Análise e filtragem de sinais de acelerômetro para monitoramento de linhas férreas

Abstract

The railways have a strong participation in both the national and world economy. However, accidents can be caused by poor track conditions. One of the main difficulties for railway transport companies is to identify the points where there are problems on the tracks. For this purpose, metric-gauge railway vehicles are used, equipped with equipment for collecting and processing information about the conditions of the railroad, called Control Cars. However, data obtained from inertial sensors installed in these vehicles contains a lot of noise, which makes it difficult to correctly interpret the signals. On the other hand, mechanical sensors have a high financial cost. With the removal of noise present in signals obtained from low-cost accelerometers installed in control cars, they can be used to monitor railways. With this, higher value sensors are replaced by accelerometers. This work proposes the analysis and filtering of signals obtained from accelerometers installed in control cars. Techniques for removing noise in sensor signals were evaluated, the sections in which the control car was curved were detected from the accelerometer data and a method was developed to detect the orientation in which the accelerometers are located. installed, thus making the automatic calibration of the inertial sensors of the car. Still, the analysis of lateral accelerations that influence the calculation of the alignment of the tracks was made. Finally, the filtered signals were compared with data from LVDT (Linear Variable Differential Transformer) mechanical sensors.