ABSTRACT
Application reliability is a critical aspect of software quality. To achieve improved reliability, software development must include a validation stage where problems can be identified and rectified. The device reliability growth model (SRGM) was converted from simply modelling the defect identification mechanism (FDP) to integrating the fault rectification process (FCP). However, intricate variables such as failure reliance and the impact of people levels, which are constrained by mathematical monitoring, are difficult to be incorporated into computational models. This limits the application of empirical models. As a result, data-driven techniques such as the artificial neural network (ANN) hold promise for modelling FDP and FCP because no assumptions are required. Based on the ANN, a step-by-step model for the FDP and FCP is proposed in this study. Our approach incorporates the test initiative since it has a significant impact on the mistake detection and correction process. The performance of several types of neural networks is compared to an analytical model using real data. The efficacy of the proposed models was confirmed by scientific investigation. In addition, to clarify the applications, the optimum release time policy is frequently mentioned.
