Research On Markov Model Of Safety Instrumented System Reliability Based On D-S Evidence Theory

Large-scale development of modern industrial production, to provide mankind with a richer material life products, there are also a catastrophic accident hazards. In order to improve the safety of industrial production, safety instrumented system (SIS) is widely used in different fields of process industry. Before the occurrence of a hazardous event can correctly perform its safety function, avoid dangerous accidents, or reduce the risk of damage caused by the accident. Safety instrumented system as a major force in the production process of industrial safety, security and reliability of its demanding. However, due to its structure, software, hardware, and maintenance of the surrounding environment and other reasons, the existence of its own security issues. To ensure safe production, to avoid disasters, the reliability of safety instrumented systems research in the field of functional safety became a hot issue.Currently, the typical safety instrumented system reliability assessment methods are five categories:fault tree, reliability block diagram, simplified formula method, PDS and Markov model. The first four model calculations simple, but once modeled considering the performance index is limited, there are too many conditions assumption, not dynamic response of the safety instrumented system dynamics between states. As Markov chain itself features, Markov model overcomes the deficiencies of the previous four models, this paper uses Markov models for safety instrumented system reliability evaluation. But in the Markov model of evaluation is not considered a safety instrumented system uncertainties Uncertainty in the SIS system reliability evaluation has a great impact, including parameter uncertainty, model imperfection and human factors. Human factors are difficult to quantify, the paper not considered. Markov model parameter uncertainties is the assumed parameter values of each state point value, but in practice with the use of each state parameters change over time, and an interval value. Safety Instrumented Systems in order to improve the safety instrumented system availability and reliability typically use redundant structure. In multiple redundant structures greatest impact on the safety and reliability is the common cause failure factor. But Markov model used in different redundant structure factor models use the same common cause failure factor, there is a model is not integrity.In order to solve Markov reliability evaluation model uncertainties, this paper analyzes the existing reliability evaluation model. For Markov model is applied to the SIS system reliability evaluation parameter uncertainty exists, this paper introduces the Markov model of DS evidence theory proposed DS-Markov model. The model DS theory of belief functions and plausibility function to calculate the composition of the recognition by the various states in the framework of the failure rate of the interval value to obtain two different Markov model state transition matrix. Depending on the state transition matrix calculation unit or the system, the average probability of failure on demand. Finally get an average probability of failure on demand interval value. Incompleteness for the model problem, this paper combined with β-factor models and multiple β-factor model proposes a new common cause failure factor model—β*model,β*model not only using the correction factor CMooN common cause failure to distinguish between the different structures of the redundant cause failure factor, also considered the safety instrumented system self-diagnostic.This paper using DS-Markov models and Markov models for five common structural unit redundant reliability assessment and analysis of the results; This structural unit of the2oo3redundancy model through V and V models to assess and analyze the results of the assessment; Then the improved model is applied to a complete verification of its safety instrumented system reliability assessment accuracy. Experiments show that the proposed DS-Markov model with interval value indicates the average probability of failure on demand. DS-Markov model assessment Markov model to assess the mean is higher than the value obtained in point. DS-Markov model shows the cross-level safety integrity level. Compared with the Markov model DS-Markov model to assess higher accuracy; In this paper,β*model both in DS-Markov model or Markov model to assess the results were higher than the β model, compared with the β model β model considers more comprehensive. Safety Instrumented System reliability evaluation more accurate.