A Prediction Method for Power Frequency Quality Based on Bayesian Theorem and Uncertainty classification

With the continuous concern on power system stability and increasing interesting on advanced PMU device application, the focus shifts from ordinary grid measurement data acquisition to system stability monitor and possible event forecast. Among these analysis, frequency quality and stability is always first priority. A proper prediction can help grid to adjust their operation quicker and increase the power reliability, however it still depend on the original data accuracy. In this paper, the proposed method presents a novel frequency quality detection and prediction method on the background of Bayesian theorem and BIPMs Guide to the expression of Uncertainty in Measurement (G.U.M). By analyzing the data stream from PMU in historical record, the proposed method draws an overall conclusion of system frequency quality with confidential level, uncertainty estimation and short time prediction value, which is easy to illustrate for control center operator. It has been tested using data from Michigan Tech Power Energy Lab. The preliminary DC signal tests the normal result for method validation. Testing results show this Bayesian-based prediction method has a great potential in power system frequency quality analysis and dynamic event prediction.