Cooperative control method and system for hoisting and lifting of transformer substation operation vehicle

Abstract

The invention discloses a substation operation vehicle hoisting and lifting cooperative control method and system, and the method comprises the steps: collecting operation environment and obstacle information, planning a non-collision cooperative trajectory of a cargo boom and a lifting arm based on an improved RRT algorithm, achieving the precise tracking of the trajectory through a model prediction control algorithm, and achieving the cooperative control of the hoisting and lifting of the substation operation vehicle. And a multi-level active safety protection mechanism is combined to complete collaborative operation. The system comprises an operation mechanism layer, a sensing and detecting layer, a control and decision-making layer and a man-machine interaction layer, and all the layers are coordinated and linked to realize intelligent coordination of hoisting and lifting actions. The method solves the problems of complex operation, poor collaboration and high safety risk in the prior art, improves the operation safety, stability and automation level, and can be widely applied to power transmission and transformation projects and other scenes.

Description

Technical Field

[0001] This invention relates to the field of substation operation technology, and more specifically, to a method and system for coordinated control of lifting and hoisting of a substation operation vehicle. Background Technology

[0002] With the continuous expansion of the power grid, the tasks of routine maintenance, troubleshooting, and upgrades of equipment in substations, including high-altitude operations, are becoming increasingly demanding. In these operations, specialized substation work vehicles (such as aerial work platforms and insulated bucket trucks) are indispensable key equipment. The safety, stability, and efficiency of their operation directly affect the reliability of the power grid, equipment safety, and personnel safety. Currently, when performing combined tasks involving equipment hoisting and personnel lifting in substations, the traditional operating mode, where the hoisting and lifting mechanisms operate independently and in shifts, is commonly used. This mode requires operators to possess high levels of skill, coordinating the movement of heavy loads and the raising and lowering of the work platform by alternately operating multiple control devices and relying on experience. This highly manual-dependent mode has significant drawbacks: 1) Poor operational coordination, low work efficiency, and high safety risks: Due to the lack of system-level coordinated control, the operation process is cumbersome and slow to respond, making it difficult to achieve precise and stable synchronization between the load and the work platform. In complex on-site environments, this lack of coordination not only leads to low work efficiency but also greatly increases the risk of safety accidents due to operational errors; 2) Significant potential risks and weak operational safety assurance capabilities: The shortcomings of traditional independent lifting and maintenance operations are reflected in two aspects: First, the entire vehicle is at risk of overturning. Independent motion control cannot perceive and calculate in real time the changes in the vehicle's center of gravity and torque distribution caused by the combined lifting and hoisting actions. When the overturning moment generated by the combined actions exceeds the stability limit, it may lead to the catastrophic consequence of the entire vehicle overturning. Second, there is a risk of collision during operation. In substations with compact structures and numerous electrical equipment, the crane boom and hoisting boom, lacking coordinated motion trajectory planning, are extremely prone to collisions with critical equipment such as busbars and instrument transformers when operating independently, causing serious equipment damage or power accidents. 3) Insufficient intelligence and passive safety assurance: Existing safety systems are mostly passive and lagging protection devices targeting a single mechanism or single parameter. They can only alarm or shut down when limits are exceeded, and cannot perform proactive and preventative safety coordination control at the vehicle system level. They lack the ability to comprehensively perceive dynamic loads, stability, and environmental information and make intelligent decisions.

[0003] Furthermore, while patent application CN114852888A discloses a solution that integrates the lifting arm and the lifting platform into the same vehicle body, improving equipment integration to some extent, its core technology lies in the combination of mechanical structures and does not address the deep coordination and dynamic safety protection of the two actuators at the control level. Therefore, this solution still cannot solve the problems of operational complexity, low efficiency, and systemic safety risks caused by the lack of control coordination.

[0004] In summary, existing technologies and published patents are insufficient to meet the high standards of efficiency, intelligence, and safety required for high-altitude operations in modern substations. Therefore, developing a control method and system capable of intelligent coordination of lifting and hoisting actions and integrating active safety protection mechanisms to fundamentally overcome the shortcomings of existing technologies has become a critical technical problem urgently needing to be solved in this field. Summary of the Invention

[0005] This invention provides a method and system for coordinated control of lifting and hoisting of substation work vehicles, which improves the RRT by introducing deep reinforcement learning. Path planning algorithms and model predictive control algorithms enable dynamic coordination and precise control of the crane boom and lifting arm. Combined with active safety protection mechanisms, this improves the safety, stability, and automation level of operations at the system level, reduces manual intervention, lowers reliance on operator skills, and meets the high-efficiency, intelligent, and safe requirements of modern substation operations.

[0006] The technical solution adopted in this invention is: A method for coordinated control of lifting and hoisting of a substation work vehicle includes the following steps: S1. Detection and assessment of internal and external working environment: Collect information on the working environment and spatial data of obstacles, construct a three-dimensional scene model of the working environment, and assess the feasibility of the working environment; S2. Crane boom and lifting arm path planning: After receiving the operation command, the path planning is performed using an obstacle detection mechanism combined with an improved RRT based on deep reinforcement learning. The path planning algorithm plans the collision-free optimal cooperative expected trajectory of the crane boom and lifting arm from the starting point to the target point. S3. Cooperative control of lifting boom and boom: Based on the generated collision-free optimal cooperative expected trajectory of lifting boom and boom, the control quantity is obtained through the output of model predictive control algorithm to achieve accurate trajectory tracking of lifting boom and boom. S4. Coordinated operation of crane boom and lifting arm: Under the action of control, the crane moves to the work position according to the planned trajectory to complete the predetermined work. During the entire operation, the operation status is monitored in real time through an active safety protection mechanism. When a safety risk is detected, the corresponding protective action is executed.

[0007] Furthermore, the obstacle detection mechanism in step S2 is as follows: 1) The obstacle and robotic arm are modeled using the bounding sphere method: ① Model spatial obstacles as spheres; the formula for a sphere is as follows:

[0008] In the formula, , , For the first The three-dimensional coordinates of the center of the sphere in the world coordinate system of the obstacle. The radius of the sphere; ② Model the crane arm and lifting arm as multiple cylindrical segments; the coordinates of any point on the robotic arm are as follows:

[0009] in,

[0010]

[0011] In the formula, , These are the coordinates of the left and right endpoints of the robotic arm's main arm; The value to be solved; 2) Conduct collision detection between obstacles and the robotic arm, as well as between the crane arm and the lifting arm: ① Substituting the coordinates of any point on the robotic arm into the obstacle sphere formula, we obtain the collision detection equation:

[0012] In the formula, It is the sum of the radius of the obstacle and the radius of the cylinder; ② Solve for t in the collision detection equation and determine whether there is a collision risk between the robotic arm and the obstacle: When there is no solution to the collision detection equation, it is determined that there is no collision between the robotic arm and the obstacle; when there is a solution to the collision detection equation and the solution is within a reasonable range, it is determined that there is a collision risk between the robotic arm and the obstacle, and the path is adjusted until there is no collision. ③ Calculate the relationship between the distance between the centerlines of the crane arm and the lifting arm and the preset threshold to determine whether there is a collision risk between the crane arm and the lifting arm: when the distance is greater than or equal to the preset threshold, it is determined that there is no collision between the crane arm and the lifting arm; when the distance is less than the preset threshold, it is determined that there is a collision risk between the crane arm and the lifting arm, and the movement trajectory of the crane arm and the lifting arm is optimized until there is no collision.

[0013] Furthermore, the improved RRT in step S2 The path planning algorithm employs an improved RRT based on deep reinforcement learning. The path planning algorithm is as follows: 1) The workspace is defined using a three-dimensional coordinate range; the workspace must meet the following conditions:

[0014] In the formula, L is the length of the workspace, W is the width of the workspace, and H is the height of the workspace; 2) Initialization and Sampling Expansion: Initialize a search tree containing only the starting point, which is the initial position coordinates of the robotic arm; set a random sampling area centered on the starting point, and perform random sampling within the area to generate a random sampling point; traverse all nodes in the search tree, find the node closest to the sampling point, and extend the node towards the sampling point with a set step size based on the closest node to generate a new node; 3) Collision detection: Perform collision detection on the path edges between the new node and the nearest node to determine whether there is a risk of collision with an obstacle or another robotic arm: if a collision is detected, the new node is abandoned and resampled; if no collision is detected, proceed to the next step. 4) Local path optimization: including reselection and rerouting operations; Among them, ① Reselection operation: Set a certain search area near the new node, find all candidate parent nodes, calculate the path cost from the starting point through each candidate parent node to the new node, and select the candidate parent node with the smallest path cost as the parent node of the new node. ② Rewire operation: Check the existing nodes near the new node, calculate the path cost from the starting point through the new node to these existing nodes. If the path cost is less than the existing path cost, redirect the parent node of these existing nodes to the new node to optimize the search tree structure. 5) Iteration and convergence: Repeat the above sampling, collision detection, and local path optimization steps to continuously expand and optimize the search tree until the collision-free optimal cooperative expected trajectory from the starting point to the target point is found.

[0015] Further, step S3 includes the following: 1) A mathematical model of the robotic arm is established based on the Euler-Lagrange equations. The model includes system state variables and control variables; the expressions are as follows:

[0016] In the formula, These are actual state quantities; This refers to the actual control quantity; 2) Obtain the generated optimal cooperative expected trajectory between the crane boom and the lifting arm without collision; the expression for the expected trajectory is as follows:

[0017] In the formula, The desired state quantity is the sequence of state parameters corresponding to the cooperative trajectory generated by path planning. For desired control quantity; 3) Define tracking error; tracking error is the difference between the actual state quantity and the desired state quantity, expressed as follows:

[0018] In the formula, For tracking error; These are actual state quantities; For the desired state quantity; By performing a first-order Taylor approximation expansion on the actual state equations and subtracting the equations for the actual trajectory and the desired trajectory, the linear time-varying tracking error equation is obtained. The expression for the linear time-varying tracking error equation is as follows:

[0019] In the formula, The first derivative of the tracking error; , This is the system coefficient matrix; This is the difference between the actual control quantity and the desired control quantity. An approximate discretization error model is used to discretize the linear time-varying tracking error equation, transforming the continuous system into a discrete system; the expression is as follows:

[0020] In the formula, Representing the Each sampling time; For the first The difference between the actual state quantity and the expected state quantity at each sampling time; For the first The difference between the actual control quantity and the expected control quantity at each sampling time , The system coefficient matrix after discretization; 4) Design the objective function and control constraints; the objective function expression is as follows:

[0021] The control constraints are as follows:

[0022] In the formula, For prediction in the time domain; and These are the weight matrices for the state variables and the control variables, respectively. By adjusting the element values ​​of the weight matrices, the accuracy of state tracking and the smoothness of control variables are balanced. Represents the transpose of a matrix; , These are the minimum and maximum values ​​of the control quantity, respectively; 5) Obtain a new state-space expression through state-space transformation;

[0023] in,

[0024] In the formula, and This represents the system coefficient matrix in the new state space. This refers to the current state of the system. To control the control increment within the time domain; 6) The objective function is transformed into a quadratic programming problem, and an adaptive dynamic programming algorithm is used to solve it to obtain the control increment sequence in the control time domain. The first control increment is taken and combined with the control quantity at the previous time step to calculate the control quantity at the current time step. 7) Repeat the above process to update the system status and control variables in real time, so as to achieve accurate trajectory tracking of the crane boom and lifting arm, and accurate tracking of the actual trajectory to the desired trajectory.

[0025] Furthermore, the active safety protection mechanism in step S4 includes overturning warning and stability control, collision prediction and obstacle avoidance control, platform stability control and multi-level safety response control, real-time monitoring of the operation status, and execution of deceleration, amplitude limiting, trajectory adjustment, emergency braking or locking actions according to changes in safety margin.

[0026] Furthermore, the multi-level safety response control includes three safety levels: warning level, restriction level, and protection level. The warning level is the lowest safety level: triggered when a minor safety risk is detected, the audible and visual alarm device emits a yellow warning signal, and the display screen shows the corresponding warning information, allowing operators to adjust their work accordingly. The restriction level is the medium safety level: triggered when the risk level increases, the system automatically limits the arm's movement speed and amplitude, prohibits high-risk actions, and simultaneously emits a continuous yellow warning signal. The protection level is the highest safety level: triggered when a serious safety risk is detected, the system performs an emergency braking or locking action, stopping all work actions, the audible and visual alarm device emits a strong red alarm signal, and the display screen shows a safety protection prompt. Work can only resume after the operator has investigated and eliminated the safety risk and deactivated the protection status via the operating terminal.

[0027] Another technical solution adopted by the present invention is: A substation work vehicle lifting and hoisting collaborative control system includes: The working mechanism layer is used to perform hoisting and lifting operations; The perception and detection layer is used to collect operation-related data and perform multi-source fusion processing. The control and decision-making layer is used to implement path planning, collaborative control, and safety decisions. The human-computer interaction layer is used for selecting work modes, setting parameters, and exchanging work information.

[0028] Furthermore, the working mechanism layer includes a lifting arm, a hoisting arm, a working platform, and a hydraulic drive unit. The lifting arm is used to lift heavy objects or equipment; the hoisting arm is used to carry workers for high-altitude maintenance; both the lifting arm and the hoisting arm are driven by a hydraulic unit and are equipped with degrees of freedom for angle, extension, and rotation.

[0029] Furthermore, the sensing and detection layer includes a load sensor, a tilt sensor, an angle encoder, a pressure sensor, an inertial measurement unit, a camera, and a radar sensing module. The load sensor is installed at the root and end of the lifting arm, collecting load data in real time by sensing changes in tension during the lifting process to reflect the weight and stress state of the lifted object. The tilt sensor is installed at the four corners of the vehicle body, using high-precision measuring elements to collect lateral and longitudinal tilt information of the vehicle body and assess the stability of the entire vehicle. The angle encoder is installed at the joints of the robotic arm, detecting changes in the joint angles of the robotic arm in real time to provide basic data for attitude calculation. The pressure sensor is installed at key parts of the hydraulic drive unit, monitoring the pressure data of the hydraulic drive unit and reflecting its working status. The inertial measurement unit is installed in the middle of the vehicle body, integrating an accelerometer and a gyroscope to acquire the acceleration, angular velocity, and attitude information of the vehicle body, capturing dynamic changes in the vehicle body. The camera uses a high-definition industrial camera to collect image information of the working environment, assisting in the identification of obstacles and work targets. The radar sensing module uses millimeter-wave radar or lidar to scan the work space and acquire information on the distance, position, and motion status of surrounding obstacles.

[0030] Furthermore, the control and decision-making layer includes a central control unit, a collaborative control unit, a safety decision-making unit, a work planning unit, and a communication interface; wherein, the central control unit is responsible for information fusion and task allocation; the collaborative control unit executes collaborative trajectory planning and outputs control signals; the safety decision-making unit assesses the safety status in real time and triggers active protection actions; and the work planning unit runs an improved RRT incorporating deep reinforcement learning. The path planning algorithm; the human-machine interface layer includes an operating terminal, a display screen, an audible and visual alarm device, and a remote monitoring interface; the operating terminal supports manual, semi-automatic, and automatic operation modes and parameter settings; the display screen shows the attitude, load, tilt angle, and safety margin in real time; the audible and visual alarm module issues a warning when the safety threshold is approached; the remote monitoring interface supports remote data transmission and operation monitoring.

[0031] Compared with the prior art, the present invention has the following advantages: 1) Improved RRT by introducing deep reinforcement learning The deep integration of path planning algorithms and model predictive control algorithms enables intelligent path planning and automatic coordinated control of the crane boom and lifting arm, reducing manual adjustment and waiting time during operation; among other things, the improved RRT (Rapid Response Time) is achieved. The algorithm can quickly generate collision-free optimal trajectories, and the model predictive control algorithm ensures that the robotic arm moves accurately and smoothly, avoiding path redundancy and repeated actions caused by uncoordinated operation. The robotic arm can move to the work position quickly and accurately, significantly shortening the overall work cycle, improving work efficiency, and meeting the needs of efficient high-altitude operations in substations. 2) Construct a multi-level active safety protection system that includes overturning early warning and stability control, collision prediction and obstacle avoidance control, platform stability control, and multi-level safety response mechanisms; by using multiple sensors to perceive the operation status in real time, it can predict potential risks such as overturning, collision, and platform instability in advance, and take active protective actions such as deceleration, amplitude limiting, trajectory adjustment, and emergency braking, thereby changing the status quo of passive protection by traditional technology, effectively preventing the occurrence of various safety accidents, protecting personnel and equipment safety, and improving the reliability of operations; 3) The system integrates multi-source information sensing, intelligent decision-making, and automatic control functions, and has the ability to fuse multi-source information and make intelligent decisions. The sensing and detection layer can comprehensively collect data on the working environment and equipment status, and the control and decision-making layer can dynamically adjust algorithm parameters and control strategies according to environmental changes, adapting to the complex and ever-changing power grid working environment. It can autonomously complete the work tasks without much human intervention, promote the intelligent development of substation work vehicles, and provide new solutions for the technological upgrading of power operation and maintenance equipment. Attached Figure Description

[0032] The present invention will be further described below with reference to the accompanying drawings and specific embodiments: Figure 1 This is a schematic diagram of the overall architecture of the collaborative control system in this invention.

[0033] Figure 2 This is a flowchart illustrating a coordinated control method for lifting and hoisting of a substation work vehicle according to the present invention. Figure 1 ; Figure 3 This is a flowchart illustrating a coordinated control method for lifting and hoisting of a substation work vehicle according to the present invention. Figure 2 . Detailed Implementation

[0034] Example 1

[0035] like Figure 1 As shown, a substation work vehicle lifting and hoisting coordinated control system includes: (1) Operation mechanism level The operating mechanism layer is the system's execution mechanism, including the crane boom, lifting arm, work platform, and hydraulic drive unit, providing mechanical support for operation. The crane boom adopts a multi-section telescopic boom structure and is equipped with lifting components such as hooks for lifting heavy objects or equipment, enabling large-scale spatial lifting operations. The lifting arm adopts a folding or telescopic structure, with the top connected to the work platform for carrying workers to perform high-altitude maintenance operations. The work platform is equipped with safety facilities such as guardrails and anti-slip mats to ensure the personal safety of workers.

[0036] Note: Both the lifting boom and the hoisting boom are driven by a hydraulic drive unit, which includes components such as a hydraulic pump, hydraulic valves, hydraulic cylinders, and hydraulic oil pipes. It features fast response, large output torque, and smooth operation, and can precisely control the angle adjustment, extension, and rotation of the robotic arm to meet the operational needs of different work scenarios. The pressure, flow, and other parameters of the hydraulic drive unit are monitored in real time by a pressure sensor, providing data support for collaborative control and safety decision-making.

[0037] (2) Perception and Detection Layer

[0038] The perception and detection layer is the core of the system's data acquisition, including load sensors, tilt sensors, angle encoders, pressure sensors, inertial measurement units, cameras, and radar sensing modules. It is responsible for comprehensively collecting various key data during the operation. Load sensors are installed at the base and end of the crane boom, sensing changes in tension during lifting to collect load data in real time, reflecting the weight and stress state of the lifted load. Tilt sensors are installed at the four corners of the vehicle body, using high-precision measuring elements to collect lateral and longitudinal tilt information, assessing the overall stability of the vehicle. Angle encoders are installed at the joints of the robotic arm to detect mechanical... The changes in the joint angles of the arm provide basic data for attitude calculation; pressure sensors are installed in key parts of the hydraulic drive unit to monitor the pressure data of the hydraulic drive unit and reflect its working status; the inertial measurement unit is installed in the middle of the vehicle body, integrating accelerometers and gyroscopes to acquire the acceleration, angular velocity and attitude information of the vehicle body and capture its dynamic changes; the camera uses a high-definition industrial camera to collect image information of the working environment and assist in the identification of obstacles and working targets; the radar perception module uses millimeter-wave radar or lidar to scan the working space and acquire information on the distance, position and motion status of surrounding obstacles.

[0039] Explanation: All data collected by the sensors are transmitted to the control and decision-making layer via the bus. The central control unit calls the multi-source fusion algorithm to integrate, complement and optimize the data from multiple sensors, eliminate data redundancy and measurement noise, calculate vehicle stability parameters and environmental safety boundaries, and provide accurate, comprehensive and reliable data support for path planning, collaborative control and safety decision-making in the control and decision-making layer.

[0040] (3) Control and decision-making level

[0041] The control and decision-making layer is the core control unit of the system, including the central control unit, collaborative control unit, safety decision-making unit, operation planning unit, and communication interface. It is responsible for realizing core functions such as data processing, task allocation, path planning, collaborative control, and safety decision-making. Among them, the central control unit adopts a high-performance embedded processor, which has powerful data processing capabilities and real-time control capabilities. It receives data transmitted from the perception and detection layer, performs information fusion and task allocation, integrates and processes various types of data, and then issues control commands to the collaborative control unit, safety decision-making unit, and operation planning unit to coordinate the working rhythm of each module.

[0042] Explanation: The collaborative control unit, based on the collaborative trajectory generated by the operation planning unit and the mathematical model of the robotic arm, calculates the control quantity through a model predictive control algorithm and outputs it to the hydraulic drive unit at the working mechanism level. This controls the robotic arm to move precisely along the planned trajectory, achieving dynamic coordination between the lifting arm and the hoisting arm. The safety decision unit receives various status data in real time, including vehicle posture, load size, environmental parameters, and robotic arm position. Based on a preset safety assessment model, it evaluates the safety status in real time, calculates the safety margin, and triggers corresponding active protection actions according to the risk level, such as warning prompts, speed limits, trajectory adjustments, and emergency braking. The operation of the operation planning unit incorporates an improved RRT using deep reinforcement learning. The path planning algorithm generates a collision-free optimal cooperative expected trajectory based on the work instructions, environmental model, and robotic arm constraints, providing a target trajectory for cooperative control. The communication interface employs multiple communication methods such as CAN bus and Ethernet to achieve data transmission and interaction between modules within the control and decision-making layer, and between the control and decision-making layer and other levels, ensuring the coordinated operation of all parts of the system.

[0043] (4) Human-computer interaction layer

[0044] The human-machine interface layer serves as the window for information exchange between the system and the operator, including the operating terminal, display screen, audible and visual alarm devices, and remote monitoring interface. The operating terminal uses a touchscreen or button-operated panel, offering three operating modes: manual, semi-automatic, and automatic. In manual mode, the operator directly controls the robotic arm's movements using a joystick or buttons. In semi-automatic mode, the system automatically completes most of the posture adjustments and coordinated movements according to the task, requiring only simple operations such as target confirmation and parameter correction by the operator. In automatic mode, the system independently completes the entire operation process based on the planning results and requests manual intervention when an anomaly is detected.

[0045] Description: The operating terminal supports setting operation parameters, such as target location, maximum lifting load, and safety threshold, allowing operators to flexibly configure settings according to operational needs. The display screen uses a high-resolution industrial display, showing real-time key information such as robotic arm posture, load size, vehicle tilt angle, safety margin, operating mode, and fault information. The graphical and digital display method allows operators to intuitively and clearly understand the operational status. The audible and visual alarm device includes a buzzer and warning lights, emitting audible and visual warning signals when safety thresholds are approached, abnormal situations occur, or malfunctions occur, reminding operators to pay attention and take appropriate measures. The remote monitoring interface supports remote data transmission and monitoring, transmitting operational status data and video images to a remote monitoring center via a wireless network. Managers can monitor the operational status in real-time through the remote terminal, enabling remote management and scheduling, facilitating unified coordination of operational resources and timely handling of emergencies.

[0046] The system of this invention includes an operating mechanism layer, a sensing and detection layer, a control and decision-making layer, and a human-machine interaction layer. Each layer realizes data transmission and coordinated operation of lifting and hoisting actions through communication interfaces, constructing a complete collaborative and safe operation system and providing hardware and software support for the implementation of collaborative control methods.

[0047] Example 2

[0048] like Figure 1 and Figure 3 As shown, a substation work vehicle lifting and hoisting coordinated control method, applied to the control system of Embodiment 1, includes: S1. Detection and assessment of internal and external working environment: Collect information on the working environment and spatial data of obstacles, construct a three-dimensional scene model of the working environment, and assess the feasibility of the working environment; Specifically, upon system startup, various sensors in the perception and detection layer comprehensively collect information on the working environment and obstacle space, providing precise data support for subsequent path planning and collaborative control. Specifically, the radar perception module performs a 360-degree all-around scan of the working space, acquiring information on the location, shape, size, and distribution of surrounding equipment to form preliminary environmental contour data; the wind speed sensor measures wind speed data in real time, continuously monitoring the impact of wind loads on operational stability and providing wind field parameters for safety decisions; the tilt sensor detects the lateral and longitudinal tilt angles of the vehicle body, assessing the initial stability of the vehicle and determining whether the ground flatness meets operational requirements; load sensors are installed at the root and end of the lifting boom, collecting lifting load data in real time; pressure sensors monitor pressure changes in the hydraulic drive unit, reflecting the force state of the robotic arm; and the inertial measurement unit simultaneously collects vehicle attitude, acceleration, and angular velocity data, capturing dynamic changes in the vehicle body.

[0049] All collected data is transmitted to the control and decision-making layer via the communication interface. The central control unit calls a multi-source fusion algorithm to integrate, filter, and calibrate the data, eliminating abnormal data and improving data reliability. Based on the processed data, a high-precision 3D scene model of the working environment is constructed, clearly presenting key information such as obstacle distribution, vehicle position, and initial posture of the robotic arm within the working space. A comprehensive evaluation of factors such as wind speed, vehicle posture, obstacle distribution, and load size is conducted to determine the feasibility of the working environment, i.e., whether it meets the safe working conditions. If the evaluation results meet the requirements, the system automatically proceeds to the subsequent work process. If there are safety hazards, such as excessive wind speed, excessive vehicle tilt angle, or dense obstacle distribution that restricts the working space, an audible and visual warning signal is issued and work is prohibited until environmental conditions improve or corresponding safety measures are taken, at which point the feasibility of the working environment is reassessed. S2. Crane boom and lifting arm path planning: After receiving the operation command, the path planning is performed using an obstacle detection mechanism combined with an improved RRT based on deep reinforcement learning. The path planning algorithm plans the collision-free optimal cooperative expected trajectory of the crane boom and lifting arm from the starting point to the target point. The obstacle detection mechanism is as follows: 1) The obstacle and robotic arm are modeled using the bounding sphere method: ① Model spatial obstacles as spheres; the formula for a sphere is as follows:

[0050] In the formula, , , For the first The three-dimensional coordinates of the center of the sphere in the world coordinate system of the obstacle. The radius of the sphere; ② Model the crane arm and lifting arm as multiple cylindrical segments; the coordinates of any point on the robotic arm are as follows:

[0051] in,

[0052]

[0053] In the formula, , These are the coordinates of the left and right endpoints of the robotic arm's main arm; The value to be solved; 2) Conduct collision detection between obstacles and the robotic arm, as well as between the crane arm and the lifting arm: ① Substituting the coordinates of any point on the robotic arm into the obstacle sphere formula, we obtain the collision detection equation:

[0054] In the formula, It is the sum of the radius of the obstacle and the radius of the cylinder; ② Solve for t in the collision detection equation and determine whether there is a collision risk between the robotic arm and the obstacle: When there is no solution to the collision detection equation, it is determined that there is no collision between the robotic arm and the obstacle; when there is a solution to the collision detection equation and the solution is within a reasonable range, it is determined that there is a collision risk between the robotic arm and the obstacle, and the path is adjusted until there is no collision. ③ Calculate the relationship between the distance between the centerlines of the crane arm and the lifting arm and the preset threshold to determine whether there is a collision risk between the crane arm and the lifting arm: when the distance is greater than or equal to the preset threshold, it is determined that there is no collision between the crane arm and the lifting arm; when the distance is less than the preset threshold, it is determined that there is a collision risk between the crane arm and the lifting arm, and the movement trajectory of the crane arm and the lifting arm is optimized until there is no collision. Improved RRT The path planning algorithm employs an improved RRT based on deep reinforcement learning. The path planning algorithm is as follows: 1) The workspace is defined using a three-dimensional coordinate range; the workspace must meet the following conditions:

[0055] In the formula, L is the length of the workspace, W is the width of the workspace, and H is the height of the workspace; 2) Initialization and Sampling Expansion: Initialize a search tree containing only the starting point, which is the initial position coordinates of the robotic arm; set a random sampling area centered on the starting point, and perform random sampling within the area to generate a random sampling point; traverse all nodes in the search tree, find the node closest to the sampling point, and extend the node towards the sampling point with a set step size based on the closest node to generate a new node; 3) Collision detection: Perform collision detection on the path edges between the new node and the nearest node to determine whether there is a risk of collision with an obstacle or another robotic arm: if a collision is detected, the new node is abandoned and resampled; if no collision is detected, proceed to the next step. 4) Local path optimization: including reselection and rerouting operations; Among them, ① Reselection operation: Set a certain search area near the new node, find all candidate parent nodes, calculate the path cost from the starting point through each candidate parent node to the new node, and select the candidate parent node with the smallest path cost as the parent node of the new node. ② Rewire operation: Check the existing nodes near the new node, calculate the path cost from the starting point through the new node to these existing nodes. If the path cost is less than the existing path cost, redirect the parent node of these existing nodes to the new node to optimize the search tree structure. 5) Iteration and convergence: Repeat the above sampling, collision detection, and local path optimization steps to continuously expand and optimize the search tree until the collision-free optimal cooperative expected trajectory from the starting point to the target point is found.

[0056] Note: This method combines obstacle detection mechanisms with an improved RRT based on deep reinforcement learning. The path planning algorithm plans the collision-free optimal cooperative expected trajectory of the crane arm and lifting arm, ensuring that they do not interfere with each other or collide with surrounding obstacles during their movement. The improved RRT path planning algorithm introduces deep reinforcement learning technology, which adaptively adjusts algorithm parameters such as the sampling area range and extension step size based on dynamically changing environmental information such as the complexity of the working environment and the distribution density of obstacles. This avoids the problem of low exploration efficiency caused by the traditional RRT algorithm relying on human experience to set fixed parameters, and improves the adaptability and efficiency of path planning.

[0057] S3. Coordinated Control of Crane Boom and Lifting Arm: Based on the generated collision-free optimal cooperative expected trajectory of the crane boom and lifting arm, the control quantity is obtained through the model predictive control algorithm to achieve precise trajectory tracking of the crane boom and lifting arm, ensuring that they move synchronously according to the planned trajectory and realizing dynamic coordination; as detailed below: 1) A mathematical model of the robotic arm is established based on the Euler-Lagrange equations. The model includes system state variables and control variables; the expressions are as follows:

[0058] These are actual state quantities, including key state parameters such as the robot arm's angle, angular velocity, extension length, and extension speed; The actual control quantity includes control signals from the hydraulic drive unit, etc. 2) Obtain the generated optimal cooperative expected trajectory between the crane boom and the lifting arm without collision; the expression for the expected trajectory is as follows:

[0059] In the formula, The desired state quantity is the sequence of state parameters corresponding to the cooperative trajectory generated by path planning. For desired control quantity; 3) Define tracking error; tracking error is the difference between the actual state quantity and the desired state quantity, expressed as follows:

[0060] In the formula, For tracking error; These are actual state quantities; For the desired state quantity; By performing a first-order Taylor approximation expansion on the actual state equations and subtracting the equations for the actual trajectory and the desired trajectory, the linear time-varying tracking error equation is obtained. The expression for the linear time-varying tracking error equation is as follows:

[0061] In the formula, The first derivative of the tracking error; , This is the system coefficient matrix; This is the difference between the actual control quantity and the desired control quantity. An approximate discretization error model is used to discretize the linear time-varying tracking error equation, transforming the continuous system into a discrete system; the expression is as follows:

[0062] In the formula, Representing the Each sampling time; For the first The difference between the actual state quantity and the expected state quantity at each sampling time; For the first The difference between the actual control quantity and the expected control quantity at each sampling time , The system coefficient matrix after discretization; 4) Design the objective function; the expression for the objective function is as follows:

[0063] In the formula, For prediction in the time domain; and These are the weight matrices for the state variables and the control variables, respectively. This is the transpose of the matrix; The expression for the control constraint is as follows:

[0064] In the formula, , These are the minimum and maximum values ​​of the control quantity, respectively; 5) Obtain a new state-space expression through state-space transformation;

[0065] in,

[0066] In the formula, and This represents the system coefficient matrix in the new state space. This refers to the current state of the system. To control the control increment within the time domain; 6) The objective function is transformed into a quadratic programming problem, and an adaptive dynamic programming algorithm is used to solve it to obtain the control increment sequence in the control time domain. The first control increment is taken and combined with the control quantity at the previous time step to calculate the control quantity at the current time step. 7) Repeat the above process to update the system status and control variables in real time, so as to achieve accurate trajectory tracking of the crane boom and lifting arm, and accurate tracking of the actual trajectory to the desired trajectory.

[0067] Explanation: The actual tracking error equation clearly describes the dynamic change law of the tracking error; since the linear time-varying tracking error equation is continuous, it cannot be directly used for model predictive controller design and needs to be discretized. Discretization transforms the continuous system into a discrete system, adapting to the discretization calculation requirements of the model predictive controller; by adjusting the element values ​​of the weight matrix, the state tracking accuracy and control quantity smoothness are balanced, enabling the system to track the desired trajectory quickly and smoothly; constraints on the control quantity are set, limiting the value boundaries of the control quantity to ensure that the control signal is within the physical tolerance range of the hydraulic drive unit, avoiding violent robotic arm movements or equipment damage due to excessive control quantity; the new state-space expression shows that the state quantity in the prediction time domain can be calculated from the current state quantity of the system and the control increment in the control time domain, which is the core implementation method of the prediction function of the model predictive control algorithm; S4. Coordinated operation of crane boom and lifting boom: Under the action of control, it moves to the work position according to the planned trajectory to complete the predetermined work. During the entire operation, the operation status is monitored in real time through the active safety protection mechanism. When a safety risk is detected, the corresponding protective action is executed. The active safety protection mechanism is multi-layered, including overturning warning and stability control, collision prediction and obstacle avoidance control, platform stability control, and multi-level safety response control. It monitors the operational status in real time and executes deceleration, amplitude limiting, trajectory adjustment, emergency braking, or locking actions based on changes in safety margin. The multi-level safety response control includes three safety levels: warning, restriction, and protection. The warning level is the lowest safety level: triggered when a minor safety risk is detected, the audible and visual alarm emits a yellow warning signal, and the display shows the corresponding warning information, allowing operators to adjust their operations accordingly. The restriction level is the medium safety level: triggered when the risk level increases, the system automatically limits the arm's movement speed and amplitude, prohibits high-risk actions, and issues a continuous yellow warning signal. The protection level is the highest safety level: triggered when a serious safety risk is detected, the system executes emergency braking or locking actions, stops all operations, the audible and visual alarm emits a strong red alarm signal, and the display shows a safety protection prompt. Operations can only resume after the operator has investigated and eliminated the safety risk and deactivated the protection status via the operating terminal. Specifically, under the control output of the model predictive control algorithm, the robotic arm moves along the planned collision-free optimal cooperative expected trajectory. During the movement, the perception and detection layer continuously collects information on the arm's posture, load changes, center of gravity position, and surrounding environment, and transmits it to the control and decision-making layer in real time. Based on this data, the safety decision-making unit evaluates the operational safety status in real time, dynamically calculates the overall vehicle safety margin, and executes corresponding protective actions in conjunction with the active safety protection mechanism.

[0068] When a risk of tipping over is detected, such as when the vehicle's center of gravity shifts and the safety margin falls below a preset threshold, the system automatically reduces the robotic arm's movement speed or limits its range of motion to minimize further shift of the center of gravity. If the safety margin continues to decrease, the hydraulic outriggers are activated for fine-tuning compensation, adjusting the outriggers' support height and force distribution to restore vehicle stability. When a collision is anticipated, such as when the distance between the robotic arm and an obstacle approaches a safety threshold, the system automatically adjusts its trajectory and replans a local path to avoid the obstacle. If the collision risk is extremely high and cannot be avoided through trajectory adjustment, the robotic arm's movement is paused, and an emergency alarm signal is issued. When platform vibration or swaying exceeds limits, the system adjusts the damping coefficient of the hydraulic drive unit through hydraulic damping control or fine-tunes the arm's posture through the collaborative control unit to suppress platform swaying and ensure the safety of operators.

[0069] After the robotic arm moves to the preset working position, the system issues a work-ready signal, and maintenance personnel can carry out equipment replacement, repair, and testing through the work platform. During the operation, the display screen of the human-machine interface layer displays key information such as the robotic arm's posture, load size, vehicle tilt angle, and safety margin in real time, allowing operators to intuitively understand the working status. When the safety threshold is approached or an abnormal situation occurs, the audible and visual alarm device will promptly issue an audible and visual warning signal to remind the operator to pay attention to the safety status. After the operation is completed, the operator issues a work-end command through the operating terminal, and the robotic arm smoothly returns to the initial position according to the planned path under collaborative control, thus ending the operation process.

[0070] The method of this invention collects information on the working environment and obstacles using various sensors, providing accurate data support for subsequent path planning and collaborative control; based on improved RRT The algorithm plans the collision-free optimal cooperative expected trajectory of the crane boom and lifting arm, ensuring that they do not interfere with each other or collide with surrounding obstacles during their movement; the model predictive control algorithm outputs control quantities to achieve precise trajectory tracking, ensuring that the two move synchronously according to the planned trajectory and achieve dynamic cooperation; combined with an active safety protection mechanism, the operation safety is guaranteed, and the cooperative operation is finally completed.

[0071] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the principles and essence of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for coordinated control of a substation service vehicle hoist lift, characterized by, Includes the following steps: S1. Detection and assessment of internal and external working environment: Collect information on the working environment and spatial data of obstacles, construct a three-dimensional scene model of the working environment, and assess the feasibility of the working environment; S2, crane boom and lifting arm path planning: after receiving the operation action instruction, the improved RRT based on deep reinforcement learning is adopted through the obstacle detection mechanism path planning algorithm, the collision-free optimal cooperative expected trajectory of the crane boom and the lifting arm from the starting point to the target point is planned; S3. Cooperative control of lifting boom and boom: Based on the generated collision-free optimal cooperative expected trajectory of lifting boom and boom, the control quantity is obtained through the output of model predictive control algorithm to achieve accurate trajectory tracking of lifting boom and boom. S4. Coordinated operation of crane boom and lifting arm: Under the action of control, the crane moves to the work position according to the planned trajectory to complete the predetermined work. During the entire operation, the operation status is monitored in real time through an active safety protection mechanism. When a safety risk is detected, the corresponding protective action is executed.

2. The substation service vehicle hoisting and lifting cooperative control method according to claim 1, characterized in that, The obstacle detection mechanism in step S2 is as follows: 1) The obstacle and robotic arm are modeled using the bounding sphere method: ① Model spatial obstacles as spheres; the formula for a sphere is as follows: wherein , , is the first spherical center three-dimensional coordinates of the is the spherical radius; ② Model the crane arm and lifting arm as multiple cylindrical segments; the coordinates of any point on the robotic arm are as follows: in, In the formula, , These are the coordinates of the left and right endpoints of the robotic arm's main arm; The value to be solved; 2) Conduct collision detection between obstacles and the robotic arm, as well as between the crane arm and the lifting arm: ① Substituting the coordinates of any point on the robotic arm into the obstacle sphere formula, we obtain the collision detection equation: In the formula, It is the sum of the radius of the obstacle and the radius of the cylinder; ② Solve for t in the collision detection equation and determine whether there is a collision risk between the robotic arm and the obstacle: When there is no solution to the collision detection equation, it is determined that there is no collision between the robotic arm and the obstacle; when there is a solution to the collision detection equation and the solution is within a reasonable range, it is determined that there is a collision risk between the robotic arm and the obstacle, and the path is adjusted until there is no collision. ③ Calculate the relationship between the distance between the centerlines of the crane arm and the lifting arm and a preset threshold to determine whether there is a collision risk between the crane arm and the lifting arm: when the distance is greater than or equal to the preset threshold, it is determined that there is no collision between the crane arm and the lifting arm; when the distance is less than the preset threshold, it is determined that there is a collision risk between the crane arm and the lifting arm, and the movement trajectory of the crane arm and the lifting arm is optimized until there is no collision.

3. The substation work vehicle lifting and coordinated control method according to claim 1, characterized in that, The improved RRT based on deep reinforcement learning is used in step S2. The path planning algorithm is as follows: 1) The workspace is defined using a three-dimensional coordinate range; the workspace must meet the following conditions: In the formula, L is the length of the workspace, W is the width of the workspace, and H is the height of the workspace; 2) Initialization and Sampling Expansion: Initialize a search tree containing only the starting point, which is the initial position coordinates of the robotic arm; set a random sampling area centered on the starting point, and perform random sampling within the area to generate a random sampling point; traverse all nodes in the search tree, find the node closest to the sampling point, and extend the node towards the sampling point with a set step size based on the closest node to generate a new node; 3) Collision detection: Perform collision detection on the path edges between the new node and the nearest node to determine whether there is a risk of collision with an obstacle or another robotic arm: if a collision is detected, the new node is abandoned and resampled; if no collision is detected, proceed to the next step. 4) Local path optimization: including reselection and rerouting operations; Among them, ① Reselection operation: Set a certain search area near the new node, find all candidate parent nodes, calculate the path cost from the starting point through each candidate parent node to the new node, and select the candidate parent node with the smallest path cost as the parent node of the new node. ② Rewire operation: Check the existing nodes near the new node, calculate the path cost from the starting point through the new node to these existing nodes. If the path cost is less than the existing path cost, redirect the parent node of these existing nodes to the new node to optimize the search tree structure. 5) Iteration and convergence: Repeat the above sampling, collision detection, and local path optimization steps to continuously expand and optimize the search tree until the collision-free optimal cooperative expected trajectory from the starting point to the target point is found.

4. The substation work vehicle lifting and coordinated control method according to claim 1, characterized in that, Step S3 includes the following: 1) A mathematical model of the robotic arm is established based on the Euler-Lagrange equations. The model includes system state variables and control variables; the expressions are as follows: In the formula, These are actual state quantities; This refers to the actual control quantity; 2) Obtain the generated optimal cooperative expected trajectory between the crane boom and the lifting arm without collision; the expression for the expected trajectory is as follows: In the formula, The desired state quantity is the sequence of state parameters corresponding to the cooperative trajectory generated by path planning. For desired control quantity; 3) Define tracking error; tracking error is the difference between the actual state quantity and the desired state quantity, expressed as follows: In the formula, For tracking error; These are actual state quantities; For the desired state quantity; By performing a first-order Taylor approximation expansion on the actual state equations and subtracting the equations for the actual trajectory and the desired trajectory, the linear time-varying tracking error equation is obtained. The expression for the linear time-varying tracking error equation is as follows: In the formula, The first derivative of the tracking error; , This is the system coefficient matrix; This is the difference between the actual control quantity and the desired control quantity. An approximate discretization error model is used to discretize the linear time-varying tracking error equation, transforming the continuous system into a discrete system; the expression is as follows: In the formula, Representing the Each sampling time; For the first The difference between the actual state quantity and the expected state quantity at each sampling time; For the first The difference between the actual control quantity and the expected control quantity at each sampling time , The system coefficient matrix after discretization; 4) Design the objective function and control constraints; the objective function expression is as follows: The control constraints are as follows: In the formula, For prediction in the time domain; and These are the weight matrices for the state variables and the control variables, respectively. By adjusting the element values ​​of the weight matrices, the accuracy of state tracking and the smoothness of control variables are balanced. Represents the transpose of a matrix; , These are the minimum and maximum values ​​of the control quantity, respectively; 5) Obtain a new state-space expression through state-space transformation; in, In the formula, and This represents the system coefficient matrix in the new state space. This refers to the current state of the system. To control the control increment within the time domain; 6) The objective function is transformed into a quadratic programming problem, and an adaptive dynamic programming algorithm is used to solve it to obtain the control increment sequence in the control time domain. The first control increment is taken and combined with the control quantity at the previous time step to calculate the control quantity at the current time step. 7) Repeat the above process to update the system status and control variables in real time, so as to achieve accurate trajectory tracking of the crane boom and lifting arm, and accurate tracking of the actual trajectory to the desired trajectory.

5. The substation work vehicle lifting and hoisting coordinated control method according to claim 1, characterized in that, The active safety protection mechanism in step S4 includes overturning warning and stability control, collision prediction and obstacle avoidance control, platform stability control and multi-level safety response control, real-time monitoring of the operation status, and execution of deceleration, amplitude limiting, trajectory adjustment, emergency braking or locking actions according to changes in safety margin.

6. The substation work vehicle lifting and hoisting coordinated control method according to claim 5, characterized in that, The multi-level safety response control includes three safety levels: warning, restriction, and protection. The warning level is the lowest safety level: triggered when a minor safety risk is detected, the audible and visual alarm emits a yellow warning signal, and the display shows the corresponding warning information, allowing operators to adjust their work accordingly. The restriction level is the medium safety level: triggered when the risk level increases, the system automatically limits the arm's movement speed and amplitude, prohibits high-risk actions, and emits a continuous yellow warning signal. The protection level is the highest safety level: triggered when a serious safety risk is detected, the system performs an emergency braking or locking action, stopping all work actions, the audible and visual alarm emits a strong red alarm signal, and the display shows a safety protection prompt. Work can only resume after the operator has investigated and eliminated the safety risk and deactivated the protection status via the operating terminal.

7. A coordinated control system for lifting and hoisting of a substation work vehicle, characterized in that, include: The working mechanism layer is used to perform hoisting and lifting operations; The perception and detection layer is used to collect operation-related data and perform multi-source fusion processing. The control and decision-making layer is used to implement path planning, collaborative control, and safety decisions. The human-computer interaction layer is used for selecting work modes, setting parameters, and exchanging work information.

8. A substation work vehicle lifting and hoisting coordinated control system according to claim 7, characterized in that, The working mechanism layer includes a lifting arm, a hoisting arm, a working platform, and a hydraulic drive unit. The lifting arm is used to lift heavy objects or equipment; the hoisting arm is used to carry workers for high-altitude maintenance; both the lifting arm and the hoisting arm are driven by a hydraulic unit and are equipped with degrees of freedom for angle, extension, and rotation.

9. A substation work vehicle lifting and hoisting coordinated control system according to claim 7, characterized in that, The sensing and detection layer includes load sensors, tilt sensors, angle encoders, pressure sensors, an inertial measurement unit, a camera, and a radar sensing module. Load sensors are installed at the base and end of the lifting boom to collect load data in real time by sensing changes in tension during the lifting process, reflecting the weight and stress state of the lifted object. Tilt sensors are installed at the four corners of the vehicle body, using high-precision measuring elements to collect lateral and longitudinal tilt information of the vehicle body, assessing the stability of the entire vehicle. Angle encoders are installed at the joints of the robotic arm to detect changes in joint angles in real time, providing basic data for attitude calculation. Pressure sensors are installed at key parts of the hydraulic drive unit to monitor the pressure data of the hydraulic drive unit, reflecting its working status. The inertial measurement unit is installed in the middle of the vehicle body, integrating accelerometers and gyroscopes to acquire the vehicle body's acceleration, angular velocity, and attitude information, capturing dynamic changes in the vehicle body. The camera uses a high-definition industrial camera to collect image information of the working environment, assisting in the identification of obstacles and work targets. The radar sensing module uses millimeter-wave radar or lidar to scan the work space and obtain information on the distance, position, and motion status of surrounding obstacles.

10. A substation work vehicle lifting and hoisting coordinated control system according to claim 7, characterized in that, The control and decision-making layer includes a central control unit, a collaborative control unit, a safety decision-making unit, a work planning unit, and a communication interface. The central control unit is responsible for information fusion and task allocation; the collaborative control unit executes collaborative trajectory planning and outputs control signals; the safety decision-making unit assesses the safety status in real time and triggers active protection actions; and the work planning unit runs an improved RRT incorporating deep reinforcement learning. The path planning algorithm; the human-machine interface layer includes an operating terminal, a display screen, an audible and visual alarm device, and a remote monitoring interface; the operating terminal supports manual, semi-automatic, and automatic operation modes and parameter settings; the display screen shows the attitude, load, tilt angle, and safety margin in real time; the audible and visual alarm module issues a warning when the safety threshold is approached; the remote monitoring interface supports remote data transmission and operation monitoring.

Images