detection unit, and an intelligent online detection unit, collects data on harmful gases, dust concentrations, vehicle location, and environmental images around the mining truck, and uploads them to the cloud database, while simultaneously accessing map database data.

[0083] S2. Dynamic pollution model construction: Based on multi-source data from a cloud database, a dynamic pollution model of the mining area is constructed.

[0084] S3. Pollution Level Classification: Based on the output results of the dynamic pollution model, the air pollution level of each area in the mining area is classified.

[0085] S4. Purification power level control: Determine the pollution level of the mine car at its current location. If it is low pollution, control the air purification system to operate at low power. If it is medium pollution, control the air purification system to operate at medium power. If it is high pollution, control the air purification system to operate at high power.

[0086] S5. High-pollution entry prediction: Summarize the pollution levels of the locations of unmanned mining trucks in the entire mining area, and predict whether the mining trucks are about to enter high-pollution areas through the dynamic pollution model prediction and analysis module.

[0087] S6. Collaborative Strategy Execution: If it is determined that the vehicle will not enter a high-pollution area, the current purification strategy and driving path will be maintained; if it is determined that the vehicle is about to enter a high-pollution area, the air purification system will be controlled to operate at high power first, and then the driving path of the mining truck will be replanned. The path planning results will be synchronously transmitted back to the cloud database to form a closed loop.

[0088] In step S1, the environmental data acquisition process is achieved through the collaborative deployment of multiple modules in a distributed manner. Specifically, gas detection units and dust detection units are deployed at key locations such as the front of the mining truck, both sides of the truck body, and the power compartment to form an omnidirectional sensing array. The intelligent online detection unit is integrated into the roof-mounted sensing platform. The gas detection unit uses a combination of electrochemical and infrared sensors to simultaneously collect the concentrations of harmful gases such as carbon monoxide, sulfur dioxide, nitrogen oxides, and volatile organic compounds. The dust detection unit uses a laser scattering principle sensor to collect the concentrations of PM2.5, PM10, and total suspended particulate matter. The intelligent online detection unit obtains centimeter-level vehicle location data through a GPS / BeiDou dual-mode positioning module and collects surrounding environmental image data through a high-definition camera. All collected multi-source data are first preprocessed by the on-board edge computing unit to complete timestamp alignment, outlier removal, and format standardization. Then, they are uploaded to the cloud database with low latency through the mining area's 5G / V2X private network. The cloud synchronously accesses the basic geographic data of the mining area's existing high-precision map database, providing a unified spatiotemporal reference data source for the subsequent construction of dynamic pollution models.

[0089] In steps S2, S3, S4 and S5, the module includes a data fusion unit, a spatial interpolation unit, a time series analysis unit and a map database update unit. The Kriging interpolation algorithm is used to perform spatial interpolation on discrete monitoring point data to generate a continuous pollutant concentration distribution map.

[0090] The spatial interpolation unit uses ordinary kriging interpolation to spatially interpolate discrete monitoring point data. The variogram model is selected based on the spatial autocorrelation characteristics of the monitoring data. Specifically, the experimental variogram is first calculated using sample data to obtain the semivariogram distribution under different lag distances. Then, a spherical model, an exponential model, or a Gaussian model are used for fitting, with the minimum sum of squared residuals, the highest coefficient of determination, and the minimum cross-validation error used as the model selection criteria. When pollutant concentration changes rapidly over short distances and spatial continuity is moderate, the spherical variogram model is preferred; when pollutant diffusion has obvious local abrupt changes and the correlation gradually decreases with distance, the exponential variogram model is preferred; when the pollutant field distribution is relatively smooth and the concentration change is relatively continuous, the Gaussian variogram model is preferred.

[0091] Simultaneously, the nugget value, sill value, and range parameter of the variogram are jointly estimated. When the ratio of the nugget value to the sill value is less than 25%, it is determined to be spatially strong correlation; when the ratio is between 25% and 75%, it is determined to be spatially moderate correlation; and when the ratio is greater than 75%, it is determined to be spatially weak correlation. For areas with weak spatial correlation, the interpolation accuracy is improved by increasing sampling points or introducing data from fixed monitoring stations to constrain the data.

[0092] To ensure the stability and reliability of ordinary Kriging interpolation results, discrete monitoring point data must meet the minimum sampling density requirements: within the area to be modeled, there should be no fewer than 5 effective monitoring points per square kilometer, and no fewer than 8 effective sampling points within the search neighborhood of any interpolation location; preferably, 8 to 15 effective monitoring points per square kilometer. For key areas with significant pollution gradient changes, such as loading and unloading areas, blasting areas, crushing stations, and intersections of main transportation routes, the local sampling density should be increased to 10 to 20 effective monitoring points per square kilometer, with the distance between adjacent sampling points preferably not exceeding 1 / 2 of the range. If the number of effective sampling points is lower than the above thresholds, only low-confidence pollution distribution results will be output, or compensation updates will be performed based on historical time-series prediction results.

[0093] The air pollution level classification module is used to classify the air pollution level of each area in the mining area according to the dynamic pollution model. The criteria for determining a low pollution level are as follows: Furthermore, the concentration of harmful gases meets the standards, corresponding to the low-power operation mode of the air purification system;

[0094] Criteria for determining medium pollution level: Or, if the level of a single harmful gas exceeds the standard by ≤50%, the corresponding power operation mode of the air purification system should be selected.

[0095] Criteria for determining high pollution levels: If the concentration of a single harmful gas exceeds the standard by more than 50%, or if multiple gases exceed the standard simultaneously, the corresponding air purification system will operate in high-power mode.

[0096] The control module adjusts the operating power of the air purification system based on the air pollution level at the current location of the mining truck. When the mining truck enters areas with different pollution levels, the operating power of the air purification system automatically switches, achieving adaptive adjustment.

[0097] The model prediction and analysis module, based on the Model Predictive Control (MPC) algorithm and combining the current position, current speed, planned path, high-precision map, and dynamic pollution model of the mining truck, performs rolling predictions of the discrete road segments the truck will traverse within the future prediction time domain. This yields the comprehensive probability that the truck will enter a high-pollution area within the next 5 minutes, and determines whether the truck is about to enter such an area. The preferred prediction time domain is 5 minutes, and the preferred control step size is 10s–30s.

[0098] Specifically, the prediction time-domain discretization is The control step, calculate the... The probability of entering a high-pollution zone at the corresponding position of each control step. ,in, For the first Predicting pollution concentrations The threshold for high pollution concentration;

[0099] The overall probability of entering a highly polluted area within the future time domain is predicted as follows: Under normal transportation tasks At that time, it was determined that the mining truck was about to enter a highly polluted area; high-value task threshold. Lowered to 0.40-0.50, threshold for time-priority tasks. Adjusted upwards to 0.70-0.80.

[0100] To convert the comprehensive probability Pin into a risk assessment result, a probability assessment threshold Pth is set. The probability assessment threshold Pth is a risk assessment parameter preset by the system and can be calibrated and determined based on historical operating data, simulation test results, equipment environmental protection safety margin, and task priority.

[0101] Preferably, a baseline threshold P0 is first determined based on historical operation samples and simulation results under ordinary transportation tasks, and then the baseline threshold is dynamically corrected according to the task priority to obtain the probability judgment threshold Pth corresponding to the current task.

[0102] when ≥ At that time, it was determined that the mining truck was about to enter a highly polluted area; when < At that time, it was determined that there was no significant risk of mining trucks entering high-pollution areas within the current prediction time domain.

[0103] The probability determination threshold The threshold is dynamically adjusted based on task priority. For ordinary transportation tasks, a baseline threshold of 0.60 is preferred; for high-value load transportation tasks, critical equipment support tasks, or tasks operating in severe weather, a threshold of 0.40–0.50 is preferred; and for time-sensitive tasks such as emergency hazard mitigation, repair, and supply guarantee, a threshold of 0.70–0.80 is preferred. In other words, the higher the task safety priority, the lower the probability determination threshold; the higher the task timeliness priority, the higher the probability determination threshold.

[0104] In step S6, the path planning in step S6 adopts a comprehensive cost function formula, specifically:

[0105] Construct the path synthesis cost function: ,in For the cost of time, For the cost of distance, As a result of environmental exposure, the environmental risk coefficients for low-pollution, medium-pollution, and high-pollution areas are 1, 2-3, and 4-6, respectively.

[0106] The improved A* algorithm or Dijkstra's algorithm is used to find the driving path with the lowest overall cost as the planning result. The path planning result is synchronously sent back to the cloud database to update the data link and complete the closed-loop operation.

[0107] The path planning specifically refers to:

[0108] S61. Define the node cost calculation rules:

[0109] When using the A* algorithm for path search, the node evaluation function satisfies: ;

[0110] in, The path from the starting point to the current node The cumulative comprehensive cost is calculated using the following formula: ;

[0111] in, For the current node The estimated total cost to the destination of the route is calculated using the following formula: ;

[0112] When using the improved Dijkstra's algorithm for path search, the cumulative cost per node is used. As a core indicator for node evaluation;

[0113] in, The path from the starting point to the current node The total number of road segments included in the planned route; , , These are preset time cost weights, distance cost weights, and environmental exposure cost weights, which are dynamically adjusted according to the priority of the unmanned mining truck transportation task. , , Each is the current node Estimated travel time to the end of the route, estimated travel distance, and estimated environmental exposure cost;

[0114] S62. Define the single-segment cost calculation rule: For any segment in the path... For each road segment, the calculation method for the three types of costs is as follows:

[0115] Time cost satisfied: ,in This is the actual length of the road section. The design speed for unmanned mining trucks on this section of road;

[0116] Distance cost satisfies: ;

[0117] Environmental exposure cost satisfied: ; represents the estimated travel time of the unmanned mining truck on this section of road. This is the environmental risk coefficient corresponding to the road section, which corresponds to the real-time pollution level of the road section.

[0118] S63. Path Search Iteration Rule: When performing path search using the improved Dijkstra algorithm, the cumulative comprehensive cost is selected in each iteration. The smallest unvisited node is expanded until the target destination is reached, thus obtaining the initial optimal driving path;

[0119] S64. Dynamic Replanning Triggering Rule: After the dynamic pollution model of the mining area completes its periodic update, the edge weights of all road segments affected by the changes in pollution distribution are automatically recalculated. If the change in the edge weight of any road segment in the planned driving path exceeds the preset threshold, the path replanning process is immediately triggered, and the calculation logic of S61 to S63 is repeated to iteratively obtain the updated optimal driving path that integrates environmental protection requirements and transportation efficiency requirements.

[0120] The present invention further discloses an unmanned mining truck on-board environmental monitoring and dynamic path planning system.

[0121] System overall architecture corresponding instruction manual attached Figure 1As shown, the system includes a data acquisition module, a dynamic pollution model construction module, an air pollution level classification module, a control module, a dynamic pollution model prediction and analysis module, a route planning module, and a strategy adjustment module. Each module interacts locally via in-vehicle Ethernet or CAN bus, and simultaneously synchronizes data with a cloud-based digital twin platform via 4G / 5G networks. The complete system operation flow is detailed in the attached instruction manual. Figure 2 As shown, the specific technical solutions for each module are as follows:

[0122] The data acquisition module is distributed and deployed in key locations such as the front, sides, and roof of the unmanned mining truck to achieve omnidirectional perception of the surrounding environment. After the collected data is preprocessed by the on-board edge unit, it is used for local real-time control decision-making and uploaded to the cloud database to participate in the construction of dynamic pollution models.

[0123] The data acquisition module includes a gas detection unit, a dust detection unit, and an intelligent online detection unit:

[0124] The gas detection unit employs a combination of electrochemical and infrared sensors, enabling simultaneous detection of carbon monoxide (CO), sulfur dioxide (SO2), and nitrogen oxides (NOx). x ), volatile organic compound (VOCs) concentration, with detection accuracy at the ppm level and a response time ≤30s;

[0125] The dust detection unit uses a particulate sensor based on the principle of laser scattering, which can detect the concentration of PM2.5, PM10, and total suspended particulate matter (TSP) in real time. The sensor has a built-in automatic calibration function to ensure measurement accuracy during long-term operation.

[0126] The intelligent online detection unit integrates a GPS / BeiDou dual-mode positioning module and a high-definition camera, which can acquire centimeter-level vehicle location data and environmental image data in real time to meet the accuracy requirements of autonomous driving scenarios. Among them, the environmental image data is only used for dust source identification, abnormal environmental event judgment, and auxiliary marking of dust diffusion areas and pollution scenes. It is not used as a direct quantitative measurement basis for pollution concentration. The quantitative data of pollution concentration is based on the detection results of the gas detection unit and the dust detection unit.

[0127] The dynamic pollution model construction module receives mobile monitoring data from multiple unmanned mining vehicles and fixed-point monitoring data from fixed monitoring stations in the mining area. It constructs a dynamic pollution model of the mining area through four steps: data fusion, spatial interpolation, time-series analysis, and map updating. Based on the output of the dynamic pollution model, the air pollution level classification module divides the mining area's pollution level into three levels:

[0128] Low pollution level: PM10 ≤ 150 μg / m³ and harmful gas concentrations meet standards;

[0129] Medium pollution level: 150μg / m³ < PM10 ≤ 500μg / m³ or single harmful gas exceeding the standard by ≤ 50%;

[0130] High pollution level: PM10 > 500 μg / m³ or single harmful gas exceeds the standard by more than 50%, or multiple harmful gases exceed the standard simultaneously.

[0131] The construction logic of the dynamic pollution model is as follows: time alignment, spatial registration, and quality verification are performed on asynchronously collected multi-source data, and outliers are removed to form a unified spatiotemporal dataset; among them, data from fixed monitoring stations are used as the baseline data source, with a preferred weight of 0.55~0.75, and mobile monitoring data from mining trucks are used as a dynamic supplementary data source, with a preferred weight of 0.25~0.45. The weights can be dynamically adjusted according to the sampling time difference, spatial distance, and data quality score; outliers are jointly determined by physical range threshold, adjacent time jump threshold, and statistical deviation threshold: when the monitored value exceeds the physical reasonable range, or the relative change rate of PM10 in adjacent sampling periods is >200% / absolute change is >300μg / m³, or the relative change rate of a single harmful gas concentration in adjacent sampling periods is >150%, or the current value deviates from the sliding window mean by more than 3 times the standard deviation, it is determined as an outlier and removed.

[0132] The ordinary Kriging interpolation algorithm is used to spatially interpolate discrete monitoring point data based on the variogram model to generate a continuous raster map of pollutant concentration distribution.

[0133] The time series analysis method is used to model the temporal variation of pollutant concentration and predict the pollution diffusion trend in the next 1-2 hours. The time series analysis method can be an ARIMA model or an LSTM neural network model. When using an ARIMA model, it is preferred to construct an ARIMA(p,d,q) model, where the preferred values ​​of p, d, and q are 1-5, 0-2, and 1-5, respectively.

[0134] When using an LSTM neural network model, it is preferable to construct a network structure with 2-3 LSTM hidden layers and 1 fully connected output layer. The number of neurons in the hidden layer is preferably 64-128, 32-64, and 16-32, respectively. The state activation function is preferably tanh, the gate activation function is preferably sigmoid, and the output layer is preferably a linear activation function.

[0135] The training dataset consists of mobile monitoring data from unmanned mining vehicles, fixed-point data from fixed monitoring stations, pollutant concentration raster data generated by spatial interpolation, and auxiliary environmental features. Input and output samples are constructed using a sliding time window method. The input sequence is preferably historical data from the previous 12 to 24 time steps, and the output sequence is preferably predicted values ​​from the next 2 to 12 time steps. Before model training, the data needs to be processed by filling in missing values, removing outliers, and normalizing, and then divided into training set, validation set, and test set.

[0136] The generated pollution distribution data is synchronized and updated to the map database in real time for use by various functional modules.

[0137] The air purification system adopts a distributed architecture, with dedicated purification modules designed to meet the protection needs of different core devices, and is driven by a control module.

[0138] Dedicated purification module for core control cabinet: An independent sealed cabinet is set up for the mine car computing unit (industrial computer, domain controller) and communication equipment. The internal part integrates a small, high-efficiency active air purification and temperature control system, which can continuously maintain a positive pressure of 50~100Pa and a clean environment inside the cabinet, ensuring the stable operation of the core control equipment in harsh environments.

[0139] Sensor cluster self-cleaning system: It equips key sensors such as lidar, camera, and millimeter-wave radar with protective covers with micro air curtains or periodic high-pressure air purging functions. The air curtain system can continuously blow clean air to the sensor surface to form a protective barrier. The purging system can be automatically activated when the vehicle enters a high-dust area to periodically remove the dust that has been attached.

[0140] Electric drive system protection unit: Equipped with an independent cooling and air filtration system for high-voltage components such as the battery pack and motor controller of the electric / hybrid mining truck, to prevent dust accumulation from causing overheating or short circuit faults, and to ensure the safe operation of the power system.

[0141] The dynamic pollution model prediction and analysis module uses a model predictive control (MPC) framework to achieve predictive pollution control. The MPC controller combines high-precision maps and transportation task plans to predict the road sections that vehicles will soon pass through. The specific control process is as follows:

[0142] First, based on the vehicle's current location and planned route, predict the road segments the vehicle will pass through in the next 5 to 10 minutes;

[0143] Then, query the dynamic pollution model to obtain the predicted pollution level distribution of the road segment;

[0144] Next, the equipment exposure risk and energy consumption cost under different control strategies are calculated;

[0145] Finally, the optimization problem is solved to determine the optimal air purification power adjustment strategy.

[0146] For example, when the system predicts that the vehicle will enter the high-dust loading and unloading area in 5 minutes, it will perform pressurization and high-efficiency purification operations on the critical equipment compartment in advance to ensure that the equipment is in the best protection state when entering the high-pollution area; in clean sections, it will automatically switch to low power consumption maintenance mode to effectively reduce the energy consumption of the whole vehicle.

[0147] The path planning module introduces an environmental exposure cost factor into the cost function of traditional path planning algorithms, realizing dynamic path planning that integrates environmental protection requirements and transportation efficiency requirements. The specific logic is as follows:

[0148] Construct the comprehensive cost function: ,in For the cost of time, For the cost of distance, As a result of environmental exposure, the environmental risk coefficients for low-pollution, medium-pollution, and high-pollution areas are 1, 2-3, and 4-6, respectively.

[0149] In peak transportation scenarios, the weighting of time cost can be increased. And correspondingly reduce the weight of environmental exposure costs. The environmental exposure cost weighting can be increased during periods of high pollution incidence, post-blast dust diffusion, high-dust operation periods in loading and unloading areas, and in severe weather scenarios. And correspondingly reduce the weight of time cost. ;

[0150] Environmental exposure costs Based on the pollution level, segment length, and expected travel time of each segment of the candidate route, the preferred route is selected based on the following criteria:

[0151] ,in Let i be the environmental risk coefficient of the i-th road segment. Let be the length of the i-th road segment. The expected travel time for the i-th road segment is denoted as ; the environmental risk coefficients for low-pollution, medium-pollution, and high-pollution zones are 1, 2~3, and 4~6, respectively. When a single harmful gas exceeds the standard or multiple gases exceed the standard simultaneously, risk gains can be added on top of the corresponding level.

[0152] Path search solution: An improved A algorithm or Dijkstra's algorithm is used to search and solve the road network in the mining area, incorporating environmental exposure costs into the edge weights. When using the improved A algorithm, the node evaluation function satisfies... ;

[0153] in ; represents the cumulative total cost from the starting point of the path to the current node n. , which is the estimated total cost from the current node n to the end of the path;

[0154] When using the improved Dijkstra's algorithm, the cumulative total cost is calculated based on the nodes. As a core indicator for node evaluation, each iteration selects the unvisited node with the lowest cumulative comprehensive cost for expansion.

[0155] The calculation rules for the three types of costs for a single road segment are as follows: Time cost Distance Cost Environmental exposure costs ,in For the first The actual length of each road segment The design speed for unmanned mining trucks on this section of road. This is the estimated travel time for this section of road. This represents the environmental risk coefficient for this road section.

[0156] Dynamic replanning trigger: After the dynamic pollution model completes its periodic update, it automatically recalculates the edge weights of all road segments affected by changes in pollution distribution. If the change in the edge weight of any road segment in the planned path exceeds a preset threshold, the path replanning process is immediately triggered to iterate and obtain the optimal driving path that integrates environmental protection and transportation efficiency.

[0157] The strategy adjustment module achieves coordinated control of air purification power adjustment and path replanning. When the dynamic pollution model prediction and analysis module determines that the mining truck is about to enter a high-pollution area, it triggers a collaborative decision-making process: the collaborative decision-making unit evaluates the comprehensive cost of two response strategies. Strategy A is to maintain the current path and only increase the air purification power to high power mode, while strategy B is to replan the driving path and adjust the air purification power. The decision is based on factors such as the environmental exposure cost of the current path, the feasibility of alternative paths, and the remaining processing capacity of the air purification system. The energy consumption optimization unit selects the response strategy with the lowest comprehensive energy consumption cost while ensuring the safety of key equipment.

[0158] The strategy adjustment module also has a built-in health prediction unit, which can predict the remaining service life of the filter element based on the pressure difference before and after the air purification system filter, cumulative operating time, and environmental pollution load, and output graded maintenance early warning information: the filter element life loss rate meets the requirements. ,in For the initial pressure difference of the filter element, For current detection of differential pressure, The differential pressure threshold for scrapping;

[0159] Combined with the cumulative running time of the filter element With pollution load factor After correction, the remaining lifespan of the filter element meets the requirements. ,in For the weighting coefficients, the preferred ones should satisfy... .

[0160] Maintenance alerts are triggered according to a tiered rule: when or This is the normal state; when or A Level 1 warning is triggered at this time;

[0161] when or A level-two warning is triggered at this time; when The system will trigger a Level 3 warning, prompting you to replace the filter element immediately.

[0162] Through the coordinated operation of the above modules, this system can achieve closed-loop control of the entire chain of unmanned mining trucks, including environmental perception, predictive protection, and dynamic path planning. It optimizes operating energy consumption while protecting core equipment from pollution and corrosion, thus meeting the comprehensive needs of unmanned mine operation.

[0163] In one embodiment, refer to Figure 3 As can be seen, the application scenario of this embodiment is as follows: After a blasting operation in an open-pit mine, a high dust diffusion zone forms downwind of the blasting area; the main transportation road A in the mining area is the commonly used route for unmanned mining vehicles to travel from the loading area to the crushing station, with a total length of 2.8km, and the alternative detour road B has a total length of 3.4km. Within 5 minutes after the blasting, monitoring data uploaded by multiple unmanned mining vehicles and fixed monitoring stations show that the PM10 concentration in the vicinity of road A rapidly increases to 650~820μg / m³, accompanied by a short-term increase in NOx concentration in some areas; the PM10 concentration along road B remains stable at 120~180μg / m³. The execution flow of the technical solution in this embodiment is as follows:

[0164] Environmental data acquisition: The environmental data acquisition module deployed on the unmanned mining truck synchronously collects PM10, TSP, and NOx concentration data, vehicle centimeter-level positioning data, and surrounding environmental image data along roads A and B. After time alignment and outlier removal preprocessing by the on-board edge unit, the data is uploaded to the cloud database through the mining area's 5G private network. The cloud database simultaneously accesses the basic geographic data of the mining area's high-precision map database.

[0165] Dynamic pollution model construction: The dynamic pollution model construction module performs weighted fusion of multi-vehicle mobile monitoring data and fixed monitoring station data, with a weight of 0.65 for fixed monitoring stations and 0.35 for mobile monitoring. After quality verification and removal of outliers, a unified spatiotemporal dataset is formed. A raster map of continuous pollutant concentration distribution in the mining area after blasting is generated using the ordinary Kriging interpolation algorithm. The time series analysis unit, based on the ARIMA model and combined with historical data from the first 15 minutes after blasting, predicts the pollution diffusion trend within the next 30 minutes, determining that Road A will remain in a high-pollution state. The pollution distribution data is updated to the map database in real time.

[0166] High-pollution entry risk prediction: The dynamic pollution model prediction and analysis module, based on the MPC framework, calculates the comprehensive probability of the unmanned mining truck entering a high-pollution area within the next 5 minutes by combining the truck's current location, speed, and planned transportation task. If the value is higher than the threshold of 0.60 for ordinary transportation tasks, it is determined that the mining truck is about to enter a highly polluted area.

[0167] Collaborative strategy execution: First, the purification system is pre-regulated, switching the core control cabinet purification module, sensor cluster self-cleaning system, and electric drive system protection unit to high-power operation mode; then, the path planning module calls the improved A* algorithm that integrates environmental exposure costs, where the environmental exposure cost weight is 0.45, the time cost weight is 0.35, and the distance cost weight is 0.20. After calculation, the comprehensive cost of road A is significantly higher than that of road B, so the system triggers path replanning and selects road B as the new transportation path.

[0168] After the mine truck detoured along road B, the transportation time for a single task only increased by 6% to 10% compared to the original route. However, the dust deposition rate on the surface of key sensors decreased significantly, and the particulate matter concentration in the core control cabinet remained within the safe threshold for a long time. This avoided the continuous erosion of lidar, cameras, and electronic control units by the highly polluted road section after blasting, and achieved a dynamic balance between transportation efficiency and equipment protection.

[0169] In one embodiment, refer to Figure 4 It is known that multiple excavating and transferring equipment in the loading and unloading area of ​​a certain open-pit coal mine operate continuously, resulting in a prolonged medium-to-high dust environment near the loading and unloading point. Unmanned mining trucks need to frequently travel between the loading and unloading area and the spoil heap, and due to road conditions, they cannot completely avoid the loading and unloading area in the short term. Monitoring data around the loading and unloading area shows that PM10 concentrations consistently remain between 420 and 560 μg / m³, and TSP concentration peaks exceed 800 μg / m³. PM10 concentrations on other sections of the transportation route remain between 80 and 150 μg / m³. The execution flow of the technical solution in this embodiment is as follows:

[0170] Risk prediction: When the unmanned mining truck enters 300m in front of the loading and unloading area, the dynamic pollution model prediction and analysis module, combined with high-precision map, historical traffic speed and dynamic pollution model, predicts that the vehicle will enter the medium-to-high pollution area within 90s. The calculated comprehensive probability of high pollution entering is 0.72, which exceeds the judgment threshold.

[0171] Collaborative Decision-Making: Since this is a continuous shipping task with limited alternative routes, the strategy adjustment module comprehensively evaluates the combined costs of the two response strategies.

[0172] Strategy A involves maintaining the current path while increasing purification power, while Strategy B involves replanning a detour route. Ultimately, Strategy A, which has a lower overall cost, was chosen.

[0173] Purification power regulation: The control module switches the core control cabinet's dedicated purification module to a high-power positive pressure mode to maintain an internal pressure of 100Pa;

[0174] The air curtain intensity of the sensor cluster self-cleaning system was increased by 50%, and the blowing interval of the lidar and camera protective cover was shortened from 30 minutes to 2 minutes;

[0175] The electric drive system protection unit simultaneously increases the filtration level and cooling intensity to reduce the risk of dust entering the battery pack and motor controller.

[0176] Filter life warning: As the vehicle passes through the loading and unloading area, the health prediction unit monitors the pressure difference before and after the filter in real time. With continuous operation over multiple shifts, the current pressure difference... Rise to Combined with cumulative running time and pollution load factors The remaining lifespan of the filter element was calculated. The system triggers a level-two warning and uploads the warning information to the fleet dispatch platform. It recommends replacing the filter element at the maintenance window after the shift ends. The mine car completes the transportation task without affecting the loading and unloading operation rhythm, avoiding unplanned shutdowns caused by filter element failure, sensor mirror contamination, or dust ingress into the core control cabinet. The equipment failure rate is reduced by more than 20% compared to the traditional solution.

[0177] Working principle: By distributing gas detection units, dust detection units, and intelligent online detection units on unmanned mining trucks, an omnidirectional environmental perception link is constructed. This simultaneously collects multi-source data such as the concentration of harmful gases, particulate matter, vehicle centimeter-level location, and environmental images. After preprocessing, the data is uploaded to the cloud and integrated with benchmark monitoring data from fixed monitoring stations in the mining area. Through multi-source data weighted fusion, ordinary Kriging spatial interpolation, and ARIMA or LSTM time-series prediction processes, a low-latency, high-precision dynamic pollution spatiotemporal distribution model of the mining area is constructed, and regional pollution level classification is completed. This is based on the current location of the mining truck. Based on the pollution level, the distributed air purification system designed for the core control cabinet, sensor cluster, and electric drive system is subject to graded power regulation. On the other hand, the model predictive control (MPC) framework is used to predict the probability of mining trucks entering high-pollution areas in the future. If it is determined that there is a risk of high pollution entering the area, the high-power purification mode is activated in advance to protect the core equipment. Then, the improved A* or Dijkstra path planning algorithm that integrates environmental exposure cost factors is called to comprehensively weigh the costs of time, distance, and environmental exposure to solve the optimal driving path. The path planning results are synchronously transmitted back to the cloud to form a data closed loop.

[0178] This approach overcomes the shortcomings of traditional solutions, which lack environmental sensing sensors and the ability to detect pollutants in mining areas. It also breaks through the limitations of traditional path planning, which focuses solely on traffic accident risk. It can proactively plan paths to avoid highly polluted areas and reduce the exposure time of core equipment to pollution. Secondly, through a combination of graded purification and predictive pre-control, it achieves a technological upgrade from passive reactive purification to proactive on-demand purification, avoiding ineffective energy consumption. At the same time, the distributed protection architecture for different core equipment can effectively block the erosion of dust and corrosive gases. Finally, through the collaborative decision-making of purification power adjustment and path replanning, it achieves a multi-objective balance of equipment reliability, energy consumption control, and transportation efficiency, meeting the comprehensive needs of unmanned mine operation.

[0179] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for onboard environmental monitoring and dynamic path planning of an unmanned mining truck, characterized in that, Includes the following steps: S1. Environmental Data Acquisition: The environmental data acquisition module, composed of a gas detection unit, a dust detection unit, and an intelligent online detection unit, collects data on harmful gases, dust concentrations, vehicle location, and environmental images around the mining truck, and uploads them to the cloud database, while simultaneously accessing map database data. S2. Dynamic pollution model construction: Based on multi-source data from a cloud database, a dynamic pollution model of the mining area is constructed. S3. Pollution Level Classification: Based on the output results of the dynamic pollution model, the air pollution level of each area in the mining area is classified. S4. Purification power level control: Determine the pollution level of the mine car at its current location. If it is low pollution, control the air purification system to operate at low power. If it is medium pollution, control the air purification system to operate at medium power. If it is high pollution, control the air purification system to operate at high power. S5. High-pollution entry prediction: Summarize the pollution levels of the locations of unmanned mining trucks in the entire mining area, and predict whether the mining trucks are about to enter high-pollution areas through the dynamic pollution model prediction and analysis module. S6. Cooperative Strategy Execution: If it is determined that the vehicle will not enter a high-pollution area, maintain the current purification strategy and driving path; If it is determined that the vehicle is about to enter a highly polluted area, the air purification system is first controlled to operate at high power, and then the driving route of the mining truck is replanned. The route planning results are synchronously transmitted back to the cloud database to form a closed loop.

2. The unmanned mining truck on-board environmental monitoring and dynamic path planning method according to claim 1, characterized in that, The data acquisition and dynamic pollution model construction in steps S1 to S2 are as follows: the gas detection unit collects concentration data of carbon monoxide, sulfur dioxide, nitrogen oxides, and volatile organic compounds; the dust detection unit collects concentration data of PM2.5, PM10, and total suspended particulate matter; and the intelligent online detection unit collects centimeter-level positioning data and environmental image data. After preprocessing, all data are uploaded to the cloud. After integrating the data from fixed monitoring stations, a continuous pollutant concentration distribution raster map is generated using a common Kriging interpolation algorithm. The map is updated to the map database with a delay of less than 10 seconds to form a dynamic pollution spatiotemporal distribution model.

3. The unmanned mining truck on-board environmental monitoring and dynamic path planning method according to claim 1, characterized in that, The pollution level classification and power regulation rules in steps S3 to S4 are as follows: Criteria for determining low pollution levels: Furthermore, the concentration of harmful gases meets the standards, corresponding to the low-power operation mode of the air purification system; Criteria for determining medium pollution level: Or, if the level of a single harmful gas exceeds the standard by ≤50%, the corresponding power operation mode of the air purification system should be selected. Criteria for determining high pollution levels: If the concentration of a single harmful gas exceeds the standard by more than 50%, or if multiple gases exceed the standard simultaneously, the corresponding air purification system will operate in high-power mode.

4. The unmanned mining truck on-board environmental monitoring and dynamic path planning method according to claim 1, characterized in that, The prediction of high-pollution zone entry in step S5 uses a probability calculation formula, specifically: Discretize the 5-minute prediction time domain as follows: The control step, calculate the... The probability of entering a high-pollution zone at the corresponding position of each control step. ,in, For the first Predicting pollution concentrations The threshold for high pollution concentration; Calculate the overall probability of entering a highly polluted area within the predicted time domain: Under normal transportation tasks When the time is right, it is determined that the mining truck is about to enter a highly polluted area; the threshold for high-value tasks is lowered to 0.40-0.50, and the threshold for time-priority tasks is raised to 0.70-0.

80.

5. The unmanned mining truck on-board environmental monitoring and dynamic path planning method according to claim 1, characterized in that, The path planning in step S6 uses a comprehensive cost function formula, specifically: Construct the path synthesis cost function: ,in For the cost of time, For the cost of distance, As a result of environmental exposure, the environmental risk coefficients for low-pollution, medium-pollution, and high-pollution areas are 1, 2-3, and 4-6, respectively. The environmental exposure cost Based on the pollution level, segment length, and expected travel time of each segment of the candidate route, the following conditions are met: ,in Let i be the environmental risk coefficient of the i-th road segment. Let be the length of the i-th road segment. Let be the expected travel time for the i-th road segment; The improved A* algorithm or Dijkstra's algorithm is used to find the driving path with the lowest overall cost as the planning result. The path planning result is synchronously sent back to the cloud database to update the data link and complete the closed-loop operation.

6. The unmanned mining truck on-board environmental monitoring and dynamic path planning method according to claim 5, characterized in that, In step S6, the path planning specifically involves: S61. Define the node cost calculation rules: When using the A* algorithm for path search, the node evaluation function satisfies: ; in, The path from the starting point to the current node The cumulative comprehensive cost is calculated using the following formula: ; in, For the current node The estimated total cost to the destination of the route is calculated using the following formula: ; When using the improved Dijkstra's algorithm for path search, the cumulative cost per node is used. As a core indicator for node evaluation; in, The path from the starting point to the current node The total number of road segments included in the planned route; , , These are preset time cost weights, distance cost weights, and environmental exposure cost weights, which are dynamically adjusted according to the priority of the unmanned mining truck transportation task. , , Each is the current node Estimated travel time to the end of the route, estimated travel distance, and estimated environmental exposure cost; S62. Define the single-segment cost calculation rule: For any segment in the path... For each road segment, the calculation method for the three types of costs is as follows: Time cost satisfied: ,in This is the actual length of the road section. The design speed for unmanned mining trucks on this section of road; Distance cost satisfies: ; Environmental exposure cost satisfied: ; represents the estimated travel time of the unmanned mining truck on this section of road. This is the environmental risk coefficient corresponding to the road section, which corresponds to the real-time pollution level of the road section. S63. Path Search Iteration Rule: When performing path search using the improved Dijkstra algorithm, the cumulative comprehensive cost is selected in each iteration. The smallest unvisited node is expanded until the target destination is reached, thus obtaining the initial optimal driving path; S64. Dynamic Replanning Triggering Rule: After the dynamic pollution model of the mining area completes its periodic update, the edge weights of all road segments affected by the changes in pollution distribution are automatically recalculated. If the change in the edge weight of any road segment in the planned driving path exceeds the preset threshold, the path replanning process is immediately triggered, and the calculation logic of S61 to S63 is repeated to iteratively obtain the updated optimal driving path that integrates environmental protection requirements and transportation efficiency requirements.

7. A system for onboard environmental monitoring and dynamic path planning for unmanned mining trucks, characterized in that, include: The environmental data acquisition module consists of a gas detection unit, a dust detection unit, and an intelligent online detection unit. It is deployed at key locations on the mine truck body to collect multi-source environmental and vehicle status data around the mine truck. A cloud-based database is used to store collected environmental and vehicle data, and synchronously accesses map database data. The dynamic pollution model construction module is used to generate a dynamic spatiotemporal distribution model of pollution in mining areas based on multi-source data in the cloud. The air pollution level classification module is used to classify the air pollution level of each area in the mining area; the graded control module is used to control the air purification system to operate in low power mode, medium power mode and high power mode according to the pollution level of the current location of the mining truck. The dynamic pollution model prediction and analysis module is used to summarize the pollution level data of all mining trucks and predict the probability that a mining truck will soon enter a high-pollution area. The collaborative execution module is used to select an execution strategy based on the prediction results: when there is no high pollution risk, the current strategy is maintained; when there is a high pollution risk, the purification system is first upgraded to high power operation, and then the driving path is replanned and the results are sent back to the cloud database.

8. The unmanned mining truck on-board environmental monitoring and dynamic path planning system according to claim 7, characterized in that, The environmental data acquisition module is specifically as follows: The gas detection unit adopts a combination of electrochemical and infrared sensors; The dust detection unit uses a sensor based on the principle of laser scattering. The intelligent online detection unit integrates a GPS or Beidou dual-mode positioning module with a high-definition camera, and uses environmental image data for dust source identification and auxiliary marking of pollution scenes.

9. The unmanned mining truck on-board environmental monitoring and dynamic path planning system according to claim 7, characterized in that, The dynamic pollution model construction module includes: The data fusion unit is used to perform time alignment, spatial registration, and outlier removal on asynchronously collected mobile monitoring data and fixed monitoring station data, and to perform weighted fusion with a weight of 0.55-0.75 for fixed monitoring stations and 0.25-0.45 for mobile monitoring data. Spatial interpolation unit, with built-in ordinary Kriging interpolation algorithm and spherical, exponential and Gaussian variogram models, generates continuous pollutant concentration distribution raster map; The time-series analysis unit, with a built-in ARIMA model or LSTM neural network model, predicts the pollution diffusion trend in the next 1-2 hours. The map update unit is used to synchronize pollution distribution data to the map database in real time. The air purification system adopts a distributed architecture, including: a dedicated purification module for the core control cabinet, which provides a 50-100Pa positive pressure clean environment for the computing unit and communication equipment; The sensor cluster self-cleaning system provides air curtain isolation and periodic purging for key sensors such as lidar and cameras. The purging interval is dynamically adjusted according to the pollution level. The electric drive system protection unit provides cooling and air filtration for the battery pack and motor controller, with the filtration level dynamically adjusted according to the pollution level.

10. The unmanned mining truck on-board environmental monitoring and dynamic path planning system according to claim 7, characterized in that, The collaborative execution module also includes a health prediction unit, which has a built-in filter cartridge remaining life calculation model. The model calculates the remaining life of the filter cartridge based on three parameters: the pressure difference across the filter, the cumulative operating time, and the environmental pollution load. Lifetime attrition rate: ,in For current detection of differential pressure, For initial pressure difference, The differential pressure threshold for scrapping; The three-level maintenance warning is triggered based on the remaining lifespan percentage: remaining lifespan > 30% is normal, 15%-30% triggers a level one warning, 5%-15% triggers a level two warning, and ≤5% triggers a level three warning to prompt replacement of the filter element.

Images