Mining area pavement maintenance deciding method and system based on pavement condition estimation, and storage medium

Abstract

A mining area pavement maintenance deciding method and system based on pavement condition estimation, and a storage medium, relating to the technical field of mining area pavement inspection. The method comprises: acquiring real-time relevant parameters of pavement condition collected by a vehicle while traveling in the current region of mining area pavement, and pre-processing the real-time relevant parameters; performing feature extraction on the pre-processed parameters to obtain a relevant parameter set; performing feature learning on the relevant parameter set on the basis of a historical sample to obtain a pavement condition estimated value of the current region, wherein the historical sample comprises a historical relevant parameter and a corresponding pavement condition value, and the pavement condition value represents the degree of pavement roughness; optimizing the pavement condition estimated value on the basis of historical condition of the current region; and generating a corresponding dispatching instruction for vehicle passage and pavement maintenance on the basis of the optimized pavement condition estimated value.

Description

A method, system, and storage medium for decision-making on road surface maintenance in mining areas based on road surface condition estimation.

[0001] Cross-references to related applications

[0002] This disclosure is based on and claims priority to Chinese application No. 202411765776.2, filed on December 4, 2024, the contents of which are incorporated herein by reference in their entirety. Technical Field

[0003] This disclosure belongs to the field of mining area road surface detection technology, and in particular relates to a mining area road surface maintenance decision-making method, system and storage medium based on road surface condition estimation. Background Technology

[0004] The definition of bumpy road surfaces in mining scenarios refers to the unevenness, potholes, cracks, and gravel on the road surface caused by long-term mining operations, crushing by mining trucks, falling slag, and erosion by rainwater. This causes the vehicle's center of gravity to fluctuate constantly during driving, and the direction and magnitude of the resistance it experiences to change continuously, thus affecting the vehicle's handling and system stability.

[0005] Although the unmanned roads in mining areas are designed with structured surfaces, their condition often becomes complex and variable due to long-term heavy loads and inherently unfavorable natural conditions. Potholes, falling rocks, and ruts created by vehicles pose serious challenges to vehicle power performance, system stability, and even overall safety. Therefore, accurate assessment and timely maintenance of bumpy road surfaces are crucial to ensuring the stable operation of unmanned driving systems. Summary of the Invention

[0006] Firstly, this disclosure provides a method for decision-making on road maintenance in mining areas based on road surface condition estimation, including:

[0007] The system acquires real-time road condition parameters collected from the current area of ​​the road surface where the vehicle is traveling in the mining area, and preprocesses the real-time parameters.

[0008] Feature extraction is performed on the preprocessed parameters to obtain the relevant feature set;

[0009] Based on historical samples, feature learning is performed on the relevant feature set to obtain the estimated road surface condition of the current area. The historical samples include historical relevant features and their corresponding road surface condition values, and the road surface condition values ​​characterize the degree of road surface bumpiness.

[0010] The road surface condition estimate is optimized based on the historical state of the current area;

[0011] Based on the optimized road condition estimates, corresponding dispatch instructions for vehicle passage and road maintenance are issued.

[0012] Optionally, the step of acquiring real-time road surface status parameters collected in the current area of ​​the mining area road surface when the vehicle is driving includes: collecting point cloud data, IMU data and wheel speed data respectively through lidar, IMU inertial measurement unit and wheel speed sensor installed on the vehicle during the vehicle's driving process.

[0013] Optionally, the preprocessing of the real-time relevant parameters includes:

[0014] Spatial processing is performed on the point cloud data of the current area collected by the lidar to obtain the ground point cloud set of the current area;

[0015] The inertial data and wheel speed data of the vehicle traveling in the current area are processed in time series to obtain time series IMU data and time series wheel speed data.

[0016] Optionally, the step of extracting features from the preprocessed parameters to obtain a relevant feature set includes:

[0017] The ground point cloud set is encoded using pointpillars to obtain a point cloud mesh map of the current area. The features of the point cloud within each mesh are calculated to obtain the discreteness of the road surface height, the degree of local road surface protrusion, and the local geometric information of the point cloud. The road surface features [H,W,C] are generated based on the calculated point cloud features, where H and W are the ranges of the point cloud, and C is the feature of the point cloud.

[0018] PCA principal component analysis is performed on the IMU data at each time step in the time series IMU data. The eigenvectors of the principal components are used as IMU data features, and the IMU data features are encoded according to the time dimension to generate an IMU feature matrix.

[0019] Based on the time-series wheel speed data, calculate the standard deviation of wheel speed and the first derivative or difference of wheel speed for each wheel during the current travel time in the current area to obtain wheel rotation stability and wheel speed change rate. Then encode the wheel rotation stability and wheel speed change rate according to the time dimension to generate a wheel speed feature matrix.

[0020] The road surface features, IMU feature matrix, and wheel speed feature matrix of the current area are combined into a relevant feature set of the road surface condition of the current area.

[0021] Optionally, the step of performing feature learning on the relevant feature set based on historical samples to obtain the road surface state estimate for the current area includes:

[0022] The relevant feature set is input into multiple pre-trained road surface state prediction models to obtain multiple road surface state prediction values;

[0023] The road condition estimate is obtained by averaging the multiple road condition prediction values.

[0024] The method for obtaining the multiple road surface condition prediction models includes:

[0025] Multiple road surface condition prediction models are obtained by training each pre-built decision tree on each historical sample in a pre-built historical sample set, wherein the road surface condition values ​​in the historical samples are in percentage form.

[0026] Optionally, during the construction of the historical sample set, the historical data related to the road surface condition of each area of ​​the mining area are sampled with replacement.

[0027] During the construction of each decision tree, some relevant features are randomly selected to perform optimal node partitioning.

[0028] Optionally, optimizing the road surface condition estimate based on the historical state of the current area includes:

[0029] Obtain the influence weights of each historical state in the current region on the current state. The formula for calculating the influence weights is: w t =λ(Tt),

[0030] Among them, w t Let λ be the weighting factor corresponding to time t, λ be the decay rate (0 < λ < 1), T be the current time, and n be the number of historical time points (Tn ≤ t). <T;

[0031] The influence weights are normalized, and the calculation formula is as follows:

[0032] in, These are the normalized weighting factors;

[0033] The road surface condition estimate is adjusted by a normalized weighting factor to obtain an optimized road surface condition estimate, calculated as follows:

[0034] Where IP represents the initial pavement condition estimate, AP is the optimized pavement condition estimate, α is an adjustment parameter used to control the influence of historical data on the initial pavement condition estimate, and S... t This represents the estimated road surface condition at time t.

[0035] Optionally, the step of issuing corresponding vehicle passage and road maintenance scheduling instructions based on the optimized road condition estimate includes:

[0036] Compare the optimized road surface state estimation value with the preset bump degree thresholds θ1 and θ2. If the bump degree S < θ1, it indicates that the road surface is flat and there is no need for maintenance temporarily. If θ1 < S < θ2, it indicates that the road surface has small bumps and undulations. Send a decision signal to the driverless mining truck to reduce the speed in the current area, and dispatch the idle grader on-site to perform ground maintenance work on the current area after the driverless mining truck passes. If S > θ2, it indicates that the road surface has large bumps and undulations. Send a decision signal to the driverless mining truck to detour in the current area, and dispatch the grader to repair the road surface condition in a timely manner.

[0037] In a second aspect, the present disclosure provides a mining area road surface maintenance decision-making system based on road surface state estimation, including:

[0038] Parameter preprocessing module: used to obtain the real-time relevant parameters of the road surface state collected when the vehicle is driving in the current area of the mining area road surface, and preprocess the real-time relevant parameters;

[0039] Feature extraction module: used to extract features from the preprocessed parameters to obtain a relevant feature set;

[0040] Road surface state estimation module: used to perform feature learning on the relevant feature set according to historical samples to obtain the road surface state estimation value of the current area. Among them, the historical samples include historical relevant features and their corresponding road surface state values, and the road surface state value characterizes the bump degree of the road surface;

[0041] Estimation value optimization module: used to optimize the road surface state estimation value according to the historical state of the current area;

[0042] Road surface maintenance decision module: used to make corresponding scheduling instructions for vehicle passing and road surface maintenance according to the optimized road surface state estimation value.

[0043] In a third aspect, the present disclosure provides a computer storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the mining area road surface maintenance decision-making method according to any step in the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] FIG. 1 shows a flowchart of the mining area road surface maintenance decision-making method based on road surface state estimation in Embodiment 1 of the present disclosure;

[0045] FIG. 2 shows a flowchart of the mining area road surface maintenance decision-making method based on road surface state estimation in Embodiment 2 of the present disclosure;

[0046] FIG. 3 shows an architecture diagram of the mining area road surface maintenance decision-making system based on road surface state estimation in Embodiment 3 of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] The technical solutions provided in this disclosure will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of this disclosure and should not be used to limit the scope of protection of this disclosure.

[0048] This disclosure provides a method, system, and storage medium for mine road maintenance decision-making based on road condition estimation. The method obtains an estimate of the bumpiness of a certain road section through feature extraction and feature learning, and optimizes the estimate using historical bump information of the road section, thereby achieving an accurate estimate of the road section's condition. Then, corresponding measures are taken based on the road condition to reduce safety hazards, keep the road surface for unmanned driving operations at an ideal level, and reduce operation interruptions.

[0049] To achieve the above objectives, this disclosure is implemented using the following technical solution.

[0050] Example 1

[0051] As shown in Figure 1, this embodiment provides a method for decision-making on road surface maintenance in mining areas based on road surface condition estimation, including:

[0052] The system acquires real-time road condition parameters collected from the current area of ​​the road surface where the vehicle is traveling in the mining area, and preprocesses the real-time parameters.

[0053] Feature extraction is performed on the preprocessed parameters to obtain the relevant feature set;

[0054] Based on historical samples, feature learning is performed on the relevant feature set to obtain the estimated road surface condition of the current area. The historical samples include historical relevant features and their corresponding road surface condition values, and the road surface condition values ​​characterize the degree of road surface bumpiness.

[0055] The road surface condition estimate is optimized based on the historical state of the current area;

[0056] Based on the optimized road condition estimates, corresponding dispatch instructions for vehicle passage and road maintenance are issued.

[0057] This embodiment of the invention addresses the detection of road bumps caused by repeated driving in a fixed work area during unmanned mining operations. By extracting features from road surface condition data collected by the vehicle while driving on a road segment, and learning these features, the bump condition is regressed. Historical bump information from the same road segment is used to optimize the predicted road surface condition, achieving accurate estimation of the same road surface condition. Different measures are taken for different road surface conditions to reduce safety hazards, improve system stability, and increase operational efficiency.

[0058] Example 2

[0059] As shown in Figure 2, based on Example 1, this example also incorporates the following design.

[0060] The road surface condition detection refers to detecting the road surface condition of each unit area, that is, the bumpiness parameter of each unit area. The road surface condition parameter value is a percentage from 0 to 100. If the road surface is completely smooth and there is no bumpiness, the road surface condition value is 0%, which also means that the bumpiness of the road surface is 0%. If the road surface condition is extremely poor, the road surface is uneven, and there are a lot of ruts and fallen rocks, the road surface condition value is 100%, which also means that the bumpiness of the road surface is 100%.

[0061] The aforementioned decision-making method for mine road maintenance based on road condition estimation is specifically divided into the following steps.

[0062] Step 1: Read the sensor data collected by the sensors installed on the vehicle during driving, and preprocess the sensor data, including lidar point cloud data, IMU data, and wheel speed sensor data.

[0063] For LiDAR data, point cloud pass-through filtering, radius filtering, and voxel filtering are performed to obtain ideal point cloud data for the target area with noise removed. Then, the CSF cloth simulation filtering algorithm is used to perform ground segmentation on the ideal point cloud, dividing the original point cloud into two parts: ground point cloud and non-ground point cloud, to obtain the ground point cloud.

[0064] After segmenting the ground points, collect ground point clouds [P1, P2, P3, ..., P] over a continuous time period while the vehicle is traveling in a certain area. n The transformation relationship between each frame of the ground point cloud is obtained through ICP point cloud registration, and then the point cloud is stitched together to obtain a continuous set of ground point clouds in this area.

[0065] For IMU data, which includes triaxial angular velocity and triaxial acceleration, triaxial acceleration is collected during travel within a certain area. Where N represents the number of time points in which the vehicle travels within this area, N = f × t, where f represents the data acquisition frequency of the IMU, and t i Indicates the time i when the vehicle is traveling in this area, a x a represents the acceleration along the x-axis. y a represents the y-axis acceleration. z This represents the acceleration along the z-axis, forming an IMU data matrix based on time.

[0066] For wheel speed sensor data, the rotational speed of each of the four wheels is read, and a wheel speed data matrix is ​​generated based on time. The first dimension of the matrix is ​​the rotational speed of the four wheels at a certain moment, and the second dimension is the number of times the data is collected at a frequency f within the time t of the region.

[0067] Step 2: Perform feature extraction on the sensor data matrix provided after data preprocessing to obtain the sensor data of vehicles in a certain area that are related to bumps.

[0068] For ground point clouds, PointPillars encoding is used, and a planar mesh is created based on the region size. The ground region of the point cloud is [x max -x min ,y max -y min ,z max -z min Set the grid size to [x] grid ,y grid ], forming a wide Gao Wei The grid map projects the ground point cloud into the corresponding grid, and extracts features from the point cloud within each grid.

[0069] For each grid, the main features of the point cloud inside it are calculated as follows: statistically analyze the z-coordinate of the point cloud inside the grid, calculate the standard deviation of the point cloud in the z-direction to represent the dispersion of ground height, calculate the angle between the point cloud normal vector and the z-direction to represent the local road surface convexity, calculate the FPFH feature descriptor of the point cloud, and describe the local geometric information of the ground point cloud.

[0070] The three features of the point cloud mentioned above are aggregated to form an expression of the road surface features, denoted as [H,W,C], where H and W are the ranges of the point cloud, and C is the feature of the point cloud.

[0071] For IMU (Inertial Measurement Unit) data, PCA (Principal Component Analysis) is performed on the IMU data at each time step. First, a covariance matrix is ​​constructed, and then the eigenvalues ​​and eigenvectors of the covariance matrix are calculated. The eigenvectors of the principal components are used as features of the IMU data, and the IMU features are encoded according to the time dimension to form the IMU feature matrix.

[0072] For wheel speed sensors, the standard deviation of wheel speed for each wheel during a certain travel time in a certain area is calculated to represent the stability of wheel rotation. The first derivative or difference of wheel speed is calculated to reflect the rate of change of wheel speed. The stability of wheel rotation and the rate of change of speed are also formed into a feature matrix with time as the dimension.

[0073] Step 3: Perform feature learning on the acquired sensor features and output the estimated state of the road surface.

[0074] The features are regressed using the random forest regression method. First, multiple decision trees are constructed. From the original training data, multiple sample sets are generated by sampling with replacement, and each sample set is used to train a decision tree. At the same time, when constructing the nodes of each tree, only a randomly selected part of all the features is used for the optimal partition.

[0075] The trained decision trees are used to predict the road surface state value for the obtained sensor features. Each decision tree will give a prediction value based on the input feature vector (features of IMU data, lidar data, and wheel speed sensor data), that is, an estimated value of the road surface bumpiness. The average value of the results of all decision trees is taken to obtain the final regression value.

[0076] Step 4: Optimize the prediction value according to the historical state of the current area.

[0077] Since the road surface state is correlated with the road surface state at this location at historical moments, for the predicted road surface state, the current state is optimized by combining the historical state. For the bumpiness at historical moments, a weight factor is introduced to represent the degree of its influence on the current prediction value. The exponential decay weight factor is used to represent the influence of the state of this area in the past time on the current state.

[0078] Let w t be the weight factor corresponding to time t, λ be the decay rate (0 < λ < 1). Since the road surface changes little over time during the unmanned driving process, λ is taken as 0.8, T is the current time point, and n is the number of historical time points. Then for each historical time point t (where T - n ≤ t < T), the weight factor w t can be calculated as: w t = λ(T - t).

[0079] Then the weight factor is normalized so that the sum of the weights of historical data points is 1. The normalized weight factor is:

[0080] Regarding the current road surface state prediction value regressed by the random forest as IP as the initial prediction value, the prediction value is adjusted using the weight factor, and the adjusted prediction value AP is: where AP is the optimized road surface state value, IP is the state value regressed by the random forest according to the current ground features, S E is the road surface state value at historical time t, and α is an additional adjustment parameter used to control the influence degree of historical data on the preliminary prediction value.

[0081] Step 5: Make a decision based on the road surface state value of the detection area to determine whether the area needs maintenance.

[0082] Set thresholds θ1 and θ2. If the bumpiness level S < θ1, it means the ground is flat and no maintenance is required for the time moment. If θ1 < S < θ2, it means there is a certain degree of bumpiness on the road surface. The remote dispatch center sends a decision signal to the driverless mining truck. When entering this area, the truck should slow down. If there is an idle grader on-site, after the mining truck passes through this area, the grader will be dispatched to perform ground maintenance work. If S > θ2, it means the road surface has excessive bumps and undulations and is impassable. At this time, the vehicle chooses to bypass the obstacle. Meanwhile, the remote dispatch center needs to dispatch the grader to repair the road surface condition. After the grader finishes the maintenance, the road surface status and maintenance request in this area are initialized.

[0083] Embodiment 3

[0084] As shown in Figure 3, this embodiment provides a mining area road surface maintenance decision-making system based on road surface state estimation, including:

[0085] Parameter preprocessing module: used to obtain the real-time relevant parameters of the road surface state collected when the vehicle is driving in the current area of the mining area road surface, and preprocess the real-time relevant parameters;

[0086] Feature extraction module: used to extract features from the preprocessed parameters to obtain a relevant feature set;

[0087] Road surface state estimation module: used to perform feature learning on the relevant feature set according to historical samples to obtain the estimated value of the road surface state in the current area. Among them, the historical samples include historical relevant features and their corresponding road surface state values, and the road surface state value represents the bumpiness level of the road surface;

[0088] Estimated value optimization module: used to optimize the road surface state estimated value according to the historical state of the current area;

[0089] Road surface maintenance decision module: used to make corresponding scheduling instructions for vehicle passage and road surface maintenance according to the optimized road surface state estimated value.

[0090] Embodiment 4

[0091] This embodiment provides a computer storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the mining area road surface maintenance decision-making method according to any step in Embodiment 2.

[0092] The beneficial effects achieved by this disclosure are as follows: This disclosure collects data on the bumpiness of a road segment in real time as the vehicle travels along the mining area. Based on historical data, it extracts and learns features from this data to estimate the current bumpiness level of the road segment. The estimated value is then optimized by incorporating historical data. Considering the influence of historical road conditions on the current condition, the accuracy of road condition prediction is improved. Finally, corresponding traffic decisions and road maintenance measures are implemented for different road conditions, achieving real-time detection of road conditions. This keeps the road surface at an ideal level for unmanned operations, reduces operational interruptions, lowers safety hazards associated with unmanned driving, and improves the stability and operational efficiency of the unmanned mining truck system.

[0093] By using multi-sensor data to detect ground conditions, ground spatial features are acquired through lidar, and the vehicle's own status is acquired through IMU and wheel speed sensors. By fusing the features from various sensors to determine the road surface condition, the road surface bump characteristics can be represented more effectively.

[0094] By quantifying road surface conditions and expressing the bumpiness as a percentage, decision tree regression can accurately predict the degree of road bumpiness, providing more detailed data support for subsequent decision-making.

[0095] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0096] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more flowchart illustrations and / or one or more block diagrams.

[0097] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0098] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

[0099] The embodiments of this disclosure have been described above with reference to the accompanying drawings. However, this disclosure is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this disclosure without departing from the spirit and scope of the claims. All of these forms are within the protection scope of this disclosure.

Claims

1. A mine site road maintenance decision method based on road condition estimation, characterized by, include: The system acquires real-time road condition parameters collected from the current area of ​​the road surface where the vehicle is traveling in the mining area, and preprocesses the real-time parameters. Feature extraction is performed on the preprocessed parameters to obtain the relevant feature set; Based on historical samples, feature learning is performed on the relevant feature set to obtain the estimated road surface condition of the current area. The historical samples include historical relevant features and their corresponding road surface condition values, and the road surface condition values ​​characterize the degree of road surface bumpiness. The road surface condition estimate is optimized based on the historical state of the current area; Based on the optimized road condition estimates, corresponding dispatch instructions for vehicle passage and road maintenance are issued.

2. The mine road surface maintenance decision method based on road surface condition estimation according to claim 1, characterized by, The acquisition of real-time road surface status parameters collected by the vehicle while driving on the mining area road includes: collecting point cloud data, IMU data, and wheel speed data respectively by using lidar, IMU inertial measurement unit, and wheel speed sensor installed on the vehicle during the vehicle's driving process.

3. The mine road surface maintenance decision method based on road surface condition estimation according to claim 2, characterized by, The preprocessing of the real-time relevant parameters includes: Spatial processing is performed on the point cloud data of the current area collected by the lidar to obtain the ground point cloud set of the current area; The inertial data and wheel speed data of the vehicle traveling in the current area are processed in time series to obtain time series IMU data and time series wheel speed data.

4. The mine road surface maintenance decision method based on road surface condition estimation according to claim 3, characterized by, The step of extracting features from the preprocessed parameters to obtain a relevant feature set includes: The ground point cloud set is encoded using pointpillars to obtain a point cloud mesh map of the current area. The features of the point cloud within each mesh are calculated to obtain the discreteness of the road surface height, the degree of local road surface protrusion, and the local geometric information of the point cloud. The road surface features [H,W,C] are generated based on the calculated point cloud features, where H and W are the ranges of the point cloud, and C is the feature of the point cloud. PCA principal component analysis is performed on the IMU data at each time step in the time series IMU data. The eigenvectors of the principal components are used as IMU data features, and the IMU data features are encoded according to the time dimension to generate an IMU feature matrix. Based on the time-series wheel speed data, calculate the standard deviation of wheel speed and the first derivative or difference of wheel speed for each wheel during the current travel time in the current area to obtain wheel rotation stability and wheel speed change rate. Then encode the wheel rotation stability and wheel speed change rate according to the time dimension to generate a wheel speed feature matrix. The road surface features, IMU feature matrix, and wheel speed feature matrix of the current area are combined into a relevant feature set of the road surface condition of the current area.

5. The mine road maintenance decision method based on road condition estimation according to any one of claims 1 to 4, characterized in that, The step of performing feature learning on the relevant feature set based on historical samples to obtain the road surface state estimate for the current area includes: The relevant feature set is input into multiple pre-trained road surface state prediction models to obtain multiple road surface state prediction values; The road condition estimate is obtained by averaging the multiple road condition prediction values. The method for obtaining the multiple road surface condition prediction models includes: Train each pre-constructed decision tree with each historical sample in the pre-constructed historical sample set to obtain multiple road surface state prediction models. Among them, the road surface state value in the historical sample is in percentage form, and the road surface state value in the historical sample is in percentage form.

6. The decision-making method for mine road maintenance based on road surface condition estimation according to claim 5, characterized in that, During the construction of the historical sample set, perform sampling with replacement on the historical relevant data of the road surface state in each area of the mining area road surface; During the construction of each decision tree, randomly select some relevant features to perform optimal partitioning on the nodes.

7. The mine road maintenance decision method based on road condition estimation according to any one of claims 1 to 6, characterized in that, The optimization of the road surface state estimated value according to the historical state of the current area includes: Obtain the influence weight of each historical state of the current area on the current state. The calculation formula of the influence weight is: w t = λ (T - t), where w t is a weight factor corresponding to time t, λ is a decay rate, 0<λ<1, T is the current time, and n is the number of historical time points, T-n≤t<T. The influence weight is normalized, and the calculation formula is as follows: wherein is the normalized weight factor; The road surface condition estimate is adjusted by a normalized weighting factor to obtain an optimized road surface condition estimate, calculated as follows: where IP represents the initial road surface state estimation value obtained, AP is the optimized road surface state estimation value, a is an adjustment parameter for controlling the degree of influence of the historical data on the initial road surface state estimation value, S t represents the road surface state estimation value corresponding to the time t.

8. The mine road maintenance decision method based on road condition estimation according to any one of claims 1 to 7, characterized in that, The scheduling instructions for vehicle passing and road surface maintenance made according to the optimized road surface state estimated value include: Compare the optimized road surface state estimated value with the preset bump degree thresholds θ1 and θ2, If the bump degree S < θ1, it means the road surface is flat and no maintenance is needed temporarily, If θ1 < S < θ2, it means the road surface has small bumps and undulations. Send a decision signal to the unmanned mining truck terminal to reduce the speed of passing in the current area, and dispatch the idle grader on site to perform ground maintenance work on the current area after the unmanned mining truck passes, If S > θ2, it means the road surface has large bumps and undulations. Send a decision signal to the unmanned mining truck terminal to bypass the current area, and dispatch the grader to repair the road surface condition in time.

9. A mine site road maintenance decision system based on road condition estimation, characterized by, Includes: Parameter preprocessing module: used to obtain the real-time relevant parameters of the road surface state collected when the vehicle travels in the current area of the mining area road surface, and preprocess the real-time relevant parameters; Feature extraction module: used to extract features from the preprocessed parameters to obtain a relevant feature set; Road surface state estimation module: used to perform feature learning on the relevant feature set according to the historical samples to obtain the road surface state estimated value of the current area. Among them, the historical samples include historical relevant features and their corresponding road surface state values, and the road surface state value represents the bump degree of the road surface; Estimated value optimization module: used to optimize the road surface state estimated value according to the historical state of the current area; Road surface maintenance decision module: used to make corresponding scheduling instructions for vehicle passing and road surface maintenance according to the optimized road surface state estimated value.

10. A computer storage medium having stored thereon a computer program, characterized in that When the computer program is executed by the processor, it implements the mining area road surface maintenance decision method based on road surface state estimation as described in any one of claims 1-8.

Images