DATA-DRIVEN UNMONITORED USER ELECTRICITY CONSUMPTION PATTERN RECOGNITION METHOD
Smart meters installed on the end user side can effectively analyze their abnormal power consumption behavior and power consumption patterns.To fill the user data missing that may exist in the transition period,a data-driven unmonitored user power pattern recognition method is proposed.We used the typical daily load curve historical data of users with smart meters to extract the typical power consumption patterns,and trained multi-time sale machine learning models to estimate the monthly consumption of users.The recursive Bayesian learning and branch current state estimation residual method were used to obtain the daily load curve from the monthly electricity bill of the unmonitored user.The simulation results on measurement data from actual systems show that the proposed method can identify the power consumption mode of unmonitored users quickly and accurately.
国家科技期刊平台
NSTL
万方数据
