A New Optimization Selection for Test Process of Equipment Manufacturing Based on Fused Algorithm

Abstract

Test optimization selection is a set cover problem, and heuristic algorithm for set covering problem is effective method. A genetic simulated annealing neural network fused algorithm was proposed by fusing the genetic algorithm, BP neural network and the simulated annealing algorithm, the genetic algorithm global search ability, strong ability of BP neural network training algorithm and fast search ability of simulated annealing algorithm were made full use of in this algorithm, the phenomenon falling into local optimum was avoided, and also the search efficiency and accuracy wad improved, the algorithm is applied to solve the test optimization selection problem. Example proves that this algorithm can effectively and quickly obtain test the optimal solution of optimization problems. Introduction In order to optimize the selection, the optimal combination test is chose in all possible combination of tests in the system, and the testability index can be met, while the minimum cost, include test time and test expense[1][2]. Perspective of mathematics, optimization selection is a test set covering problem, but we know set cover problem is a NP problem, when the system is on a big scale, to obtain the optimal solution is very difficult[3][4]. At present there are many more effective to solve the set covering problem of heuristic algorithm, such as genetic algorithm (GA) and simulated annealing algorithm (SA), neural network (ANN), tabu search (TS), ant colony system (ACO), etc., but these algorithms have their own advantages and defects, the effect is not ideal to solve alone, to complement each other mutual fusion has become the focus of future research[5][6][7]. This article is based on this idea, BP neural network and simulated annealing algorithm and introduced on the basis of traditional genetic algorithm, and a genetic simulated annealing neural network fusion algorithm has been formed, optimization selection can be solved faster and more accurate. Fusion algorithm Operation process of fusion algorithm The operation process of fusion algorithm is shown in Figure 1. Extreme value and dead zone analysis of algorithm In the process of solving the traditional heuristic algorithm, the maximum and minimum values appear easily. These extreme values are very close to the optimal value of algorithm, but the fusion algorithm is based on genetic algorithm, and the advantages of genetic algorithm is strong global search capability, which can avoid the occurrence of extreme value in the whole search space[8][9]. Therefore, the occurrence of extreme value can be avoided in the fusion algorithm can the occurrence of extreme value. Similarly, the genetic algorithm is good at global search, which can search to every corner of the space in the process of search, and the dead zone can be avoided in the fusion algorithm. Because of strong global search ability of the genetic algorithm, extreme value and dead zone can be avoided, but the disadvantage of slow speed is also caused in the genetic algorithm. Moreover, International Conference on Advances in Mechanical Engineering and Industrial Informatics (AMEII 2015) © 2015. The authors Published by Atlantis Press 402 The BP neural network and simulated annealing algorithm are introduced in the fusion algorithm, the search speed can be improved, and the optimal solution can solved quickly and effectively[10][11].