A laser displacement meter based on the principle of optical triangulation is applied to measuring distance or displacement to an object and its shape. The dispersion of the obtained data is caused by the affection of measuring surface. The three dimensional shape measurement system was assembled by using the laser displacement meter and computer controlled scanning machine CDC-3DN. In this paper we consider the on-line measurement with 5μm resolution for mounting IC chips on printed circuit boards. The moving average of the measurement data is calculated to delete their dispersion in the computer. And the CG (Computer Graphics) drawings of the obtained date are made by a software program Mathematica. The measurement results show the possibilities to test the shape of IC pins and to control the shape of wiring solder on boards.

View full abstract

High-resolution heterodyne Michelson interferometer has been built using the intermode beat of a two-mode stabilized He-Ne laser. To facilitate the precision phase measurement, the high-frequency beat signal of 724MHz is down-converted to the intermediate frequency by means of single sideband suppressed carrier modulation technique. The digital phasemeter based on sampling the saw-tooth wave of the intermediate frequency provides nanometer resolution. In addition, periodic nonlinearity error caused by leakage of the polarization components in the interferometer optics is evaluated.

View full abstract

This paper is concerned with the optimization of observations with feedback. In the information theory, the problem is also discussed as that of optimal transmission of a stochastic signal through a noisy channel with feedback. For Gaussian signal and noise, it is well-known that the optimal feedback signal is the least-squares estimate of the signal, and the observation is optimal when the estimation error multiplied by the largest admissible gain is observed with the additive noise. In this paper, we show that this well-known result for Gaussian signal and noise is still vaild for Non-Gaussian signal and noise which are a square-integrable process and a square-integrable martingle, respectively.

View full abstract

The problem of robust output tracking of uncertain single-input and single-output nonlinear systems is considered. Based on Lyapunov methods in conjunction with input/output linearization approach, a new approach to designing a class of continuous and bounded state feedback controllers using the possible bound of the uncertainty for uncertain nonlinear systems is proposed. It is shown that the state feedback controller developed in this paper can guarantee exponential stability instead of uniform ultimate boundedness for the trajectory error system of nonlinear systems with uncertainties. Furthermore, the proposed controller can guarantee the existence of the solution to the dynamical system in usual sense. Therefore, since we can also easily select the control parameters, the method can be expected to have good transient response. Finally, a numerical example is presented to demonstrate the effectiveness of the approach developed in this paper.

View full abstract

Recently, for linear ω-periodic discrete-time systems, authors proposed a new control method which is based on infrequently observation and a time-varying feedback, that is called Sampled Incomplete State Periodic Hold Control (SISPHC). It includes Sampled State Periodic Hold Control (SSPHC) as a special case. When using SSPHC, it is possible to assign the closed-loop transition matrix over n ω steps arbitrarily, where n is the dimension of a state vector, if and only if an open-loop system is reachable. But, when using SISPHC, it is impossible to assign the transition matrix. This paper proposes a new method for assigning the transition matrix by using observers. Necessary and sufficient conditions for assigning the transition matrix of the closed-loop system and simple procedures to calculate the observer gain are given. Simple numerical examples are also given.

View full abstract

This paper proposes a design method of gain scheduling controllers for descriptor systems whose A and E matrices may share the common scheduling parameters. By using parameter dependent Lyapunov functions which include E matrix explicitly, it is shown that the controller design problem can be reduced to LMI (Linear Matrix Inequality) problem, and the gain scheduling controllers are parameterized. The effectiveness of the proposed method is shown via a numerical example.

View full abstract

An algorithm is proposed to solve the linear quadratic optimal control problem subject to terminal state constraints, state and control constraints and interior point constraints. The algorithm is based on parameterizing the system state variables using a finite length Chebyshev series of unknown parameters, and the control variables are obtained as a function of the approximated state variables such that the system differential equations are satisfied. After reformulating the problem using the proposed algorithm, the constrained optimal control problem is converted into a quadratic programming problem which can be solved easily. Moreover, by using the proposed algorithm there is no need to integrate the system or costate differential equations.

View full abstract

In this paper, we propose an improved approximation scheme of the solution of the Hamilton-Jacobi inequalities, which fully exploits the conventional approximation methods. We also show several numerical examples including the practical case of stabilization of an inverted pendulum to verify the effectiveness of our approach.

View full abstract

An optimal synchronous path control system for industrial mechatronic servo systems is proposed in this paper. The synchronous path control system is improved from the synchronous positoning control system which has alrealy been proposed. In this method, the synchronous ratio is included in state variables and it is equivalent to perform the coordinate transformation of each axis so as to contain the unity synchronous ratio. Then, the synchronous positoning control for variable synchronous ratios can be realized by using the same control parameters. The proposed control system is always stable except only the changing point, and the synchronous path control is realized by decreasing the synchronous error for the desired path which is linear interpolated in three dimensional space. It is applied to a linear X-Y table and its effectiveness is confirmed in experimental studies.

View full abstract

In this paper, we focus on the property that the system remains stable even if some failures of the pre-specified actuators (or sensors) occur. This property is defined as “l-partial integrity, ” where l indicates the maximum number of tolerable failures and “partial” means only pre-specified actuators (sensors) have possibility to fail. We derive a sufficient condition for l-partial integrity from the scaled small-gain theorem, and reduce the condition to linear matrix inequalities(LMIs) in the state feedback case. This LMI condition is useful to design control systems that achieve not only this property but also other performances. A numerical example is also provided to show its usefulness.

View full abstract

This paper presents an approach to design robust nonlinear H_∞ controller without any solution of Hamilton-Jacobi inequality or equations. The uncertainty is assumed to be described by nonlinear function on state variable, which is unknown but gain bounded. It is shown that if the zero dynamics of the given system is stable or stabilizable, then the Lyapunov function ensuring robust stability of the overall system and the state feedback controller with changes of the coordinate can be designed by recursive algorithm based on the Lyapunov function of zero dynamics.

View full abstract

It is well known that the ILQ (Inverse Linear Quadratic) method is a powerful tool for continuous-time control system design. The paper generalizes the existing polynomial ILQ method to the design of discrete-time regulator, and then studies its performance from a sampled-data point of view. First, given a discretized system, a state feedback gain is designed which places part of the closed-loop poles exactly at specified points inside the unit circle, and is LQ optimal for some weightings at the same time. This is achieved by placing the rest of the poles sufficiently close to the origin, thereby satisfying a modified circle criterion, a solution to the inverse problem of discrete-time LQ control. Secondly, it is checked whether thus obtained (pure) discrete-time feedback is again optimal as a sampled-data system; i.e., whether it minimizes some continuous-time performance index.

View full abstract

A model reference adaptive control system (MRACS) based on dynamic certainty equivalence (DyCE) principle taking into consideration realization of high order tuner (HOT) is proposed. In conventional scheme using DyCE principle and HOT, the knowledge of upper bound of high frequency gain of controlled object was needed for stable MRACS. However, it is generally difficult to get exactly such a knowledge a priori and its conservative estimate often leads to numerical instability in realizing HOT. In this paper, the problem is solved by introducing a projection type of HOT and an adaptive control law with division. Stability proof of the proposed MRACS is given and its validity is verified by simple numerical simulations.

View full abstract

In this paper, we propose a design methodology for multivariable time-delay control systems on the basis of the generalized stabilizing controller, which satisfies the 2-disk mixed sensitivity specifications. The key point is that, the existence of a solution depends on the existence of a stable rational polynomial function with respect to a free parameter.The robustness problem reduces to adjusting the parameter. The design of the generalized controller involves inner transformation of the strictly proper system and a sensitivity compensator. Finally, we present some examples and show that the methodology is easy to handle with and of pratical use.

View full abstract

In this paper, a self-organizing Fuzzy Neural Network is proposed to control chemical feeding, which is one of the most important problems in water purification process. In the case of the learning according to raw water quality, the self-organizing Fuzzy Neural Network, which can be driven by plant operator, is very effective. For this purpose, we have contrived the dynamical tuning method of learning rate and the setting method of initial membership function of premise part using FCM algorithm, therefore propose parameter-free Fuzzy Neural Network. As compared to standard backpropagation, the convergence rate and learning stability are improved. Simulation results of the proposed method using the data of water purification plant show good performance. This algorithm is included to chemical feeder, which is composed of PLC, magnetic flow-meter and control valve, so the intelligent control of chemical feeding is realized.

View full abstract

Abduction is the procedure to derive a set of hypotheses that explains a set of observed events under given knowledge. The derived set of hypotheses is called an explanation. We have already proposed a theory of fuzzy abduction that was an expansion of the abduction with fuzzy theory, and shown the way to derive fuzzy explanations. In the theory, observed events and hypotheses are expressed in fuzzy sets, and a set of implications with a truth value between zero and one is used as the given knowledge. However, it is not guaranteed that fuzzy explanations always exist. In this paper, we specify the necessary and sufficient conditions of existence of fuzzy explanations and propose a method to obtain approximate explanations when the conditions are not satisfied. Fuzzy abduction is a procedure similar to the inverse problem of fuzzy relational equations. However, unlike methods proposed so far in the inverse problem of fuzzy relational equations, our method has a merit that it does not need iterative calculation to obtain approximate solutions.

View full abstract

This paper considers an adaptation of rhythmic movements to changes in environments. Perturbed locomotion has been proposed as a new paradigm of motion learning. Here, we formulate this phenomenon with CPG model and give a rule for parameter changing in locomotion patterns. Our rule is derived from a criterion that less interaction is more natural in CPG oscillation. Then extending our approach, we discuss general features and necessary conditions for rhythmic motion and propose a mathematical framework for its adaptation.

View full abstract

Control systems using neural networks have been used recently in many fields, but some problems remain to be solved. One of the problems which should be overcome is to enhance the robustness of the neural networks control systems. In this paper, a new robust control method is proposed, which is based on the second order derivatives of Universal Learning Network and fuzzy criterion function.

View full abstract

This paper presents an evolutional algorithm for the development of a behavior formation of simulated autonomous mobile robot, that is assumed to be equipped with some range sensors. We discuss that the key for the emergence of skillful behaviors exists in a open-loop cycle consisting of a robot's perception, action, and behavior intervened by an environment. We model this cycle by introducing an evolutional computing method, which is applied to organize a robot's behavioral rules associating between a robot's perception and actions. Especially, for promoting a flexible behavior formation we stress on the importance of maintaining diversity of behavioral rules during the evolution. We consider about the preferred instruction strategies needed when an external human designer intervenes in a robot's behavior formation and makes it be guided in some way by indirect instruction.

View full abstract

This paper proposes a new method of constructing a repetitive control system by using “Parallel Integrators”. We derive a necessary and sufficient condition for stability of a closed loop system which has Parallel Integrators and verified the effectiveness of the proposed system by experiments.

View full abstract

Many classes of non-linear systems can be represented by Volterra kernels. In order to obtain the Volterra kernels, in this paper, the authors propose a new method for identifying of odd- and even-order Volterra kernels by use of M-sequences and correlation function.
Simulations are carried out for system which contains 1st, 2nd, and 3rd-order Volterra kernels. The results show a good agreement with the theoretical considerations.

View full abstract