Research On Intelligent Detection Method For Train Slide/slip

Slide/slip of train is an important factor which affects the accuracy of odometer for train positioning.Accurate and prompt detection of slide/slip can effectively avoid potential danger by compensatory measurements,which is very important for the safety of the train.Poor wheel-rail contact or big traction/braking force may lead to slide/slip.A big slide/slip can be detected by fixed threshold method,but such method is unable to detect weak slide/slip.Although one weak slide/slip may have little effect on train positioning,but accumulation of multiple times of slide/slip may cause large positioning error.Therefore,intensive study of sensitive and reliable method for train slide/slip detection is essential to train safety and the development of railway traffic.Firstly,research on detection and compensation algorithms of slide/slip is reviewed.Secondly,an introduction to the principle of slide/slip is given.Main factors that cause slide/slip are analyzed.The detection methods which have been applied on the current trains from different countries are listed.A variety of fixed-threshold detection methods based on odometer were studied in detail,including detection with single odometer,detection with odometer combined with Doppler radar or accelerometer,and detection with all three sensors.The experimental results showed that fixed threshold methods can only detect big train slide/slip and may ignore weak slide/slip.A fuzzy adaptive algorithm for slide/slip detection is proposed in order to detect both big and weak slide/slip in terms of measurements from Odometer,Doppler radar,and accelerometer.The fuzzy rules for detection are established based on actual data and professional experience.The experimental results showed that the performance of fuzzy adaptive reasoning method is better than any fixed threshold methods.Furthermore,it is capable to detect weak slide/slip which is usually ignored by fixed threshold detection.In addition,multi-sensor fusion algorithm is also studied in order to achieve more accurate estimation of train speed.A federated Kalman filter algorithm with adaptive weight fusion is proposed.The experimental results showed that the algorithm can improve the accuracy of train speed estimation.The work in the thesis is of great significance to improve performance of current slide/slip detection and laying a foundation for the future application on the train.