An Adaptive Monitoring Method Based on Improved Random Forest Algorithm for Power Stealing Behavior of Low-voltage Users
The electricity stealing behavior of users occurs frequently.Aiming at the problems of difficult monitoring process and low accuracy of daily power consumption,voltage and three-phase current changes after the electricity theft,an adaptive monitoring method of low-voltage users'electricity stealing behavior based on improved random forest algorithm is proposed.The distributed structure is used to collect the power consumption information of low-voltage users.Through the cooperation between the control oscillator(CCO)and the single threaded unit(STA),the 96 point meter curve data collection in one day is realized.It extracts the deviation degree characteristics of power consumption curve of electric quantity,current and voltage of power stealing users from the collected power consumption information.These features are input into the random forest algo-rithm to start training.After training,the adaptive monitoring samples are obtained,and the penalty factor is introduced into the random forest algorithm to improve the training ability and make it adaptive,so as to complete the adaptive monitoring of low-voltage users'electricity theft behavior.The experimental results show that this method can accurately collect the daily power consumption of users,quickly find the changes of voltage and three-phase current after the power theft,and realize the monitoring of a variety of power theft behaviors.
improved random forest algorithmlow-voltage usermonitor of stealing electricitythree-phase current
万方数据
维普
