The extensive use of energy generation processes presents a severe challenge to the environment and makes indispensable to focus the research on the maximization of the energy efficiency and minimization of environmental impact like NOx and CO emissions. The proposed idea describes an approach, based on an artificial life environment, for on-line optimization of complex processes for energy production. Such an approach is based on the evolutionary control methodology which, by emulating the mechanism of the biological evolution, composes the capability of sophisticated models with the continuous learning. In order to work with MSWC (Municipal Solid Waste Combustion) it was necessary to improve the stability of the optimizer to obtain a good compromise between stability and reactivity. In this way, a specific MSWI performance function has been properly defined in order to quantitatively characterize the current status of the process. The evolutionary control approach has been successfully tested on a MSWC simulator and subsequently installed on a real MWSC plant which produce electricity and heat for a small Italian town (Ferrara). The paper reports the first promising experimental tests on the real plant for optimization of energetic efficiency and pollutant emission reduction.