Software Reliability Growth Model Based on FABP
Lu Liu · Jisuanji fangzhen · 2015
Prediction accuracy of software reliability prediction model( SRPM) based on a single neural network is not high and this SRPM has low adaptability,and network structure of SRPM based on advanced neural network is too complex. In order to improve the adaptability of SRPM and reduce the neural network structure in the case of high prediction accuracy,a new prediction model is proposed. The weights and thresholds of the BP neural network are optimized by Firefly Algorithm( FA) in the training process of BP neural network by software defect data. At the same time,in order to reduce the fluctuation of prediction by BP network,averaging method is used to deal with predicted results. Based on those,software reliability prediction model is established by FABP. According to 3 groups of software defected data,and the mean value of error and sum of squared errors are taken as measurement to compare prediction performance. Simulation results show that the SRPM based on FABP which has a relatively simple network structure can improve the prediction accuracy and adaptability.