Genetic Algorithm Improvements Used In Test Case Generation

With the rapid development of the software industry, people are increasingly demanding of software quality, how to assess and ensure the quality of software becoming an urgent problem.so software testing arising. The traditional software testing is basically done by hand. When the size of today's software increases, it becomes a difficulty and huge amount of work. How to generate test cases and reduce test cost becomes the key to software testing.View of the huge workload and human error and other factors, people start to explorer automated software testing methods, the most importtent thing of it is automated test cases generation, it becomes a hot spot in this research now. Genetic algorithm,as a random search algorithm simulates natural evolution, it has implicit parallelism inherent in the global optimization and a good ability to adaptively adjust the search direction and guidance to optimize the search space. So it has been used in the automatic generation of test cases. And some inherent drawbacks of genetic algorithms, such as slow convergence, premature convergence is easy to form, full search capability is weak, and vulnerable parameters. The fitness function and genetic operators are the key to the efficiency of test case generation. Therefore, how to improve the design on the fitness function and genetic operators,and combine it with the test case generation bocomes the key to automated test case generation.View Genetic algorithm's shortcomings to generate test cases, the paper proposed a weighted fitness function transformation which based on branch function., and then design an adaptive mutation operator, Self Crossover operator and the effective cross-points operator in to improve the global search capability.This paper focused on the line of how to use genetic algorithms to generate test cases.Firstly, ayalyzes the shortcomings of the genetic algorithm's generating test cases, and ananalyzes the traditional test generation methods on test case generation, then the basic theory of genetic algorithms a detailed discussion of analysis of genetic algorithm and test case generation phase combination of theoretical basis. Then, proposed process and design methods of improved genetic algorithm (IGA). Finally, the improved algorithm (IGA) used in test case generation, and use the triangle program to ayalyze the effects of it. Experiment results show that, IGA compared to the traditional genetic algorithm, has a higher efficiency,and also hase a certain extent to avoid premature convergence of the algorithm.