Bad data tracking identification by method of time series analysis

Just as we know, in the local power system, the network is always open, so, many existing methods for bad data identification (BDI) can´t work well. We present a new method of time series analysis for bad data measurements identification. In this method, a local system is divided into several small areas. For each area, we form two data series, one is formed by the hourly load data of this area, the other is the 168th order difference of the above series; for each of the series, we build three AR (autoregression) models using various methods. With these six models, we can identify the bad data of real-power measurements of lines which are connected to this small area. This method has been tested in a real local system, the results shows that it can satisfy a real-time application.<>