Disaggregation of Cold Appliance Loads from Smart Meter Data Processing

In a context where an increasing flexibility is required from the demand side in distribution systems, cold appliances such as refrigerators can offer a continuous and non-negligible flexibility potential. Nevertheless, efficient demand response schemes based on cold appliances inevitably rely on the accurate estimation of their actual load not only at an aggregate level, but directly at the household level. Therefore, we propose two novel approaches to disaggregate the load profile of cold appliances from residential smart meter data. Both approaches are complementary and exhibit a MAE generally lower than 40W with 1- to 15-minute resolution data at the household level. Their applicability is finally demonstrated on more than 4000 loads.