HI, Mr. Singh, If you go for optimisation, you need to first prepare the ranges for the influencing parameters. GA, ANN are not optimizing techniques to my knowledge. They are tools for mathematical modelling which are done after experimenting. U can use response surface methodology. that is the optimization technique.
GA and SA can help you locate the global optimum. There is no exact way to say which is more suited for a particular problem. However, the advantage with GA is that the search proceeds in multiple directions and can be easily parallelized.
Generally, it is common to couple GA/SA with a local optimum solver such as conjugate gradient method once you have narrowed the solution domain to a narrow space.