Study Of Evolutionary Algorithm For Software Project Scheduling And Application

Evolutionary Algorithms(Evolutionary Algorithms, EAs) are a class of optimization algorithms which apply the basic principle of evolution such as natural selection and genetics.The classical EAs include genetic algorithms, evolutionary programming, evolutionary strategies and genetic programming. Since EAs have the ability of high robustness and self-adapting, they can be widely used in function optimization, pattern recognition, machine learning, intelligent control, production scheduling, etc. The project scheduling problem(PSP)of software development process is receiving more and more attention. A main task in PSP is to minimize the project duration and cost with proper approaches. EAs have become an efficient way to solve the optimization problem in PSP.The paper is studied on the PSP in the software engineering from the performance improvement of the algorithm, the improvement of the task scheduling model and the application of the method.Aiming at the premature convergence problem in genetic algorithm, the selection operator,crossover operator, mutation operator are modified in order to improve the performance and LNGA algorithm is therefore proposed. According to PSP in the software engineering, the optimization solutions with normalization, the limit of the workload, and the method of matching the skills of employees, the skills level and the rules of learning are added in the scheduling model, in order to enrich the scheduling model, which is termed as LNSTSM model. Finally, the LNGA algorithm with the LNSTSM model is verify by comparing with other algorithms based on the experimental results.Further study is carried out on the genetic algorithm and the software project scheduling problem, ICLIAGA algorithm is proposed for improving the performance of the adaptive genetic algorithm. The task scheduling model is further improved and therefore the ICLSTSM model is proposed. In the ICLSTSM model, the recruitment intern system is firstly introduced,by verifying the formula, and the skill level and learning progress turn out to be continuous,so that the project scheduling model is more close to the reality. Finally ICLIAGA algorithm is applied to ICLSTSM model. Through the experiments, ICLIAGA combined with the more practical ICLSTSM model shows better performance.Finally, the paper is summarized and the prospect on the scheduling model and the optimization problem are drawn.