Study on deviation correction and target positioning of intelligent operation robot for high-voltage switchgear

Yan, Haipeng

DOI: https://doi.org/10.1007/s40430-024-05254-6

IF: 2.361

2024-11-05

Journal of the Brazilian Society of Mechanical Sciences and Engineering

Abstract:High-voltage switch control cabinet plays a key role in power generation, transmission, distribution, and energy conversion in the power system, including on/off, control, or protection of power supply. During their operation, partial discharge may occur. Therefore, it is necessary to develop a high precision positioning robot to reduce personnel risks. This study focuses on the deviation of AGV (Automated Guided Vehicle, AGV) path guidance, the accuracy of the robotic arm to positioning electrical components, and the detection of contact between robotic arm and target, which designs an AGV path correction controller, proposes a target positioning method based on residual BP neural network, and a collision detection method based on friction disturbance compensation. Firstly, the kinematics and dynamics of the diagonal steering wheel robot system are analyzed. Secondly, based on the kinematic and dynamic models of the AGV, the first motion fuzzy controller is designed with the AGV position and angle deviation as input and the AGV steering angle as output. The variable domain scaling factor is inferred through speed grading, and the second motion fuzzy controller is designed. A variable domain fuzzy controller based on speed grading is developed by combining the two fuzzy controllers. Then, the dimensional features of the target are fused with the camera's imaging model to derive the world coordinate system and pixel coordinates. By utilizing the advantage of residual networks to avoid gradient disappearance and reducing the workload of a single neural network, three residual BP neural networks are used to map coordinate transformation functions and obtain the three-dimensional coordinate information of the target. Finally, considering friction force as a disturbance term in the joint motion of the robotic arm, the joint friction force is estimated by a second-order disturbance observer, and an external torque observer based on friction disturbance compensation was established for collision detection. The results show that the speed grading variable domain fuzzy controller can achieve rapid and smooth path deviation correction across various speeds of AGV, effectively. The residual BP neural network has an average positioning error of 0.629 mm. The collision detection method of the robotic arm with friction disturbance compensation demonstrates high accuracy in detection. This study can effectively improve the safety and accuracy of operation for high-voltage switch control cabinets.

engineering, mechanical