Kharitonov's theorem appeared originally in the late seventies in a Russian literature but at the beginning so much attention was not paid to it at least in the western countries. Recently the potentiality of this far striking result in the analysis and synthesis of robust systems or systems with uncertainty has been gradually recognized and several research works using this theorem have begun to be reported. However, it seems that the original proof of the theorem might give no clear insight into why one could obtain such a marvelously simple result, especially to people who are in the engineering side.
Motivated by this point, the present paper makes an attempt to provide an interpretation of the theorem from a little different viewpoint. Discussions form an alternative way of proving the theorem. Moreover, these results are explained as well in terms of Mikhailov's stability criterion, a familiar tool to engineers for analyzing and designing control systems. This geometric interpretation makes it possible to grasp the central idea behind the theorem intuitively. The whole discussions of the present paper are expected to give some suggestions for developing further “Kharitonov-like” theorems for equations of other types of systems such as discrete systems, time-delay systems and so forth.

抄録全体を表示

Until now, an optimum finite preview control system utilizing finite future desired signals has been proposed in order to improve a control system performance. However, since the control system is constructed by only the design method for the optimum tracking control system, the robustness of the control system for the parameter variations of the controlled object and the disturbance signals is not acceptable for the practical applications. In this paper, new synthesis method for multiinput and multioutput robust preview control system is proposed. This method is based on the already proposed “error system method” which has been already applied for the design of the usual robust control system. After transforming the original state equations of the controlled object into the error system and defining a quadratic performance index, a tracking control problem consisting of the error system and the performance index is solved by usual optimal tracking control technique utilizing the priniciple of optimality. The designed preview control system consists of feedback part and feedforward part. The feedback loop consists of proportional actions and integral actions same as in the usual robust control system. Therefore, this preview control system is robust because of these proportional actions and integral actions for the parameter variations of the controlled object and the disturbance signals. Moreover, the compensating actions for the input delay time due to the processing time required to execute the employed control algorithm in the controller are included in the feedback loop. The feedforward loop represents the utilization of the future desired signal which is most important part in the control system. Some simulational results for this robust preview control system are presented in the paper for showing the effectiveness of the design method. As might be expected, the robustness of the control system for the parameter variations of the controlled object and the disturbances is satisfactory and it is shown that the utilization of the future desired signal is very effective to improve the control system performance. Extension of this method to the case of polynomial disturbance signals is easy.

抄録全体を表示

This paper investigates the necessity of the internal model principle for the so-called repetitive control systems. The paper demonstrates that the commonly accomodated “periodic function generator” which makes use of a puredelay operatox is necessary for realization of repetitive control systems. This is done via an effective use of the realization theory for pseudo-rational impulse responses developed by the first author. Then stabilizability of repetitive control systems is considered, and it is proved that the current restriction on the plant, namely the 0-relative degree property, is inevitable for the precise realization of repetitive control. The study is expected to offer a rigid ground for the study of repetitive control systems.

抄録全体を表示

There are many plants which should be assumed to be a linear time-varying system in order to improve their transient properties. The plant can be handled with the MRAC (Model Reference Adaptive Control) and the learning control, which decreases a control error through trials when the control can be applied repeatedly. However, the former requires a priori information about the time-varying structure of the plant and the latter requires a great part of the plant characteristics in order to design a stable learning control system. These requirements can be satisfied rarely in the practical situation.
We redesign MRAC from the view point of the learning control and call it MRLC (Model Reference Learning Control), which can control the discrete-time linear time-varying system without any knowledge except it's system order and time delay in the input. Since the input is generated through the feed back manner, the control system can cope with a change of the target responses between trials unlike the learning control where it is done through the feed forward manner. This paper presents a detail description and a stability analysis of the MRLC system. Numerical simulations are also performed for the verification of the analysis.

抄録全体を表示

This paper proposes a new method for knowledge acquisition, knowledge representation, generalization and reasoning in learning control. A learning controller to exploit an analogy between a current unsolved problem and a similar but previously solved problem to simplify its search for a solution is outlined. As an example, this controller is applied to contlol of multi-links systems and by simulation it is indicated that by using an analogy it can speed the search of input sequences to transfer initial states to goal states.

抄録全体を表示

This paper proposes a new “Coverage Monte Carlo” for evaluating a failure probability of a system described by a monotonic Boolean function. All the minimal cut sets are assumed to be known. It is first clarified that the Karp-Luby's coverage Monte Carlo evaluates effects of the second and the higher terms of an inclusion-exclusion formula. The first and the second terms, however, can often be calculated by ordinary, deterministic algorithms, and thus the new Monte Carlo is derived to evaluate only contributions of the third and the higher terms. The Monte Carlo estimate always falls in between the known upper (the first term) and lower (the first minus the second term) bounds. It is proven that the proposed Monte Carlo estimator has a smaller variance than the Karp-Luby estimator under reasonable assumptions. The estimator variance and the coefficient of variation can be evaluated before and after Monte Carlo executions. In contrast to the direct Monte Carlo, the coefficient of variation is proven to approach to zero as the system becomes more reliable and hence rare-event characteristics become more dominant. A numerical example is given.

抄録全体を表示

In this paper, we introduce a simple behavior of reliable system with replacement interpreted by Petri net and derived its probabilistic behavior by using Markov renewal processes. We considered two replacement polices which are preventive replacement and corrective replacement. We derived the mean firstpassage time, the expected number of visiting a certain state and limiting probabilites by applying Markov renewal processes. Finally we confirmed the effective of replacements for the system in numerical examples.

抄録全体を表示

Electron microscopes, which are indispensable to the research and development of materials, pose some problems in their focus adjustment. Detection of defocus has been so far carried out by visual inspection, a tedious process even for skilled operators. In particular, accurate determination of astigmatism depends on time-consuming photographic processes or 2-dimensional Fourier transform (2D-FT), making use of a fact that ellipses drawn by the spectrum power band of amorphous materials images represent the property of astigmatism.
This paper describes a fast method for detecting astigmatism, based on the steady randomness of amorphous materials. Because parameters of ellipses can be detected from spectra on at least three lines passing the origin of the frequency domain, there is no need to calculate the whole 2D-FT. The linear spectra are rapidly derived by the use of Radon transform, yielding the computation time of O(Nlog₂N) for N×N-pixel images contrasted against the 2D-FT method with O(N²log₂N).
The prototype employs eight lines on which data reduction from 128×128 pixel images into linear arrays are accomplished.
Test runs with 8086/8087 microcomputer demonstrate the computation time of 2s, compared with 40s by the 2D-FT method. The accuracy of detection is approximately ±10% for strength and ±10° for direction. The speed can be improved up to the order of 100ms by employing some circuitry and a signal processor.

抄録全体を表示

It is known that the stem diameter changes in short term according to water content in the stem and in long term according to growth of plant. It is tried to make the apparatus which can measure stem diameter without contact on it in arbitrary period in long term. As the growth of plant is usually measured by the sensing part that touchs on plant, it is influenced by touching pressure. Especially there is no literature about measuring a little young plant just after germination. This apparatus uses beam of a L.E.D. and a photo diode for detection of stem. When the L.E.D. and the photo diode move parallel at linear stroke, a stem acrosses the beam of L.E.D. and the stroke length covered with the stem is in proportion to length of the diameter of it. This apparatus can measure an stem diameter as minimam diametre of 0.3mm with 0.002mm resolution.

抄録全体を表示

Described in this paper, is a pneumatic logic device which functions as an exclusive-OR gate. Two similar sets of bellows are built in an enclosure and they divid the enclosure into three chambers. The movable ends of both of the sets of bellows are closed and connected with a plate each other. The inside chamber of the enclosure opens to the atmosphere with a vent orifice and connected to an air source through a nozzle with rect-angular section provided on the enclosure wall. The nozzle faces a side plane of the plate at a very small distance. Air is supplied from a constant pressure source into the enclosure through the nozzle and flows out to the atomosphere through the vent orifice. Input pressures are introduced in the sets of bellows and the pressure in the enclosure serves as output of the device. When the pressure of the same level are put in both of the set of bellows, the plate positions as to block the outlet and the output pressure maintains a low level, because of very large fluid circuit resistance of the nozzle outlet. When a high pressure is put into one side of the sets of bellows and a low pressure is put in the other side, the plate displaces as to open the nozzle outlet and the output pressure maintains high level. When the source pressure is supplied from the orifice to the enclosure and the nozzle is opened to the atmosphere, this device can function as an exclusive-NOR gate.

抄録全体を表示

In order to reduce the positioning time in a positioning system without introducing deviation into the parameters (motor torque constant, friction), a deadbeat control is applied.
However, in an actual system, as system parameters deviate, applying this control to the system gives rise to positioning error. Consequently, the commercial use of deadbeat control is very rare.
In this paper, a new deadbeat control (modified deadbeat control) is proposed, and both experimental and analytical results on positioning characteristics are reported.
In this new control, the control time interval is divided into two parts, the estimation interval and the precise positioning interval.
During the former interval, the system parameters incorporating static deviation are estimated by both the motor control force and the motor angular position. During the latter interval, the motor control force, which is determined by the estimated system parameters, is fed back to the motor. In the new control, as the system parameters can be estimated precisely, the positioning error due to the static deviation of parameters can be reduced to zero.
According to the analytical results, in which the dynamic deviation of parameters are taken into account, the positioning error is within ±0.2% under commercial motor conditions. This positioning error is sufficiently small for the control to be applied for practical use.
The new control has been implemented in a microcomputer and applied to a positioning system, in which both the torque constant and the friction deviate statically, from 0 to -50% for the torque constant, and from 0 to 500% for the friction.
Under such conditions, the estimation error for the torque constant ranges between -10% and +10%, and that for the friction ranges between -30% to +30%. Moreover this new control reduces the positioning error to about one tenth that of the conventional deadbeat control.

抄録全体を表示

In a direct-drive arm, motors are directly coupled to their arm linkages without reducers. This eliminates some defects of traditional gearing such as backlash, friction and deflection, and therfore enables the drive system to have improved control accuracy. The D.C. motors used for the direct-drive arm, however, have a significant amount of torque ripples and deadband, which limits the control accuracy, particularly for compliance and force control.
In this paper, a method for the torque feedback control of a direct-drive joint is presented. The motor output torque is directly measured by a strain gage sensor built into the motor. It is found that closed loop torque control can compensate for the complicated nonlinearities and improve the overall control accuracy.
First, the design of the torque sensor is presented. The effect of the structural vibration mode which results from the insertion of the flexible torque sensor is analysed, and a design method to place the natural frequency of this mode far above the operating frequencies is shown. Torque feedback control using the sensor is then discussed. The sensor design is verified experimentally, and the control performance using the direct torque feedback is evaluated.

抄録全体を表示

This paper describes three-wheeled vehicles for monitoring internal piping. The vehicle is basically composed of two-hinged arms called scissors structure by authors. Each end of the arms is joined with links crossing together at their middle points. The wheel located at the fulcrum of the scissors structure is driven by DC motor and the other wheels located at the ends of the links are sphere bearings.
Relations between the stretch force and the configuration of the vehicle are analyzed to optimize the length of the lever and the direction in which the lever should be attached for force generation. To obtain the condition in which the vehicle can move stably, authors investigated the shape of the pipe and friction exerted on the vehicle. Results of the simulation of locomotion in the two dimensional area proved the self-adjustability of the vehicle to various shapes and sizes of pipe. Simplified vehicle, MOGRER, is fabricated to carry a CCD video camera in his body for monitoring internal piping. Experimental results show the MOGRER can monitor while locomoting in inclined pipes with a deep angle without the influence of the gravity and the change of the pipe's size and shape.

抄録全体を表示

The ammonia (NH₃) injection control of non-catalytic exhaust gas denitrification system in a refuse incinerator plant has been developed.
While saving NH₃ consumption, the control system reduces the nitrogenoxide (NO_X) concentration in the exhaust gas effectively.
Because of the injection system structure, generated NO_X concentration from the refuse combustion process cannot be measured in the period of NH₃ injection. Moreover, generated NO_X concentration fluctuates widely and rapidly due to the refuse quality and the combustion condition. These characteristics make it difficult to analyse the dynamics of the process and to design the control system.
The statistical estimation of generated NO_X concentration applied to combustion process data has solved the problem. It has been found that the denitrification ratio calculated with the estimated NO_X concentration has few noise components.
From the NH₃ injection step response analyses to the actual plant, the following results are obtained.
i) The process can be regarded as a first-order delay system with deadtime.
ii) The order of the process delay is some minutes.
iii) The temperature of the injection point influences the process gain and delay to a great extent.
From the experiments, it was found that the conventional PID control could not be effective such large process delay and large process parameter variation.
Finally, combination of the gain scheduled PID control and the modified Smith predictor was adopted to the NH₃ injection control.
Implemented on a micro-computer, this control system has been applied to an actual refuse incinerator plant and has shown good performance.

抄録全体を表示