A direct-drive robot of a comparatively large size has been developed for practical applications in industry, such as laser cutting. It was designed based on the semi-direct-drive concept, and has 5 degrees of freedom with a 700 mm-long shoulder arm, and a 1050 mm-long forearm, covering a working range as large as a car body. In this paper, control strategies for the full scale model are proposed through basic analysis. Unfavorable behavior of the arm due to its flexibility was supressed by using the pole-cancellation to obtain high servo-stiffness of the arm. Furthermore, feedforward control was applied to a conventional controller to practically remove its tracking-error.
In the conventional optimizing control system the optimal policy determined at the static optimization level is transfered to the feedback control system as the set-point. The follow-up property to the set-point, however, becomes an issue when the catalyst deactivates rapidly and the optimal policy changes considerably at the next optimizing period and also when the process response delays. In a proposed optimizing control system an optimal operational schedule until the next optimizing period is determined by using a reaction rate and a deactivation model to solve the issue. The resultant optimal policy during the period is effectively used to improve the follow-up property of the feedback control system by using the model predictive control algorithm. It works as an interface between the optimizer and the feedback control system. Application of the proposed 3-level on-line optimizing control system to an experimental catalytic rector showed excellent performance in preservation of the optimality.
The linear optimal regulators in linear mathematical models of physical systems have proved useful in numerous technical applications, although most differential equations describing the actual system behavior are generally nonlinear. However, a nonlinear optimal regulator is necessary in the inherent nonlinear system which shows remarkable nonlinearity. In this paper we propose a new method, using a Liapunov function, which both asymptotically stabilize the nonlinear system and minimizes a certain cost function. This method is considered to be an extension of the optimal regulator which uses a quadratic cost function. We confirm the validity of this method by applying it to the transient stability control of a generator in a power system.
Though the H∞ control has been studied vigorously, the existing H∞ control theory is somewhat complex so that comparison of its control effect with other conventional controls is difficult. Therefore, in this paper, we formulate some state feedback controls using H∞ norm criterion and give suboptimal solutions using only bounded real lemma. The results show that robust control has superior stability margin to the LQ optimal control, for example.
The purpose of this paper is to investigate the invariant domain with fractal structures of a class of one-dimensional nonlinear systems described by difference equations. The invariant domain treated here is the set of initial conditions satisfying a condition that the state variable is included in the normalized interval] 0, 1 [for any time. Under the condition that the nonlinearity of the system is described by the unimodal nonlinear function, the mechanism yielding fractal invariant domains and conditions for its existence are clarified by using the notation of the symbolic dynamics. First, it is shown that the fractal structure of the invariant domain is generated by the inverse mapping of the nonlinear function. Furthermore, the mechanism yielding the fractal invariant domain is classified to the five different mechanisms by using the notation of the symbolic dynamics. One of them is similar to that yielding the Cantor centor 1/3 set. The other four types of mechanisms are newly demonstrated by theorems of this paper.