This study establishes detailed neural network modeling procedures for the combustion optimization of a power generation unit, equipped with a Selective Catalytic Reduction (SCR) system. In particular, artificial intelligence neural networks were applied due to its capability in treating large pools of data of great nonlinearity. The objective of this study was to reduce nitrogen oxide (NOx) emissions in the stack gas, lower the cost of emissions compliance and minimize the net unit heat rate at this power generation unit at full-load conditions. These procedures include building a database with real coal power plant operating data, modeling the database and using the genetic algorithm to optimize the operating conditions. The operation is optimized with respect to the steam temperature, selective catalyst reactor, and separated over-fire air conditions. Verifications of the results demonstrated that by carrying out the reported study, the pollutant concentrations of the boiler are significantly decreased and the catalyst layer is effectively preserved, while the operation can be maintained stable near the full-load condition. It was found that the average ammonia (NH3) flow to the SCR under optimized conditions was 229 kg/h, while the average NH3 flow under non-optimal operational conditions was 260 kg/h, resulting in a reduction in NH3 consumption of approximately 12 % from the non-optimized operational condition. In the meantime, the SCR inlet (boiler exit) NOx was reduced from 523 mg/Nm3 (non-optimized operational condition) to 444 mg/Nm3 (optimized operational condition), which represent about 15 % reduction. Based on the experience with SCR from manufactures and operators, the SCR catalyst life expectancy has a direct relationship with the SCR inlet NOx concentration and NH3 injection rate. The lower the NOx concentration at the SCR inlet and the less the NH3 injection rate, the longer of the SCR catalyst life will be. Thus, it is expected that the SCR catalyst layer replacement frequency could be reduced by 10 to 20 % over the normal catalyst replacement cycle, while meeting the same environmental NOx emissions limit. Additionally, a Delta Heat Rate was also calculated using best available data from the optimized operational conditions, including boiler load, boiler excess oxygen (O2), fly ash unburned carbon and attemperation flow. Results were compared with the average Delta Heat Rate calculated from baseline tests. It was found that the average Delta Heat Rate from baseline tests are −60 kJ/kWh, while the delta heat rate from the optimized operation condition hours is −118 kJ/kWh. This represents about 0.57 % improvement on unit heat rate from the baseline average.