Prediction and Comparative Analysis of Software Reliability Model Based on NHPP and Deep Learning

Over time, software has become increasingly important in various fields. If the current is more dependent than past and broken owing to large small issues, such as coding system errors, it expected cause significant damage entire industry. To address this problem, field of reliability crucial. In past, efforts were made develop models by assuming a nonhomogeneous Poisson-process model (NHPP); however, became complex, there many special cases which fit well. Hence, study proposes using deep learning that relies on data rather mathematical statistical assumptions. A based recurrent neural networks (RNN), long short-term memory (LSTM), gated units (GRU), are most basic networks, was constructed. The dataset divided into two, Datasets 1 2, both used 80% 90% data, respectively. Using 11 criteria, estimated learned results these datasets proved excellent capabilities. GRU showed satisfactory results.