抽象的な

Study for annealing and genetic algorithm in automatic process of data collection in computer software test run

Rui Ping Wang, Sun Gao Fei Sun

The concept of simulated annealing algorithm is derived from the organic integration of optimization and thermal equilibrium in statistical mechanics. This algorithm simulates the process in thermal equilibrium of natural cooling after being heated. In the process of finding the optimal solution, this algorithm shows a big advantage as it regards to be a technical approach to find the optimal solution. In this algorithm, the extremes should be considered as a function of dynamic equation. Although simulated annealing algorithm can avoid focusing on the local optimal solution, there are still some drawbacks, such as large amount of calculation and poor efficiency. Genetic algorithm (GA) simulates the natural law in biology: survival of fittest. It runs a “global optimized” algorithm, which appeared several years earlier than simulated annealing algorithm. Actually, GA is a set of arithmetic algorithm used in a group. It is necessary, at beginning, to choose a group of original population. Then, by crossover and mutation, some new population will be produced. This process goes on generation by generation, and always chooses the optimized ones to survive. As a result, the global optimized solution will be worked out. In this research, simulated annealing algorithm and genetic algorithm are combined to develop their biggest advantages to obtain the optimal solution. It is proved by test that this method is more suitable for seeking the optimal solution, especially, in automatic process of data collection in computer software test run. Its main advantages are high accuracy, convergence speed and high practical value, etc.