A new spatial filtering method is proposed for a narrow band elimination of spatial frequency spectrum. The spatial filter is an array of solar cells which consists of same sized parallel slits manufactured by lithography technique on a single silicon wafer.
Output signals from each solar cell are electrically weighted through the external do amplifiers to realize the weighting distribution so that the weighting value of the center-cell is equal to the sum of the weighting values of the other cells. This spatial weighting gives a characteristics of a narrow band spectrum elimination.
It is expected to apply this spatial filter to the defect detection of knitted fabrics which originally have a narrow spatial frequency spectrum yielded by the periodic knitting. To examine the effectiveness of this spatial filtering method, the experiments are performed to detect some typical defects occurred in the knitted fabrics. And the designing parameters are discussed for the total number of slits, the pitch and the width of slits to realize the better filtering characteristics matched to the individual defect spectrum.
As the results, it is found out that this narrow band elimination spatial filtering method can be applicable to the on-line defect detection of knitted fabrics, e. g. underwears and stockings.

View full abstract

This paper presents new pointing device based on the principle of supersonic position measurement by phase difference. Any kind of motor action by users are available to pointing by the device, and it allows computer users various pointing styles including “mouse-like” one.
The principle and the formulae of supersonic position measurement are introduced. Trial construction of the pointing system is described. And performance of the pointing system is investigated using test system constructed with X-Y plotter. Finally some seamples of pointing devices with various pointing styles are realized, and experiments on the efficiency of pointing tasks with computer users are carried out to validate that the pointing system is suitable for practical use.

View full abstract

We propose a new method for a clamp-on ultrasonic flowmeter.
A conventional clamp-on ultrasonic flowmeter has a problem that the determination of transit path of ultrasonic waves is complex, because there is a pipe wall on the transit path. In this paper, this problem is studied by a spatial frequency analysis. It is found that the transit path depends on the conditions of incident waves; frequency, incident angle and beam width. It is also found that especially plate mode excitation affects the transit path seriously, and that a wide beam would be very effective in reducing the effect due to the plate mode excitation.
According to the above mentioned result, we propose a new idea of “wide beam model” for determining the transit path, and a new method for an ultrasonic flowmeter employing a wide beam. Some experiments are carried out to confirm that the transit path can be estimated exactly by this new model. The wide beam ultrasonic flowmeter provides a measurement with high accuracy, even if temperature changes in transit path would occur. This new flowmeter can be applied to any fluids whose sound velocities are unknown.

View full abstract

This paper deals with the frequency domain analysis and design of the differential gas pressure measuring systems with compact transducer. The system model is based on the Brown's line model and adiabatic cavity model without the effect of diaphragm displacement. From this model and experiments, the following results are obtained; 1) The left line-cavity subsystem is independent of the right subsystem. 2) The relation between the gain of the left subsystem at ω=ω_1L, |G(jω_1L)|, and the non-dimensional parameter Ω_1L(=a_1L√ω_1L/ν) is obtained, where a_1L is the radius of line, ω_1L the fundamental natural angular frequency and ν the kinematic viscosity. 3) When ω_1L=ω_1R and |G(jω_1L)|=|G(jω_1R)|=0.707(Ω_1L=Ω_1R=2.44), the differential gas pressure can be measured at the frequency bounds 0≤ω<0.6ω_1L.

View full abstract

This paper is concerned with the design of controllers which achieve both the given H₂ control performance and the robust regulation in the presence of a class of disturbances. A necessary and sufficient condition for the existence of such a controller is given by exploiting the free parameters of the H₂ suboptimal controllers. In addition, some simulation results are shown to demonstrate the validity of the proposed controller.

View full abstract

The time-invariant LQ (linear-quadratic) optimal regulator problem is studied from a view point of learning control based on a gradient method, and fast convergent learning is discussed.
A key relation is that some important sensitivity functions related to the performance function of the unknown system are given by inner products of response signals. It is also shown that the matrix Riccati differense equation is rewritten by the sensitivity functions.
A fast convergent learning algorithm to optimize the state feedback gains for the unknown system is derived on the bases of these relations. Its convergent condition and the convergent speed are given by those of the Riccati difference equation.
From a numerical simulation for a discrete-time second order system, the system is stabilized typically within several steps of the system motion, and the gain is nearly optimized within 20∼40 steps even if the original system is unstable.

View full abstract

In many conventional studies on the geometric structure of the space of linear systems, only the transfer functions of systems have been discussed. However, for example, if a plant has a specific physical structure with perturbed parameters, then the approximation problem must be analyzed with the geometry on the space of the realizations. In this paper, we analyze the geometric structures of the space of linear systems which depend on their realizations. We define a Riemannian metric on the manifolds of Gramians, in order to measure the difference of the effects from the inputs to the state variables and from the state variables to the outputs between two systems. The Riemannian metric is equal to the inverse of Fisher information matrix and it represents the degree of distinction between the parameters of two systems. Moreover we introduce two connections on the manifolds and two measures corresponding to these connections, that is, Riemmanian distance and Kullback divergence.
To analyze the structures of the geometries, we present the realizations of two systems described by transfer functions, which attain the minimum values of the distance and the divergence between them. This is a main subject of this paper. The minimum values are derived from their Hankel singular values and the balanced realization is one of the realizations which attain the minimum values. Moreover the space of pairs of controllability Gramians and observability Gramians has a parallel structure.

View full abstract

In this paper, high speed synchronization of two axes positioning systems under adaptive feedforward control is considered. The adaptive feedforward control system for one axis consists of a proportional feedback controller and an adaptive feedforward controller. To achieve high speed synchronization between the two positioning systems, a coupling controller, which responds to the synchronization error, is introduced. The synchronization error and the tracking errors are used as the adaptation error signals in the two adaptive feedforward controllers. When a disturbance input is applied to one axis, the tracking errors appear in the undisturbed axis as well as in the disturbed axis. This allows a quick removal of the synchronization error. The stability of the proposed control system is analyzed and the effectiveness of the system is demonstrated by experiment.

View full abstract

Recently there has been an increased demand for large-capacity magnetic hard disk drives. To meet this demand, a fast and precise head positioning servo system is required. A mode switching control (MSC) system, which includes several different servo structures and a switching function, is widely used to meet this requirement. In an MSC system, each servo system can be optimally designed to satisfy a certain performance index. On the other hand, a design method for the switching function has not yet been perfected.
In this paper, the initial value compensation method has been applied to this problem. When the servo mode is switched from one mode to another, certain initial values are input for the state variables of the compensator in the new servo mode. Two approaches to determining the initial values are proposed. One is to minimize a linear quadratic function of state variables, which is called the J-min method in this paper, and the other is to cancel the unfavorable poles of the closed loop with zeros, which is called the zero-assign method.
In actual systems, several filters are used to compensate for mechanical resonance gain peaks. Since applying the initial value compensation method to the filters is fairly complicated, a 1st-order approximate time delay model can be applied to represent the filters. The initial values in the compensation method are chosen for this. Several simulation results show that this method, especially the zero-assign approach, is an effective way of settling the head swiftly, and provides robustness for velocity fluctuations during mode switching.

View full abstract

Many studies have been undertaken in order to apply both the flexibility and learning ability of neural networks to control systems. With regard to servo control, an adaptive type controller and an learning type controller are proposed, and verified by applications to a force control system. An adaptive type identifier has also been studied. The convergence speed of this type of identifier is fast because learning is performed within one trial. In contrast, although the convergence speed of the learning type identifier is slow because its learning requires after several trials, it offers the possibilities of greater robustness when the plant dynamics are unknown and there is external noise. Jordan has stuided several types of identifiers using neural networks. However, learning type identifiers for discrete systems have not been studied yet. They should be studied because of the advantages to the offer of software servo control.
Therefore, this paper proposes a practical design method for both a learning type direct transfer identifier and an inverse transfer function identifier for discrete systems. The direct transfer function identifier constructs the forward dynamics of the object plant in the neural network. The inverse transfer function identifier constructs the inverse dynamics of the object plant in the neural network. Analytical approach describes the local stability condition of the proposed identifiers when both the object plant and the neural networks are linear. Simulation results for a second order plant confirm both the characteristics, and the nonlinear saturated function (sigmoid function) effect.

View full abstract

This work presents a robust control methodology for hybrid position/force control. It develops a method for the design of controllers of constrained manipulators in the presence of model uncertainties. The control input of this methodology consists of a nonlinear and a linear part. The nonlinear control part gives a new coordinate system which can be defined for any constraints. By this coordinate system robot dynamics is decoupled and input spaces are devided into two orthogonal spaces of position and force control. Position control gives desired velocity of the end-effector through the constraints and force control gives desired reaction force on the constraints. The linear control part improves robustness properties of the proposed nonlinear hybrid control scheme. According to the nonlinear part the linear part is devided into two parts of position and force control. Model feedback control scheme is applied to the linear part of position control and integral control is used in the linear part of force control in order to suppress tracking errors due to model uncertainties. The robustness of the proposed hybrid control scheme is demonstrated through simulation of a constrained 2 degrees of freedom manipulator.

View full abstract

In this paper, we develop a method to plan a suboptimal trajectories for cooperative multiple manipulators which transfer an object along a specified path. The dynamic equations of manipulators and the object are expressed by a path parameter‘s’which does not depend on time explicitly. A quantity λ(s) which connects the time‘t’and the path parameter‘s’is introduced. A suboptimal λ(s) is determined by B-spline approximation considering the boundary conditions of both end points and the driving force/torque constraint of each joint of each manipulator. The proposed method is applied to two-manipulator system. Numerical results show the effectiveness of the method.

View full abstract

This paper discusses a self-tuning method of fuzzy controllers. The self-tuning procedure consists of two parts. One is a real time tuning by δ-rule which is a basic learning law in neural networks. The δ-rule of neural network is used for minimizing integral of squared error between a reference signal and output of controlled object. The other is an iterative adjustment based on evaluation of transient characteristics using compensation rules. The compensation rules are derived from the viewpoint of effectively improving frequency characteristics, that is, gain and phase characteristics, of control system. We point out from simulation results for several controlled objects that two self-tuning mechanisms are useful and that the self-tuning fuzzy controller constructed by combining these self-tuning mechanisms is more effective.

View full abstract

In order to reduce labor and careless mistakes in plant control operations, supervisory control systems using expert system technologies have been developed. These control systems can handle foreseen situations, however, due to a lack of the appropriate control rules, they cannot handle unforeseen situations. Up until now, such situations have had to be handled by skilled operators. To overcome such unforeseen situations, the authors have developed model-based reasoning technologies, such as diagnosis with qualitative causal model, knowledge compiler generating control knowledge from plant models (structure, function and general rules) and fuzzy qualitative reasoning. This paper proposes an overall system concept and architecture which integrates these technologies. This system was developed by first observing how skilled operators handle unforeseen situations, and then emulating such actions. This paper highlights the importance of predicting and monitoring the results of applying generated control rules to the plant, and re-generating new control rules in the case that the unforeseen situation could not be restored to a normal situation. The proposed architecture has been successfully applied to an experimental thermal power plant control system.

View full abstract

We developed a production scheduling system for batch plants. This system supports human intelligent works by using the auto scheduling on the system and the manual modification or the rescheduling cooperatively.
The scheduling system which works well in various situations at factories must have some flexibilities as follows. 1) A scheduling system has a frame to reflect user requests easily. 2) A scheduling system has a function to modify scheduled results to satisfy user's expectation.
They are realized as follows. 1) We add a customize function to incorporate a individual plant's know-hows into the auto scheduling strategy. 2) We develop the flexible user interface to cooperate with user and the function of manual modification.
In this paper, the problems for practical use are summarized and the effects of this system are demonstrated by the field test in a chemical plant.

View full abstract

The calculated control input can diverge because of the saturation of plant input even if the control system is internally stable in the linear sense. In this paper a controller configuration that avoids this divergence is proposed.

View full abstract

This paper presents a neural network system for chromatic adaptation which is one of the functions of human color vision. It is proved that the learned hidden units show almost the same response as human color vision does.

View full abstract