Research On CMMI Oriented Software Development Quality Management Method

With the continuous expansion of the information technology industry,software has became an indispensable element in people's life,the computer is gradually replacing the manual operation and play an important role.The software used in all walks presents the characteristics of large scale,complexity,and internationalization.Functional requirements and non-functional requirements of software also enriched and changed with the development of society.However,the existing software project development situation is not optimistic.A large number of software projects have problems,such as project delay,project budget exceeding and project failure,the quality management level of software project is not optimistic.Unlike industrial production,software is difficult to analyze its quality because of its characteristics.How to analyze and manage the quality of software project has beacame an urgent problem.In this paper,the CMMI model is analyzed.Then,the quality of software project is quantized and analyzed.To improve the quality controllability of the software product,we analyze the quality management method from three dimensions: product,process and resource,so that the software project can deliver high quality software in time.The specific research includes:1.From the perspective of product,aiming at the complex and changeable features of software project environment,based on self learning,a real-time early warning model of software development quality is proposed.First collect the evaluation indexes in software projects,and then analyze the quality of each module.In order to solve the problem of the variability of the development environment,we iteratively update the classification knowledge of the model,realizeing the accurate real-time evaluation of the product quality of the software project.At the same time,the algorithm has been improved,the efficiency of the algorithm is improved,and the problem of noise and unbalanced data can be better handled,.The experimental result shows that the model can classificate accurately and could detect the quality of modules in time.2.From the perspective of software project process,we find out that log records in the existing software projects are too tedious.To improve the readability of the project development log,a parallel log clustering model of software project is proposed.The experimental results show that the efficiency of the algorithm has been improved,and the clustering analysis of large-scale data sets can be carried out.3.From the perspective of the software project resource,the problem of the unscientific allocation of software development resources is analyzed,A optimization configuration model of the software project resource based on the constraint configuration is proposed.First,it defines the activities,resources,and optimization goals of the software project,then analyzes and defines the constraints between the resources and the activities in the software project.After that,we proposesed an algorithm to solve the problem of optimal resource allocation scheme.Experiments show that the algorithm has certain advantages in solving the optimal solution,and it can better configure all resources in software projects.