To achieve a transition to smart clean grid and low carbon energy system, integrated energy system (IES), which includes multi-energy synergy, has better renewable energy accommodation and provides high energy utilization efficiency, will play a significant role. However, the feasibility of IES planning is challenged by the intermittent nature of renewable generation, typical day scenarios and high computation requirements. This paper proposes a Gaussian mixture model clustering method for IES planning. The dimensionality of multi-energy data is reduced by the PCA method and automatically grouped into sets to generate typical scenarios based on the Gaussian mixture model. A numerical study of a real IES is implemented using this clustering method on the CloudPSS IESLab platform to investigate device selection and capacity options. The performance of the algorithm was evaluated.