System Design Methods for Simultaneous Optimal Control of Combustion Instabilities and Efficiency

摘要

The work under this contract primarily involved the design, analysis and testing of control algorithms aimed at the suppression of thermoacoustic instabilities and the optimization of combustor performance. Three different types of algorithms were considered in this work: Pattern search, explicit gradient, and least-mean-square based feedback, which we have designated filtered-E. The algorithms are listed in order of increasing speed and increasing amount of a priori information required. Both pattern search and explicit gradient algorithms are useful for the optimization of combustion performance as well as the suppression of thermoacoustic instabilities. Filtered-E is intended solely for the fast suppression of thermoacoustic instabilities. All of the algorithms were successful in suppressing thermoacoustic instabilities in an experimental combustor. In addition, we provide an analysis of the mechanism for achieving control using on-off actuators pulsed subharmonically, validate the analysis with experimental results and propose a variable- subharmonic controller. Control systems using on-off actuation can be adaptively tuned using pattern search or explicit gradient algorithms with little modification. The filtered-E algorithm has also been effective when applied to on-off actuation, even though the implicit gradient is not quite correct.