Defects Prediction of Aerospace Software Based on BP Neural Network

Aerospace software is required to possess the quality of high reliability and high security, because the software quality directly determines the quality of the satellite. Software defects may result in the failure of the satellite mission. Defects prediction of the aerospace software will help discover and modify software defects in the early stage, which makes test resources allocate more reasonably, and thus it can improve the reliability and security of software and reduce the costs of software development. This paper presents a defect prediction method on aerospace software based on BP neural network, adopting the static attributes of software to predict the number of potential defects, and the method turns out to achieve good results.