Issam Abu-Mahfouz

Drilling wear detection and classification using vibration signals and artificial neural network

In automated flexible manufacturing systems the detection of tool wear during the cutting process is one of the most important considerations. This study presents a comparison between several architectures of the multi-layer feed-forward neural network with a back propagation training algorithm for tool condition monitoring (TCM) of twist drill wear. The algorithm utilizes vibration signature analysis as the main and only source of information from the machining process. The objective of the proposed study is to produce a TCM system that will lead to a more efficient and economical drilling tool usage. Five different drill wear conditions were artificially introduced to the neural network for prediction and classification. The experimental procedure for acquiring vibration data and extracting features in both the time and frequency domains to train and test the neural network models is detailed. It was found that the frequency domain features, such as the averaged harmonic wavelet coefficients and the maximum entropy spectrum peaks, are more efficient in training the neural network than the time domain statistical moments. The results demonstrate the effectiveness and robustness of using the vibration signals in a supervised neural network for drill wear detection and classification.

A comparative study of three artificial neural networks for the detection and classification of gear faults

Artificial neural networks (ANN) have been recognized as a powerful tool for classification and pattern recognition in various fields of applications. This paper presents an overview of three ANN architectures and the results of applying those ANNs for the detection and classification of malfunction, wear and damage of a gearbox operating under steady state conditions. The ANN models studied are: feed forward back propagation (FFBP), functional link network (FLN) and learning vector quantization (LVQ). Three artificial defects were deliberately introduced to the gearbox and these are: (1) loose key, (2) single tooth flank wear and (3) full tooth breakage (missing tooth). Vibration signals, collected from extensive experimentation, were analyzed using time and frequency domain descriptors that were used as feature vectors to feed the ANNs. The results show that, for this study, the FLN learns more quickly and is more accurate in operation than the FFBP or the LVQ. The LVQ algorithm exhibits faster rate of convergence than the FFBP but suffers more from misclassifications.

Numerical Study of Some Nonlinear Dynamics of a Rotor Supported on a Three-Pad Tilting Pad Journal Bearing (TPJB)

The use of tilting pad journal bearings (TPJBs) has increased in the recent past due to their stabilizing effects on the rotor bearing system. However, in this paper two mechanisms capable of producing instabilities in terms of subharmonic and chaotic motions are suggested. The first one is that of a centrally loaded pad with rotor unbalance excitations. The second one represents a concentric rotor (or a vertical rotor) acted upon by centering sprigs and large unbalance excitations. Extensive numerical experimentation shows, for certain parameters, subharmonic, quasi-periodic, and chaotic motions. The pad state space trajectory, in many cases, resembles that of the two-well potential case as in Duffing’s oscillator. Time trajectories, Poincare maps, fast Fourier transform (FFT) plots, and the max Lyapunov exponent are utilized to examine the periodicity (order) of the nonsynchronous rotor orbits and pad trajectories. The TPJB problem belongs to a family of nonlinear rotor-dynamical phenomena that are potentially of a considerable value as diagnostic tools in assessing rotating machinery condition monitoring.

Drill flank wear estimation using supervised vector quantization neural networks

Drill wear detection and prognosis is one of the most important considerations in reducing the cost of rework and scrap and to optimize tool utilization in hole making industry. This study presents the development and implementation of two supervised vector quantization neural networks for estimating the flank-land wear size of a twist drill. The two algorithms are; the learning vector quantization (LVQ) and the fuzzy learning vector quantization (FLVQ). The input features to the neural networks were extracted from the vibration signals using power spectral analysis and continuous wavelet transform techniques. Training and testing were performed under a variety of speeds and feeds in the dry drilling of steel plates. It was found that the FLVQ is more efficient in assessing the flank wear size than the LVQ. The experimental procedure for acquiring vibration data and extracting features in the time-frequency domain using the wavelet transform is detailed. Experimental results demonstrated that the proposed neural network algorithms were effective in estimating the size of the drill flank wear.

On the Investigation of Nonlinear Dynamics of a Rotor with Rub-impact Using Numerical Analysis and Evolutionary Algorithms☆

Abstract In this paper the dynamics of a rotor-stator system with mass imbalance induced rub-impact interactions is investigated with particular attention on the routes to chaos. The rub-impact interaction is modelled by a Hertz contact radial force and a Coulomb friction tangential force. Extensive numerical experimentation for a wide range of parameters shows the resulting response to be rich in subharmonic, quasiperiodic and chaotic motions. Parameter identification of chaotic systems has become an important topic of research in the past decade. Of particular interest is the problem of identifying or estimating system parameters when the quasiperiodic or chaotic responses of the system are known. The problem of identifying parameters can be cast as an optimization problem and non-traditional optimization methods such as evolutionary algorithms, simulated annealing and others have been developed to identify system parameters. In this paper, three evolutionary algorithms – particle swarm optimization, differential evolution and firefly algorithm are presented and compared for the problem of identifying parameters of a rotordynamical system given a chaotic response. The results of this analysis can potentially be of a considerable value as diagnostic tools in assessing condition monitoring signals that are routinely taken on modern rotating machinery.

Drill Wear Feature Identification under Varying Cutting Conditions Using Vibration and Cutting Force Signals and Data Mining Techniques

Abstract Monitoring drill wear is a major topic in automated manufacturing operations. This paper presents an effective drill wear feature identification scheme based on robust clustering techniques. Three types of drill wear (namely; chisel wear, chipped edge and flank wear) are artificially induced on the drill point. The drill cutting edge wear related features are extracted experimentally by processing the force signals from a three-axis piezoelectric load cell in both the time and frequency domains. Techniques based on the short time Fourier transform (STFT), wavelet transform (WT) and statistical parameters are utilized for feature extraction. The sensitivity of the proposed method is tested under different cutting feed and speed conditions. The computational study is conducted using the features extracted from three dimensional vibration and cutting force signals. The type of drill wear and related variations in the cutting forces are identified using robust clustering methods. The objective is to isolate regions in the feature space, each region corresponding to one of the drill wear types. Results show that power spectral density data clusters better than data obtained using wavelet coefficients. The clustering results can be used to design classifiers for real time monitoring of wear conditions while drilling.

Surface Roughness Identification in End Milling Using Vibration Signals and Fuzzy Clustering

The study presented involves the identification of surface roughness in Aluminum work pieces in an end milling process using fuzzy clustering of vibration signals. Vibration signals are experimentally acquired using an accelerometer for varying cutting conditions such as spindle speed, feed rate and depth of cut. Features are then extracted by processing the acquired signals in both the time and frequency domain. Techniques based on statistical parameters, Fast Fourier Transforms (FFT) and the Continuous Wavelet Transforms (CWT) are utilized for feature extraction. The surface roughness of the machined surface is also measured. In this study, fuzzy clustering is used to partition the feature sets, followed by a correlation with the experimentally obtained surface roughness measurements. The fuzzifier and the number of clusters are varied and it is found that the partitions produced by fuzzy clustering in the vibration signal feature space are related to the partitions based on cutting conditions with surface roughness as the output parameter. The results based on limited simulations are encouraging and work is underway to develop a larger framework for online cutting condition monitoring system for end milling.Copyright

Statistical Comparison of Feature Sets for Time Series Classification of Dynamic System Response

The use of classification techniques for machine health monitoring and fault diagnosis has been popular in recent years. System response in form of time series data can be used to identify type of defect, severity of defect etc. However, a central issue with time series classification is that of identifying appropriate features for classification. In this paper, we explore a new feature set based on a delay differential equations (DDEs). DDEs have been used recently for extracting features for classification but have never been used to classify system responses. The Duffing oscillator, Van der Pol–Duffing (VDP-D) oscillator, Lu and Chen oscillators are used as examples dynamic systems, and the responses are classified into self-similar groups. Responses with the same period should belong to the same group. Misclassification rate is used as an indicator of the efficacy of the feature set. The proposed feature set is compared to a statistical feature set, a power spectral coefficient feature set and a wavelet coefficient feature set. In work described in this paper, a density estimation algorithm called DBSCAN is used as the classification algorithm. The proposed DDE-based feature set is found to be significantly better than the other feature sets for the classifying responses generated by the Duffing, Lu and Chen systems. The wavelet and power spectral coefficient data sets are not found to be significantly better than the statistical feature set for these systems. None of the feature sets tested are discerning enough on the VDP-D system.

Bearing Fault Parameter Identification Under Varying Operating Conditions Using Vibration Signals and Evolutionary Algorithms

This paper presents an effective bearing fault parameter identification scheme based on evolutionary optimization techniques. Three seeded faults in the rotating machinery supported by the test roller bearing include inner race fault, outer race fault and a single ball defect. The fault related features are extracted experimentally by processing the acquired vibration signals in both the time and frequency domain. Techniques based on the power spectral density (PSD) and wavelet transform (WT) are utilized for feature extraction. The sensitivity of the proposed method is investigated under varying operating speeds and radial bearing load. In this study, the inverse problem of parameter identification is investigated. The problem of parameter identification is recast as an optimization problem and two well known evolutionary algorithms, differential evolution (DE) and particle swarm optimization (PSO), are used to identify system parameters given a system response. For online parameter identification, differential evolution outperforms particle both in terms of adaptability and tighter convergence properties. The distinction between the two methods is not distinctively obvious on the offline parameter identification problem.Copyright

Fatigue Damage Diagnosis Using Statistical, Spectral, and Wavelet Analysis Techniques

The objective of this work is to illustrate a comparative investigation into the signatures of crack initiation and propagation using vibration and displacement signals. Experimental tests were performed on low carbon steel specimen under fully reversed bending cycles in a cantilever support configuration. Accelerometer and displacement-sensor vibration signals were collected using a National Instrument data acquisition system. Several methods, such as statistical moments, FFT, Wavelets, and short-time-frequency analysis techniques, were used to analyze the collected signals. It was found that the wavelet analysis gave the most consistent patterns in tracking crack initiation and propagation. For the wavelet analysis, extensive comparative investigation was conducted to select the optimum combination of wavelet packets for crack detection and monitoring.© 2013 ASME