Serhat Şeker

Comparisons between the various types of neural networks with the data of wide range operational conditions of the Borssele NPP

This paper addresses a trend monitoring in operating nuclear power plant by use of two types of Recurrent Neural Networks (RNN). The interesting feature of the RNN is intrinsic dynamic memory that reflects the current output as well as the previous inputs and outputs are gradually quenched. The first one Elman type of RNN which has a feed-back from hidden layer to the input layer neurons while in the Jordan type, from the outputs of the neural net to the inputs of the neural net. In this paper the theoretical assessment of the both RNNs is given. Both topological structures including Back Propagation (BP) neural network were implemented to the Borssele NPP. Learning achieved from 30% to 100% nominal power at the starting period of the new core 30 September 2001. After learning period the reactor operation is followed by the neural network. Paper will present the reactor system, the real time data collection and the merits of the three types of the neural network applied while in the learning and continuous processing of the changing of the operational conditions.

Fault detection based on continuous wavelet transform and sensor fusion in electric motors

Purpose – The purpose of this paper is to extract features from vibration signals measured from induction motors subjected to accelerated aging of bearings by fluting tests.Design/methodology/approach – Aging tests were performed according to IEEE test procedures. The data acquisition involved the measurement of vibration signals using accelerometers that were installed on the bearings and on the motor casing. In this application, only two accelerometers, which were placed near the process end of the motor bearing, are used for data analysis and feature extraction studies. After the data collection, information from the two sensors was combined using simple sensor fusion method under the linearity conditions, and then spectral analysis and time‐scale analysis were performed. The fused vibration signal is decomposed into several scales using continuous wavelet transform (CWT) and its first scale is used to indicate the bearing degradation.Findings – Bearing damage characterization was determined between 2‐...

A Reliability Model for Induction Motor Ball Bearing Degradation

In this study a reliability model of accelerated aging processes applied to motor bearings is presented. For this purpose, bearing damage is created through several accelerated aging processes for 5 HP induction motors, and motor vibration signals are examined by statistical and spectral analysis techniques. In this sense standard deviation values of the considered signals are determined as the most effective statistical parameter in reflecting the bearing degradation behavior. A new definition, which indicates the reliability level of the bearing, is given as a correlation degree defined between the initial case and any aged case. This relationship is used to establish the reliability model of motor bearing degradation through the exponential variation of the standard deviation for the overall process.

Wavelet based Neuro-Detector for low frequencies of vibration signals in electric motors

This study presents a Wavelet based Neuro-Detector approach employed to detect the aging indications of an electric motor. Analysis of the aging indications, which can be seen in the low frequency region, is performed using vibration signals. More specifically, two vibration signals are observed for healthy and faulty (aged) cases which are measured from the same electric motor. Multi Resolution Wavelet Analysis (MRWA) is applied in order to obtain low and high frequency bands of the vibration signals. Thus for detecting the aging properties in the spectra, the Power Spectral Density (PSD) of the subband for the healthy case is used to train an Auto Associative Neural Network (AANN). The PSD amplitudes, which are computed for the faulty case, are applied to input nodes of the trained network for the re-calling process of AANN. Consequently, the simulation results show that some spectral properties defined in low frequency region are determined through the error response of AANN. Hence, some specific frequencies of the bearing damage related to the aging process are detected and identified.

Real time reactor noise diagnostics for the Borssele (PWR) nuclear power plant

Abstract After the upgrade of Borssele NPP in 1997, core cycle 24, the power plant operated three years more with 91% availability. The authority of the power plant decided to enhance and upgrade the reactor trend monitoring and plant information recording system with higher frequencies than the plant data processing system (PPS) as well as installing a flexible and multiple-purpose reactor noise analysis system which may support the reactor maintenance group with on-line and off-line capabilities for several different signal processing applications. Two measuring and monitoring systems were built in 2001 and fully taken in implementation during the start-up of the new core 28. In this sense, the new system was used in power operation during the 29 th of September 2001. This paper will introduce the measuring system, the operational tasks, and the results obtained so far on the real-time core-barrel motions (CBM) and the two-primary coolant pump vibrations measured through the reactor noise analysis.

Neuro-detector Based on Coherence Analysis for Stator Insulation in Electric Motors

Abstract This research describes the monitoring of the fundamental spectral features of stator insulation damage through accelerated aging studies for induction motors with a power rating of 5 HP. In order to accomplish this goal, even-harmonic values of the line frequency defined between the 4th and the 16th harmonics, which are computed by the coherence approach between the stator currents and vibration signals, are determined as indicators of stator insulation damage. After this determination, a neuro-detector based on the auto-associative neural structure is trained in the frequency domain. This uses coherence variations and even-harmonic values as indicators of the insulation damage of an induction motor by observing the changes in the errors (residuals) generated by the neural net.

Anfis model for vibration signals based on aging process in electric motors

In this study, the aging process of an electric motor is accomplished by adaptive neuro-fuzzy inference system (ANFIS) using vibration signals. Different ANFIS models are compared for representing the aging process in the best possible way. An artificial aging experiment is performed and vibration data taken from the initial (healthy) and final (faulty) cases are used to identify the aging process. Four different ANFIS models are presented. Moving average (MA) filters are applied to the input and output pairs for different lagging factors to change the smoothness degree of the data and thus the performance of system identification. The success of the models is evaluated on three conditions; the performance of the ANFIS and the linear correlation between expected output (faulty case data) and aging model output, in time and frequency domains.The study also evaluates the influence of preprocessing using MA filtering on the ANFIS performance for vibration data which have stochastic characteristics.

Artificial neural networks for dynamic monitoring of simulated-operating parameters of high temperature gas cooled engineering test reactor (HTTR)

Abstract This paper addresses to the problem of utilisation of the artificial neural networks (ANNs) for detecting anomalies as well as physical parameters of a nuclear power plant during power operation in real time. Three different types of neural network algorithms were used namely, feed-forward neural network (back-propagation, BP) and two types of recurrent neural networks (RNN). The data used in this paper were gathered from the simulation of the power operation of the Japans High Temperature Engineering Testing Reactor (HTTR). For the wide range of power operation, 56 signals were generated by the reactor dynamic simulation code for several hours of normal power operation at different power ramps between 30 and 100% nominal power. Paper will compare the outcomes of different neural networks and presents the neural network system and the determination of physical parameters from the simulated operating data.

Monitoring the Aging of Industrial Motors by Geometric Trending of Frequency Domain Signatures

The characterization of the aging of electric motors using vibration measurements and a geometric interpretation of spectral domain signatures is presented. For comparison, two vibration signal records from a three-phase induction motor were acquired during degradation due to bearing fluting, following thermal and chemical aging sequences. Power spectral densities (PSD) of the vibration signals were calculated and linear approximations to the spectral signatures were used to observe the overall trend. The intersection point and slopes of these linear representations were used to extract certain critical frequencies in order to define convex regions (CR) and convex hulls (CH) as representatives of the aging (degradation) process. Furthermore, using these CRs and CHs, the functional readiness of motors could be determined by comparison of the signatures for initial and aged conditions.

Lyapunov exponent for aging process in induction motor

Nonlinear systems like electrical circuits and systems, mechanics, optics and even incidents in nature may pass through various bifurcations and steady states like equilibrium point, periodic, quasi-periodic, chaotic states. Although chaotic phenomena are widely observed in physical systems, it can not be predicted because of the nature of the system. On the other hand, it is known that, chaos is strictly dependent on initial conditions of the system [1-3]. There are several methods in order to define the chaos. Phase portraits, Poincare maps, Lyapunov Exponents are the most common techniques. Lyapunov Exponents are the theoretical indicator of the chaos, named after the Russian mathematician Aleksandr Lyapunov (1857-1918). Lyapunov Exponents stand for the average exponential divergence or convergence of nearby system states, meaning estimating the quantitive measure of the chaotic attractor. Negative numbers of the exponents stand for a stable system whereas zero stands for quasi-periodic systems. On the...