In the future environment of ubiquitous electric Internet of Things, smart home system will further utilize the Internet of Things to endow common household electrical equipment with good perception, decision-making and adaptability. In order to improve their intelligent degree, through real-time access to household load information to make a reasonable power consumption strategy in time, to minimize energy waste and reduce the cost of electricity for users. The authors propose a design scheme of Internet of Things platform based on the analysis of customers' electricity consumption behavior. Firstly, the authors studied the power consumption habits of different customers, then Logistic regression and probability density mathematical models for different customers have been established. A small volume of 86 boxes of sensors were designed, and the gateway of multi-protocol access for edge computing was added to complete the design of energy analysis platform. This scheme could solve the inconvenience caused by the large size of traditional sensors and the inaccuracy of traditional energy saving calculation. Because of the single communication protocol, the gateway can not be fully connected to electrical equipment, and other problems have been solved. The experiment of practical office buildings show that the proposed system is reliable and feasible, and has certain application and reference value in the field of intelligent buildings.