Research On Optimization Of Methods Of Experimental Design And Implement Technology Based On Evolutionary Computation

The ideal and correct test results can be got fast through the scientific and rational design experiment which involves the less tests and cost in a short period during the process of designing innovative technology and developing experiment program. Experimental design greatly expands the scope of science as well as technology and receives the scientific experiment program to solute more and more problems which relied mainly on experience and test of people in the past. Focusing on the question of optimizational Experiment designing programs, the research on optimizational experiment designing methods based on evolutionary computation and implement technology is done in this paper through the analysis on the relationship of test target, effective element and test results to construct the mathematical models to turn the problem which seeks the optimal experimental scheme into the one that solving the optimizational function.Several common methods of experimental design and the theory & relevant knowledge of the orthogonal experimental design and uniform experimental design method are studied in this paper. With booming of the evolutionary computation algorithm, evolutionary computation has been able to solve the optimization problem well. Starting from the initial population, evolutionary algorithm makes the population evolution to better area of the search space by the processes of selecting, variation update and evolution restructuring. Because there is a certain gap between some commonly used results which designed by simple experiment method and the best ideal value, according to the related features of evolutionary computation, evolutionary computation with PSO and GA is adopted into optimize experiment design method to make the best test results which arrange scientifically according to experimental design method more close to the objective nature of the system itself by analyzing and optimizing the test results.In the reservoir numerical simulation, the test correction method is still widely used at home and abroad for history matching, which depends on repeatedly and artificially calculating to modify reservoir parameters according to the experience of calculators. This method is more subjective arbitrary and time-consuming, the optimization theory is introduced in this paper and the scientific experimental program is designed to establish the objective function, which transforms the reservoir numerical simulation history matching into solving the function extreme value problem defined in the Multidimensional space within nonlinear programming. It combines with the optimizational experiment methods based on evolutionary computation to find a group of conforming and practical dynamic parameters, which makes the calculated value and the actual measured value most closely to each other with same distribution rules in order to get an optimal history matching solution ultimately.