Reliability Analysis On Some Complex System

Reliability engineering receives much recognition by expert and scholar all over the world after its naissance. Even in some country, reliability engineering is regarded as a matter which causes a country to be rise and fall. Reliability engineering of system encompasses a range of issues related to the design, analysis and calculation, maintenance of system, one of the most basic is system reliability analysis, which is devoted to research how to calculate the reliability of a system subject to their components reliability parameter.With the development of the technology, the more complex the system is built in reality, the more difficult the reliability index is obtained. In this paper, some reliability issues existing in complex system have been studied from analysis method and theory models. The main results are listed in the following:1. Because of the difficulty to obtain the availability of the consecutive k-out-pf-n: G systems, two methods to estimate the availability of this system have been put forward. A maximum likelihood estimator is adopted when it is generally not feasible to make assumptions regarding its failure and repair time distributions. Bayes point estimator for availability is obtained by assuming prior distributions for the parameters involved in failure time and repair time distributions.2. One of the main issues to apply the Markov modeling method to reliability and availability analysis is the challenge called largeness, i.e., the explosive number of state. For consecutive k-out-of-n: F repairable system with multi-state components, two methods have been suggested to cut short the large number of states. First availability analysis is based on the Markov modeling method for one multi-state component, then, a tight approximation of availability of the system is obtained. Another way is also based on Markov modeling method in a smaller state space when a criterion of state truncation is presented. A numerical experiment shows that the proposed two methods are efficient.3. For repairable system, the assumption that the repair time follows an exponential distribution may be unreasonable or may be violated. A model has been put forth based on repair time following any distribution for consecutive k-out-of-n: F repairable system with multi-state components. A set of differential equations are obtained. For theconsecutive n-1-out-of-n: G system with multi-state components, the exact formulas of the system reliability (or its Laplace transform) and the system MTTFF (mean time to first failure) can be obtained.4. The methods to estimate the upper and lower bounds of network reliability with multistate components are studied. According some criterions, the k most probable states which have a proportation over 90 percent of state space have been generated in order of nonincreasing probability. The lower bound for multistate network two-ternimal reliability can be obtained when algorithm is used to judge if the k most probable state is connected or not. The upper bound is get on the basis of the multistate minimal cut set. Numerical example demonstrates the bound issuperior to MESP, MEDP and MLQ. The upper bound for multistate network all-ternimal reliability is obtained on the basis of the cut set separating individual nodes from a network.5. In a general repairable system, the failure rate and the repair rate of the component is unlikely constant because of the weather and environmental conditions, using fuzzy number to represent failure rate and the repair rate is more realistic. A repairable system with fuzzy failure rate and fuzzy repair rate is studied in this paper. A fuzzy Markov modeling method is put forward and an algorithm to calculate the fuzzy steady availability is established. A network is analyzed based on this model and the fuzzy steady availability of the network is gained.6. Imprecise reliability models of common systems, such as a bridge system structure, compound system and the consecutive k-out-of-n system are put forward in this paper when there is only some partial information about the component lifetime distributions including points of unknown distributions and probabilities on nested intervals. The effect of the component independence condition on the reliability of systems is studied. The formulas of the unreliability of the bridge system structure, compound system and consecutive 2-out-of-n: system, are obtained in the explicit form under different conditions. An algorithm is put forward for consecutive k-out-of-n: F system. Finally, a model is obtained based on imprecise probability theory to estimate the network reliability such as star network, ring network and bus network.