Research On Speed Control Of Automatic Train Operation System Based On Fuzzy Neural Network

With the high-speed development of cities and urban transportation, the urban rail transitdevelops rapidly in recent years. Presently the development of the urban rail transit with thecharacteristics of large capacity, high speed, low energy consumption, less pollution, highreliability has become the preferred solution to solve the traffic congestion throughout theworld. The modernization level of the urban rail transit has become one of the importantmarks of the urban modernization. Automatic train operation system (ATO) is a key techniqueto rail transit, and the algorithm of speed controller directly determines the ATO’sperformance. In this dissertation, the structure of ATO has been studied by analyzing thetrain’s operational characteristics; on the study of fuzzy control and neural network algorithm,the combination of fuzzy control and BP neural network algorithm has been applied into thetrain’s automatic operation system, and leaded to the design of speed controller based onfuzzy neural network algorithm.Firstly, the dissertation introduced the status of studies on ATO and the advantages anddisadvantages of the various algorithms used in the speed controller of ATO. The paperfocused on the combination of fuzzy control and neural network algorithm, and twoalgorithms can complement with respective advantage.Secondly, the establishment of the mathematical model of train motion, thesimplIfication of circuit data, the setting of the speed limit, the summarization of the optimaldriving strategy, the generation of the speed-distance and speed-time curve, all support for theanalyzation of the speed tracking simulation. Considering the rail conditions, train’sadditional resistance, speed-limit, in the process of designing target curve, combined with thetrain’s operational strategy and equations of motion, researchers respectively dealt with eachstage in the course of train operation, designed the corresponding programme, and generatedthe target curve by simulation; then, they designed the fuzzy neural speed controller,established the fuzzy sub-sets, fuzzy domain and the input and output parameters,and impliedthe BP neural network into the fuzzy algorithm in the reasoning process. Each of theperformance indexes of the metro’s ATO has been summarized to test the effect of fuzzyneural speed controller. In addition, establishing the speed controller of PID model, thesuperiority of fuzzy neural controller has been demonstrated in the dynamic and stableperformance by step response.Finally, the dissertation has finished the MATLAB simulation of the ATO speedcontroller which is based on fuzzy neural network algorithm. Through the comparison andanalysis of the results of simulation, the speed controller based on this algorithm can welltrack the target curve and achieve each of the performance indexes of the train operation.