Research Of Complicated Multi-state System Reliability Evaluation Approach

The traditional reliability analyzing theory and approach are based on two states, "failured completely" and "functioning perfectly". However, with the progress of science and technology and the continuouse studying on the failure principles, people have found that the system and its components, in their lifetime cycle, are of some intermediate states except the two states metioned above, viz., are characterized by multi-state. This situation urges that people start to study the reliability analyzing theory and approach of multi-state system and component in order to more satisfy the realistic requrements of reliability engineering.With the more complexity structure and more complete function of modern industry system, reliability analyzing theory and approach based on multi-state are becoming the research hotspot. The related study includes many aspects such as reliability analyzing, maintenance decision making and reliability optimizinig, etc. Aiming to the emering difficulties and challengings coming from the multi-state system reliability engineering, this thesis will research the reliability assessment approach of multi-state system combing related theories and technologies in order to provide the evidence on mastering system functioning rules precisely. According to the characters of different systems, the main research contents and novel outcomes are presented as following.1% In order to resolve the difficulty on ascertaining the structure fuction of monotone coherent complicated multi-state system when its reliability needs to be assessed, a simulated approach for calculating multi-state system state probability distribution is put forward based on the tranditional binary system reliability theory. According to the set of minpath or mincut, system reliability can be assessed without system structure function. Based on the multi-state component, three types of indices can be calculated combining Markov stochastic process, namely, the probability of component failed until time T, the probability of component failed until or before time T and the probability of component transmitted to specific state at the given time. This established the basis for system reliability assessment at different time. Examples validate the simulated method and indices.2、A model including multi-factor for multi-state degrading system reliability analyzing has been established. This model includes Poisson failure state caused by occasional cause, system can be restored to the state before Poisson failure state by maintenance. Because of the existence of user requirement level, the system state should not be below a given state. This model can be configured flexibly according user requirement, some irrelevant states can be incorporated and resolving of the model will be simplified furthermore. In order to keep systems running on the appropriate state, some preventative maintenance activities will be adopted to restore the system state when the system state is below the given state. According to this model including several factors, system reliability indices can be calculated and numerial examples validate this model.3、Aiming to the question that the number of system state will reach so huge that the model can not be resoved when the system is directly modeled, a system reliability assessing method based on the combining of multi-state component stochastic model and universal generating function (UGF) technology is put forward. The Markov stochastic model of component with minor failure and minor repair can be established so that the UGF expression of component can be ascertained. According to the physical structure of system, the combinational operator among components may be choosen and then UGF expression of system will be calculated. The related reliability indices of system can be evaluated according to the system UGF. Examples analyzing validate the method and precision and approximate calculation can be chosen based on the system dimensions.4、Based on UGF technoloty a method is put forward for reliability assessment of multi-state series-parallel system with common cause failure (CCF). The CCF is caused by some components and the components affected by the CCF may be independent or overlap partly. The CCF can be incorporated in the UGF expression of component. According to the combination of components leading to CCF, system reliability indices can be calculated with total probability theorem. For some important components, the protechtive mechanism can be added in order to reduce the probability of components failed. Then the system reliability indices can be reevaluated. In order to improve the system reliability to the greatest extent under the limited cost, which CCF lead by components should be eliminated, namely the sensitivity of components can be further determined. Examples results show that the method is validate.5、A reliability evaluation platform for power plant DCS system has been build up in order to provide evidence of chosing different DCS system for power plant. According to the reliability theory reseaching outcomes partly, the entire DCS system and its subsystem can be modeled respectively. Then the models establish the basis of reliability indices calculating. In order to ensure the system to operate saftely a three-level hazard list has been founded and the update algorithm for the failure data is also set up. The trial operating of the platform shows that it is effective and corrective and is of value of generizing and applying.