The Improvement Of Gene Expression Programming And Its Application In Software Reliability Modeling Research

With the popularity of software in the social from all walks of life, the software quality has attracted more and more attention. Software reliability is one of the important evaluation features to the software quality, so the study of software reliability model can guide the software development process and predict software reliability in the future. At present there have been hundreds of software reliability models, but the application field of these models is limited, and the universality is poor. Gene expression programming(GEP) is a new machine learning technique, it’s structure is special which is different from the traditional evolutionary algorithm, and the ability in data mining is high. GA has difficulty in complicated problems, while GP owns a low efficiency. But GEP can overcome these shortcomings. GEP is suitable for the processing of highly nonlinear systems, and now has become a research hot spot in the field of intelligent computing.This article describes the mechanism of the GEP algorithm, in view of the insufficiency of standard GEP, this paper puts forward an improved gene expression programming algorithm based on hybrid strategy(HSI-GEP). Firstly, HSI-GEP using mirror and reset mechanism to replace the worst individuals of population, to improve the quality and the diversity of population; Secondly, HSI-GEP introducing the clonal selection before tournament selection operator in order to improve the mining ability of algorithm about high qualities. Finally, in order to verify the effectiveness and superiority of HSI-GEP, a large number of experiments have been carried on the function finding problems, and the results show that the algorithm is of high quality, has fast convergence rate and obvious competitiveness compared with the improved GEP in authoritative literature.Taking three groups of classic software failure sequences as experimental samples, the paper applies the standard GEP and HSI-GEP algorithm in the software failure sequence mining and the software reliability models are established respectively. Experimental results show that software reliability model based on these two algorithms can dig up the function relation between software failure sequences in the absence of prior knowledge, improve the poor universality and low accuracy problem of existing models to a certain extent, and have a great practical significance in the study of software reliability modeling. Moreover, the software reliability model based on HSI-GEP algorithm has better fitting degree and higher precision compared with the classical software reliability model.Finally, develop a software reliability modeling prototype system by using Java and Matlab mixed programming technology. It brings great convenience to the algorithm research in the paper, and has a good performance and user experience.