A control method and system for a tunnel inner pipeline transport vehicle to self-help go up and down steps

Abstract

This invention belongs to the field of pipeline construction technology, specifically a control method and system for a self-service pipeline transport vehicle to and from steps within a tunnel. The system includes: a perception module for accurately identifying and geometrically modeling the step structure within the tunnel to obtain step structure information; a decision module for inputting the step structure information into a proximal strategy optimization model to optimize the behavior strategy for ascending and descending steps and the control actions of the auxiliary wheels, and outputting a control strategy; a control module for constructing an attitude adjustment mechanism using position sensors, pressure sensors, and an IMU, combining the output control strategy to achieve PID closed-loop control, and outputting control commands; and an execution module for controlling hydraulic cylinders using electro-hydraulic proportional valves, driving the main drive wheel set and auxiliary support wheel set to work collaboratively according to the output control commands, enabling the pipeline transport vehicle to ascend and descend steps self-servicely within the tunnel. The technical solution of this invention can effectively solve the problems of low mechanization and poor safety of pipeline transport vehicles ascending and descending steps within tunnels.

Description

Technical Field

[0001] This invention relates to the field of pipeline construction technology, and in particular to a control method and system for a pipeline transport vehicle to move up and down steps independently in a tunnel. Background Technology

[0002] In recent years, tunnel water diversion projects have become a focus of water conservancy engineering construction due to their significant advantages, such as water sources being less susceptible to pollution and low water evaporation loss.

[0003] Because the tunnel needs to withstand a large internal pressure, secondary lining is required after the shield tunneling is completed. The inner lining pipes are large in diameter, long in length, and heavy in weight. When encountering obstacles in the tunnel, it is extremely difficult to transport them up and down steps. Although the steel inner lining pipes can be laid by using roller sliding, winch traction, and internal combustion engine wheel-rail trolley transportation, there are problems such as low mechanization and poor safety when working up and down steps. Summary of the Invention

[0004] To address the aforementioned technical problems, the present invention aims to provide a control method and system for self-service ascending and descending steps of pipeline transport vehicles within tunnels. This invention effectively solves the problems of low mechanization and poor safety associated with ascending and descending steps of pipeline transport vehicles within tunnels.

[0005] This invention discloses a control system for a pipeline transport vehicle to move up and down steps in a tunnel, comprising:

[0006] Perception module: used to accurately identify and geometrically model the step structure inside the tunnel to obtain step structure information;

[0007] Decision module: Used to input the step structure information output by the perception module into the proximal policy optimization model to optimize the behavior strategy of going up and down the steps and the auxiliary wheel control action, and output the control strategy;

[0008] Control Module: A posture adjustment mechanism is constructed using position sensors, pressure sensors, and an IMU. Combined with the control strategy output by the control module, closed-loop PID control is achieved, and control commands are output. The position sensors are located at the front of the pipeline transport vehicle, the pressure sensors are located at the contact points between the hydraulic cylinders, the running wheel assembly, the auxiliary support wheel assembly, and the steps, and the IMU is located in the middle of the pipeline transport vehicle.

[0009] The execution module includes an electro-hydraulic proportional valve, a hydraulic cylinder, a set of traveling wheels, and an auxiliary support wheel set. Based on the control commands output by the control module, the electro-hydraulic proportional valve controls the hydraulic cylinder, which drives the set of traveling wheels and the auxiliary support wheel set to work together to enable the pipeline transport vehicle to move up and down the steps in the tunnel independently.

[0010] This invention also discloses a control method for a pipeline transport vehicle to move up and down steps independently in a tunnel, comprising the following steps:

[0011] S1. An improved target detection model is used to fuse binocular depth estimation to accurately identify and geometrically model the step structure in the tunnel, thereby obtaining step structure information.

[0012] S2. Input the step structure information obtained in step S1 into the proximal strategy optimization model to optimize the behavior strategy of going up and down the steps and the control action of the auxiliary wheel, and output the control strategy.

[0013] S3. An attitude adjustment mechanism is constructed using position sensors, pressure sensors, and an IMU. Combined with the output control strategy, closed-loop PID control is achieved, and control commands are output.

[0014] S4. According to the control command output in step S3, the hydraulic cylinder is controlled by the electro-hydraulic proportional valve to drive the traveling wheel group and the auxiliary support wheel group to work together, so as to realize the self-service up and down of the pipeline transport vehicle in the tunnel.

[0015] Furthermore, step S1 includes:

[0016] Collect data on the existing steps within the tunnel to create a custom step dataset;

[0017] The improved object detection model was trained using a proprietary step dataset;

[0018] An improved target detection model is used for target detection inference. For images that meet the confidence level, preliminary step detection results are obtained.

[0019] Based on the preliminary detection results of the steps, the steps inside the tunnel are accurately identified and geometrically modeled, and the structural information of the steps is output.

[0020] Furthermore, based on the preliminary detection results of the steps, the steps inside the tunnel are accurately identified and geometrically modeled, and the output step structure information includes:

[0021] An improved target detection model was used to perform real-time target detection on the steps inside the tunnel to determine the position and outline of the steps.

[0022] The depth information of each point on the step is obtained by using binocular depth estimation. Combined with the preliminary detection results of the step, a three-dimensional geometric model of the step is constructed to obtain the step structure information, which includes the height, width and slope of the step.

[0023] Furthermore, step S2 includes: using the step structure information as input, using the stability, safety and efficiency of the pipeline transport vehicle going up and down the steps as the reward function, and training and optimizing the control actions of the pipeline transport vehicle going up and down the steps through a proximal policy optimization model to obtain the optimal control strategy adapted to different step structures; wherein the control actions include behavioral strategies and the extension and angle adjustment of the auxiliary wheels.

[0024] Furthermore, in step S2: the objective function of the proximal policy optimization model is:

[0025]

[0026] In the formula, For the new policy function updated during training, Here, A(s,a) is the old policy function, and A(s,a) is the advantage function output by the value network. The clip function is used to control the ratio of the old and new policy functions within a certain range. .

[0027] Furthermore, step S3 includes:

[0028] A position sensor installed at the front of the pipeline transport vehicle is used to detect the position data of the step on which the pipeline transport vehicle is located.

[0029] Pressure sensors, respectively installed at the contact points between the hydraulic cylinder, the traveling wheel assembly, and the auxiliary support wheel assembly and the step, are used to detect the pressure data of the pipeline transport vehicle in contact with the step;

[0030] An IMU located in the middle of the pipeline transport vehicle is used to detect the attitude data of the pipeline transport vehicle;

[0031] An attitude adjustment mechanism is constructed based on position data, pressure data, and attitude data. The deviation between the actual attitude and the desired attitude is calculated, and the deviation is adjusted by a PID controller.

[0032] Output control commands for adjusting the driving force and attitude of the running wheel set and the auxiliary support wheel set.

[0033] Furthermore, step S4 includes:

[0034] The three-position four-way electro-hydraulic proportional valve controls the oil inlet, oil return, pressure, extension and retraction of the hydraulic cylinder and its speed according to the output control command. The extension and retraction of the hydraulic cylinder adjusts the descent height of the auxiliary support wheel assembly. The position sensor, pressure sensor and IMU collect relevant data in real time, sense the real-time status of the hydraulic cylinder, and feed the relevant data back to step S2 to form a closed-loop control to correct the execution action.

[0035] The auxiliary support wheel set is controlled to drop when it encounters an obstacle, while the traveling wheel set is raised, enabling the pipeline transport vehicle to move up and down steps independently inside the tunnel.

[0036] Furthermore, the relationship between the piston rod's extension / retraction amount x and the descent height of the auxiliary support wheel assembly is as follows:

[0037]

[0038] In the formula, the horizontal distance from the fixed hinge point of the hydraulic cylinder to the fixed hinge point of the auxiliary support wheel connecting rod is L; the length of the auxiliary support wheel connecting rod is R; in the initial state (i.e., when the piston rod is not extended), the length of the hydraulic cylinder is L0; after the piston rod extends by x, the length of the hydraulic cylinder becomes L0+x; the descent height of the auxiliary wheel is... .

[0039] The beneficial effects of this invention are as follows: The perception module employs the YOLO model algorithm fused with binocular depth estimation, enabling accurate identification and geometric modeling of the step structure within the tunnel. This provides accurate environmental information for subsequent decision-making and control, solving the problem of inaccurate step structure identification in existing technologies. The decision-making module introduces the PPO algorithm, which optimizes the behavior strategy for ascending and descending steps and the control actions of the auxiliary wheels based on different step structures. This gives the pipeline transport vehicle AI self-learning capabilities, allowing it to adapt to different step environments and improving its adaptability and intelligence. The control module incorporates position sensors, pressure sensors, and an IMU to construct an attitude adjustment mechanism, achieving PID closed-loop control. This enables real-time monitoring of the pipeline transport vehicle's attitude and position information, allowing for timely adjustments to the control strategy and ensuring the stability and safety of the pipeline transport vehicle during step ascent and descent. The execution module uses electro-hydraulic proportional valves to control hydraulic cylinders, achieving coordinated control of the traveling wheel set and the auxiliary support wheel set. This results in fast response speed and high control accuracy, ensuring the coordination of the actions of each wheel set and further improving the stability and reliability of the pipeline transport vehicle when ascending and descending steps. The embodiments of the present invention enable pipeline transport vehicles to move up and down steps independently in tunnels, reducing manual intervention, improving transportation efficiency, and lowering the incidence of safety accidents, thus having significant practical application value. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0041] Figure 1 This is a flowchart of a control method for a pipeline transport vehicle to move up and down steps in a tunnel according to the present invention.

[0042] Figure 2 for Figure 1 Flowchart of the mid-sensing stage;

[0043] Figure 3 for Figure 1 Flowchart of the decision-making stage;

[0044] Figure 4 A schematic diagram of a pipeline transport vehicle adapted to the control method of the present invention;

[0045] Figure 5 This is a control principle diagram of the electro-hydraulic proportional valve and hydraulic cylinder in the execution module of this invention.

[0046] In the diagram: 1-1 is the chassis frame; 1-2 is the running wheel assembly; 1-3 is the auxiliary support wheel assembly; 1. PID controller; 2. Comparator element; 3. Hydraulic cylinder; 4. Position sensor; 5. Three-position four-way electro-hydraulic proportional valve; 6. Oil tank; 7. Low-voltage motor; 8. Fixed displacement oil pump. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with specific embodiments and accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of protection.

[0048] Please see Figures 1-5 This invention discloses a control system for a pipeline transport vehicle to move up and down steps in a tunnel, comprising the following modules:

[0049] Perception module: used to accurately identify and geometrically model the step structure inside the tunnel to obtain step structure information;

[0050] Decision module: Used to input step structure information into the proximal policy optimization model to optimize the behavior strategy for going up and down the steps and the control action of the auxiliary wheel, and output the control strategy;

[0051] Control module: The attitude adjustment mechanism is constructed using position sensors, pressure sensors and IMU. Combined with the control strategy output by the decision module, it realizes PID closed-loop control and outputs control commands. The position sensor is set at the front of the pipeline transport vehicle, the pressure sensors are set at the contact points between the hydraulic cylinder, the running wheel set and the auxiliary support wheel set and the step, and the IMU is set in the middle of the pipeline transport vehicle.

[0052] The execution module includes an electro-hydraulic proportional valve, a hydraulic cylinder, a set of traveling wheels, and an auxiliary support wheel set. Based on the control commands output by the control module, the electro-hydraulic proportional valve controls the hydraulic cylinder, which drives the set of traveling wheels and the auxiliary support wheel set to work together to enable the pipeline transport vehicle to move up and down the steps in the tunnel independently.

[0053] Based on the aforementioned control system, this invention also discloses a control method for a pipeline transport vehicle to move up and down steps independently in a tunnel, specifically including the following steps:

[0054] S1. Perception Stage: An improved target detection model is used to fuse binocular depth estimation to accurately identify and geometrically model the step structure inside the tunnel, thereby obtaining step structure information.

[0055] Specifically, the perception phase employs an improved object detection model based on deep learning. The object detection model is a YOLO series model, at least YOLOv5, with a confidence level of 0.8. The core principle of the YOLO model is to achieve object detection through convolutional neural networks, simultaneously predicting the object's category and bounding box information, thus transforming the object detection problem into a regression problem. The improved object detection model is characterized by adding an angle detection branch to the YOLO model, enabling it to detect curved steps within the tunnel.

[0056] Please see Figure 2 Step S1 specifically includes:

[0057] S11. Collect data on the existing steps within the tunnel to create a dataset of existing steps.

[0058] S12. The improved target detection model is trained using a proprietary step dataset;

[0059] S13. An improved target detection model is used for target detection inference. For images that meet the confidence level, preliminary step detection results are obtained.

[0060] S14. Based on the preliminary detection results of the steps, accurately identify and geometrically model the steps in the tunnel, and output the step structure information.

[0061] Specifically, step S11 includes using a binocular camera to identify step data and construct a custom step dataset; step S12 includes using the YOLOv5 model in conjunction with the custom step dataset for training, and using multiple threads (usually 16) to call multiple thread algorithms to improve computation speed.

[0062] Step S13 specifically includes using an improved object detection model for object detection inference. For step images with low recognition confidence, step data is re-acquired and input into the YOLOv5 model for training until the object detection inference result meets the confidence level (confidence level greater than or equal to 0.8). Then, based on the preliminary step detection results, the step structure in the tunnel is accurately identified and geometrically modeled to obtain step structure information. For images that meet the confidence level, the preliminary step detection results are directly obtained, and based on the preliminary step detection results, the steps in the tunnel are accurately identified and geometrically modeled to output step structure information.

[0063] The specific process of accurately identifying and geometrically modeling the steps in the tunnel in step S14 is as follows: using the YOLOv5 model to perform real-time target detection on the steps in the tunnel to determine the position and outline of the steps; using binocular depth estimation to obtain the depth information of each point on the steps, and combining the preliminary detection results of the steps to construct a three-dimensional geometric model of the steps and obtain the step structure information; among which, the step structure information includes the height, width, slope and other information of the steps.

[0064] S2, Decision-making stage: Input the step structure information obtained in step S1 into the proximal policy optimization model to optimize the behavior strategy of going up and down the steps and the auxiliary wheel control action, and output the control strategy.

[0065] Please see Figure 3 In the decision-making stage, the step structure information obtained in the perception stage needs to be input into the proximal policy optimization model (PPO algorithm). The core of the PPO algorithm is to ensure that each policy update is within a confidence region, obtain self-reward through environmental interaction, and use this to learn itself.

[0066] Step S2 specifically includes taking the step structure information as input, using the stability, safety, and efficiency of the pipeline transport vehicle going up and down the steps as the reward function, introducing a proximal strategy optimization model, and training and optimizing the control actions of the pipeline transport vehicle going up and down the steps through the proximal strategy optimization model to obtain the optimal control strategy adapted to different step structures; wherein the control actions include the behavior strategy of going up and down the steps and the extension and angle adjustment of the auxiliary wheels.

[0067] The objective function of the near-end policy optimization model is:

[0068]

[0069] In the formula, For the new policy function updated during training, The old policy function, A(s,a), is the dominance function output by the value network, obtained using the generalized dominance method (GAE) by estimating the reward signal and value function of the environment; the parameters in the new policy function... After a period of time, it will replace the old strategy function. The `clip` function is used to control the ratio of the old and new policy functions to a certain value. This is used to limit the magnitude of updates, preventing excessive policy updates and ensuring the stability of parameter updates. Among these, the parameters... The update formula is:

[0070]

[0071] In the formula, These are the strategy parameters to be updated; These are the parameters from the old strategy function; Indicating the old strategy Expected state-action distribution in the induced state; This means finding the parameter that maximizes the expected value E. .

[0072] S3. Control Stage: A posture adjustment mechanism is constructed using position sensors, pressure sensors, and an IMU. Combined with the control strategy output from step S2, closed-loop control of the PID controller is achieved, and control commands are output.

[0073] Specifically, the control phase introduces position sensors, pressure sensors, and an IMU to construct an attitude adjustment mechanism. Position sensors are installed at the front of the pipeline transport vehicle to acquire real-time positional information such as the distance and relative angle between the front of the vehicle and the step. Pressure sensors are installed at the contact points between the hydraulic cylinders, the running wheels, and the auxiliary support wheels with the step, detecting pressure data from the hydraulic cylinders and the pressure data between the wheels and the step. Each hydraulic cylinder, running wheel, and auxiliary support wheel has a pressure sensor at its contact point with the step. The IMU (Inertial Measurement Unit) is located in the middle of the pipeline transport vehicle to collect attitude information such as tilt angle, angular velocity, and acceleration. The position information from the position sensors, the pressure information from the pressure sensors, and the attitude information from the IMU are compared with the desired pressure range and attitude parameters to calculate the deviation. The deviation is adjusted by a PID controller, and combined with the control strategy output from the decision-making phase, control commands are output to adjust the driving force and attitude of the running wheels and auxiliary support wheels, achieving closed-loop control of the PID controller.

[0074] S4. Execution phase: According to the control command output in step S3, the hydraulic cylinder is controlled by an electro-hydraulic proportional valve to drive the main drive wheel set and the auxiliary support wheel set to work together, so as to realize the self-service up and down of the pipeline transport vehicle in the tunnel.

[0075] Please see Figure 4 , Figure 4This is a schematic diagram of a pipeline transport vehicle adapted to the control method described in this invention. When transporting pipelines, a pipeline transport vehicle needs to be installed at both ends of the supporting beam. The pipeline transport vehicle includes a chassis frame 1-1, a set of running wheels 1-2 suspended on the chassis frame, and an auxiliary support wheel set 1-3 connected to the chassis frame via a hydraulic cylinder 3. The running wheels 1-2 provide the main power for the pipeline transport vehicle to move forward. The auxiliary support wheel set 1-3 is used to assist in supporting and adjusting the vehicle's posture when going up and down steps. The hydraulic cylinder drives the extension of the auxiliary support wheel set to support and adjust the vehicle's posture. The rear running wheels provide power to the vehicle, causing the front running wheels to lift, thus completing the self-service step-climbing function of the pipeline transport vehicle in the tunnel. Specifically, the pipeline transport vehicle can climb steps and load / unload into and out of pipelines by coordinating the traveling wheel set 1-2 and the auxiliary support wheel set 1-3. When the pipeline transport vehicle needs to go up a step, the auxiliary support wheel set 1-3 is lowered, the traveling wheel set 1-2 is raised, and the pipeline transport vehicle moves towards the step. When the traveling wheel set 1-2 is completely on the step, the traveling wheel set 1-2 is lowered, and the auxiliary support wheel set 1-3 is retracted. The action of going down a step can be deduced in the same way.

[0076] For further details, please refer to Figure 5 The rod chamber of the hydraulic cylinder 3 is connected to port B of the three-position four-way electro-hydraulic proportional valve through a pipeline, and its rodless chamber is connected to port A of the three-position four-way electro-hydraulic proportional valve through a pipeline. The P port of the three-position four-way electro-hydraulic proportional valve is connected to the oil tank through the quantitative oil pump 8, and its T port is connected to another oil tank 6. The position sensor 4, IMU and pressure sensors at the contact points between the hydraulic cylinder, the traveling wheel group and the auxiliary support wheel group and the step are all electrically connected to the PID controller 1. The PID controller is electrically connected to the three-position four-way electro-hydraulic proportional valve.

[0077] The specific process of using an electro-hydraulic proportional valve to control the hydraulic cylinder during the execution phase is as follows: The three-position four-way electro-hydraulic proportional valve 5, according to the control commands output during the control phase, controls the oil inlet and outlet flow, pressure, and piston rod extension and retraction speed of the hydraulic cylinder 3. The height of the auxiliary support wheel assembly's descent is adjusted based on the piston rod's extension and retraction. Specifically, the three-position four-way electro-hydraulic proportional valve drives the piston rod to extend or retract according to the control commands output during the control phase to control the raising and lowering of the auxiliary support wheel assembly. When it is determined that an ascent to a step is required, the three-position four-way electro-hydraulic proportional valve is in the left position. In this position, port A is connected to port P, and port B is connected to port T. The fixed displacement pump 8 delivers hydraulic oil to the rodless chamber of the hydraulic cylinder through the PA passage. Simultaneously, the hydraulic oil in the rod chamber returns to another oil tank through the BT passage, making the oil pressure in the rodless chamber greater than the oil pressure in the rod chamber. This drives the piston rod of the hydraulic cylinder to extend, causing the auxiliary support wheel assembly 1-3 to fall, lifting the traveling wheel assembly 1-2, and the pipeline transport vehicle to move towards the step. When the traveling wheel assembly 1-2 completes its ascent... When all the wheels are on the steps, the travel wheel set 1-2 is lowered. At this time, the PID controller switches the three-position four-way electro-hydraulic proportional valve to the right position. In this position, port A is connected to port T and port B is connected to port P. The fixed displacement oil pump sends hydraulic oil to the rod chamber of the hydraulic cylinder through the PB passage. At the same time, the hydraulic oil in the rodless chamber returns to another oil tank through the AT passage, so that the oil pressure in the rod chamber is greater than the oil pressure in the rodless chamber, driving the piston rod of the hydraulic cylinder to retract and retract the auxiliary support wheel set 1-3. The corresponding action process when going down the steps can be deduced by analogy.

[0078] Meanwhile, the decision-making module incorporates relevant data from position sensors, pressure sensors, and IMU feedback into the training samples using the PPO algorithm, optimizes subsequent action strategies, and accurately and safely controls the auxiliary support wheels to fall when encountering obstacles, supporting the pipeline transport vehicle to lift the running wheel set, thereby enabling the pipeline transport vehicle to go up and down steps.

[0079] The relationship between the piston rod's extension / retraction x and the distance ∆h that the auxiliary support wheel assembly descends is:

[0080]

[0081] Wherein, the horizontal distance from the fixed hinge point of the hydraulic cylinder to the fixed hinge point of the auxiliary support wheel connecting rod is L; the length of the auxiliary support wheel connecting rod is R; in the initial state (i.e., when the piston rod is not extended), the length of the hydraulic cylinder is L0; after the piston rod extends x, the length of the hydraulic cylinder becomes L0+x; the descent height of the auxiliary support wheel is... (i.e., vertical displacement).

[0082] This invention's perception module employs the YOLO model algorithm fused with binocular depth estimation, enabling accurate identification and geometric modeling of the step structure within tunnels. This provides accurate environmental information for subsequent decision-making and control, solving the problem of inaccurate step structure identification in existing technologies. The decision-making module introduces the PPO algorithm, optimizing the behavior strategies for ascending and descending steps and the control actions of the auxiliary wheels based on different step structures. This gives the pipeline transport vehicle AI self-learning capabilities, allowing it to adapt to different step environments and improving its adaptability and intelligence. The control module incorporates position sensors, pressure sensors, and an IMU to construct an attitude adjustment mechanism, achieving PID closed-loop control. This enables real-time monitoring of the pipeline transport vehicle's attitude and position information, allowing for timely adjustments to the control strategy and ensuring the stability and safety of the pipeline transport vehicle during step ascent and descent. The execution module uses a three-position four-way electro-hydraulic proportional valve to control the hydraulic cylinder, achieving coordinated control of the traveling wheel set and the auxiliary support wheel set. This results in fast response speed and high control precision, ensuring the coordination of the actions of each wheel set and further improving the stability and reliability of the pipeline transport vehicle when ascending and descending steps. The embodiments of the present invention enable pipeline transport vehicles to move up and down steps independently in tunnels, reducing manual intervention, improving transportation efficiency, and lowering the incidence of safety accidents, thus having significant practical application value.

[0083] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort, such as modifications to the technical solutions described in the following embodiments or equivalent substitutions of some technical features, are within the scope of protection of the present invention.

Claims

1. A control system for a pipeline transport vehicle to automatically move up and down steps in a tunnel, characterized in that, include: Perception module: used to accurately identify and geometrically model the step structure inside the tunnel to obtain step structure information; Decision module: Used to input the step structure information output by the perception module into the proximal policy optimization model to optimize the behavior strategy of going up and down the steps and the auxiliary wheel control action, and output the control strategy; Control module: The attitude adjustment mechanism is constructed using position sensors, pressure sensors and IMU. Combined with the control strategy output by the decision module, it realizes PID closed-loop control and outputs control commands. The position sensor is set at the front of the pipeline transport vehicle, the pressure sensors are set at the contact points between the hydraulic cylinder, the running wheel set and the auxiliary support wheel set and the step, and the IMU is set in the middle of the pipeline transport vehicle. The execution module includes an electro-hydraulic proportional valve, a hydraulic cylinder, a set of traveling wheels, and an auxiliary support wheel set. Based on the control commands output by the control module, the electro-hydraulic proportional valve controls the hydraulic cylinder, which drives the set of traveling wheels and the auxiliary support wheel set to work together to enable the pipeline transport vehicle to move up and down the steps in the tunnel independently.

2. A control method for a pipeline transport vehicle to move up and down steps independently in a tunnel, characterized in that, Includes the following steps: S1. An improved target detection model is used to fuse binocular depth estimation to accurately identify and geometrically model the step structure in the tunnel, thereby obtaining step structure information. S2. Input the step structure information obtained in step S1 into the proximal strategy optimization model to optimize the behavior strategy of going up and down the steps and the control action of the auxiliary wheel, and output the control strategy. S3. A posture adjustment mechanism is constructed using position sensors, pressure sensors, and an IMU. Combined with the control strategy output in step S2, closed-loop PID control is achieved, and control commands are output. S4. According to the control command output in step S3, the hydraulic cylinder is controlled by the electro-hydraulic proportional valve to drive the traveling wheel group and the auxiliary support wheel group to work together, so as to realize the self-service up and down of the pipeline transport vehicle in the tunnel.

3. The control method for a self-service up-and-down movement of a pipeline transport vehicle in a tunnel according to claim 2, characterized in that, Step S1 includes: Collect data on the existing steps within the tunnel to create a custom step dataset; The improved object detection model was trained using a proprietary step dataset; An improved target detection model is used for target detection inference. For images that meet the confidence level, preliminary step detection results are obtained. Based on the preliminary detection results of the steps, the structure of the steps inside the tunnel is accurately identified and geometrically modeled, and the step structure information is output.

4. The control method for a self-service up-and-down movement of a pipeline transport vehicle in a tunnel according to claim 3, characterized in that, Based on the preliminary inspection results of the steps, the structure of the steps inside the tunnel is accurately identified and geometrically modeled. The output step structure information includes: An improved target detection model was used to perform real-time target detection on the steps inside the tunnel to determine the position and outline of the steps. The depth information of each point on the step is obtained by using binocular depth estimation. Combined with the preliminary detection results of the step, a three-dimensional geometric model of the step is constructed to obtain the step structure information, which includes the height, width and slope of the step.

5. The control method for a self-service up-and-down movement of a pipeline transport vehicle in a tunnel according to claim 2, characterized in that, Step S2 includes: using the step structure information as input, the stability, safety and efficiency of the pipeline transport vehicle going up and down the steps as the reward function, and training and optimizing the control actions of the pipeline transport vehicle going up and down the steps through the proximal policy optimization model to obtain the optimal control strategy adapted to different step structures; wherein the control actions include behavioral strategies and the extension and angle adjustment of the auxiliary wheels.

6. The control method for a self-service up-and-down movement of a pipeline transport vehicle in a tunnel according to claim 5, characterized in that, In step S2: the objective function of the near-end policy optimization model is: , In the formula, For the new policy function updated during training, Here, A(s,a) is the old policy function, and A(s,a) is the advantage function output by the value network. The clip function is used to control the ratio of the old and new policy functions within a certain range. .

7. The control method for a self-service ascending and descending step of a pipeline transport vehicle in a tunnel according to claim 2, characterized in that, Step S3 includes: A position sensor installed at the front of the pipeline transport vehicle is used to detect the position data of the step on which the pipeline transport vehicle is located. Pressure sensors installed at the contact points between the traveling wheel assembly and the auxiliary support wheel assembly and the step are used to detect the pressure data of the pipeline transport vehicle in contact with the step. An IMU located in the middle of the pipeline transport vehicle is used to detect the attitude data of the pipeline transport vehicle; An attitude adjustment mechanism is constructed based on position data, pressure data, and attitude data. The deviation between the actual attitude and the desired attitude is calculated, and the deviation is adjusted by a PID controller. Output control commands for adjusting the driving force and attitude of the running wheel set and the auxiliary support wheel set.

8. The control method for a self-service up-and-down movement of a pipeline transport vehicle in a tunnel according to claim 2, characterized in that, Step S4 includes: According to the control command output in step S3, a three-position four-way electro-hydraulic proportional valve is used to control the oil inlet, oil return, pressure, piston rod extension and retraction speed of the hydraulic cylinder. The lowering height of the auxiliary support wheel assembly is adjusted by the piston rod extension. The position sensor, pressure sensor, and IMU collect relevant data in real time to sense the real-time status of the hydraulic cylinder and feed the relevant data back to step S3 to form a closed-loop control to correct the execution action. The auxiliary support wheel set is controlled to drop when it encounters an obstacle, while the traveling wheel set is raised, enabling the pipeline transport vehicle to move up and down steps independently inside the tunnel.

9. A control method for a pipeline transport vehicle to move up and down steps in a tunnel according to claim 8, characterized in that, The relationship between the piston rod's extension / retraction amount x and the descent height of the auxiliary support wheel assembly is as follows: ， In the formula, the horizontal distance from the fixed hinge point of the hydraulic cylinder to the fixed hinge point of the auxiliary support wheel connecting rod is L; the length of the auxiliary support wheel connecting rod is R; in the initial state, the length of the hydraulic cylinder is L0; after the piston rod extends x, the length of the hydraulic cylinder becomes L0+x; the descent height of the auxiliary support wheel is... .

Images