Office Appliance Data Classification Based on Non-intrusive Load Monitoring

As building digitalization is accelerating, one of the challenges is making intelligent use of building energy consumption data. Since appliance-level energy consumption data is not generally available, Non-Intrusive Load Monitoring (NILM) provides novel ways to disaggregate total energy consumption into appliance-level data while ensuring the privacy of customers. This work focuses on a NILM algorithm that is applicable to common appliances and widespread smart metering infrastructure. The energy consumption data of an office was collected with a 1-min resolution and used for analysis. A fuzzy c-means clustering NILM algorithm and inter-cluster entropy were used to classify data and verify different categories of office appliances. The algorithm was proven to be able to disaggregate and classify office appliance energy consumption data with a satisfactory accuracy.