With the rapid development of big data technology, it has greatly changed the way people get information, and also improved the speed and quality of information. In this context, smart sports has become a new trend in sports development. This paper creates an intelligent learning environment and builds a smart sports platform through advanced concepts and technical means, which can effectively optimize the integration and sharing of sports resources. Starting from the overall architecture design of smart sports, the key technologies of machine learning model to realize smart sports are sorted out. Through the five basic linking stages with machine learning model as the core, the value innovation path of platform construction structure is analyzed. The current status of sports resources application is studied, and the data mining algorithm is used to calculate the user usage data of the smart sports platform and improve the construction of the smart platform. Through the construction of the smart sports platform, people shift from traditional reading books and watching TV programs to getting information through intelligent mobile terminals, and the proportion of attention to sports information is as high as 58.6%. This shows that by building a smart sports platform, it can provide support and guarantee for the sustainable development of sports.