Intelligent fault diagnosis in microprocessor systems for vibration analysis in roller bearings in whirlpool turbine generators real time processor applications

Abstract

Large steam turbines used for electrical power generation demand governing systems of very high integrity (safety) and availability. The latest generation of electronic governors uses microprocessors in a distributed, two level architecture to achieve the required integrity and availability and in addition provides greater configuration flexibilities and wider facilities than earlier governors. Rolling element bearings are one of the major machinery components used in industries like power plants, chemical plants and automotive industries that require precise and efficient performance. Vibration monitoring and analysis is useful tool in the field of predictive maintenance in small hydro electric power plants. Health of rolling element bearings can be easily identified using vibration monitoring because vibration signature reveals important information about the fault development within them. Numbers of vibration analysis techniques are being used to diagnosis of rolling element bearings faults. This paper proposes a new signal feature extraction and fault diagnosis method for fault diagnosis of low-speed machinery. Initially, the proposed work explores the Continuous Wavelet Transform (CWT) to adaptively remove the exact noises from vibration analysis and then feature extraction is performed by exploiting the noise removed pre-processed data. Statistic filter (SF) and Hilbert transform (HT) are combined with moving-peak-hold method (M-PH) to extract features of a fault signal, and Special bearing diagnostic symptom parameters (SSPs) in a frequency domain that are sensitive to bearing fault diagnosis are defined to recognize fault types. The SF is first used to adaptively cancel noises, and then fault detection is performed by exploiting the optimum symptom parameters in a time domain to identify a normal or fault state. For precise diagnosis, the SSPs are calculated after the signals are processed by M-PH and HT.

Introduction

The evolution of microprocessor architecture depends upon the changing aspects of technology. As die density and speed increase, memory and program behaviour become increasingly important in defining architecture tradeoffs. While technology enables increasingly complex processor implementations, there are physical and program behaviour limits to the usefulness of this complexity. Physical limits include device limits as well as practical limits on power and cost A Whirlpool turbine generator plays an irreplaceable role in modern power industry. Over the past decades, the safety of equipment has received more and more attention and the fault diagnosis of rotary machinery has become a hot research topic. Faults not only include the imbalance of the rotor itself but also may occur at the bearings, gear boxes and couplings, which determines the variation and complexity of faults. Hence, how to describe faults is key to fault diagnosis. Various sources such as vibration [1], electric current and acoustic signals [2,3] are used in diagnosis. Generally, vibration signals are important sources of faults and contain abundant information about running states of rotary machinery, which are widely used to extracted features in fault description.

Fault diagnosis and estimation has been done using different techniques in different domains [4]. Bearing faults have been diagnosed mostly using techniques, which diagnose bearing defects by analyzing different types of signals, such as the vibration acceleration signal of a bearing's housing measured through accelerometers [5], the stator current of the induction motor [6], the acoustic emission (AE) signals [7], and the stray flux spectra [8]. Techniques that analyze the vibration acceleration signal and the motor stator current are effective in diagnosing bearing defects at high rotational speeds. However, at low rotational speeds, bearing defects, especially incipient defects, are more effectively diagnosed using AE-based methods, as they are sensitive to the low energy acoustic emissions released by a developing crack in the bearing even if it is sub-surface [9]. Hence, in this paper, AE signals are used to diagnose incipient bearing defects under variable operating speeds. The diagnosis of bearing defects under variable operating speeds is an important problem. Many studies [10] have considered similar problems in different contexts. For instance, the authors of [11] have studied the application of traditional vibration-based techniques for the diagnosis of various rotor faults in machines operating at different speeds and with different foundation supports.

AE-based methods mostly diagnose bearing defects either by using envelope analysis [12], or by constructing discriminative models for features extracted from the bearing fault signals using discriminative classifiers such as support vector machines (SVM) [13]. Envelope analysis-based methods diagnose bearing defects by looking for peaks at characteristic frequencies associated with each defect type in the power spectrum of the envelope signal. However, these characteristic defect frequencies (CDFs) are functions of the bearing's rotational speed, which renders these techniques ineffective under variable operating speeds. Similarly, feature extraction-based methods are also not effective in diagnosing bearing defects under variable operating speeds, as variations in the operating speed result in inconsistent features that yield poor discriminative models. Hence, these methods have predominantly been used to diagnose bearing defects under constant operating speeds. Moreover, since feature extraction-based methods use the statistical properties of the time and frequency domain AE signal, and the complex envelope signal; the diagnostic performance of these methods depends upon the quality of the extracted features. The selection of appropriate features requires both expert domain knowledge and feature selection algorithms to eliminate redundant and irrelevant features [13]. In summary, so far, the literature on the application of vibration or AE analysis to automatically diagnose lowspeed bearing fault has not been found.

This work carries out the automatic diagnosis of low-speed bearings using vibration analysis, which has widely been used in production plants at a low cost. The motivation of the work is as follows.

• In order to automatically extract the weak fault signal of a low-speed bearing from the vibration signal contaminated by strong noises, the self-adaptive signal processing methods based on M-PH, SF, and HT are proposed [14]. Also before that, a pre-processing process is carried out using Continuous Wavelet transform to remove the false signal from the data samples.

• In order to sensitively reflect the features of the extracted fault signal, the special bearing diagnostic SPs (SSPs) are newly defined for precision diagnosis [15].

• In order to precisely and automatically determine the fault type of low-speed bearings, the construction method for the intelligent diagnostic system is proposed by introducing fuzzy neural network with the use of the SPs in time and frequency domain. The design of FNN includes the development of the fuzzy rules that have IF-THEN form [16].

The fault states of a bearing can be classified into early stage (spot flaw), middle stage (multiple localized defects), and final stage (generalized defects) [17]. This paper emphasizes early fault diagnosis, which is beneficial in real-world industries, because potential catastrophic failure can be prevented by successful early fault detection [18]. In condition-based maintenance, early fault detection can also provide important information to carry out the state trend control [19]. The focus of this paper is early fault diagnosis of the roller bearing at low speeds, but the bearing diagnosis method will be tested during the middle and final stages in a forthcoming study [20].