Research And Application Of Automatic Test Case Generation Based On Improved Genetic Algorithm

Along with the continuous increase of software complexity and scale, test automation is an inevitable trend in the development of software testing. However, due to the diversity of test cases and the uncertainty of the test program, become the technical difficulties of the automatic generation of test cases. Most automated testing tools are mainly aimed at the implementation and management of the automatic generation of test, the test data is not perfect, when the program of large scale, hand written test cases of the workload is very big, structural test case coverage is very difficult, resulting in the testing efficiency is not high. The test case generation is an un-decidable problem. The general search algorithm is extremely limited.Standard genetic algorithm such as premature and other defects in the traditional test case is generated, in finding the optimal solution in the process often need to spend a higher price. Aiming at these problems, the improvement and optimization of genetic algorithm:in the individual adaptability evaluation standards, the fitness is greater than the population average fitness of individual fitness and average, get a new mean value are preferred and elimination of the individual, to maintain the diversity of population, and avoid the premature convergence phenomenon in the cross; when in operation, the crossover strategy and a dynamic control cross point position, expand the search space; in the mutation operator, the best individual preservation method to improve the genetic algorithm in local search ability and overall efficiency. The experimental data show that, the improved genetic algorithm compared with the traditional genetic algorithm has obvious advantages in efficiency and effectiveness of test case generationThe improved genetic algorithm is applied to the online evaluation system of Collegiate Programming Contest in the actual development, online evaluation system based on Web in C kernel based evaluation, the game management, compile online, online evaluation and performance ranking functions, test ability improves the online evaluation system, extends C evaluation of the function of the system, solves the design competition management undergraduate program of the school problems, provides a source program evaluation and exchange platform for the teachers and students. At present, the stable running of the system, program design competition has been successfully held the school.