Method On Safety Risk Prediction And Analysis Based On Dynamic Event Tree

Modern engineering systems take on some significant dynamic behaviors, and their safety-risk has uncertainty. With the development of PSA, a variety of techniques for dynamic safety analysis have been proposed. Most of them are used to handle the polymorphism, non-monotonic, software affect and human reliability of systems; but few of them do some works about the dynamic behavior of systems, i.e. the interact between their process variables and components. which is a common factor for almost all the engineering systems. Therefore, it has important theoretical and practical significance to take a study on the dynamic behavior of system.. Furthermore, for the demand of risk control, if we can predict or analyze the evolution path and time of a possible accident, and the critical state or event of safety risk of system, it would provide important basis for decision making of the safety risk control.Dynamic Event Tree (DET), which is a tool for Probabilistic Safety Assessment (PSA) based on the accident sequence, can describe the dynamic behavior of the system very well. Particularly, it has some advantages for describing the interaction between the process variables and components; Also, we can better depict the evolution of the accident sequence and rootle the evolution information of the potential accident by using DET. So DET can be used for predicting and analyzing the safety risks.In this thesis, the basic theory of dynamic event tree is introduced at first, which is based on the theory of the traditional event tree and continuous event tree. The basic concepts and definitions of DET, and the mathematical description of the evolution of the system state are given. At the same time, a variety of constraint rules during building the DET are discussed in detail. Two different operational strategies for simulation are studied, and the corresponding comparative analysis is presented; then the selection of some parameters in the simulation process is discussed. Secondly, the basic concepts of safety risks are explained and the method for risk prediction and analysis based on the DET is studied. On the basis of DET, we discuss the describing method and dynamic characteristics of risk, and then introduce the basic concepts and accident evolution mechanism of the accident sequence. The corresponding mathematical definitions of the evolution path of the accident are proposed. For the problem of forecasting the evolution time of accident, two methods are given which are static forecasting methods based on the ET / FT and dynamic method based on the DET. In the following, the definitions of a class of critical states and the two search algorithms which are based on the sub-tree decomposition and logical operations are given. Finally, taking a cooling buffer system as example to verify the effectiveness and feasibility of the methods proposed in this thesis.