Kriging Auxiliary Agent Model For Solving The Problems Of Expensive Single Objective Constrained Optimization

The fields of product designing, robot tracing, energy planning, plastic molding,corporate finance, administrative decision-making, car collisions, wing design,investment allocation, urban planning, airflow analysis and so on, may be exist aproblem of single objective constrained optimization (The Problems of SingleObjective Constrained Optimization, TPOSOCO), which is very complex andexpensive. Optimization algorithm is directly applied to solve the expensiveTPOSOCO, which is difficult to obtain a satisfactory solution in a short time.Therefore, the Kriging auxiliary agent model (Kriging Auxiliary Agent Model,KAAM) optimization algorithm is put forward to solve expensive TPOSOCO in thisarticle. The KAAM optimization algorithm can not only ensure the accuracy of theoptimal solution, but also impoving efficiency of the expensive TPOSOCO. In thispaper, specific studies are as follows:(1) Kriging surrogate model is used to replace the original expensive objectivefunction and the expensive total constraint function. Adaptive Latin of HypercubeSampling (Latin of Hypercube Sampling, LHS), the insert guidelines of the bestindividual, the insert guidelines of the mean square error, the delete standards of therelevant points as well as repeated LHS optimization strategy, are used to improve thecalculation accuracy of KAAM.(2) This article combine Kriging surrogate model with the original expensiveobjective function, the total constraint function, genetic algorithm, zero constraints-non-dominated selection mechanism (Zero Constrained-Non-Dominated, ZCND)technology. The dynamic Kriging auxiliary agent model optimization algorithm is putforward to solve expensive TPOSOCO, and its can speed up the convergence ofexpensive TPOSOCO.(3) This article combine Kriging surrogate model with the original expensiveobjective function, the total constraint function, niche, ZCND. A layered mixedKAAM optimization algorithm is put forward to solve the expensive TPOSOCO, andits can speed up the convergence of expensive TPOSOCO. A layered mixed KAAMoptimization algorithm is verified that it has good sound effects, and the methodavoids the conditional be set up of the using auxiliary agent model.The KAAM optimization algorithm is put forward to solve the expensive TPOSOCO. The higher precision and better results are proved by experiments and thenumber of evaluations of the original expensive model is significant reduced.