Use of Monte Carlo method in planning expensive experiments

Determining the optimal operating conditions of newly designed engineering complexes (e.g. nuclear power plants) or technological processes is the aim of numerous applied scientific investigations. The basic difficulty arising in the search for optimal values of the controlling parameters is the lack of a precise mathematical model which would determine the relation between the controlling parameters at the disposal of the designer and the quality criteria. Usually, on the basis of a knowledge of the essence of the physical phenomena occurring in the complex or process, it is possible to determine the structure of the generalized relation between the quality criteria and the controlling parameters. However, within the framework of the known structure, there are parameters whose values are not accurately known to the designer. At the same time, these parameters significantly influence the choice of optimal conditions in the part being designed. The numerical method proposed here for constructing the experiment plan is intended for the construction of a mathematical model of a technological process with the aim of 'adjusting' it in optimal conditions. The distinctive feature of the method is the use of the information value parameter as the quality index of the experiment plan. The advantage of the method is the possibility of determining the optimal parameters of the technological process with a small number of experiments, which is especially important when the experiments are expensive