| As a large and complex electromechanical system,the high-speed train is increasingly important in transportation.In order to ensure the safety and reliability of trains,the requirements for the control performance of its component systems are getting higher and higher.The running gear system is the core component that affects the running quality and running speed of high-speed trains,and its mechanical properties have a direct impact on the overall health of the train.Based on the mechanism analysis of the running gear system,its complex operating environment makes the operation data unable to reflect the state information of the running gear system intuitively and accurately.The technology of health status assessment is an effective method to ensure stability and operation safety of trains,and it can achieve the purpose of monitoring the running gear system status.This paper uses the historical data of the running gear system in high-speed trains to extract the fault features,and proposes a processing method to analyze the evaluation state by using the slowness feature.That is a kind of health state assessment technology based on slow feature analysis(SFA)and hidden Markov probability distribution.This paper mainly studies from the following two aspects:(1)This study proposes a method combining SFA and hidden Markov to evaluate the health status of the running gear system.In the feature analysis part,by analyzing the influencing and controlling factors of the running gear system performance,the measurement points of the feature data are selected under the condition of ensuring the physical meaning of the monitoring data.The utilization of SFA can filter out the slowest features,and complete the data analysis and dimensionality reduction.And the correlation function fitting is introduced for feature matching to achieve the purpose of feature selection,which ensures the accuracy of classification.In the building a classification model section,based on historical information and expert knowledge,the data is planned and classified in advance,and a hidden Markov health state evaluation model is established to complete the training and preliminary evaluation of the data.The effectiveness of the method is verified by the case analysis and simulation.(2)In order to further evaluate the interval of state change,this paper proposes a performance evaluation method of the running gear system based on state difference optimization,and proposes the slow feature maximum entropy analysis(SFMA)method.The process monitoring characteristics of slow feature statistics are used to detect and analyze the test data.At the same time,the method combines the maximum entropy function to find the stable/disturbed interval of the system state change.Finally,a state difference optimization module is designed to optimize the evaluation model.After comparison,the result shows the proposed method can enhance evaluation performance. |
