In the context of distributed renewable energy access and implementation of time-of-use tariff, a home energy management strategy based on model predictive controlis proposed to reduce electricity cost, improve user comfort and increase renewable energy consumption. This paper establishes a home energy system consisting of distributed photovoltaic and various types of electric loads, analyzes working characteristics of various devices and puts forward corresponding comfort evolutionindices. For the typical power variable load of air conditioner, an indoor temperature prediction model is established by combining the thermal dynamic characteristics of the building. Based on the architecture of the home energy system, the optimal management is carried out using genetic algorithms, which are continuously implemented and updated under the model prediction control framework. The experimental comparison results show that the home energy management strategy based on model predictive control proposed in this paper can effectively achieve the optimal scheduling of energy and has strong robustness under the prediction uncertainty environment.