Optimum system modelling using recent gradient methods

Three gradient optimization methods have been applied to the problem of determining on optimum low-order model for a high-order system. A comparison has been made between the three methods (the highly regarded Fletcher—Powell method, the new Fletcher method and the Jacobson-Oksman method) by obtaining second- and third-order models of a seventh-order system so that the stop-response of the model approximates that of the system in the least-squares sense.