The integration of a high proportion of renewable energy into the power grid brings lots of uncertainty factors. Applying traditional deterministic planning will lead to problems such as low utilization of system equipment and serious wind and photovoltaic curtailment. Based on the Internet a comprehensive, real-time and shared planning data system was established to realize the transparent interaction of information between the power supply side and the load side and avoid blindness in planning. Probabilistic modeling was taken into account the characteristics of the source load. Applying multi-scenario analysis to simulate the uncertainty of the source load and the improved M-K clustering algorithm to determine the typical scenarios of the source and load. Founding a bi-level programming model in which the upper layer aims at the lowest comprehensive cost of planning and the lower layer optimizes from the three perspectives of economy, technology and safety. The model was solved by applying immune genetic algorithm. Finally, the comparison and verification of the planning area and the overall calculation example show that the planning mode based on the Internet technology can effectively reduce the capacity ratio and the comprehensive planning cost, while reducing the phenomenon of wind and light curtailment.