Identification And Intelligent Control Research Of Electric Load Simulator For Gun Control System

In the modern warfare,there is a higher requirement for artilleries to take advantages of the maneuverability,rapidity,reliability and accuracy performance.The gun control system(GCS)with modern ammunition works as a core of the artillery,which plays a vital role in realizing the "First shoot,first hit".As the torque in the load side of the GCS is always changing when the artillery is istransferring or shooting,the disturbance torque palys negative effects in the GCS performance.The electric load simulator(ELS)can simulate the complex and time-variable torque of the load side of the GCS dynamically,and the GCS performance can be debugged and evaluated before the overall system is established.Moreover,the development and production cycle of the GCS can be reduced effectively.However,common control methods can't achieve great performance because of the complex nonlinearities existing in the ELS.Therefore,the model identification and control strategies are further studied to improve the torque tracking precision of the ELS,which also shows theoretical significance and engineering application values.The main research work of this dissertation focuses on the following aspects:(1)The structure feature and working principle of the ELS for the GCS are analyzed,and the mathematical expression of permanent magnet synchronous motor(PMSM)is established based on the vector control method.With studying the current loop,speed loop and position loop of the ELS for the GCS,the model of the position and torque motor of the ELS for the GCS is set up respectively.The existing uncertainty factors and negative effects are discussed,which can lay the theoretical foundation for the identification,control and semi-physical simulation.(2)Owing to complex nonlinearities,the exact model of the ELS can't be achieved.Thus,an identification method named as adaptive differential evolution wavelet neural network with variable structure(ADE-VSWNN)is proposed.Pseudo random multilevel and chrip signals are chosen as identification inputs,and the t-test is used to evaluate the significance of relevant performance indicators and identification precision,which proves the effectiveness and practicability of the proposed identification algorithm.In addition,the established model can be used as the simulation platform,which evaluates the effectiveness and application of relevant controllers.(3)On the basis of the fuzzy,sliding mode variable structure,particle swarm optimization(PSO)and WNN intelligent algorithm,the intrinsic advantages of proposed algorithms and ELS characteristics are taken into consideration comprehensively,which contribute the construction of the fuzzy multiresolution WNN with dynamic compensation(DCFMWNN)controller and double sliding modes structure learning of PSO WNN(2S-PSOWNN with SL)controller,and the stability of proposed controllers is analyzed in the sense of Lyapunov stability.Finally,simulation results of the convergence,step response and sinusoidal tracking show that the proposed controllers can satisfy the system controller index requirements.(4)The hardware component and software design of the ELS are introduced,and the semi-physical simulation platform is established.Then,the proposed DCFMWNN and 2S-PSOWNN with SL controllers are applied to the test of suppressing the surplus torque,variable gradient loading and robust performance.Experimental results indicate that two controllers can satisfy the requirement of the system performance index,and the 2S-PSOWNN with SL controller shows better than the DCFMWNN controller in the control precision and robustness.The position motor of the GCS is also made work in the step response,constant speed and sinusoidal tracking when the torque motor of the ELS is working in different motion styles,which meets the control index requirement and provides the function of guidance and reference for the practical engineering application.