作者: J. Douglas Birdwell
DOI:
关键词:
摘要: An expert system (CASCADE) has been implemented for the design of multivariable controllers linear plants. using quadratic Gaussian/loop transfer recovery (LQG/LTR) methodology [1,2,3]. The development and testing knowledge base uncovered several inadequaces in LQG/LTR theory design. This paper focuses on problems associated with selection a state weighting matrix regulator (or process noise covariance filter design) to satisfy constraints singular values design's return ratio (the "balancing" problem [4]). availability allowed rapid generation designs example systems, modifications base. provided close study interaction between effects nominal plant poles at origin (system type), augmentation control steady error performance, placed ratio, method used meet those constraints. tracks evolution our philosophy toward design, summarizes current approach