Research Of Software Project Resource Scheduling Based On CGA

Resource constrained project scheduling problem (RCPSP) is a class of combinatorial optimization problem. There are many methods to solve the problem at home and abroad, and mainly are intelligent optimization algorithms. But there haven't a algorithm with good convergence, and the problem model often have a big gap with reality. For these problems, this thesis builds a scheduling model which is more realistic, and tries to solve the problem model by cloud-based genetic algorithm.This thesis first summarized current research status of resource-constrained project scheduling problem at home and abroad, and compared several methods which to solve the problem, also generalized the research status of GA and the application of cloud model on GA. Then, build a software project scheduling model based on time and cost, this model used the skill matrix, so that the person who are not fully meet the skill constraints can also be assigned tasks, next selected the cloud-based GA and adaptive genetic algorithm to solve the model, and used serial scheduling algorithm for decoding. Finally, random initialized the population and completed the program to iterate the population in Matlab, at last output the final scheduling plan and the cost, also compared the performance of two algorithms.The experiment result show that the model meets the needs of practical applications, and the cloud-based genetic algorithm to solve the model is effective, this algorithm is better than adaptive genetic algorithm on convergence speed and precision. The research result has a certain advising significance for human resource scheduling of software project.