Evolutionary learning of fuzzy logic controllers and their adaptation through perpetual evolution

Abstract

This paper presents an adaptive control architecture, where evolutionary learning is applied for initial learning and real-time tuning of a fuzzy logic controller. The initial learning phase involves identification of an artificial neural network model of the process and subsequent development of a fuzzy controller with parameters obtained via a genetic search. The neural network model is utilized for evaluating trial fuzzy controllers during the genetic search. The proposed adaptive mechanism is based on the concept of perpetual evolution, where parameters of the fuzzy controller are updated at each time step with solutions extracted from a continuously evolving population of trials. There are two mechanisms that accommodate the real-time changes in the control task and/or the process into the continuous genetic search: a scheme that dynamically modifies the fitness evaluation criteria of the genetic algorithm (GA) and the online learning of the neural network model used for evaluating trial controllers. The potential for using evolutionary learning for real-time adaptive control is illustrated through computer simulations, where the proposed technique is applied for a chemical process control problem.

Improvement of Transient Response of a DC Motor Controller Based on Non-Linear Optimization Techniques

“…The open loop and closed loop transfer function of the system with PID controller is given by (10) and 11  Mamdani-type and  Sugeno-type. These two types of inference systems vary somewhat in the way outputs are determined [10,11]. Mamdani-type inference requires finding the centroid of a two-dimensional shape by integrating across a continuously varying function.…”

“…Further, a PID controller was designed and tuned based on the conventional methods [10,14]. To achieve smoother control, a fuzzy logic controller with two inputs and one output including several rules was also designed [11]. All the four controllers were implemented in the simulation.…”

Intelligent Adaptive Dynamic Matrix Control

“…3.1 Discrete dynamic model structure Generally a nonlinear system with multiples inputs and multiple outputs (MIMO), with a vector input u(t)=[u 1 (t)...u n (t)] and an output vector y(t)=[y 1 (t)...y m (t)] T can be described by the following function (Rajapakse, et al, 2002):…”

Evolutionary Computation for on-Line and Off-Line Parameter Tuning of Evolving Fuzzy Neural Networks

“…[11][12][13][14][15][16][17] In the example references, online-EC are characterised by small population size (6-50) and high selective pressure (use of elitist selection and low selection ratio), which aims to minimise computational intensity and to accelerate convergence (at the cost of lesser exploration in the search space) respectively. Their applications involve optimising only a small number of parameters (2-10), which are often the high-level tuning parameters of a model.…”

“…The approach is similar to that in Ref. 16. Relative to the learning rates, MFs are static control parameters that remain constant over time and hence require less frequent updating.…”