|Controller Design Based on Wavelet Neural Adaptive Proportional Plus Conventional Integral-derivative for Bilateral Teleoperation Systems with Time-varying Parameters
Soheil Ganjefar, Mohammad Afshar, Mohammad Hadi Sarajchi, and Zhufeng Shao*
International Journal of Control, Automation, and Systems, vol. 16, no. 5, pp.2405-2420, 2018
Abstract : "In this study, a new controller method based on wavelet neural adaptive proportional plus conventional
integral-derivative (WNAP+ID) controller through adaptive learning rates (ALRs) for the Internet-based bilateral
teleoperation system is developed. The PID controller design suffers from dealing with a plant with an intricate dynamic
model. To make an adaptive essence for PID controller, this study uses a trained offline self-recurrent wavelet
neural network as a processing unit (SRWNN-PU) in parallel with conventional PID controller. The SRWNN-PU
parameters are updated online using an SRWNN-identifier (SRWNNI) in order to reduce the controller error in realtime
function. Using feedback linearization method and a PID controller, the presented control method reduced the
tracking error in the subsystems of the teleoperation system, i.e., master and slave which are stabilized, respectively.
Additionally, time-varying delay in teleoperation systems is considered as noise making the master signals
be modulated because wavelt neural networks have a high susceptibility to remove the noise, thus the WNAP+ID
controller is able to eliminate the noise effect. In this paper, we concentrated on the efficiency and stability of
the teleoperation system with time-varying parameters through simulation outcomes. Moreover, the results of the
WNNs are compared with those of multi-layer perceptron neural networks (MLPNNs)."
Adaptive PID control, bilateral teleoperation systems, time-varying delay, wavelet neural network.