System State Estimation Considering EV Penetration with Unknown Behavior Using Quasi-Newton Method

The growing population of electric vehicles (EVs) is resulting in the aggregate stochastic charging demand which puts additional pressure on the peak load. Therefore, the importance of having an accurate system state estimation (SSE) arises as some EV user behavior is unknown. In this paper, a new approach is proposed for forecasting EV charging load with both predictable and unknown user behaviors. The forecast charging load is then integrated with predictable base power load (load without EVs) and converted into system state forecast. An effective SSE algorithm based on quasi-Newton (QN) method is proposed to obtain a faster, more accurate and yet more reliable state estimation under potential forecast and measurement errors. The efficiency of the proposed approach is assessed with IEEE 14-bus and 30-bus systems using actual travel survey statistics and base load records. Finally, the estimation accuracy and computation time required are compared with weighted least square (WLS) method and extended Kalman filter (EKF) method. It is shown that the proposed QN method has the best performance under most scenarios.