Research on parallel control of CMAC and PD based on U model

Abstract

In this paper, the nonlinear U model with time-varying coefficients is investigated and the transformation of the nonlinear model is accomplished by the Newton iterative algorithm. Based on the nonlinear U model, a control algorithm with cerebellar model articulation controller and proportional derivative (PD) in parallel is proposed. The algorithm learns online through a neural network while optimizing the output of the PD, which ultimately enables the actual output of the system to track up to the desired output. Considering that the nonlinear object has the characteristic of rapid change with time, the article improves the PD algorithm to nonlinear PD control algorithm to complete the design of the system. The algorithm automatically adjusts the weights according to the error magnitude to complete the controller parameter adjustment, thus reducing the error of the system. The simulation results show that the nonlinear PD algorithm is better than the PD algorithm, meanwhile, the tracking speed and control precision of the system are improved.

Dual cerebella model neural networks based robust adaptive output feedback control for electromechanical actuator with anti‐parameter perturbation and anti‐disturbance

“…k 2 is the proportion gain of the controller, which can influence the tracking performance in the same way as k p influences the state estimation performance in the observer. 19 Therefore, it is necessary to select appropriate parameter values according to engineering experience through continuous debugging.…”

“…Increasing the differential gain could help to reduce overshoot, overcome oscillation and accelerate the response speed of the system, but the differential term often amplifies the system noise, which would lead to system instability. k 2 is the proportion gain of the controller, which can influence the tracking performance in the same way as $$$ {k}_p $$$ influences the state estimation performance in the observer 19 . Therefore, it is necessary to select appropriate parameter values according to engineering experience through continuous debugging. Remark In most industrial applications, only the position sensors are installed in the electromechanical actuators due to the limitation of cost and volume, and thus only position signal can be measured, while the velocity signal is usually obtained by differentiating the position signal which would amplify the noise 22 .…”

“…It has both stronger generalization ability and faster learning speed than many other neural networks, and it is suitable for approximating complex nonlinear functions. 18,19 Neural networks based output feedback control schemes have been deeply studied in theory and most methods are relatively complex and create large computation burden, which restrict its application in the real-time control filed. 20,21 Thus, most of the current research is in the simulation stage and there are few experimental studies on this method.…”

“…CMAC uses its unique addressing method to complete nonlinear spatial mapping. It has both stronger generalization ability and faster learning speed than many other neural networks, and it is suitable for approximating complex nonlinear functions 18,19 . Neural networks based output feedback control schemes have been deeply studied in theory and most methods are relatively complex and create large computation burden, which restrict its application in the real‐time control filed 20,21 .…”

Adaptive Single-Input Recurrent WCMAC-Based Supervisory Control for De-icing Robot Manipulator

“…Thanh Quyen Ngo, Adaptive Single-Input Recurrent WCMAC-Based Supervisory Control for De-icing Robot Manipulator control system. In [31]- [34], the proposed single-input CMAC controllers are not only solely used to control the plant, so the input space dimension can be simplified, and no conventional controller. However, the disadvantage of the proposed CMAC control system adopts two learning stages, an off-line learning stage, and an on-line learning stage.…”