In [13], some of the authors of this paper proposed the fictitious reference iterative tuning (which is abbreviated as FRIT in the following) as one of the useful and powerful methods of parameter tuning of an implemented controller in the sense that we require only one-shot experiment for the parameter tuning in order to achieve the minimization of the squared error between the real output and the desired one. However, feedback properties like sensitivity or noise reductionability those are also required for a control system to be more practical have not been addressed yet. On the other hand, it is well known that the two-degree of freedom of control scheme is one of the effective ways those enable us to deal both of tracking property and feedback ones simultaneously. From this reason, in this paper, we expand FRIT into the two-degree of freedom control scheme in order to tune parameters of impelmented controllers in this structure for the sake of the acheivement of given specifications on not only tracking property but also feedback properties. We consider the correspondence of the minimization between desired cost function and the fictitious ones. We then provide effective procedures of FRIT in the two-degree of freedom of control scheme with two-shot experiments. Finally, we also give illustrative examples in order to show the validity of our proposed methods.

抄録全体を表示

This paper studies a procedure for identifying a Hammerstein model equipped with a static nonlinear odd function, where the only output is measured and the input is white Gaussian. A statistical property of the process generated by the static nonlinear odd function with a white Gaussian input is analyzed by means of a moment generating function. A proposed procedure for identifying the Hammerstein model consists of two steps : The first step is to identify the linear dynamical part via a stochastic subspace identification method based on a block LQ-decomposition; the second step is to estimate the static nonlinear odd function by using distribution functions. Numerical simulation results are also included.

抄録全体を表示

In this paper, we present a new identification method based on the fictitious reference iterative tuning (which is abbreviated to as FRIT in the following). FRIT is one of the parameter tuning methods for controllers based on the direct use of the experimental data without mathematical models, particularly, this method enables one to obtain the optimal parameter with only one-shot experiment. Motivated by such practical advantages, we expand this method to a facile and fast-track identification by using only one-shot experiment data of the closed loop implemented with the existing controller in the one-degree of freedom control system. As one of the real systems in which it is effective for such an identification method to be applied, we consider the identification of the opening-and-closing speed dynamics of an elevator door. For this system, it is important to obtain the actual state of an elevator door for retaining a safety and quick motion. Thus, an identification of this dynamics is one of the most important factors for an effective maintenance. However, it is difficult to obtain the actual parameter because of the fact that there are many cases in which an experiment for the identification spends much time and it is undesirable to use sufficiently excited signals for the identification. We apply our identification method to solve these practical problems. Finally, in order to show the validity of our result in this paper, we also give an experimental result on the identification of the opening-and-closing speed dynamics of the door of an elevator.

抄録全体を表示