Abstract
Keywords - GPS, ORD, accelerators, railway, signal processing
The paper presents the results of the research on the application of GPS and other ORD (on-board recording device) sensors for the detection of vehicle state, particularly in rail transport. Recent years saw an increasing interest in rail transport and the future development in this area depends on the application of new technology in order to improve the management and security. The detection of critical situation became a crucial component of modern transport management systems. The project considered the usage of sensors such as GPS, accelerometers and gyroscope MEMS rotation sensors. The use of GPS signals in transport is by all means justified because it provides such data as exact geographical position and vehicle speed. Vehicle state is described from many points of view. This are: engine malfunctions, leaks, vibrations, speed, geographical position, track conditions and the like. The signals taken from GPS and ORD sensors were analyzed through special signal processing and artificial intelligence methods. The researches on GPS application are on the preliminary stage. They are concentrated on equipment selection and testing. Some registrations were done to estimate the capabilities of proposed and established solutions. The introductory researches on the application of various ORD devices (accelerometers, MEMS gyroscopes, GPS) have shown a large complexity of sensors and circuit fitting and adjustments. In the case of critical situation detection the crucial point is the choice of the onset of recording. Some accelerometers were assigned to operate together with GPS in order to establish the vehicle position while others were used to acquire vibroacoustic signals for the assessment of the engine state, particularly in terms of potential failures. To discern between the engine states (correct engine operation versus faulty engine operation) some unique artificial intelligence methods were used (e.g. support vector machines). Calculating the given parameters for each signal which can represent both correct and incorrect engine state gave a simple two-class multidimensional classification problem that can be solve using support vector machine. The solutions presented in the paper considered many kinds of the parameters describing a signal - some spectral parameters, nonlinear parameters (Lyapunow exponents), the global statistical parameters like mean, median or standard deviation and the parameters acquired in the short time analysis. Some of the measurements were performed on locomotive diesel engine before and after rebuild. These measurement were carried out under load (water resistor) for adopted powers at a given measurement point while the sensors were fitted on engine block at points, ensuring proper signal reception from the engine crankshaft bearing.