An investigation into the application of evolutionary algorithms on highly constrained optimal control problems and the development of a graphical user interface for comprehensive algorithm control and monitoring

In this thesis we investigate how intelligent techniques, such as Evolutionary Algorithms, can be applied to finding solutions to discrete optimal control problems. Also, a detailed investigation is carried out into the design and development of a superior execution environment for Evolutionary Algorithms.

An overview of the basic processes of an Evolutionary Algorithm is given, as well as detailed descriptions for several genetic operators. Several additional operators that may be applied in conjunction with an Evolutionary Algorithm are also studied. These operators include several versions of the simplex method, as well as 3 distinct hill -climbers, each designed for a specific purpose. The hill -climbing routines have been designed for purposes that include local search, escaping local minima, and a hill -climbing routine designed for self -adaptation to a broad range of problems.

The mathematical programming formulation of discrete optimal control problems is used to generate a class of highly constrained problems. Techniques are developed to accurately and rapidly solve these problems, whilst satisfying the equality constraints to machine accuracy.

The improved execution environment for Evolutionary Algorithms proposes the use of a Graphical User Interface for data visualisation, algorithm control and monitoring, as well as a Client/Server network interface for connecting the GUI to remotely run algorithms.