Do-Hwan Lee’s research while affiliated with Korea Hydro and Nuclear Power - Central Research Institute and other places

A support vector machine (SVM) is a very powerful classification algorithm when a set of training data, each marked as belonging to one of several categories, is given. Therefore, SVM techniques have been used as one of the diagnostic tools in machine learning as well as in pattern recognition. In this paper, we present the results of classifying ball bearing fault types and severities using SVM with an optimized feature set based on the minimum distance rule. A feature set as an input for SVM includes twelve time-domain and nine frequency-domain features that are extracted from the measured vibration signals and their decomposed details and approximations with discrete wavelet transform. The vibration signals were obtained from a test rig to simulate various bearing fault conditions.

Several studies on the use of Support Vector Machines (SVMs) for diagnosing rotating machinery have been successfully carried out, but the fault classification depends on the input features as well as a multi-classification scheme, binary optimizer, kernel function, and the parameter to be used in the kernel function. Most of the published papers on multiclass SVM applications report the use of the same features to classify the faults. In this study, simple statistical features are determined on the basis of time domain vibration signals for various fault conditions, and the optimal features for each fault condition are selected. Then, the optimal features are used in the SVM training and in the classification of each fault condition. Simulation results using experimental data show that the results of the proposed stepwise classification approach with a relatively short training time are comparable to those for a single multi-class SVM.

For the purpose of monitoring by ultrasonic test of the ball bearing conditions in rotating machinery, a system for their diagnosis was developed. ultrasonic technique is used to detect abnormal conditions in the bearing system. And various data such as frequency spectrum, energy and amplitude of ultrasonic signals, and ultrasonic parameters were acquired during experiments with the simulated ball bearing system. Based on the above results and practical application for power plant, algorithms and judgement criteria for diagnosis system was established. Bearing diagnosis system is composed of four parts as follows : sensing part for ultrasonic sensor and preamplifier, signal processing part for measuring frequency spectrum, energy and amplitude, interface part for connecting ultrasonic signal to PC using A/D converter, graphic display and software part for display of bearing condition and for managing of diagnosis program.

Stem friction coefficient and valve factor are very important parameters for the evaluation of valve performance. In this study, the characteristics of stem friction coefficient and valve factor are analyzed, and thor bounding value is determined. The hydraulic testing is performed for many flexible wedge gate valves in the plant and statistical method is applied to the determination of bounding value. According to the results of this study, stem friction coefficient does not change much with differential pressure, and the bounding value of closing stroke is higher than that of opening stroke. The valve factor of valves with high differential pressure is higher than that of valves with medium differential pressure. It means valve factor is more sensitive to the differential pressure than the stem friction coefficient. Valve factor of the closing stroke is higher than that of opening stroke due to piston effect.

It is necessary to monitor periodically the operability of safety-related motor-operated valves (MOVs) in nuclear power plants. However, acquiring diagnostic signals for MOVs is very difficult, and doing so requires an excessive amount of time, effort, and expenditure. This paper introduces an accurate and economical method to evaluate the performance of MOVs remotely. The technique to be utilized includes electrical measurements and signal processing to estimate the motor torque and the stem thrust, which have been cited as the two most effective parameters in diagnosing MOVs by the US Nuclear Regulatory Commission. The motor torque is calculated by using electrical signals, which can be measured in the motor control center (MCC). Some advantages of using the motor torque signature over other signatures are examined. The stem thrust is calculated considering the characteristics of the MOV and the estimated motor torque. The basic principle of estimating stem thrust is explained. The developed method is implemented in diagnostic equipment, namely, the Motor Operated Valve Intelligent Diagnostic System (MOVIDS), which is used to obtain the accuracy of and to validate the applicability of the developed method in nuclear power plants. Finally, the accuracy of the developed method is presented and some examples applied to field data are discussed.

Stem-stem nut coefficient of friction(COF) in motor-operated gate/globe valves is one of the important factors which determine the performance of the valve/actuators. The COF is affected greatly by the type and condition of the stem-stem nut lubricants, environmental parameters, surface condition of the stem/stem-nuts, and the number of strokes after the lubrication. In this paper, the measured data of the COFs at stem threads of some safety-related motor-operated gate/globe valves in domestic nuclear power plants are presented. In addition, the performance of the lubricants is evaluated by comparing the COFs among those valves. The results show that the measured COF at torque switch trip are higher than the unwedging COF and conservatively applicable to the unwedging COF. It is also shown that the lubricating performance based on the measured COFs varies with the lubricants.