Tuning of Fuzzy Logic Controller by Genetic Algorithm—In case of concentration controls in resources processing

摘要

Mineral processing plants show a lot of difficultiesas control objects. As they involve many variables, nonlinearityand large disturbances, they can hardly be mathematically modeled,which is required for application of deterministic control theories.It is therefore hard to control those plants by the conventional,especially linear control algorithms.The authors have therefore conducted a series of studies onfeasibility of the fuzzy logic control in mineral processing. As aresult, the fuzzy logic control yielded sufficient controlperformances. Usually a trial and error method, however, musthave been practiced in fuzzy logic control design to optimize thecontroller parameters, such as scaling factors, fuzzy rule tables and so on.The authors intended in this paper to develop a way to automatea part of the above procedure by applying the genetic algorithm tooptimize the scaling factors. The genetic algorithm is a uniqueoptimum-seeking method that comes from an evolution process,and has recently been highlighted in different engineering areas. Itis the algorithm to search the optimal value by mapping the searchspace structure into the dynamics of the evolution of the living beings.The target control system for this study is concentration controlsystem based on actual resources processing plant. From theresults of the computer simulations, it has been demonstrated thatthis system can determine the fuzzy controller scaling factorsautomatically, and can be an effective method for a control systemdesign of mineral processing plants.