A microprocessor system for surveillance of sensors via autoregressive identification

This paper presents an on-line sensor tester based on autoregressive models. The noisy signal delivered by the sensor is analysed as a response of a linear system to a white noise excitation. This model is identified and then used to obtain an estimation of the frequency and step responses of the sensor.After this identification stage, the model and its characteristics are used to determine whether or not the sensor is in working order. This decision concerning the sensor state is taken following statistical criteria.The main part of the paper discusses the theoretical background of the realized device: choice of the model and order determination, model validation, responses calculation and decision on the sensor state. Results of the preliminary study in simulation are shown. Then the hardware and software of the realized device are described as well as the obtained performances.