In light of the goals set by the peak carbon and carbon neutrality targets, the share of renewable energy in the energy mix is expected to grow. This shift, however, increases the variability and unpredictability of renewable energy generation and its impact on the power grid, presenting challenges for effective power planning. This paper addresses the carbon emission evaluation system and the green certificate–carbon trading mechanism in response to these challenges. First, to achieve low-carbon power generation, this paper integrates the uncertainty of renewable energy generation with carbon emissions. It introduces and constructs uncertain carbon emission and carbon reduction supply indicators, enabling a more nuanced evaluation of carbon emissions during power system planning. This approach helps fill a research gap in the field. Next, to promote equitable carbon reduction practices and avoid an overreliance on rigid numerical standards for green certificate–carbon trading rewards and penalties, the paper introduces the carbon trading quota compliance rate and the green certificate trading quota compliance rate. These metrics address the inflexible rewards and penalties in traditional green certificate–carbon trading, allowing for dynamic adjustments in the benefits of green certificate–carbon trading. Finally, a two-layer power planning model incorporating uncertain carbon emission indicators and a flexible reward and punishment system for green certificate–carbon trading is developed. The model consists of a decision-making layer that determines the number and type of units to construct and their connection to the power grid at each node, and a simulation layer that replicates the production output on a typical day based on the upper-layer results. Using the NSGA-II two-layer solution algorithm, the model is validated on the HPR-38 node system and the improved IEEE-30 node system. In the HPR-38 system, Scenario 1's O&M costs are RMB 608 million and RMB 1233 million lower than Scenarios 2 and 3, reflecting its higher new energy capacity. However, its green certificate–carbon trading costs are 31.52% higher than Scenario 2, though it has less wind and solar curtailment. In the IEEE-30 system, introducing uncertainty and reward-punishment factors increased renewable energy share by 16.72% and 14.49%, respectively, and by 46.31% when combined. The model also reduced carbon emissions by 23.99% and system costs by 77.71%. These results show that the model can significantly increase the installed proportion of renewable energy, reduce carbon emissions and system costs, better solve the limitations of rigid rewards and penalties in traditional green certificate–carbon trading, and provide a practical and effective method for low-carbon power planning.
