Application of regression analysis for data processing of inertial track monitoring system

Issues of application of regression analysis to data processing in inertial track monitoring system (ITMS) during multiple runs on the controlled track section are considered. Rail track diagnostics is usually ran rarely, so much time passes from the moment of the defect appearance before it is revealed. As a result of multiple runs of railway trains, the defect can reach the critical value by that time. The solution of this problem is to equip passenger and freight trains with ITMS implementing the so-called unmanned technology. ITMS was developed and successfully implemented by the scientific staff of Laser Inertial and Navigation Systems dpt. of Saint-Petersburg State Electrotechnical University ‘LETI’. One of its main subsystems is a set of MEMS sensors mounted on the wheel pair's axle-boxes, measuring their kinematic characteristics. Fusing the data from these sensors from multiple runs on the same track section enables systematic high speed track diagnostics, and allows predicting the development of the rolling surface defects as well as judging the effectiveness of repairs.