Preliminarily Research On The Intelligent Analysis Technology For Train Derail Accident

Train derailment is a joint result of interactions between subsystems of railway system under certain conditions. Train derailment is an important factor to railway running safety, which greatly affects the economic and social benefit of railway transport. In order to avoid train derailment, the only approach is to find out the real reasons of derailment accidents and then take corresponding measures.In this paper, artificial intelligence methods (such as computation intelligence and fuzzy logic) and intelligent metasynthesis methods are employed to analyze the actual causes of train derailments form qualitative to quantitative. The man-machine fused system is established and knowledge meta-synthesize is realized.In this paper, the methods that based on fuzzy logic or neural network or fuzzy- neural or fuzzy pattern recognition are put forward. These methods are lucubrated in the different chapter. The approaches of set intelligent systems up are studied in detail.For the purpose of synthesizing all analysis methods together, the basic structure of prototype of the intelligent derailment analysis support system (IDASS) is established. The development approaches of IDASS are researched in this paper. The IDASS is an IDSS that is based on three bases ?database, method base and knowledge base. The prototype of IDASS is developed by the method of object-oriented technology, which classify the field of derail accident clearly and simplify the issue of derailment analysis. Firstly, the derailment problem solving system based on distributed multi-expert system is designed and built, which is the core of IDASS. Secondly, the technologies of man-machine dialog system are researched in this paper, and the interface of man-machine is realized soundly.A subsystem to analyze the derailment at small radius of curve is established by the methods and technologies, which are researched in detail in the paper. Using some actual derail accidents, the correctness of the results that are obtainedby the subsystem of ID ASS is proved. Therefore, the practicability and feasibility of the intelligent analysis methods and ID ASS presented in this paper are proved.