The invention discloses an 
electric heating subdivision method based on historical load recognition data under a cloud edge collaborative architecture. The method comprises the steps of collecting 
power load historical recognition result sample data, extracting 
electric heating recognition result data, summarizing and counting, obtaining recognition data of resident users in the next day, extracting the 
electric heating recognition result data, and obtaining electric heating recognition results. And performing statistical calculation on the data, judging whether the data meets an electric heating 
subdivision rule or not, if so, refining the electric heating 
label of the resident user in the day into a specific 
label, and updating and summarizing statistical data to enable the statistical data to participate in the electric heating refining of the next day and the like. Combined electric kettle, the invention discloses power of an electric 
cooker, an electric oven and an electric waterheater. According to the cloud-side cooperative architecture, the electric heating large class is refined into the small class of electric appliances under the cloud-side cooperative architecture, sothat a user can know the power utilization condition, reasonably arrange the starting time of the electric appliances and respond to the step 
electricity price, the 
electric energy consumption is reduced to the maximum extent, and corresponding auxiliary judgment can be made for fault diagnosis of the electric appliances.