A fault/failure detection algorithm based on a grid of virtual sensors for engine health monitoring

A network of virtual sensors from a set of N real sensor is derived by using a model reduction approach. Starting from the output of a real sensor, analytical correlations are derived an used for linking the corresponding set of virtual sensors. By applying this procedure to all real sensors, a matrix of sensors is derived. The system can monitor its health by comparing the sensor outputs and diagnose if a failure is occurring. The application to a sensor system for a jet engine configuration is illustrated.