The uncertainties of the wind power generation, solar power generation, electricity load, and heat load bring difficulties to the optimal scheduling of the urban integrated energy system (UIES), which influence its secure and reliable operation. To promote the utilization of renewable energy and enhance the scheduling capacity of UIES, a two-stage stochastic robust optimization model with the day-ahead market and the real-time market is developed. The goal is to optimize the UIES’s profit in the day-ahead market and minimize the total system cost in the real-time market, accounting for the most unfavorable situations of uncertainties. To address multiple uncertainties, a four-dimensional ambiguity set considering multiple uncertainties is constructed by using the Dirichlet process mixture model (DPMM) and the variational inference algorithm. Meanwhile, an accelerated column-and-constraint generation algorithm (AC&CG) is presented for iterative solution of the proposed model, with the second stage reformulated via Lagrange dual theory. Besides, a novel acceleration strategy is adopted, which contributes the algorithm to converging faster which can also improve the iteration speed. Finally, simulation examples are used to demonstrate the effectiveness of the proposed model in comparison to stochastic programming and robust optimization. This paper may provide theoretical guidance for relevant research and application.