Study On Automatic Leveling Control System Of Vehicle Platform Based On Neural Network

Vehicle platform leveling control system is now widely used in engineeringand military arena. With the ever-growing demands of platform level precision andleveling speed, the research of platform leveling system has been in progress. Thereare two major factors that will impact the performance of leveling system: one istracking performance of the platform support leg, the other is the platform structureand leveling method. This paper mainly makes study from these two aspects aimingto improve the speed and accuracy of leveling system.Firstly, this paper introduces and analyzes the current development of levelingsystem, and it establishes the platform support structure and driving mode bycomparing with different kinds of platform structures. We know the leveling systemhas coupling characteristic through analyzing and studying the platform position,and through analyzing comparing several different leveling methods we finallydetermine to take the chasing the highest point leveling method.Secondly, we establish the static analysis model in this issue to get therelationship between horizontal angle and support legs stress, and we design thesupporting system of platform, including the design of permanent magnetsynchronous motor servo system and transmission system, and on that basis theleveling system model is set up. We use classic PID control to track, however, thesimulation results show that the system contains coupling elements and systemresponse is slow. According to the characteristics of coupling and multipleparameters, we propose a program that use PID neural network to control theleveling system, According to the simulation results, PID neural network controllerhas better decoupling capability than classic PID controller, and the leveling speedimproves significantly.Finally, in order to further improve the comprehensive performance of theleveling system, an improved PSO algorithm is proposed to optimize the initialweigh of PID neural network, from the simulation results we know the optimizedPID neural network controller has better control effect, and the feasibility of themethod given is validated with significant improvements in accuracy and speed ofthe leveling system.