Abstract
Network reliability is very useful decision support information. The squeeze response surface methodology (SqRSM) and artificial neural networks (ANNs) are two of the most useful types of optimal algorithms to estimate network reliability for different kinds of network configurations. The SqRSM method integrates cellular automata (CA)-based Monte Carlo simulation (MCS) and the Box-Behnken design (BBD) to simulate symbolic networks. The estimate response of the MCS is then separated into analytical and stochastic components, and the response surface methodology (RSM) is used to build the approximate symbolic network reliability function (SNRF). In this study, the proposed squeeze ANN (SqANN) approach combines the squeezed concept with depth-first search (DFS)-based MCS, BBD, ANN, and Taguchi method (TM) to evaluate the two-terminal binary-state network reliability. According to the experimental results of the benchmark example, the comparison appears to support the superiority of the proposed SqANN method over the traditional ANN. The finding also suggests that the SqANN method is better than the SqRSM approach for most applications.