Intelligent Safety Engineering And Its Application Research On Train Key Equipment Safety State Identification

At present, as China's rail transit system continueing to expand throughout the country, rail transportation system constantly improving, and the quality of the train and key equipment design requirements being more stringent, the structure complexity,function coupling and behavior loss of train equipment have obvious promotion.Therefore, the safety and reliability ensuring technology of rail trains should be improved to meet the needs of modern train safety.Safety is the premise and basic of service capacity of rail transportation. The healthy service state of key equipment of rail train such as walking rail train system,door control system, braking system, traction power supply system is closely related to the train safety operation state. In the train service process, how to quickly and accurately identify the health status of the key systems is part of the most important aspects of train safety ensuring technology system. This paper analysis the general rule of safety engineering field based on the above description, and puts forward the theoretical framework of safety intelligent engineering which provides a theoretical guidance for the development of active safety technology. Specifically, first of all, do research of the technical route based on obtaining state parameters, feature extraction,feature fusion, intelligent prediction and identification, and then fuse these methods to become a rapid, accurate and visible state identification method of components.Meanwhile, it does the groundwork for research of the state influence and fault propagation effect among different components. After then,the realization of decoupling of fault propagation within the system and the achievement of confidence ranking of hidden source can be reached though the known historical statistical data and the temporal information of running state changes in state identification and hidden trouble mining links in the system level. It provides reliable technical support for the optimization and formulation of the intelligent maintenance strategy. This technology train idea of train safety warning from component level to system level provides a scientific theoretical basis and technical support for the active protection and maintenance of train operation safety. Based on the above methodology analysis, the summaries of the work of this paper are shown as follows:(I) The framework of safety intelligent engineering theory is put forward, and under the framework, the basic system safety problem is proposed. Measurable region model is used as a framework to solve the active safety state identification of rolling bearings of train wheel set. Causal chain model is applied to solve fault causal chain decoupling and hidden trouble excavation problem.(2) State representation approach under measurable region framework.Concretely, based on the existing techniques, the nonlinear signal processing technique is extended, and the correntropy characteristics of rolling bearings with robustness to operation condition and high sensitivity to early faults are put forward. And then based on principal component analysis to achieve correntropy matrix dimensionality reduction so that the visualization of the phase space consisted by integrated correntropy matrix can be bring about.(3) The state identification method under the framework of measurable domain is proposed. Based on the characterization, using support vector machine and its extension method, safety state measurable boundary solution in the phase space constructed by integrated correntropy can be get to realize safety state identification of rolling bearing in boundary constraints. The extended support vector machine (SVM)algorithm is studied. Firstly, through the combination strategy of two classification support vector machines, the measurable region of multi-class safety states is obtained in phase space, and the multi state synchronous estimation should be implemented.Secondly, two-class support vector machine under the Bias framework is adopted to obtain the blurred boundary of the measurable domain, which aims to quantize the membership margin of the safety state to realize the quantitative expression of the rolling bearing safety state.(4) The compound fault chain decoupling and hidden trouble mining under the model of causal chain model are put forward. Through the analysis of the structure and function of the train door system, a functional Petri net is constructed to express the composite fault chain of the train door system. Using Bayesian posterior probability and timely order parameters to complete the confidence ranking of components with hidden dangers. Firstly, the extended time Petri net is constructed, and then the time series backward inference based on ODDT is applied to obtain the time constrained possibility Petri net to decouple the causal chain of the fault. Based on the Bayesian posterior probability inference, the confidence ranking result of the hidden parts can be calculated in the analytic single chain, and the hidden trouble of the train door system can be mined.