Improving the Accuracy of Bus Load Forecasting by a Two-Stage Bad Data Identification Method

As a critical data input of security constraint unit commitment, bus load forecasting is currently conducted on a bus-by-bus fashion to improve the forecasting accuracy in most provinces in China. On such substation level forecasting, data quality is much worse than the aggregated power demand for the whole system, and identifying and restoring the inaccurate measurement and abnormal disturbance to retrieve the historical trend of load is urged to improve the forecasting accuracy. In this paper, a two-stage identification and restoration method is presented. The typical patterns of inaccurate measurement and abnormal disturbance are detected in the first stage based on statistical criteria independent with normal distribution. Historical trend is further retrieved in the second stage using frequency domain decomposition, and a typical daily curve is generated to compare with the data measurements. The deviations of the data measurements from the typical daily curve obey normal distribution and are used as criteria in the second stage. The effectiveness of the proposed methodology has been confirmed by examples in real bus load forecasting systems in this paper.