A new method for residential side non-intrusive load monitoring

Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 2023
Thesis identifier
  • T16662
Person Identifier (Local)
  • 201792528
Qualification Level
Qualification Name
Department, School or Faculty
  • This thesis proposes a new non-intrusive method for residential load monitoring. The proposed method can detect appliance switching events, separate appliance electric features, and identify appliance types. Compared with other non-intrusive monitoring methods, the proposed method improves the monitoring accuracy and decreases the monitoring response time. Firstly, the monitoring hardware was designed and constructed to sample and acquire the aggregated electric data of one residential area. Secondly, the sampled data were processed and analysed with the proposed method, which achieves the monitoring of individual appliance running conditions and power consumption in this area in a non-intrusive way. The data analysis process includes three steps, 1) the appliance switching event is detected by the Heuristic detection method. 2) the working current of the switched appliance is separated according to the difference method, 3) the type of switched appliance is identified with the K-nearest neighbour method according to the appliance’s current harmonic components, and the identification result is modified and corrected according to appliance operation pattern with the aid of a Back Propagation Neural Network. Thirdly, the proposed NILM method was tested through offline and online applications. The offline application involves three days of pre-recorded data which were processed and analysed. The online application consists of two parts. The first part is a direct application for four domestic homes during one day (24 hours). As for the second part, the proposed monitoring method was applied to one domestic home for ninety days. All the online and offline tests, the running conditions and the power consumption of appliances were monitored and recorded. Due to the test results, the proposed method is reliable and offers a powerful monitoring method for demand side management.
Advisor / supervisor
  • Lo, Kwok L
Resource Type