Point source emissions account for over two-thirds of anthropogenic CO₂ emissions, and accurately assessing these emissions is crucial for understanding their impact on global climate change. The Orbiting Carbon Observatory-2 (OCO-2) satellite is capable of monitoring global CO₂ concentrations. Under specific orbital conditions, OCO-2 can effectively identify strong point sources and assess their CO₂ emission intensities, thereby providing near-real-time emission reports. To improve the accuracy and applicability of point source emission evaluations, this study introduces a novel model, EMI-GET. This model combines genetic algorithms with sequential quadratic programming methods, significantly reducing uncertainties introduced by predefined empirical parameters and meteorological data in existing models. The model was applied to 11 cases across China, comparing the CO₂ emissions estimated by OCO-2 with those reported in emission inventories. The results show that the emission differences ranged from 1.2 to 73.1 kg/s, with the EMI-GET model reducing uncertainty by approximately 14.4% compared to existing approaches, especially in point source assessments. This model provides valuable support for policymakers, aiding in the management of fossil fuel usage and the formulation of effective CO₂ emission reduction policies.