In order to make full use of the potential regulation capacity of high energy-consuming industrial loads and alleviate the pressure on the supply-demand balance of power systems caused by the large-scale access of renewable energy resources, this paper proposes a dual-time-scale optimal scheduling method for high energy-consuming industrial parks considering uncertainty of photovoltaic (PV). Firstly, to characterize the uncertainty of PV during the optimal scheduling process of industrial parks, the typical scenarios of PV power outputs are generated based on the conditional generation adversarial network (CGAN). Secondly, considering the coupling mechanism and regulation characteristics of various regulation resources in the industrial park, a dual-time-scale optimal scheduling model is established for high energy-consuming industrial parks. In this scheme, the production plan of high energy-consuming industrial users is optimized to maximize the operation economy of the industrial park in the day-ahead stage. While in the intra-day stage, the power fluctuation of the industrial park is further smoothed by coordinating the tap position of electric arc furnaces and the charging/discharging power of energy storage resources. Finally, the optimal scheduling of an industrial park containing two typical high energy-consuming industries (i.e., iron and steel plants and cement plants) is carried out to verify the effectiveness of the proposed method.