In this paper, H∞ optimal model reduction for singular fast subsystems will be inves-tigated. First, error system is established to measure the error magnitude between the original andreduced systems, and it is demon...In this paper, H∞ optimal model reduction for singular fast subsystems will be inves-tigated. First, error system is established to measure the error magnitude between the original andreduced systems, and it is demonstrated that the new feature for model reduction of singular systemsis to make H∞ norm of the error system finite and minimal. The necessary and su?cient conditionis derived for the existence of the H∞ suboptimal model reduction problem. Next, we give an exactand practicable algorithm to get the parameters of the reduced subsystems by applying the matrixtheory. Meanwhile, the reduced system may be also impulsive. The advantages of the proposedalgorithm are that it is more ?exible in a straight-forward way without much extra computation, andthe order of the reduced systems is as minimal as possible. Finally, one illustrative example is givento illustrate the e?ectiveness of the proposed model reduction approach.展开更多
文摘In this paper, H∞ optimal model reduction for singular fast subsystems will be inves-tigated. First, error system is established to measure the error magnitude between the original andreduced systems, and it is demonstrated that the new feature for model reduction of singular systemsis to make H∞ norm of the error system finite and minimal. The necessary and su?cient conditionis derived for the existence of the H∞ suboptimal model reduction problem. Next, we give an exactand practicable algorithm to get the parameters of the reduced subsystems by applying the matrixtheory. Meanwhile, the reduced system may be also impulsive. The advantages of the proposedalgorithm are that it is more ?exible in a straight-forward way without much extra computation, andthe order of the reduced systems is as minimal as possible. Finally, one illustrative example is givento illustrate the e?ectiveness of the proposed model reduction approach.