| Safe and reliable traction power supply system is vital to the operation of high-speed railway, while the maintenance pressure of traction power supply system will increase gradually with the construction of high-speed railway. Quick and accurate diagnosis on occurred fault of traction power supply system will shorten fault treatment time and reduce fault effect. Health evaluation of current state will help the operating personnel to understand the real state of traction power supply system, provide the basis for condition based maintenance. Estimation and prediction of future state will reflect the future trends of traction power supply system, provide the basis for preventive maintenance, so to reduce or even avoid the occurrence of fault. Therefore, health diagnosis and risk assessment of traction power supply system is helpful to improve the fault treatment, provide the basis for condition based maintenance and preventive maintenance.There are many uncertainty factors in health diagnosis and risk assessment of traction power supply system, they are must be taken into account in the process of health diagnosis and risk assessment.Methods of health diagnosis and risk assessment of traction power supply system are analyzed in thesis, including fault diagnosis method for error state, health evaluation method for normal state, and reliability prediction and risk assessment method for a long term operation, involving multiple stage of running, with the aim of whole process management of traction power supply system. Uncertainty mathematics, including random mathematics, fuzzy mathematics and interval mathematics, are proposed to deal with the uncertain factors in the process of fault diagnosis, health evaluation, reliability prediction and risk assessment separately.Uncertainty factors existing in the process of health diagnosis and risk assessment of traction power supply system are analyzed first, and they are classified into three areas: uncertainty of information, uncertainty of cognitive and uncertainty of environment.Fault diagnosis method based on Petri net considering service features of device is proposed to deal with the uncertainty of information. Weibull distribution in random mathematics is proposed to model the reliability of relay and circuit breaker, distribution parameters are estimated by the least square method, so their reliability can be obtained according to the service time. Associatting with the confidence of initial place of fuzzy Petri net, the reliability of relay and circuit breaker makes the confidence determination is more practical and more in line with the actual situation. The weight of places in reasoning is determined by their relative relationship of reliability, there is higher confidence and higher weight for information came from device with higher reliability, that can reduce the effect of information uncertainty on fault diagnosis result. A section of traction power supply system is proposed as a case to test the method, it shows that the confidence of diagnosis result vary with the service time of device, which coincides facts much more.Health evaluation method based on fuzzy comprehensive reasoning is proposed to deal with the cognitive uncertainty. A fuzzy comprehensive evaluation model for health status of traction power supply system is proposed, the evaluation factor set, evaluation set and weight set of evaluation model are analyzed, the analytical hierarchy process(AHP) and entropy weight are used to determine the weight of evaluation index of substation and catenary correspondingly. A section of traction power supply system is proposed as an example, evaluation indexes are transformed into membership by fuzzy transformation, substation and catenary health are evaluate separately, on this basis the overall health state of traction power supply system is obtained. The evaluation result is simple and clear, especially the health state is numerical quantified, which make an intuitive understanding of working state and development trend of traction power supply system.Reliability prediction method based on statistics and probability is proposed to deal with the uncertainty of environment. History data is used to mining the law of fault. Meteorological factors are selected by stepwise regression, relation model between faults and meteorological factors is established by regression analysis, then the future possible fault under certain conditions can be predicted by using this model. Weibull distribution is proposed to describe the reliability of traction power supply system, which is a typical chain structure. The method of minimum sum of level resident squares and the method of minimum sum of vertical resident squares are compared. Reliability function of traction power supply system is fitted by history fault data, then the reliability at future specific time can predict by using this function. Prediction results can provide reference for the operation and management of traction power supply system, and can provide basis for preventative maintenance, which is significant helpful to reduce or avoid the fault.Risk assessment method based on interval mathematics is proposed to deal with the uncertainty of environment. Lightning, as the greatest impact factor according to the regression, is proposed to assess its risk. The model of lightning risk assessment is established and the influence factors are analyzed. Interval numbers, instead of specific numbers, are proposed to describe the density of annual lightning, the area of lightning and the loss caused by lightning. The risk assessment of a section of traction power supply system shows that interval method can reflect the influence of uncertainty, and can quantify the risk degree of lightning, and can determine the equipment with high risk, and can provide reference for different measures for lightning protection.This thesis is supported by National Natural Science Foundation of China:’Health diagnosis and service evolution of key equipment of traction power supply system of high speed railway’ (No. U1234203). |
PDF Full Text Request