Abstract: In order to realize the simple and accurate measurement of carbon emissions of thermal power units and grasp the influence of operating factors on carbon emission intensity, basing on the emission factor method technology, referring to the traditional q-γ-τ matrix structure form, according to the needs of solving carbon emissions of thermal power units, and grasping the influence of operating parameters on carbon emission intensity of power generation, a general matrix model of carbon emissions and carbon sensitivity of thermal power units was established, and the matrix filling rules were determined. The accuracy of the model was verified by the heat balance method combined with the material balance algorithm, and the calculation results were compared with the emission factor method. The carbon emission of a thermal power unit for 24 h was calculated, and the disturbance ΔM_CO2 of carbon emission intensity of power generation was analyzed when the smoke oxygen, main steam temperature and pressure fluctuated. Results show that, compared with the emission factor method, the proposed model can shorten the time span of carbon emission accounting and improve the accuracy of carbon emission accounting. The carbon emission of the unit for 24 h is 5 780.644 t; When the smoke oxygen is reduced by 0.1%, ΔM_CO2 is 1.772 6 g/(kW·h); When the main steam temperature increases by 0.5 K, ΔM_CO2 is 3.020 6 g/(kW·h); When the main steam pressure increases by 0.2 MPa, ΔM_CO2 is 0.378 8 g/(kW·h).