A Trajectory Tracking Control Algorithm for Pneumatic Muscle Drive System

Abstract

The invention provides a trajectory tracking control algorithm for a pneumatic muscle drive system, the algorithm comprising: inputting the expected trajectory of the pneumatic muscle drive system; determining the tracking error according to the actual trajectory of the pneumatic muscle drive system fed back; The tracking error derivative is obtained from the error; the control signal is obtained by combining the tracking error and the tracking error derivative according to the layered fusion deep fuzzy neural network; the control signal is transmitted to the pneumatically driven joint; the pneumatically driven joint is based on the control signal Do trajectory tracking. The present invention effectively solves the problems existing in the prior art scheme through the layered fusion deep fuzzy neural network controller without pre-training, and realizes the trajectory tracking control of the pneumatic muscle driving system.

Description

technical field [0001] The invention relates to the technical field of trajectory tracking control of a pneumatic driver, in particular to a trajectory tracking control algorithm for a pneumatic muscle drive system. Background technique [0002] The pneumatic muscle drive system controls the joints driven by the pneumatic muscles, and the exoskeleton robot moves stably and accurately according to the desired trajectory. The hysteresis and nonlinear characteristics of the pneumatic drive system have brought great challenges to the control of this type of pneumatic drive system. [0003] For pneumatically driven robot systems, the existing control methods can be divided into two categories: model-free control algorithms and model-based control algorithms. Model-free control algorithms are represented by PID control and fuzzy control. Such algorithms do not need to obtain the dynamic model of the controlled object, and can be designed and put into use based on existing knowled...

AI Technical Summary

Problems solved by technology

[0005] However, the methods of literature [1] and literature [2] are based on the precise dynamic model of aerodynamic muscles, and the model-based T-S fuzzy control in literature [1] is more precise and robust than the model-free fuzzy control. The design of the sliding mode controller and the disturbance observer in [2] requires an accurate dynamic model of the aerodynamic muscle; while the control algorithms based on neural networks and deep networks in [3] and [4] , extracting the data of the controlled object and pre-training the neural network are all indispensable steps before this type of controller is put into use, which brings great inconvenience to the practical application of the controller

Images