Research Of Hydraulic Pressure Servo System On Variable Vector Propeller

Variable Vector Propeller of ocean submarine, as a new kind impellent equipment, is still at the stage of R&D and has a very large potential application. This paper proposes three freedom degrees parallel platform, which has four hydraulic pressure cylinders, studies direct and inverse solutions, designs its hydraulic pressure servo system and gives the simulation results by using NN control theory.Three freedom degrees parallel platform is made up of left fixed platform and right unfixed platform. The two platforms are connected by four hydraulic pressure cylinders. The right platform connects with screw propeller of variable vector propeller. The position of right part changes by the movement of four hydraulic pressure cylinders make screw propeller form each kind angle. It produces variable thrust, pushes the variable vector propeller to omnidirectional directions.We can acquire the platform position by direct solution and the distance of hydraulic pressure cylinders by inverse solution. By means of the error analysis of direct and inverse solutions, we can strengthen the control on platform to make it accurately and rapidly in real environment.The hydraulic pressure servo system is easier to be stepless speed regulating than other kinds of servo system. It has steady movement, high precision, low position error and is easier to achieve movement in line to-and-fro. It is better than electric servo system in celerity. Besides it is smaller, ethereal, compact framework and it accelerates quickly. All these excellence make it widely used.PID control has many advantages such as intuitionistic, easy to tune and good robust. But, in the real life, working conditions are often very restricted and it's hard to tune the PID parameters to optimum state. And due to the time varying and non-linearity of control object, adaptive control also can not achieve good performance. But Neural Network can approach nonlinearity under given conditions, so combines Neural Network methods and PID control and generates Neural Network PID.