Fumio Fujisawa

Vibration diagnosis method and apparatus for rotary machines

A vibration diagnosis method and apparatus for rotary machines adapted for measuring the vibration of rotor shaft of a rotary machine, extracting the vibration component synchronous with the revolution of the rotor shaft from the measured vibration data, comparing the amplitude of the extracted vibration component with the amplitude of the overall vibration, analyzing the causes of the synchronous vibration component when the result of the comparison meets a predetermined reference, judging, in accordance with the result of the analysis, whether the vibration due to mass unbalance is attributable to an abrupt generation of mass unbalance or to a thermal bow of the rotor shaft, and displaying the result of the judgment. The analysis of the vibration component is made in real time or making use of data stored in the past.

High Frequency Resolution Technique of FFT.

FFT (Fast Fourier Transfomation) is used for frequency analysis in many areas. Frequency resolution is given as the inverse of window length, i.e. observing time. Therefore, if high frequency resolution is required, a long window length is needed. This means that it is difficult to analyze short duration signals. This paper proposes a high frequency resolution technique of FFT in which interpolations to the data obtained in one sampling process and extractions of frequency components are applied. Simulations using this technique show higher frequency resolution, more accurate amplitude and phase compared to those of conventional FFT.

Vibration Control of a Hydraulic System using a Learning Fuzzy Logic Controller.

This paper describes a pump-controlled hydraulic system with a variable displacement pump to control the acceleration of a hydraulic cylinder. The system has non-linearity due to viscosity and compressibility of oil, packing friction, etc. So it is difficult to find a mathematical model of the process as opposed to conventional control systems and control them. Fortunately this system can be controlled by Fuzzy Logic, needless to find the troublesome mathematical model that has been proved applicable by our experimental results. However, it is not easy to acquire the control rule from our experimental results. For simple and effective tuning the fuzzy controller, we apply the Fuzzy Neural Network. It is shown that good results are achieved by use of the learning fuzzy logic controller.

Active Vibration Control of Stationary Vibration. Effect of Number and Location of Control Forces.

Recently, active vibration control has been applied to some bridges and buildings. However, the control has only been used to reduce vibration of lower modes. Control methods and spillover problems have also been discussed, but little has been done in terms of analysis of the number and location of actuators. For reduction of the vibration of higher modes, it is necessary to use many actuators. In terms of industrial application, it is important to know the necessary number and optimum location of actuators in order to obtain the required vibration reduction. In this report, we discuss the relationship between the number and location of actuators and vibration reduction for stationary vibration of a simple supported uniform beam. The following results were obtained. First, vibration reduction is strongly affected by the location of the actuators. Second, vibration reduction varies according to the distribution pattern of forces. Based on the findings, an estimation technique for the vibration reduction was proposed.

Balancing for a Rigid Rotor with Measurement Errors in Initial Unbalanced Vibration. Comparison of Balancing Conditions between the Least Squares Method and the Influence Coefficient Method.

This paper presents numerical simulations for balancing a rigid rotor. The influence coefficient and least squares balancing methods are compared, under the assumption that the initial vibration values used to calculate correction masses contain errors. The errors in the initial vibration values are : ( 1 ) errors in amplitude, ( 2 ) errors in phase angle, ( 3 ) errors in both amplitude and phase angle. In the three cases, errors are given in initial vibration at one measuring position. A number of numerical simulations are carried out and a comparison is made between correction masses and residual vibration values obtained by the two balancing methods. The results show that the vibration amplitude obtained with the least squares method is smaller than that obtained with the influence coefficient method.

Vibration Reduction of Power Transformer Core.

Recently, demand has increased for low-noise power transformers as well as large-capacity and small-size power transformers. To reduce transformer noise, it is necessary to reduce vibration of their iron cores, which is caused by magnetostrictive forces of silicon iron sheets. In this paper, a method for calculating core vibration was introduced and techniques for reducing the vibration were investigated. First, magnitudes and phases of higher-harmonic components of the magnetostriction were clarified. Then, a calculation method of core vibration using analytical solutions for a uniform beam was introduced. Calculated values for natural frequencies agreed with measured ones within 5%. Also, the vibration characteristics of a single-phase two-legged core, a three-phase three-legged core, and a three-phase five-legged core were clarified. Finally, it was found that the participation factors of magnetostriction in various portions of the core had different values for the displacement of core, and that arbitrary vibration modes could be eliminated by adjusting the magnetostriction values. Optimum combinations of magnetostriction which reduce core vibration were obtained. The preferable ratios for dimensions such as the ratio of the sectional areas of the core and yoke, or the ratio of the area moments of the inertia of the core and yoke, were proposed. In many cases, it was found that participation factors of magnetostriction in yokes were larger than those of legs, although the yokes were shorter than the legs.