Research On Key Technology For Passive Torque Servo System

Passive torque servo system(PTSS) is an indispensable important system developed in hardware-in-the-loop simulation of rudder system. It can simulate the torques of rudder control system loaded in real flight, to analyze and research the dynamic performance of rudder control system. PTSS has a wide range of applications in many areas such as aviation industry, shipbuilding industry and scientific experiments. With the rapid development of economy and the need of aircraft, the requirements of each performance index of rudder system are more and more high. Accordingly, accuracy and stability of PTSS are put forward higher requirements. Therefore the research on control strategy of PTSS is very necessary and meaningful.The key problems of PTSS control are followed: in the torque loading process, the loading motor runs passively with rudder. With the improvement of load frequency, the redundant torque is greater. It interferes with the normal loading torque. In a small torque loading process, the rudder may be loaded by friction torque and loading torque together, causing actual loading torque more than or less than loading instruction. After running for some time, vibration, wear and other reasons make backlash occur, and the system model parameters change. In short, redundant torque problem, mechanical nonlinear problems and uncertainties problem seriously hampered the improvement of system performance. In response to these problems, the research work is carried out.Firstly, the mathematical models of PTSS are established. Lu Gre model is applied to describe the friction torque. Identification of the model parameters were based on the least squares method and step response of second-order damping system. Dead-zone model is used to describe backlash and the measurement method of backlash is given. A comprehensive model of PTSS including rudder system, loading system, the backlash and friction is established. Theory derivation and simulation analysis of redundant torque are done to analyze the generation mechanism and related factors.Secondly, the controllers based on backstepping control were designed respectively to restrain the nonlinear aspects of PTSS loading at low frequency. For the friction torque makes the torque output of the system generate dead zone and unsmooth output phenomena, feed forward compensation method and backstepping adaptive control method with observer were used respectively. For backlash makes system response have the phase lag, low servo accuracy and poor stability, inverse model series compensation and backstepping control were used respectively. For system uncertainties problems, backstepping adaptive control based on RBF neural network was used.Thirdly, the methods based on neural networks control are used to solve the redundant torque problems and nonlinear problems of system loading at high frequency. For small torque loading, the friction torque is compensated by the torque produced from inner stator of double-stator motor according to Lu Gre model. The fuzzy neural networks PID controller adopts the algorithm of GA+BP. For big torque loading at high frequency, Dynamic Fuzzy Neural Network(DFNN) is used to identify the redundant torque. The redundant torque is compensated by the torque produced from inner stator according to redundant torque model identified. Single neuron PID controller which has simple structure and strong robustness is used. Its weights are adjusted according to the system error and Jacobian information given by RBF neural networks. For complex nonlinear system loading at high frequency, double DFNN are used for identification and compensation. When DFNN identification system is fully approximate inverse model, DFNN controller will eliminate the role of the PID controller to achieve the output torque tracking the load instruction.Finally, the experimental platform for PTSS is established and used to verify the control strategies. The results of these experiments show that the system indicators meet the design requirements and the problems are resolved well.