Control strategy and experimental research on girder hoister steering system based on RBF neural network PID

Order to meet the needs of construction of high-speed railway passenger dedicated line, 900t rubber-tyred girder hoister is researched and manufactured. Steering system of the hoister is controlled by electro-hydraulic proportional displacement control system, and segment-PID control strategy with dead zoon is used. However with the passage of time, steering performance of girder hoister begin to get worse and worse. Then, valve control cylinder model, equilibrium equation of the steering actuator and system model is built up. Through analyzing and referring advantage of the neural network, one kind of steering control strategy based on RBF neural network PID is presented. Based on similarity principle, this control strategy is used in the steering system of 100t ship yard transporter. The experiment has been lasted 3 months, and the result shows that the system settling time reduced by 5%, over shoot doesn´t change, and performance doesn´t get worse. Validity of the control strategy is validated.