A cooperative antagonistic control method of a rope-driven parallel antagonistic bionic joint

Abstract

This invention discloses a cooperative antagonistic control method for a rope-driven parallel antagonistic bionic joint, comprising: acquiring the system control torque and joint pose of the parallel mechanism bionic joint; grouping the acquired data to construct a dataset, and dividing the dataset into a training set and a validation set; simplifying the system dynamics model into a second-order nonlinear model, using the system control torque, joint pose, and their first derivatives as inputs to a neural network, and using the neural network to estimate the nonlinear mapping function and inertia matrix of the second-order nonlinear model to obtain the system model; using the validation set to verify the rationality of the training results; establishing error variables based on the established system model, designing a fast nonsingular terminal sliding surface, and discretizing the system model; designing the cost function of the model predictive controller; calculating the analytical solution of the control law of the cooperative controller based on the cost function of the model predictive controller; and designing an antagonistic compliant switching control strategy.

Images