Sensor data validation using gray models

A new method based on the gray model is described for the online validation of measurements. A gray model is a differential equation describing the behavior of an accumulate generating operation (AGO) data sequence. First-order gray models are fitted to measuring data records using the recursive orthogonal least-squares algorithm. Predictions derived from the fitted gray model are then compared with the actual measurements to generate a prediction error sequence. The quality of the measured value is determined by the prediction errors and variance of the prediction error sequence. Experimental results for detecting the quality of measurements from a thermistor are presented.