Performance modeling and maintainability of fault tolerant machining system using soft computing techniques

Abstract

Stochastic models and performance analysis have emerged as vital tools for the system designers and organizations to address the unavoidable interruptions in the fault-tolerant machining systems (FTMS). The queueing and reliability models of FTMS hold significant industrial applications across various domains, including computer and communication networks, nuclear and power plants, traffic control and power distribution systems, production and assembly lines, etc. The main goal of the system designers in an industrial setting is to create machining systems capable of withstanding errors/faults. The appropriate selection of redundancy and maintainability can significantly enhance the capacity and availability of many FTMS that are prone to failures and operate in machining environments. The main goal of the present study is to examine both transient and steady-state queueing and reliability models for fault-tolerant machining system with service interruption subject to imperfect repair. Several notable features have been implemented to address the disruptions in the machining environment. The worth-noting concepts incorporated in the thesis works include the standby support, vacation, working vacation, N-policy, imperfect repair, server failure, standby support, reboot and/or recovery mechanism, retrial orbits, etc. The present study on stochastic models for the FTMS focuses on the derivation of various performance metrics, including long-run probabilities, expected number of failed units, mean time to failure, reliability, availability, throughput, etc. To investigate the performance metrics of the concerned fault-tolerant machining systems, recursive methods are applied in specific chapters. For steady state and transient behavior, matrix recursive and spectral methods respectively, are employed. In the case of non-Markovian models, supplementary variable techniques and Laplace transforms are utilized to obtain analytical results. Additionally, the cost analysis is carried out using appropriate analytical and computational techniques. For the cost optimization and evaluation of optimal decision values for the concerned machine repair models, quasi-Newton method, and metaheuristic approaches like genetic algorithms, particle swarm optimization, and harmony search techniques, etc. are used. The soft computing techniques, namely fuzzy logic, and adaptive neuro-fuzzy inference systems are also employed to explore the feasibility of developing AI-based controllers of complex systems and to quantify the performance indices of the concerned FTMS.