In the present paper, we propose a computationally efficient order test which first determines the process order and then the noise order. The process order is determined comparing the instrumental residuals derived from the input instrumental product moment matrix. The noise order is evaluated by the statistics yielded from the cross-covariances of the observed output and the composite noise which is a linear combination of the noise with the coefficient of the process parameter.
The noise covariances may be the useful information for determination of the noise order. Under the situation required for the process identification, however, the estimates of the noise covariances can not give the proper order because of the poor accuracy of them, even applying them to the order test. We, therefore, take advantage of the cross-covariance estimates of the composite noise and observed output since they have good accuracy for the purpose. We calculate the order test statistics from them by the recursive algorithm and determine the order.

View full abstract

In recent years, an adaptive control problem which has the unknown parameters and states has been actively examined, and many approaches have proposed recently. But most of those studies are done without any limitation of input amplitude.
In this paper, we deal with the problem of constructing an adaptive control system for a single-input single-output time-invariant discrete system with the limitation of input amplitude. In order to able to discuss this as linear problem for the plant with an amplitude limitation, we introduce a compensator called the dispersion filter, and the structure and its feature are explained in detail. Based on the prior information of the dispersion filter, two adaptive control schemes are proposed. At the end of this paper, the results of the simulation of a third-order unstable system are included in order to show the effectiveness of the proposed schemes.

View full abstract

It has been known that the high-gain feedback is very effective to reduce the parameter sencitivity, the effect of disturbances and so on. But the analyses of high-gain feedback systems are numerically stiff because of different magnitudes of their coefficients. In this study we remove this stiffness with singular perturbation method.
We consider regulators with high-gain feedback. In general it is not clear that we can get a highly accurate regulator when a scalar gain tends to infinity. Moreover the systems are sometimes unstable and even if the systems are stable, the responses often have remarkable peeks. Thus it is important to decide a class of systems such that a high-gain feedback is permitted.
First we show the conditions of gain matrices under which the regulator is stable and has arbitrary accuracy. Next we get a class of systems for which we can design the highly accurate regulator. Such systems are following:
i) A dimension of input is greater than or equals to that of output.
ii) They have minimum phase.
iii) CB has full rank.
In addition authors present the procedure to design a highly accurate regulator by showing the algorithm for HRP (High-gain Regulator Problem) and demonstrate it for a example.

View full abstract

A new method for measuring the center position and the total intensity of a two-dimensional distribution is proposed and the basic experimental results for its verification are shown.
The two-dimensional distribution of a physical values such as the optical image and the pressure are usually detected with an array of matrix positioned sensors. The proposed method deals with the outputs of these sensors not by scanning operations but by a parallel processing circuit without a computer. The circuit has an unique stable equilibrium state. In the state, using a discrete version of Green's theorem, the center position and the total intensity are obtained by the simple operation of the edge data in the circuit.
As the circuit has a repeating simple structure, it can make into LSI and smart sensors. By level slice or filtering operations, this method can be used to measuring various characteristic scalar values about the distribution, and applied to automated visual inspection and tactile or visual sensors of robots.

View full abstract

This paper describes a function of a splitter on flow mechanism of a switching of jet in a wall-attachment fluid amplifier.
A very simplified device was used as an experimental model and the switching mechanism in it was studied by visualizing the flow using a motion picture and by measuring pressure distributions on walls.
The mechanism of the interaction between jet and splitter was analyzed. An attachment distance, an amount of switching flow, and switching time period in the switching process were calculated from the simulation model derived from experiments. It was found that both the initial attachment distance and an amount of switching flow were varied with splitter position, since the low pressure distribution nearby the tip of the splitter eased the jet curvature. The switching of jet was restrained due to the low pressure distribution caused by the separation of streamline at the tip of the splitter in the output duct.

View full abstract

In the internal pipe wall of magnetic flowmeters should be nonconductive to prevent generated electromotive force from short circuiting. Usually inside of metalic pipe is lined by insulating material. The lining material limits applicable temperature range of measured fluid and also reliability of meter. The objectives of this investigation are to eliminate the lining and to improve reliability and performance of magnetic flowmeter.
A potential distribution is formed on the pipe wall by applying electric current proportional to fluid flowrate. If the potential distribution on the wall is kept almost identical to the flow induced potential in the liquid, then no current flows across the boundary between fluid and wall.
The principle of the proposed magnetic flowmeter is to realize a equivalent boundary condition between fluid and metalic pipe wall. The boundary condition is kept equivalent to that of nonconducting pipe wall automatically by the use of electronic servo techniques. Therefore the output signal is exactly same as that of conventional magnetic flowmeters. An experimental model is built and tested using stainless steel pipe (inner diameter=27mm, outer diameter=30mm).
According to the test results of tap water flow measurement, the output is reduced to about 60% of theoretical value without applied potential, however, it can be increased to 99% of theoretical one with wall potential applied. The theoretical value is calculated from measured value of flux density, flow velocity, and pipe diameter.
The output current of the servo amplifier is proportional to flowrate and is about 2 Ampere at flow velocity=0.8m/s for this model.
Theoretical analysis and test results indicate that flowmeter output of the model is highly dependent on liquid conductivity without wall potential. However, no effect of liquid conductivity is observed if the potential is applied. These results suggest that the proposed wall potential distribution is very effective.
In this paper theoretical analysis and experimental results are described.

View full abstract

The possibility of automatic melody recognition for testing musical boxes in a production line is studied. The inspection and classification of musical boxes in production lines has been done by trained persons. They hear the sound, and check and classify the music number within about 2 seconds. To replace their work by an automated system, there exist some problems to be solved. Comb teeth in a musical box vibrate simultaneousely producing musical chords and the sound waves of the box are composed of the superposition of sinusoidal damped oscillations which differ from the speech sound wave. The speed of the music playing is different for each product and the speed also changes during one musical performance because of unwinding of the spring. In these circumstances a real-time processing is needed in the line. These features must be taken into account in analyzing the sound wave. The system we designed is under the control of a microcomputer and it includes an FFT real-time spectrum analyzer and a pulse counter which measures the rotation angle of the drum to use it as the clock pulse for data acquisition. An algorithm for rejecting noise is built in the system. The experiments for encoding and recognition of music melody from the box were performed. In the experiment, of 176 product-samples (18 melodies), 10 samples of inferior quality were all correctly rejected, and only 4.5% of acceptable products were rejected. A melody recognition rate of 100% was obtained. Since the experiments were in a noisy lab., the algorithm is believed to be effective in the factory as well.

View full abstract

The adaptive blood glucose control system is presented to compensate in an on-line manner individual differences and parameter variations in the blood glucose regulation of diabetic subjects The glucose regulation behavior is described on a minute-by-minute basis as a linearized model with insulin and glucose control inputs and a blood glucose output. The blood glucose level is controlled to track a desired profile specified by a physician. The algorithm includes two kinds of controllers: a self-tuning regulator to adaptively perform a model-matching to the reference model, and a feedback controller to minimize the output error between the actual glucose level and the desired one, even in the presence of the modelling error as well as disturbances. It is clarified in numerical simulations that the control algorithm can work very efficiently in the case with imperfect knowledge on the dynamics of diabetics, compared with a conventional control scheme with fixed parameters. The efficiency is also examined in animal experiments using dogs.

View full abstract

Incubator oxygen treatments have been used to prevent newborn respiratory distress. When the oxygen concentration of inhalation (FiO₂) is fixed at a rather high level to improve the distress, the newborn arterial oxygen blood pressure (PaO₂) sometimes becomes extremely high to bring about the retrolental fibroplasia. This paper investigates the feedback control schemes in an optimal FiO₂ inhalation system to maintain PaO₂ within normal range. The ventilation behavior of a newborn is shown to be described approximately by a first-order linear model. By modelling the air diffusion in an incubator, as the overall system we have the second-order linear model with dead time lying in a tcPO₂ sensor which can monitor PaO₂ transcutaneously. The optimal control schemes with the integral action can be given in terms of only the input (FiO₂)-output (tcPO₂) data sequence. The algorithms are examined through animal experiments in which we simulate anoxia dogs by keeping them in various low ventilated conditions. It is clarified that PaO₂ can be maintained within the range ±1.33kPa even if the ventilation is reduced to the one third of the normal rate.

View full abstract

In this paper, fundamental characteristics of redundancy in motion degrees of freedom of articulated robot arms are made clear and principles for utilizing the redundancy are investigated.
Analyzing small motions around specified reference conditions for robot arms with articulated joints of general construction, the redundancy of the articulated robot arms is defined as degree of freedom for determining joint angles corresponding to certain specified operation at those conditions. The redundancy is also investigated by using null spaces and ranges of Jacobian matrices which are defined for the robot arm motion. Then, we formulate the utilization of the redundancy based upon the newly introduced concept “tasks with priority”. Finally, we show experimental results of the trajectory tracking of an articulated robot arm with provisions for an obstacle and discuss the effectiveness of the utilization of the redundancy.

View full abstract