Toshitaka Oki

THE SWING UP CONTROL FOR THE PENDUBOT BASED ON ENERGY CONTROL APPROACH

Abstract This paper studies the energy based control of an underactuated two-link robot called the Pendubot. After having investigated the characteristics of the closed-loop system with the energy based control law (Fantoni et al. , 2000) for swinging the Pendubot up, this paper proposes a sufficient condition about parameters in the control law such that the total energy of the Pendubot will converge to the potential energy of its top upright position. This paper gives an answer to the unsolved issue in (Fantoni et al. , 2000) whether the total energy of the Pendubot will converge to the potential energy of its top upright position. Moreover, with the aid of the proposed condition, the parameters in the control law are easy to be chosen.

A design of neural-net based predictive PID controllers

PID control schemes have been widely used for various real control systems. However, in practice, since it is difficult to find suitable PID gains, a lot of research has been reported with respect to tuning schemes of PID gains. In this paper, a design scheme of neural net-based PID controllers is proposed for nonlinear systems. The modeling error between a linear nominal model and a controlled object is compensated by a multilayered neural network whose outputs are given as additive PID gains. Furthermore, the fixed PID gains are calculated for the linear nominal model based on a generalized predictive control scheme.

Intelligent tuning PID controllers

Recently, neural network techniques have been widely used in adaptive and learning control schemes for nonlinear systems. However, generally, it costs a lot of time for learning in the case applied in control systems. Furthermore, the physical meaning of neural networks constructed as a result, is not obvious. In this paper, a design method of intelligent tuning PID controllers is proposed, which has a fusional structure of self-tuning and neural network schemes. This method enables us to understand a physical meaning of the control parameters, and also to adjust PID gains quickly.

An intelligent PID controller with a neural supervisor

It is difficult to control nonlinear systems using a PID controller with fixed gains. Many types of PID controllers with variable gains have been proposed based on advanced control techniques. In this paper, an advanced PID controller for nonlinear systems is proposed. Since the nonlinear input/output characteristics can be explained using a set of linear input/output characteristics, we consider that a set of linear system self-tuning PID controllers are used through switching based on signals from the control system. The switching is carried by a neural network based supervisor. This supervisor selects the linearized model which is most accurate for the next step. The effectiveness of this method is investigated through two numerical simulations.

Discrete‐time advanced PID control systems for unknown time delay systems and their applications

PID control schemes have been used widely in most process control systems represented by chemical processes for a long time. Indeed, they are easy for most engineers in many industries to understand and for the use of various systems, whose properties are obscure. However, conventional PID control schemes have not been so available for the systems whose system parameters and time delays are unknown. The purpose of this paper is to consider advanced PID control schemes in discrete-time systems with unknown time delays. Two control schemes, that is, compound PID control systems with a local feedback compensator and a pre-compensator are proposed, and they are extended to explicit-type self-tuning PID control schemes. Furthermore, these methods are applied to a pressure-control system to show their effectiveness.

A skill-based PID controller using artificial neural networks

Recently, there have been many papers on skill-based controllers in which human skills are represented by artificial neural networks in the literature. This technique for representing the human skills plays an important role in various industries, for example, in manufacturing systems and process systems. A skill-based PID control scheme is proposed, which represents the skills of human experts as PID gains. This controller is designed using three-layered artificial neural networks. The effectiveness of the proposed skill-based PID control scheme is investigated in an application to a pressure control system. The characteristics of the skill-based PID controller are analysed in the frequency domain.

A Self-Tuning PID Controller Fused Artificial Neural Networks

Abstract Recently, neural network techniques have widely used in adaptive and learning control schemes for nonlinear systems. However, generally, it costs a lot of time for learning in the case applied in control systems. Furthermore, the physical meaning of neural networks constructed as a result, is not obvious. In this paper, a design method of self-tuningPID controllers is proposed, which has a fusional structure of self-tuning and neural network schemes. This method enables us to understand a physical meaning of the control parameters, and also to adjust PID gains quickly.

A design of non-linear electro-pneumatic servo systems by using neural-net based linearizer with off-line learning algorithm

Many electro-pneumatic servo systems have strong nonlinear property because of compressible air in cylinders. To control their systems, a neural network(NN) based linearizer with off line learning is proposed. The augmented system which consists of the controlled object and the linearizer is controlled by a linear controller. The point of the learning algorithm of the NN is to estimate from the actual input of the controlled object to the input of the augmented object directly.

A learning algorithm for a neural network in a linearlizer for nonlinear systems

The purpose of this study is to give a design method of a linearlizer by using a neural network (NN) with off-line learning algorithm. This linearlizer works so that the input-output property of the augmented system which consists of the system and the NN may be equivalently equal to that of the linear model. The learning of the NN is performed off-line by using the input and output data of the system. Finally, a numerical simulation is demonstrated to illustrate how to use it in the control problem.

A Design of Generalized Minimum Variance Controllers for Nonlinear Systems

Abstract The generalized minimum variance control (GMVC) method was proposed by Clarke, and it was developed by Allidina taking into account of the closed-loop properties. However, these methods could not be applied to nonlinear systems. In this paper, a new design method of GMVC for nonlinear systems is proposed. This method is a kind of the hybrid controllers by using linear controllers and neural networks (NN). The GMVC law is derived for the known linear model of the controlled object, and the NN learns to compensate the modeling error. Finally, a numerical simulation example is illustrated in order to show the effectiveness of the proposed method.