Performance of global optimisation algorithm EVOP for non-linear non-differentiable constrained objective functions
Sayeed Nurul Ghani · 2002
The applicability and performance of a new optimisation algorithm has been presented. Few current optimisation algorithms cope with real world problems involving discontinuous objective and constraining functions where there is a mix of continuous, discrete and integer arguments and the global minimum is sought. For noisy data, solutions are possible with genetic algorithms but costly parallel processing would be needed to locate the global minimum. Solutions remain illusive with genetic algorithms for problems with hard real-time constraints. The author's robust algorithm EVOP surmounts these difficulties with a much faster and more accurate solution. No gradient information is needed, and there is a high probability of locating the global minimum with a small number of automatic restarts as specified by the user.