Detection of Outliers in Sensor Data Based on Adaptive Moving Average Fitting

In the sensor data of the dynamic positioning reference system for voyaging ships, there are often some abnormal deflection values which are called outliers. Discarding or replacing the outliers in real time is very important for the improvement of accuracy. In this paper an adaptive moving average fitting method is proposed to detect the outliers to be discarded or replaced in the sensor data of the reference system. This method defines an outlier detection criterion, adjusts the real-time outliers and judges the bandwidth through the adjustment to standards of parameters. The experimental results showed that our method can quickly and effectively solve the problem of real-time elimination of outliers.