A mine-used trackless rubber-tyred vehicle control system and method for obstacle self-adaptive discrimination

Abstract

The application discloses a mine trackless rubber-tyred vehicle control system and method for obstacle adaptive identification, and the obstacle adaptive identification module establishes a trackless rubber-tyred vehicle adaptive obstacle identification scheme, and identifies static obstacles and dynamic obstacles according to acquired roadway environment information; the dynamic obstacle trajectory prediction module is used for establishing a dynamic obstacle trajectory prediction network based on a generative adversarial network, and predicting a future dynamic obstacle trajectory according to dynamic obstacle information; the intelligent early warning module is used for setting different levels of trackless rubber-tyred vehicle early warning mechanisms; the driving stability identification and compensation module is used for identifying the stability of the trackless rubber-tyred vehicle during driving, and if instability occurs, torque compensation is performed on the trackless rubber-tyred vehicle to eliminate the instability of the trackless rubber-tyred vehicle. Through the above process, the unmanned mining condition in the coal mine is improved, and finally, the trackless rubber-tyred vehicle has good intelligence and transportation efficiency during driving in the coal mine.

Description

Technical Field

[0001] This invention belongs to the field of intelligent equipment for coal mines, specifically a control system and method for an obstacle adaptive discrimination trackless rubber-wheeled vehicle for mining. Background Technology

[0002] Trackless rubber-tired vehicles are a common auxiliary transportation device in underground coal mines, mainly used to transport materials, personnel, and equipment to designated locations. These vehicles typically use explosion-proof diesel engines or electric motors for power and are equipped with flame-retardant rubber tires as their running gear. The use of trackless rubber-tired vehicles has increased the mining rate of coal mines, significantly improved mining productivity and operational safety, and is of great significance to underground mining operations.

[0003] Due to the complex and variable environment of coal mines, characterized by high concentrations of smoke and water mist, dim lighting, and narrow tunnels with numerous obstacles along the way, drivers are prone to losing focus or having their vision obstructed by obstacles, leading to rear-end collisions and other safety accidents. However, current trackless rubber-tired vehicles have low levels of intelligence; their control systems cannot accurately predict the appearance of obstacles in advance, posing a significant obstacle to unmanned operation. Furthermore, the undulating terrain of the tunnels and the distribution of obstacles can cause vehicles to deviate from their trajectory during operation, even leading to lateral stability issues such as sideslip and fishtailing, which also affect the stability and safety of trackless rubber-tired vehicles in coal mines.

[0004] Therefore, providing a new system and method that can accurately identify and predict obstacles during the operation of trackless rubber-tired vehicles, while solving the problem of trackless rubber-tired vehicle instability caused by the tunnel environment, is of great significance for improving unmanned mining in coal mines and enhancing the intelligence and transportation efficiency of trackless rubber-tired vehicles. Summary of the Invention

[0005] To address the problems existing in the prior art, the present invention provides a control system and method for an obstacle adaptive discrimination trackless rubber-tired vehicle for mining, which can effectively solve the above-mentioned technical problems.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: a trackless rubber-tired vehicle control system for mines with adaptive obstacle discrimination, comprising an adaptive obstacle discrimination module, a dynamic obstacle trajectory prediction module, an intelligent early warning module, a driving stability discrimination and compensation module; the adaptive obstacle discrimination module is used to establish an adaptive obstacle recognition scheme for the trackless rubber-tired vehicle, dynamically adjusting the weights of information collected by different sensors according to different roadway environmental information, and distinguishing between static and dynamic obstacles; the dynamic obstacle trajectory prediction module is used to establish a generative adversarial network (GAN) based system... The system employs a dynamic obstacle trajectory prediction network to predict future dynamic obstacle trajectories based on dynamic obstacle information. An intelligent early warning module is used to set different levels of early warning mechanisms for the trackless rubber-wheeled vehicle based on the distance between static obstacles and the vehicle, environmental information, and future trajectory information of dynamic obstacles. Simultaneously, it controls the trackless rubber-wheeled vehicle to perform corresponding actions based on different levels of early warning information. A driving stability judgment and compensation module is used to judge the stability of the trackless rubber-wheeled vehicle during operation. If instability occurs, torque compensation is performed on the trackless rubber-wheeled vehicle to eliminate instability.

[0007] Furthermore, the different sensors specifically include a camera, a lidar, and a millimeter-wave radar, wherein the camera is used to acquire images of the surrounding environment, the lidar is used to acquire three-dimensional point clouds of the surrounding environment, and the millimeter-wave radar is used to obtain the distance between the trackless rubber-wheeled vehicle and the surrounding environment.

[0008] The working method of the above-mentioned obstacle adaptive discrimination control system for trackless rubber-tired mining vehicles includes the following steps:

[0009] Step 1: Acquire surrounding environmental information through different sensors, establish an adaptive obstacle recognition scheme for trackless rubber-wheeled vehicles in complex underground coal mine environments based on the FUTR3D algorithm, dynamically adjust the weight of information acquired by different sensors according to different roadway environmental information, achieve accurate obstacle detection in unstructured scenarios, output obstacle speed information through millimeter-wave radar, and distinguish between static and dynamic obstacles through multimodal fusion classification.

[0010] Step 2: Establish a dynamic obstacle trajectory prediction network based on generative adversarial network (GAN) to predict the future trajectory of the dynamic obstacles obtained in Step 1.

[0011] Step 3: Establish a layered obstacle warning mechanism. Based on the static obstacle location information obtained in Step 1 and the dynamic obstacle future trajectory information predicted in Step 2, establish a trackless rubber-wheeled vehicle driving warning mechanism. Different sensors on the trackless rubber-wheeled vehicle collect surrounding environmental information in real time. When the distance between the trackless rubber-wheeled vehicle and the static or dynamic obstacle is less than or equal to the danger value, the trackless rubber-wheeled vehicle will brake immediately. If it is greater than the danger value, proceed to Step 4.

[0012] Step 4: Establish an end-to-end unmanned driving model for the trackless rubber-tired vehicle based on UniAD (Planning-oriented Autonomous Driving). According to the static obstacle position information, dynamic obstacle future trajectory prediction information, and early warning mechanism output information obtained in Steps 1 to 3, the above information is input into the trackless rubber-tired vehicle unmanned driving model by the state encoder; then the trackless rubber-tired vehicle unmanned driving model controls the movement of the trackless rubber-tired vehicle.

[0013] Step 5: Establish a dynamic model of the trackless rubber-tired vehicle. During the operation of the trackless rubber-tired vehicle, use the phase plane stability discrimination scheme to evaluate the dynamic state of the trackless rubber-tired vehicle in real time and obtain the vehicle stability coefficient.

[0014] Step 6: Establish a stability compensation scheme for the trackless rubber-tired vehicle based on fuzzy PID control. Determine whether the trackless rubber-tired vehicle is unstable based on the vehicle stability coefficient obtained in Step 5. If it is determined to be unstable, adjust the compensation torque of the trackless rubber-tired vehicle in the unstable state to restore its stability and achieve safe and stable operation of the trackless rubber-tired vehicle.

[0015] Furthermore, step two specifically involves:

[0016] Historical trajectory X = {x1, x2, ..., x} T}, future historical trajectory Generator G θ Discriminant D is used to generate future trajectories. φ Used to determine the trajectory generated by the generator;

[0017] The objective function minimizes the loss:

[0018]

[0019] Diversity loss: Forced generation of multiple differentiated trajectories,

[0020] L2 regression loss: Constraints generate trajectories that closely approximate the true trajectories.

[0021] Physical loss:

[0022] in, For environment masking functions, specifically manifested as according to Determining whether it falls within the feasible area. For point Projection point to the nearest feasible region boundary;

[0023] Complete objective function:

[0024] Training process: Initialize generator parameters θ and discriminator φ;

[0025] Sampling real trajectory And generate virtual trajectory

[0026] z (i) ~N(0,I);

[0027] With a fixed generator G, calculate the loss and update the discriminator D:

[0028]

[0029] Fixed discriminator D, compute function update generator G:

[0030]

[0031] The process continues until the discriminator and generator converge, thus obtaining the predicted future trajectory information of the dynamic obstacle.

[0032] Furthermore, step five specifically includes:

[0033] The phase plane region is divided into stable and unstable regions using the double-line method based on two stability boundary lines; the stability boundary is represented as:

[0034]

[0035] Among them, B1 and B2 are stability boundary parameters, and B1, B2 > 0. They are determined by vehicle parameters, vehicle speed, front wheel steering angle and road adhesion coefficient. The slope and intercept of the stability boundary line are determined by the combination of B1 and B2.

[0036] The stability coefficient η is the ratio of the distance from the current state point to the nearest stability boundary line:

[0037] Where d1 and d2 are the Euclidean distances from the current state point to the two stability boundary lines, respectively; d is the Euclidean distance from the static equilibrium point to the nearest stability boundary; when η is less than 1, the trackless rubber-wheeled vehicle becomes unstable.

[0038] Furthermore, step six specifically includes:

[0039] Calculate the yaw rate deviation: e(t)=γ des (t)-γ(t);

[0040] Calculate the sideslip angle deviation: e β (t)=β des (t)-β(t);

[0041] Input variable e(t) and Mapping to fuzzy linguistic variables: μA i (e), B j Let be a fuzzy set, where the inputs are the deviation and the rate of change of the deviation;

[0042] Rules are designed based on expert experience, and the PID parameter K is dynamically adjusted. p ,K i ,K d ΔK p =C ij ,ΔK i =D ij ,ΔK d =E ij ;

[0043] Based on the fuzzy inference model, the fuzzy quantity ΔK is output. p ,ΔK i ,ΔK d ;

[0044] Defuzzification:

[0045] Real-time updates of PID parameters:

[0046]

[0047] The compensation torque T is output through the PID controller. c :

[0048] The compensation torque of the trackless rubber-tired vehicle is obtained through the above formula to restore its stability and achieve safe and stable operation of the trackless rubber-tired vehicle.

[0049] Compared with existing technologies, this invention adopts a combination of an obstacle adaptive discrimination module, a dynamic obstacle trajectory prediction module, an intelligent early warning module, and a driving stability discrimination and compensation module. The obstacle adaptive discrimination module establishes an adaptive obstacle recognition scheme for the trackless rubber-tired vehicle, dynamically adjusting the weights of information collected by different sensors based on different roadway environmental information to distinguish between static and dynamic obstacles. The dynamic obstacle trajectory prediction module establishes a dynamic obstacle trajectory prediction network based on a generative adversarial network (GAN) and predicts future dynamic obstacle trajectories based on dynamic obstacle information. The intelligent early warning module sets different levels of early warning mechanisms for the trackless rubber-tired vehicle based on the distance between static obstacles and the trackless rubber-tired vehicle, environmental information, and future trajectory information of dynamic obstacles, and controls the trackless rubber-tired vehicle to take corresponding actions based on different levels of early warning information. The driving stability discrimination and compensation module performs stability discrimination during the trackless rubber-tired vehicle's operation; if instability occurs, torque compensation is performed to eliminate the instability. Through the above process, the present invention can accurately identify and classify obstacles during the operation of trackless rubber-tired vehicles, and effectively solve the problem of trackless rubber-tired vehicles becoming unstable due to the tunnel environment. This can improve the unmanned mining situation in coal mines and ultimately ensure that trackless rubber-tired vehicles have better intelligence and transportation efficiency when operating underground. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the overall process of the working method of the present invention. Detailed Implementation

[0051] The present invention will be further described below.

[0052] A trackless rubber-tired vehicle control system for mining applications, featuring adaptive obstacle discrimination, includes an adaptive obstacle discrimination module, a dynamic obstacle trajectory prediction module, an intelligent early warning module, and a driving stability discrimination and compensation module. The adaptive obstacle discrimination module establishes an adaptive obstacle recognition scheme for the trackless rubber-tired vehicle based on deep learning methods. It dynamically adjusts the weights of information collected by different sensors according to different roadway environmental information to distinguish between static and dynamic obstacles. The different sensors specifically include a camera, a lidar, and a millimeter-wave radar. The camera is used to acquire images of the surrounding environment, the lidar is used to acquire three-dimensional point clouds of the surrounding environment, and the millimeter-wave radar is used to obtain the distance between the trackless rubber-tired vehicle and its surrounding environment. The dynamic obstacle trajectory prediction module is used to establish a dynamic obstacle trajectory prediction network based on generative adversarial networks (GANs) and predict future dynamic obstacle trajectories based on dynamic obstacle information. The intelligent early warning module is used to set different levels of early warning mechanisms for the trackless rubber-wheeled vehicle based on the distance between static obstacles and the trackless rubber-wheeled vehicle, environmental information, and future trajectory information of dynamic obstacles, and simultaneously control the trackless rubber-wheeled vehicle to make corresponding actions based on different levels of early warning information. The driving stability discrimination and compensation module is used to discriminate the stability of the trackless rubber-wheeled vehicle during driving. If instability occurs, torque compensation is performed on the trackless rubber-wheeled vehicle through a fuzzy PID control model to eliminate the instability phenomenon.

[0053] like Figure 1 As shown, the working method of the above-mentioned obstacle adaptive discrimination trackless rubber-tired vehicle control system includes the following steps:

[0054] Step 1: Acquire surrounding environmental information through different sensors, establish an adaptive obstacle recognition scheme for trackless rubber-wheeled vehicles in complex underground coal mine environments based on the FUTR3D algorithm, dynamically adjust the weight of information acquired by different sensors according to different roadway environmental information, achieve accurate obstacle detection in unstructured scenarios, output obstacle speed information through millimeter-wave radar, and distinguish between static and dynamic obstacles through multimodal fusion classification.

[0055] Step 2: Establish a dynamic obstacle trajectory prediction network based on Generative Adversarial Network (GAN) to predict the future trajectories of the dynamic obstacles obtained in Step 1. Specifically:

[0056] Historical trajectory X = {x1, x2, ..., x} T}, future historical trajectory Y = {y1, y2, ... y} T}, Generator G θ Discriminant D is used to generate future trajectories. φ Used to determine the trajectory generated by the generator;

[0057] The objective function minimizes the loss:

[0058]

[0059] Diversity loss: Forced generation of multiple differentiated trajectories,

[0060] L2 regression loss: Constraints generate trajectories that closely approximate the true trajectories.

[0061] Physical loss:

[0062] in, For environment masking functions, specifically manifested as according to Determining whether it falls within the feasible area. For point Projection point to the nearest feasible region boundary;

[0063] Complete objective function:

[0064] Training process: Initialize generator parameters θ and discriminator φ;

[0065] Sampling real trajectory And generate virtual trajectory

[0066] z (i) ~N(0,I);

[0067] With a fixed generator G, calculate the loss and update the discriminator D:

[0068]

[0069] Fixed discriminator D, compute function update generator G:

[0070]

[0071] The process continues until the discriminator and generator converge, thus obtaining the predicted future trajectory information of the dynamic obstacle.

[0072] Step 3: Establish a layered obstacle warning mechanism. Based on the static obstacle location information obtained in Step 1 and the dynamic obstacle future trajectory information predicted in Step 2, establish a trackless rubber-wheeled vehicle driving warning mechanism. Different sensors on the trackless rubber-wheeled vehicle collect surrounding environmental information in real time. When the distance between the trackless rubber-wheeled vehicle and the static or dynamic obstacle is less than or equal to the danger value, the trackless rubber-wheeled vehicle will brake immediately. If it is greater than the danger value, proceed to Step 4.

[0073] Step 4: Establish an end-to-end unmanned driving model for the trackless rubber-tired vehicle based on UniAD (Planning-oriented Autonomous Driving). According to the static obstacle position information, dynamic obstacle future trajectory prediction information, and early warning mechanism output information obtained in Steps 1 to 3, the above information is input into the trackless rubber-tired vehicle unmanned driving model by the state encoder; then the trackless rubber-tired vehicle unmanned driving model controls the movement of the trackless rubber-tired vehicle.

[0074] Step 5: Establish a dynamic model of the trackless rubber-tired vehicle. During the vehicle's operation, use a phase-plane stability discrimination scheme to evaluate the vehicle's dynamic state in real time and obtain the vehicle's stability coefficient. Specifically:

[0075] The phase plane region is divided into stable and unstable regions using the double-line method based on two stability boundary lines; the stability boundary is represented as:

[0076]

[0077] Among them, B1 and B2 are stability boundary parameters, and B1, B2>0, which are jointly determined by vehicle parameters, vehicle speed, front wheel steering angle and road adhesion coefficient. The slope and intercept of the stability boundary line are determined by the combination of B1 and B2.

[0078] The stability coefficient η is the ratio of the distance from the current state point to the nearest stability boundary line:

[0079] Where d1 and d2 are the Euclidean distances from the current state point to the two stability boundary lines, respectively; d is the Euclidean distance from the static equilibrium point to the nearest stability boundary; when η is less than 1, the trackless rubber-wheeled vehicle becomes unstable.

[0080] Step Six: Establish a stability compensation scheme for the trackless rubber-tired vehicle based on fuzzy PID control. Determine whether the trackless rubber-tired vehicle is unstable based on the vehicle stability coefficient obtained in Step Five. If it is determined to be unstable, adjust the compensation torque for the trackless rubber-tired vehicle in the unstable state, specifically as follows:

[0081] Calculate the yaw rate deviation: e(t)=γ des (t)-γ(t);

[0082] Calculate the sideslip angle deviation: e β (t)=β des (t)-β(t);

[0083] Input variable e(t) and Mapping to fuzzy linguistic variables: μA i (e), B j Let be a fuzzy set, where the inputs are the deviation and the rate of change of the deviation;

[0084] Rules are designed based on expert experience, and the PID parameter K is dynamically adjusted. p ,K i ,K d ΔK p =C ij ,ΔK i =D ij ,ΔK d =E ij ;

[0085] Based on the Sugeno model, the output ambiguity ΔK p ,ΔK i ,ΔK d ;

[0086] Defuzzification:

[0087] Real-time updates of PID parameters:

[0088]

[0089] The compensation torque T is output through the PID controller. c :

[0090] The compensation torque of the trackless rubber-tired vehicle is obtained through the above formula to restore its stability and achieve safe and stable operation of the trackless rubber-tired vehicle.

[0091] 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 working method of a mine trackless rubber-tyred vehicle control system with obstacle self-adaptive discrimination, characterized in that, Includes the following steps: Step 1: Acquire surrounding environmental information through different sensors, establish an adaptive obstacle recognition scheme for trackless rubber-wheeled vehicles in complex underground coal mine environments based on the FUTR3D algorithm, dynamically adjust the weight of information acquired by different sensors according to different roadway environmental information, achieve accurate obstacle detection in unstructured scenarios, output obstacle speed information through millimeter-wave radar, and distinguish between static and dynamic obstacles through multimodal fusion classification. Step 2: Establish a dynamic obstacle trajectory prediction network based on generative adversarial networks to predict the future trajectories of the dynamic obstacles obtained in Step 1. Step 3: Establish a layered obstacle warning mechanism. Based on the static obstacle location information obtained in Step 1 and the dynamic obstacle future trajectory information predicted in Step 2, establish a trackless rubber-wheeled vehicle driving warning mechanism. Different sensors on the trackless rubber-wheeled vehicle collect surrounding environmental information in real time. When the distance between the trackless rubber-wheeled vehicle and the static or dynamic obstacle is less than or equal to the danger value, the trackless rubber-wheeled vehicle will brake immediately. If it is greater than the danger value, proceed to Step 4. Step 4: Establish an end-to-end unmanned driving model for the trackless rubber-tired vehicle based on UniAD. According to the static obstacle position information, dynamic obstacle future trajectory prediction information, and early warning mechanism output information obtained in Steps 1 to 3, the above information is input into the unmanned driving model of the trackless rubber-tired vehicle by the state encoder; then the unmanned driving model of the trackless rubber-tired vehicle controls the driving of the trackless rubber-tired vehicle. Step 5: Establish a dynamic model of the trackless rubber-tired vehicle. During the operation of the trackless rubber-tired vehicle, use the phase plane stability discrimination scheme to evaluate the dynamic state of the trackless rubber-tired vehicle in real time and obtain the vehicle stability coefficient. Step 6: Establish a stability compensation scheme for the trackless rubber-tired vehicle based on fuzzy PID control. Determine whether the trackless rubber-tired vehicle is unstable based on the vehicle stability coefficient obtained in Step 5. If it is determined to be unstable, adjust the compensation torque of the trackless rubber-tired vehicle in the unstable state to restore its stability and achieve safe and stable operation of the trackless rubber-tired vehicle.

2. The working method according to claim 1, characterized in that, The control system employed includes an obstacle adaptive discrimination module, a dynamic obstacle trajectory prediction module, an intelligent early warning module, a driving stability discrimination and compensation module; The obstacle adaptive discrimination module is used to establish an adaptive obstacle recognition scheme for trackless rubber-wheeled vehicles. Based on different environmental information of the tunnel, it dynamically adjusts the weight of information collected by different sensors to distinguish between static and dynamic obstacles. The dynamic obstacle trajectory prediction module is used to establish a dynamic obstacle trajectory prediction network based on generative adversarial networks and predict future dynamic obstacle trajectories based on dynamic obstacle information. The intelligent early warning module is used to set different levels of early warning mechanisms for the trackless rubber-wheeled vehicle based on the distance between the static obstacle and the trackless rubber-wheeled vehicle, environmental information, and future trajectory information of the dynamic obstacle. At the same time, it controls the trackless rubber-wheeled vehicle to make corresponding actions based on the early warning information at different levels. The driving stability discrimination and compensation module is used to discern the stability of the trackless rubber-tired vehicle during its operation. If instability occurs, torque compensation is performed on the trackless rubber-tired vehicle to eliminate the instability.

3. The working method according to claim 1, characterized in that, The different sensors specifically include a camera, a lidar, and a millimeter-wave radar. The camera is used to acquire images of the surrounding environment, the lidar is used to acquire three-dimensional point clouds of the surrounding environment, and the millimeter-wave radar is used to obtain the distance between the trackless rubber-wheeled vehicle and the surrounding environment.

4. The working method according to claim 1, characterized in that, Step two specifically involves: Historical trajectory , The future trajectory of history , generator Used to generate future trajector Used to determine the trajectory generated by the generator; The objective function minimizes the loss: ； Diversity loss: Forced generation of multiple differentiated trajectories, ; L2 regression loss: Constraints generate trajectories that closely approximate the true trajectories. ; Physical loss: ; in, For environment masking functions, For point Projection point to the nearest feasible region boundary; Complete objective function: ; Training process: Initialize generator parameters With discriminator ; Sampling real trajectory And generate virtual trajectory , ; With a fixed generator G, calculate the loss and update the discriminator D: ， ； Fixed discriminator D, compute function update generator G: ， ； The process continues until the discriminator and generator converge, thus obtaining the predicted future trajectory information of the dynamic obstacle.

5. The working method according to claim 1, characterized in that, Step five specifically involves: The phase plane region is divided into stable and unstable regions using the double-line method based on two stability boundary lines; the stability boundary is represented as: in, and These are stability boundary parameters, determined by vehicle parameters, vehicle speed, front wheel steering angle, and road adhesion coefficient. The slope and intercept of the stability boundary line are determined by... and Combinations determined; Stability coefficient The ratio of the distances from the current state point to the nearest stability boundary line: ; in, and These are the Euclidean distances from the current state point to the two stability boundary lines, respectively. The Euclidean distance from the static equilibrium point to the nearest stability boundary; when When the value is less than 1, the trackless rubber-wheeled vehicle becomes unstable.

6. The working method according to claim 1, characterized in that, Step six specifically involves: Calculate the yaw rate deviation: ; Calculate the sideslip angle deviation: ; Input variables and Mapping to fuzzy linguistic variables: , , Let be a fuzzy set, where the inputs are the deviation and the rate of change of the deviation; Rules are designed based on expert experience, and PID parameters are dynamically adjusted. : ; Based on the fuzzy inference model, output fuzzy quantities. ; Defuzzification: ; Real-time updates of PID parameters: The PID controller outputs the compensation torque. : ; The compensation torque of the trackless rubber-tired vehicle is obtained through the above formula to restore its stability and achieve safe and stable operation of the trackless rubber-tired vehicle.

Images