As the extension of smart grid in demand side, smart home energy optimization is an important branch of smart home. Smart home energy optimization aims to optimally schedule the home appliances to satisfy the comfort requirements and save the electricity cost. However, the comfort requirements are closely related to the human behavior, which has great subjectivity and uncertainty. Thus profiling the comfort requirements is one of the challenging problems. This paper presents a smart home energy management method based on the sensors data of smart wearable devices, which contains the human behavior analysis; updates the comfort requirements through creating the mapping model between human behavior and the comfort requirements by neural network; establishes the system dynamic models; and the parameters are estimated by using the sensor network data. Finally, the smart home energy optimization is solved by model predictive control. Based on the proposed method, the smart home platform is set up and the smart home energy optimization systems are developed to support the smart phone. The experiment presents promising performance on electricity cost saving and comfort improvement in four scenarios of user behaviors.