مراد شادفر

The passage of trains over railway track and the resulting forces due to interaction of wheels and rails is the main cause of deterioration of the track quality. Frequent inspections with considerable costs should be done to ensure the safety and quality of the tracks. At present, track inspections are generally done with the help of track recording vehicles (TRVs) and visual inspections which has the challenge of human resource, possibility of human error and track occupancy. Also, determining the exact severity level of the inspected defects is always come with challenge. Measuring train accelerations is a method by which railway track condition monitoring can be set. If a track deteriorates, its interaction with the train and consequently its dynamic response and accelerations are changed. Therefore, it is possible to identify system defects by measuring and studying the system acceleration and its change. It is also possible to investigate velocity-dependent/independent components of the signal, their excitation source, and their effect on the dynamic contact forces of the wheel-rail system. In this thesis, the axle box acceleration from a train in Tehran metro was measured. Short-time Fourier transformation (STFT), EMD and EEMD were used to study the signal. Also, the natural frequencies of the wheelset were extracted using the finite element method to investigate its relationship with the measured acceleration. The results show that wheel structural vibration has a great effect on the measured acceleration and can be used to detect the severity of rail defects. A noticeable innovation in this thesis is study the relation between acceleration amplitude and longitudinal velocity, which was found linear. Also, determining optimum sampling frequency and determining the amount of new information can be obtained with increase in signal processing complexity is also investigated, which is another major innovation in this study. It is found that the first six mode shapes up to 680 Hz should be considered for this issue. Also, the use of EEMD method in comparison with EMD due to reduction of mode mixing phenomenon showed minor improvement in extracting the impact from the signal. Finally, by considering the vertical and lateral acceleration, speed and wheel structural mode shapes, an index is proposed to eva‎luate the impact intensity and prioritize track inspection. The detected defects in this study includes weld impact and rail spalling. Also, rail corrugation can be identified using first wheel mode-shape. An odometry algorithm is also developed using speed and vertical acceleration data to accurately position the defect location for future monitoring purposes.

عيوب موضعي ريل , شتاب جعبه ياتاقان , پردازش سيگنال , تجزيه مود ذاتي , مكان‌يابي

Rail local defect , axle box acceleration , signal processing , EMD , odometry

Morad Shadfar

Dr. Habibollah Molatefi, Dr. Asghar Nasr