Yoshitaka Togari

Two-valued PID controller

A discrete-time proportional integral derivative (PID) controller, the manipulated variable of which takes two values, i.e., ON and OFF, is proposed and analyzed. Oversampling technology is employed in the controller design. As to the structure of the controller, a cascade type and a built-in type are proposed. An experimental example is presented.

Design method of a quadratic performance index using a reference model

An algorithm is presented, describing how to choose the weighting matrix of a quadratic criterion function so that the dynamics of the step responses to the set point have the specified decoupled form. By employing a reference model, the performance of the controller can be pre-estimated more visually than by pole assignment. LQ-controllers are robust and not very sensitive to process variations. Therefore, the controller design procedure has the advantages of both optimal control and model matching.

Piecewise linear controller improving its own reliability

Abstract Although the capability of neural networks in nonlinear dynamics modelling is well-established, the reliability of the output heavily depends on the training data. The reliability is a serious problem in applying it to real problems. In this paper, we propose a radial basis functions network (RBFN) which evaluates its own reliability and improves itself recursively. This network approximates the input-output relationships with a piecewise linear regression. An adaptive internal model control algorithm in which the reliability of the model is used to tune the controller performance, is also proposed.

Anti-saturation discrete-time nonlinear controllers and observers

Abstract In this paper, a nonlinear controller design method based on a nonlinear state space model is proposed. By adding differentiable saturation elements, this controller design scheme can deal with the saturation of the manipulated variables. Linear approximation is successively executed at each sampling time. The velocity-type controller and observer are designed based on the linear model. For the calculation of linear approximation, the information of the states at the previous sampling time must be known. A nonlinear observer using integration of nonlinear state equations is adopted for the state estimation. The observer gain of the linear observer is applied to the nonlinear observer. This control system can stabilize open-loop unstable processes and non-minimum phase systems. The performance is demonstrated by its successful application to control problems of a CSTR, which has two stable points and one unstable steady-state point.

Abnormal situation correction based on controller reconfiguration

Abstract It is desirable to support operator activity in abnormal correction of chemical plants. Although fault diagnosis systems have very fertile grounds for theoretical and industrial development, multiple possibilities of faults are diagnosed. A countermeasure against one cause might be dangerous when used against another cause. A countermeasure might be effective to multiple causes. Countermeasure planning is necessary to complete fault diagnosis. An operator manipulates one variable while observing one variable in each countermeasure. It resembles the controller configuration of a single loop controller. An algorithm for controller reconfiguration is proposed for the first stage of countermeasure planning.

Model following controller with specified sensitivity function

A design method for a model-following servo control system whose output sensitivity function is specified with a reference model is presented. Using this method, the plant output responses to reference input change and the output sensitivity function can be specified separately and directly. >

Non-interacting Control System Design of a Distillation Column by the Partial Model Matching

Abstract A non-interacting dual composition control scheme is developed for a distillation column by using the partial model matching technique and it is realized with four conventional PI controllers. The partial model matching technique developed by Kitamori enables to design decoupler and main controller at once, with partial knowledge of the process to be controlled. Also, the features of the closed loop response are easily adjustable by choosing a suitable reference model. It is shown through simulation calculations that I-P non-interacting controller system designed by Kitamoris method is effective as a dual composition control of a distillation column.

MODEL VALIDATION TEST

Abstract This paper proposes a method to determine whether or not the real plant model is in a domain which is defined with a given nominal plant model and a given perturbation bound, when an input sequence and an output sequence of the plant are available. This procedure allows one to judge whether a perturbation bound which is estimated with a statistical method can be used as a deterministic one or not, and whether a tighter bound can be used for the design of a control system.

On evaluation of model error bound from experimental data

This work deals with a method to test whether the model error of a system is in a predetermined bound given with a weight function using frequency response data of the plant obtained by an experiment. The model error bound problem is converted to a stability problem of a set of systems and the stability test is performed.

MULTI-LOOP CONTROLLERS USING WAVELET PACKET DECOMPOSITION FOR FILM PROCESSES

Abstract In film processes, many actuators and scanning sensors are utilized to control film properties. Multi-loop controllers are applied to the pairs of actuators and scanned data, because they can be tuned without the precise process models. The selection of the measurement location corresponding to an actuator is not easy because the film is stretched and its edges are cut off. It is illustrated that inadequate pairing causes deterioration of the multi-loop controller performance. A method based on wavelet analysis is proposed to detect inadequate pairing and to prevent the deterioration of the controller performance.