Model-free Backstepping Control And Optimization Study Of Parameter For Nonlinear Mechanical Systems

Nonlinear mechanical systems commonly exist in modern industrial fields,such as aerospace,robotics,manufacturing systems,etc.The typical nonlinear characteristics,such as time-varying,time-delay and variable coupling,result in certain difficulties in system modeling and control design.Existing control methods have certain limitations in practical applications,such as model dependence,insufficient stability,and poor robustness.Designing a stable model-free nonlinear controller has practical significance for the control of nonlinear mechanical systems.This thesis studies a model-free backstepping control strategy from the perspective of data driven and control stability.On this basis,the control design seeks to compromise the multiple performance objectives.In the control parameter space,the multi-objective optimization problem is formulated.The outline of the thesis is as follows:(1)Combining traditional backstepping control and model estimation methods based on system measurable data,a model-free backstepping control strategy was proposed.An overall control was formed while the stability was proven by the Lyapunov theory.Finally,the equivalence analysis of the MFBS control with unknown control coefficients was carried out.(2)The experimental helicopter system with time-varying parameters was taken as the actual object to solve the control design problem of the time-varying and coupled nonlinear mechanical system.The advantages of tracking control were discussed through the study of numerical simulation and physical experiment.(3)Under the requirements of multiple control performance objectives of the closed-loop system,a multi-objective optimization study was performed on the model-free backstepping control parameters,to provide experimenters with multiple sets of feasible control parameters that meet the requirements of the working conditions.Based on the design of model-free backstepping control of rotary flexible link,multiple performance targets were designed,and controller parameters were optimized using a hybrid multi-objective optimization algorithm(Gene Algorithm-Simple Cell Mapping,GA-SCM).The effectiveness of the MOP design was verified through physical experiments.