Neural Networks to recognize industrial characters are required to achieve high reading reliability. To achieve this high reliability, a method to control the structure of neuron network's hidden layer has been introduced. The neural network is applied to an assembled PCB (Printed Circuit Board) visual inspection system and proved its high performance. The analysis of this network showes it has a desired structure.

View full abstract

This paper presents a high speed image processor for automated visual inspection systems (AVISs), which can continuously process images (512×512×8 bits per frame) through TV cameras at the NTSC video-rato (30 frames per second). The processor has been developed aiming at the solution of a trade-off requisition problem between high speed processing capability and system construction flexibility to cope with various AVIS requirement specifications. The processor bases on a “buildingblock” typed pipeline architecture which enables image processing hardware modules to combine each other in the form of cascade, parallel, and circular connection. This paper discusses a synchronization control method that supports the architecture, which transmits not only image data but also the attributive control signals (sync signal and image identification number, etc.) into image bus, like a data flow machine. This paper describes the processor's system configuration and image processing functions, and also gives some examples of AVIS using the processor. The effectiveness of the processor can be proved by the fact that the number of kinds of AVISs which have been realized by the processor exceeds more than forty.

View full abstract

The super-conducting type MAGLEV, which is directly driven by the VVVF inverters or the cycloconverters, is well-known as the new traction system for the future super-rapid mass transportation.
Since the frequency converters generate harmonic currents possibly leading to the interference problem on utility's transmission system, the conventional AC filter of LC (reacter-capacitor) branch and/or the active filter is necessary to prevent the harmonic disturbances.
We have developed a novel harmonic compensation system consisting of shunt LC filters on load side, a step-down transformer on receiving end, and a small-capacity active filter which one end is connected to the secondary winding of the transformer and the other to the load bus.
The active filter is current-controlled to give adequate damping to the harmonic resonance condition between the power system and the passive filter equipment especially at lower harmonic frequencies.
We will describe the principle and the characteristics of the new harmonic compensation system, and its labolatory test results which show satisfactory performance.

View full abstract

This paper treats the vibration control of a motor system which has a motor and a load connected with a flexible shaft. This system often generates a shaft torsional vibration. Traditional methods for this motor system are specifically to use notch filter to suppress the resonance of the system, or to adjust PID controller so that the closed loop frequency response is slower than that of the vibration mode. On the other hand, we have already proposed one method by disdurbance obserber to suppress the vibration. In this paper, we propose a new approach based on H∞ control theory. For comparison, we also construct a PI control system based on Classical control theory.
The results of several experiments show us that the H∞control system is effective in suppressing the vibration as compared with the PI control system. Further, this study shows that the H∞ controller obtained consists of a PI controller and a series compensator that functions as an active vibration controller.

View full abstract

For a variable speed induction motor drive, the control system without a speed sensor and a PWM current source inverter have been proposed. For this system, a voltage sensor and a current sensor are required instead of the speed sensor. The motor speed can be estimated by the measured instantaneous values of the voltage and the current. Since the PWM current source inverter is clearly superior to the PWM voltage source inverter in terms of voltage and current waveforms, i.e., sinusoidal waveforms, a more accurate estimation of the speed can be achieved.
In this paper, the PWM current source inverter-induction motor drive without a speed sensor is presented, The used inverter can regulate the output current by PWM control under a fixed input dc current. The quick regulation of the ac current is achieved in this system because of the no influence of the dc link parameters. The effect of the filter capacitor in the inverter circuit on the current regulation is discussed. To eliminate the ac current sensors of the motor, the relationship between the inverter currents and the motor currents is made clear. The experiments show that the system without the ac current sensors has the same steady-state and transient characteristics as that of the system with the voltage and the current sensors.

View full abstract

In this paper, on classification problems, dependency of generalization capability for multi-layered neural networks on their number of hidden units is discussed. First, a simple model representing the relationship between generalization capability and the number of hidden units is established, which has two characteristic behaviors, namely,
(1) generalization capability increases with the number of hidden units, then saturates; and
(2) the saturation number of hidden units increases with noise level of the generalized data. According to this model, the saturation number of hidden units can be defined to be optimal, which is the smallest and sufficient for the maximum generalization capability. Then a statistical algorithm based on linear regression analysis proposed in the previous paper is extended to accurately determine the optimal number of hidden units.
By the simulation using two typical neural network classification systems, i.e. alphabet recognition and practical number recognition systems, the developed model is found to describe the above relationship properly. Also the extended algorithm is demonstrated to be effective to estimate the optimal number of hidden units.

View full abstract

Permanent-magnet brushless direct-current motors are easy of maintenance and control, compaired with direct-current motors. Therefore, they are attracting attention today. In this paper, the authors discuss non-salient pole brushless direct-current motors. Usually, the quadrature armature current which contributes to torque generation is supplied in this motor. But, in the case that the source voltage cannot be increased, by supplying the direct armature current, that is, equivallently weakening field, a quadrature armature current can be increased. Consequently, the motor torque increases.
In the case that the load characteristics are settled, the motor torque can have a maximum T_m by adjusting a direct armature current. At this time the rotational speed and output are maximum also. T_m changes with the change of the motor parameters. Consequently, the motor parameters which maximize T_m are optimal. In this paper, the auhors have clarified an optimal parameter determination method considering of the load torque-speed characteristics and armature resistance, and the power factor and efficiency where the motor has optimal parameters and generates the maximum output. Further, the authors have verified the obtained theory through experiment.

View full abstract

This paper discribes about a new algorithm for digital length measurement. In this algorithm, the way of data access is Skipped Dual Lines raster Scanning, so it is named “SDLS” method. Using SDLS, the contour length of binary image can be measured with high speed and high accuracy. Theoretical expression of contour length based on SDLS method is derived for the figure which consists of linear edge. In case of square and circle figure, it is proved by the computer simulation that the most accurate data is obtained when skip number=1. As these results, SDLS method is very useful for industrial application, especially for measuring moving objects.

View full abstract

In this paper, the pulling-into performance of synchronous motors taking account of influence of field circuit is analyzed using Lyapunov's direct method. The method enables one to directly judge the stability without solving the nonlinear differential equations representing motor involved. The influence of all machine constants on synchronization of synchronous motors is assessed. The analyzed results show that field constants have effectively large influence on the pulling-into performance.

View full abstract

Magnetic frequecy tripler has a configuration with star-connected primary windings and open-delta-connected secondary windings of three saturable reactors. This paper describes a tripler comprised of saturable reactor using an iron amorphous magnetic material for higher efficiency. In a machine of this type, the iron loss is not always constant, in that the cores are operated in the saturation region, but it is subject to change according to the load. Conventionally, this loss is estimated via a method where the steady state periodic solution of the circuit has obtained, this, solution is substituted into a empirical loss calculation formula, and thus the loss is determined. This method has been used often as it is relatively easy to apply. However it is difficult to examine directly the effects of the loss on efficiency of the device with such a method.
In ths paper in order to ascertain the efficiency and other characteristics of the tripler in more detail, a nonlinear differential equation of the circuit was induced on the basis of the winding resistance and exciting conductance of the saturable reactors.
The harmonic balance method was applied to ths equation, in order to obtain the characteristics of the device, while verification was made with a proto type. As a result, it was shown that an efficiency of 96% could be realized with the proto type. The relationship between load and loss was also revealed.
The information contaned in this paper will offer useful data for the operation and design of a magnetic frequency tripler using amorphous cores.

View full abstract

The induction of zero-phase-sequence voltage by negative-phase-sequence current in a salient pole synchronous generator is proved experimentally and theoretically. In the method of symmetrical coordinates, it has been believed that positive-, negative- and zero-phase-sequence components are independent. But an experiment driving a synchronous generator with unbalanced load shows that zero-phase-sequence voltage is induced when only negative-phase-sequence current is applied. The amplitude of the induced voltage is proportional to that of the current. The phase difference between the voltage and the current is not constant, but the sum of phases is constant. So, relation between the voltage and the current is not represented with the impedance. The induction of zero-phase-sequence voltage is proved theoretically by magnetic flux distribution in the air-gap of synchronous generator. The m. m. f. by negative-phase-sequence current generates the space 3rd harmonic flux distribution by the saliency of the rotor and the field current 2nd harmonics. This flux distribution component interlinks the space 3rd harmonics of the armature winding distribution and induces zero-phase-sequence voltage.

View full abstract

Direct linear motion drive can be obtained by the use of linear pulse motor (LPM), Miniaturized LPMs have been applied to 3.5" floppy disk drive (FDD) units. It is necessary to control through feedback of position informations for higher performance. However, to miniaturize the units, sensorless detection of position of LPM is required.
In this paper, we dicussed a detection method using exciting winding of permanent magnet type LPM, and dealt with the followings.
(1) An equation of equivalent inductance contributing to flux by permanent magnet was led from magnetic equivalent circuit on permanent magnet type LPM. As a result, the relation between the inductances for each flux by permanent magnet and those by exciting winding was clarified.
(2) We showed the principle of position detection using LPM' s exciting winding as detection device, and showed the equivalent inductance can be detected from exciting voltage and current of LPM' s exciting winding.
(3) A good experimental result was obtaind through the application of the position detection method discussed in this paper to miniature LPM for 3.5" FDD.

View full abstract

In recent years, the number of three-phase squirrel-cage induction motors fed by inverters has increased due to the benefits of high efficiency and variable speed control. Motors driven by inverters produce electromagnetic vibrations caused by many time-harmonic voltages as well as by the fundamental voltage. To reduce the electromagnetic vibration, it is necessary to determine the relationship between time-harmonic voltages and electromagnetic vibration.
In this paper, two kinds of time-harmonic voltages superimposed on the fundamental voltage with various phase angles are supplied to the motor at no-load, and the causes of the dominant electromagnetic vibration are analyzed. First, the effects of amplitudes and phase angles of harmonic voltages on the electromagnetic vibration are theoretically studied and characterized. Next, the theoretical results are experimentally verified. It is determined that the characteristics of the dominant electromagnetic vibration can be classified into the following three categories:
(1) the vibration is proportional to the amplitude of one harmonic voltage,
(2) the vibration is proportional to the product of amplitudes of two harmonic voltages,
(3) the vibration varies complicatedly with amplitudes and phase angles of two harmonic voltages.

View full abstract