Optimisation enhancement using self‐organising fuzzy control

Abstract

Purpose

To show the successful use of self‐organising fuzzy control in enhancing dynamic optimisation, a controller is used to direct the type of optimisation appropriate in each new dynamic problem. The system uses its experiences to determine which approach is most suitable under varying circumstances.

Design/methodology/approach

A knowledge extraction tool is used to gain basic information about the solution space with a simple computation. This information is compared with the fuzzy rules stored in the system. These rules hold a collection of facts on previous successes and failures, which were acquired through the performance monitor. Using this system the controller directs the algorithms, deciphering the most appropriate strategy for the current problem.

Research limitations/implications

This procedure is designed for large scale dynamic optimisation problems, where a portion of the computational time is sacrificed to allow the controller to direct the best possible solution strategy. The results here are based on smaller scale systems, which illustrate the benefits of the technique.

Findings

The results highlight two significant aspects. From the comparison of the three algorithms without the use of the controller, a pattern can be seen in how the algorithms perform on different types of problems. Results show an improvement in the overall quality when the controller is employed.

Originality/value

This paper introduces a novel approach to the problem dynamic optimisation. It combines the control ability of self‐organising fuzzy logic with a range of optimisation techniques to obtain the best possible approach in any one situation.

Citation

Tighe, A., Smith, F.S. and Lyons, G. (2005), "Optimisation enhancement using self‐organising fuzzy control", Kybernetes, Vol. 34 No. 9/10, pp. 1440-1455. https://doi.org/10.1108/03684920510614759