Structure and design method of a dynamic surface controller for a parallel single-stage double inverted pendulum

Abstract

The invention discloses a dynamic surface controller structure and design method of a parallel single-stage double inverted pendulum. The controller structure includes two controllers, and each controller includes two-stage sub-controllers. The first-stage sub-controller It is composed of a first comparator unit, a combination unit, an approximation unit and a nonlinear control unit, and the second-level sub-controller is composed of a filter unit, a second comparator unit, a combination unit, an approximation unit, a nonlinear control unit, and an auxiliary The system unit and the third comparator unit constitute. The invention solves the control problem of the parallel single-stage double inverted pendulum containing time-delay items; the introduction of the neural network approximator does not need to predict the uncertain dynamics of the complex system in advance and the relevant prior knowledge of the delay items; it also introduces the dynamic surface control technology, It overcomes the complex derivation problem of the intermediate control function of the backstepping method, and at the same time adjusts the norm of the weight matrix of the neural network instead of the weight itself, reducing the number of online adjustment parameters, thereby reducing the calculation load of the controller.

Description

technical field [0001] The invention relates to a dynamic surface controller structure and design method of a parallel single-stage double inverted pendulum, in particular to a parallel single-stage double inverted pendulum and this type of multi-input multi-output lower triangular uncertain time-delay nonlinearity with input saturation The system controller structure and design method belong to the technical field of automatic control. Background technique [0002] The parallel single-stage double inverted pendulum system connected by springs is a typical multi-input and multi-output lower triangular time-delay nonlinear system with uncertainty. The Lyapunov-Krasovskii functional is one of the tools for analyzing time-delay nonlinear systems. The adaptive controller constructed by combining the functional with backstepping and neural network approximation technology has been widely valued and developed in solving the control problems of complex lower triangular uncertain no...

AI Technical Summary

Problems solved by technology

[0003] Adaptive backstepping technology provides a powerful tool for the control of the spring-connected parallel single-stage double inverted pendulum system, but from the perspective of the controlled object's structural characteristics and controller design, there are the following problems: The parallel single-stage double inverted pendulum system has time-delay characteristics, and the controller designed based on the time-delay-free model has certain limitations in practical application; second, in engineering practice, the amplitude of the input generally has certain restrictions, Most of the existing controller designs do not consider the input saturation characteristics, it may happen that the control quantity is at the saturation upper / lower limit for a long time, and the actuator wears seriously; thirdly, there is a defect in the controller based on the backstepping method. That is, the "computational complexity" problem. The backstepping method needs to derive the virtual control law in the process of each step of recursion, resulting in the explosive growth of the number of controller items with the increase of the system order. At the same time, the number of neural network weights increase, the amount of data that needs to be adjusted is large, the calculation load increases, and the controller structure is complicated, which is not conducive to practical application

Images