A novel approach for event detection in non-intrusive load monitoring

Monitoring household electricity consumption is important to help reduce energy usage. To improve the detecting accuracy of switch on and off events, this paper proposed a new improved event monitoring method based on CUSUM, which aims for better detection performance particularly in terms of low power appliances. Furthermore, introducing confidence level as the criteria and the bootstrapping algorithm effectively reduce detection errors and missed detections. The experiment results show that the new approach achieves higher accuracy compared to the traditional method of non-intrusive load monitoring (NILM).