The Memristor-Based Intelligent PID Controller With Applications

In the modern industrial control system, the conventional proportional-integral-derivative (PID) controller has advantages of simple architecture, strong practicability, convenient adjustment and wide application, etc. However, with the increasing complexity of the controlled objects and large numbers of the nonlinear, time-delay and time-variant systems, the conventional PID controller cannot obtain the ideal control results. The new technologies and new devices are needed to improve or overcome the defects of PID controller urgently, which expected to realize the precise, effective and real-time control. Memristor inherently possesses the passive, nonlinear, nonvolatile, synapse-like, high-density, highly integrated and other properties. Memristor is employed in PID control system for its synapse-like characteristic to replace the proportional, integral and derivative coefficients to construct the PID controller based on memristor. The new controller has quicker response speed, slighter overshoot, smaller error and more flexible and convenient application than that of the conventional ones. This thesis is based on theoretical research of memristor and the conventional PID controller to explore their combining point and analyze the operating principle and control effect of their effective combination. The feasibility and practicability of the controller are proved by the application in the motor system and the inverted pendulum system and other actual control systems. The main content of the thesis consists of the following four aspects:(1) The working principle and classifications of PID controller are analyzed in detail. The defects and advantages of PID controller are presented in this thesis. The working principle and the development situation of memristor are introduced, especially the flux-controlled memristor used in this thesis.(2) The working principle of single-neuron PID controller is introduced and the innovation is that the proportional, integral and derivative coefficients are replaced by memristor synapses in this thesis to construct the adaptive single-neuron PID controller based on memristor. The mathematical model and the structure diagram of the memristor-based PID controller are established, the numerical simulation experiments under the different learning rules are presented. On this basis, the application of memristive PID controller in the electric motor control is analyzed and the new-type controller has good fitting effect of input-output curve, small overshoot and fast convergence speed.(3) Based on the analysis of the structure and principle of adaptive single-neuron multivariate decoupling PID controller, which is used to realize the multivariate control. In order to reach the control requirements of time-variant, nonlinear, high-quality control and multivariate complex systems, the memristor is introduced into the control system and the adaptive single-neuron multivariate decoupling PID controller based on memristor is built up in this thesis. The numerical simulations show that the controller has fast learning speed, good curve fitting effect, small overshoot and the system is stable; the validity and practicability of the new-type controller are proved by the control researches of the pendulum angle of the inverted and the car’s moving trajectory.(4) According to the working principle and mathematical model of PID compound controller based on CMAC, the synapse weight is replaced with memristor for its synapse-like property in the CMAC network to construct the PID compound controller based on memristor CMAC. The comparison of the compound controllers between with and without memristor is made to demonstrate that the PID compound controller based on memristor possesses faster approximation speed, better approximation accuracy, better control effect and with the obvious advantages.The feasibility of replacing the proportional, integral and derivative coefficients of PID controller with memristors is proved by the theoretical analysis and numerical simulations. Not only to realize the combination of memristor and PID controller, but also to show the memristive PID controller has faster approximation speed and better approximation accuracy through comparing simulation results between it and conventional one. Meanwhile, the research conclusion provides a broad developing space for the further study and improvement of PID control system, meanwhile the possibility is given to the realizations of VLSI controller and microcontroller.