Optimization of Reliability in Complex Systems Using an Artificial Neural Network Approach

Reliability-based design optimization (RBDO) is used for improving systems with vulnerabilities in structure factors, framework parameters, or both. RBDO incorporates a dependability examination that needs various computational battles, particularly for real-world issues. Here, a high performance Surrogate-Assisted RBDO approach is given. Computational insight and RBDO decomposition-based techniques are consolidated to build up a speedy RBDO. This creative methodology depends on the artificial neural networks (ANNs) as a surrogate model and Sequential Optimization and Reliability Assessment (SORA) strategy as a RBDO procedure. The issue in SORA is isolated into two segments of successive deterministic streamlining and dependability evaluation. In this paper, to upgrade the computational effectiveness and to extend the application scope of the SORA procedure, an Augmented SORA (ASORA) adaptation is proposed. In the created strategy, to recognize the latent probabilistic imperatives and separate the fulfilled limitations from reliability evaluation, a paradigm is utilized to diminish the related computational costs. Besides, the adjustments in the move vectors acquired for fulfilled requirements are controlled to get the estimation of zero for the following RBDO cycle. To exhibit the viability and exactness of the proposed techniques, some scientific models with various degrees of multifaceted nature and a viable designing model are fathomed and the outcomes are analyzed.