A Method for Handling Uncertainty in Evolutionary Optimization With an Application to Feedback Control of Combustion

摘要: We present a novel method for handling uncertainty in evolutionary optimization. The entails quantification and treatment of relies on the rank based selection operator algorithms. proposed is implemented context covariance matrix adaptation evolution strategy (CMA-ES) verified test functions. independent distribution, prevents premature convergence well suited online optimization as it requires only small number additional function evaluations. algorithm applied an experimental setup to feedback controllers thermoacoustic instabilities gas turbine combustors. In order mitigate these instabilities, gain-delay or model-based H infin sense pressure command secondary fuel injectors. parameters are usually specified via trial error procedure. demonstrate that their with methodology enhances, automated fashion, performance controllers, even under highly unsteady operating conditions, also compensates uncertainties model-building design process.