Software Project Risk Assessment Based On BP Neural Networks

With the rapid development of computer technology and the rapid expansion of the number of software, but software technology has not been corresponding development. Its quality, cost and performance, etc. can not be synchronized with the hardware. It leads to software technology what is complicated, technical, and fast updates have a high risk. Currently, risk prediction software project is also becoming part of the range of management software.In this paper, firstly we give an outline of the present situation and the development of risk management software. Secondly, we summarize the software project risk knowledge, including risk implications, the concept of software projects, the risk of major software projects, software project risk management, some of the model, risk identification, risk assessment methods and the theory of BP neural network basis. The past, the connection weights and training samples of artificial neural network are random numbers. Two random processes will lead to lower training accuracy. In this paper, it is using positive and negative sequence training methods, and gives the proof of the solution is uniqueness. Avoid must be generated sequence randomly each time.On this basis, create a system of risk assess, the system of indicators is the result of risk factors derived from screening .Finally, building a software risk assessment models based on BP neural network. And using the sample data training and testing, which gets by experts ways. It is prove the model is feasibility.