Rotary Machinery Research Articles

Identifying Faults of Rolling Element Based on Persistence Spectrum and Convolutional Neural Network With ResNet Structure

The task of accurately bearing fault diagnosis of the rotary machinery from the measured signal remains a major problem that attracts a lot of attention. This paper proposed a new approach to build an efficient bearing fault diagnostic model for rotary machinery. The model is based on the persistence spectrum image and convolutional neural network (CNN) with ResNet structure. The persistence spectrum is extracted from the envelope of the raw vibration signal. Then, the persistence spectrum image is constructed based on short-time Fourier transform, which presents a new relationship between the frequency, magnitude, and energy of each signal with time, which the traditional spectrum analysis methods have not been given before. To explore the discriminant features from the persistence spectrum image of the envelope signal, an improved CNN with ResNet structure allows direct connection feature maps from the lower-level layer to the higher-level layer. That helps to exploit the granularity features in the low-level layer which can be lost when feedforward through adjacent layers in a traditional CNN. As a result, the proposed model operates efficiently with high accuracy not only under various working loads but also under noise conditions. Besides, its performance is very satisfactory compared to other types of two-dimensional images and other state-of-the-art diagnosis models. Overall, the proposed approach is highly feasible for an intelligent bearing fault diagnostic model.

Novel Convolutional Neural Network (NCNN) for the Diagnosis of Bearing Defects in Rotary Machinery

This work presents the development of novel convolutional neural network (NCNN) for effective identification of bearing defects from small samples. For effective feature learning from small training data, cost function of convolution neural network (CNN) is modified by adding additional sparsity cost in the existing cost function. A novel trigonometric cross-entropy function is developed to compute the sparsity cost. The proposed cost function introduces sparsity by avoiding unnecessary activation of neurons in the hidden layers of CNN. For identification of bearing defects from small training samples, NCNN-based transfer learning is applied in the following manner. First, the raw vibration signals as well as envelope signals from source domain machine are obtained. Thereafter, these envelope signals are applied to NCNN for the learning of features from the big training data acquitted from the source domain. After feature learning, knowledge gained from NCNN is transferred to do fine-tuning of NCNN from small training samples of target domain. Thereafter, defect identification is carried out by applying the test data of target domain to fine-tuned NCNN. The experimental result validates that the proposed cross-entropy function introduces sparsity in CNN and, hence, creates an effective deep learning which can even work under a situation when training data are not available in abundant.

Research on the Method of Rotary Machinery Fault Diagnosis based on PCA and DBN

Aiming at the difficulty of complex vibration signal transmission path, great influence of different sensor positions on diagnosis results and difficulty in feature extraction of rotary machinery fault diagnosis, a new fault diagnosis method based on Principal Component Analysis (PCA) and Deep Belief Network (DBN) is proposed. The framework of the method is constructed. The theory of PCA and DBN are introduced. And the validity and superiority of the proposed method are verified by the experimental data of typical rotary machinery.

Early Bearing Fault Diagnosis of Rotating Machinery by 1D Self-Organized Operational Neural Networks

Preventive maintenance of modern electric rotating machinery (RM) is critical for ensuring reliable operation, preventing unpredicted breakdowns and avoiding costly repairs. Recently many studies investigated machine learning monitoring methods especially based on Deep Learning networks focusing mostly on detecting bearing faults; however, none of them addressed bearing fault severity classification for early fault diagnosis with high enough accuracy. 1D Convolutional Neural Networks (CNNs) have indeed achieved good performance for detecting RM bearing faults from raw vibration and current signals but did not classify fault severity. Furthermore, recent studies have demonstrated the limitation in terms of learning capability of conventional CNNs attributed to the basic underlying linear neuron model. Recently, Operational Neural Networks (ONNs) were proposed to enhance the learning capability of CNN by introducing non-linear neuron models and further heterogeneity in the network configuration. In this study, we propose 1D Self-organized ONNs (Self-ONNs) with generative neurons for bearing fault severity classification and providing continuous condition monitoring. Experimental results over the benchmark NSF/IMS bearing vibration dataset using both x- and y-axis vibration signals for inner race and rolling element faults demonstrate that the proposed 1D Self-ONNs achieve significant performance gap against the state-of-the-art (1D CNNs) with similar computational complexity.

Bearing fault diagnosis based on flexible analytical wavelet transform and fuzzy entropy approach

Abstract For fault detection of bearings to avoid unexpected downtime in rotating machinery (RM), their monitoring is continuously required. Whenever a fault develops in any of the rotating machinery components, the characteristics vibration signature of such a mechanical system will change. Revealing such characteristics in a vibration signature is the main aim of any signal processing technique. Wavelet transforms and its variants have been widely used for their ability to perform under different non-stationarity nature of vibration signals. Among them, a new variant known as Flexible analytic wavelet transform (FAWT) enjoys attractive characteristics such as flexible time–frequency (TF) covering, better shifting property in which the same amount of input shifted corresponds to the same amount of shifted output, and adjustable periodic nature of the wavelet, providing appropriate shape and wavelet window to meet the weak fault component frequencies. After that, features are extracted using fuzzy entropy (FzEn) to determine the complexity of these vibration signals. After fault feature extraction by fuzzy entropy (FzEn), several state-of-the-art classifiers are used to validate the effectiveness of the proposed method. The results show that the proposed method is advantageous in identifying the fault type and severity of bearings.

マルチモーダルディープラーニングによる一般回転機器の診断自動化とクラウドオンライン学習システムの構築

マルチモーダルディープラーニングによる一般回転機器の診断自動化とクラウドオンライン学習システムの構築

A New Generative Neural Network for Bearing Fault Diagnosis with Imbalanced Data

Intelligent bearing fault diagnosis has received much research attention in the field of rotary machinery systems where miscellaneous deep learning methods are generally applied. Among these methods, convolution neural network is particularly powerful because of its ability to learn fruitful features from the original data. However, normal convolutions cannot fully utilize the information along the data flow while the features are being abstracted in deeper layers. To address this problem, a new supervised learning model is proposed for small sample size bearing fault diagnosis with consideration of imbalanced data. This model, which is developed based on a convolution neural network, has a high generalization ability, and its performance is verified by conducting two experiments that use data collected from a self-made bearing test rig. The proposed model demonstrates a favorable performance and is more effective and robust than other deep learning methods.

Manufacturing deformation impact on the performance of electrical generator for the wind turbine application

The system of systems approach is the representation of several distributed and independent systems, for a larger and complex system. The system can be defined as interacting and interdependent components forming a complex system. System perspective considers every single stator bar as a complex system, made up of several component strands and vent tubes. Strands are interacting and interdependent in orientation, to cancel out unbalanced strand voltage and minimizing circulating current. Visualizing the electrical generator with the system approach, the issue of incorrectly aligned interdependent and interacting components of the system has been noticed. The study presents the effective flux calculation due to strand tilt, and a nondestructive testing technique to measure the strand tilt. This paper first time presents a unique system perspective of the electrical generator; this would open further dimensions of system research, in electrical rotatory machinery and beyond.

Effect of misalignment and rarefaction effect on the comprehensive characteristic of MEMS gas bearing

PurposeOwing to the development of the smaller-sized rotational machinery, the demand for the high-speed and low-resistance gas bearing increases rapidly. The research of micro gas bearing in the condition of rarefied gas state is still not satisfied. Therefore, the purpose of this paper is to present a numerical investigation of the effect of misalignment and rarefaction effect on the comprehensive performance of micro-electrical-mechanical system (MEMS) gas bearing.Design/methodology/approachThe Fukui and Kaneko model is expanded to 2D solution domain to describe the flow field parameters. The finite element method is used to discretize the equation. Newton–Raphson method is used to solve the nonlinear equations for the static performance of gas bearing, and partial deviation method is adopted for the solution of dynamic equations.FindingsThe static and dynamic characteristics of MEMS gas bearing are calculated, and the comparison is made to study the influence of rarefaction effect and misalignment. The results show that the rarefaction effect will decrease bearing load capacity compared with traditional solution of Reynolds equation, and the misalignment will reduce the stability of bearing. The influence of misalignment on gas film thickness is also analyzed in this paper.Originality/valueThe investigation of this paper emerges the change regularity of comprehensive performance of MEMS gas bearing considering rarefaction effect and misalignment, which provides a reference for the actual manufacturing of MEMS gas bearing and for the safety operation of micro dynamic machinery.Peer reviewThe peer review history for this article is available at: https://publons.com/publon/10.1108/ILT-01-2020-0023/

A Review of Artificial Intelligence Methods for Condition Monitoring and Fault Diagnosis of Rolling Element Bearings for Induction Motor

The fault detection and diagnosis (FDD) along with condition monitoring (CM) and of rotating machinery (RM) have critical importance for early diagnosis to prevent severe damage of infrastructure in industrial environments. Importantly, valuable industrial equipment needs continuous monitoring to enhance the safety, reliability, and availability and to decrease the cost of maintenance of modern industrial systems and applications. However, induction motor (IM) has been extensively used in several industrial processes because it is cheap, reliable, and robust. Rolling bearings are considered to be the main component of IM. Undoubtedly, any failure of this basic component can lead to a serious breakdown of IM and for whole industrial system. Thus, many current methods based on different techniques are employed as a fault prognosis and diagnosis of rolling elements bearing of IM. Moreover, these techniques include signal/image processing, intelligent diagnostics, data fusion, data mining, and expert systems for time and frequency as well as time-frequency domains. Artificial intelligence (AI) techniques have proven their significance in every field of digital technology. Industrial machines, automation, and processes are the net frontiers of AI adaptation. There are quite developed literatures that have been approaching the issues using signals and data processing techniques. However, the key contribution of this work is to present an extensive review of CM and FDD of the IM, especially for rolling elements bearings, based on artificial intelligent (AI) methods. This study highlights the advantages and performance limitations of each method. Finally, challenges and future trends are also highlighted.

Algorithm for screws profiling of a conical trilobed compressor - analytical solution

Screw compressors, as rotary machinery, are capable of working with high compression ratios for gas compression. They transform the mechanical work done by an electric motor or turbine into the potential energy of a gas, by reducing its volume and increasing the pressure. The essential elements of these compressors are the two helical rotors with lobes, mounted in a housing. Existing rotors in the construction of compressors are working bodies whose composite profiles include circle arcs, line segments and cycloidal curves, while also often including non-analytical curves. Such geometry of profiles gives the gear processes superior characteristics, in the sense of reducing friction and improving sealing conditions. This paper analyses, by analytical approach, the possibility of constructing a multilobated conical screw compressor with a number of lobes Z, Z + 1, of the male and female rotor. The parametric equations of the contained worm and the kinematics of the working process of the conical screw compressor shall be defined. The theory of surface wrapping comprises analytical methods of study, which were developed on the basis of the fundamental theorems on wrapping of known surfaces in differential geometry and the kinematics laws on rotor movements. Thus, this paper proposes an algorithm based on the approach to the problems analyzed under the “virtual focal points theorem” between the complex surfaces of the conical surface lobes, which constitute the active lobes of the male and female rotors. Normal to active surfaces are defined, and also the speed in relative movement, in the “gear” process of the conical screw compressor. Regarding rotors in conical screw compressors, the circular pitch of the teeth is constant. Also, the screw pitch of the conical compressor rotors is a constant value. In these conditions, the rotors of the screw compressors are acknowledged as active surfaces, conical screw surfaces with constant axial screw pitch. The cross section (in the normal plane to the driving rotor axis) determines a plane gear. This defines the specific winding condition, and the shape of the conical screw surface of the female rotor.

Numerical Study for Comparison of Pseudo Modal and Direct Method in Predicting Critical Speed of Coaxial Dual Rotor System

Machines with high speed rotation, high power, and lighter weight are the challenge in design of modern rotation machinery. In analysis of coaxial dual rotor system, the problem leads to a multi-degree of freedom system (1multi-DOF). The resulting equation of motion cannot be solved analytically. For multi-DOF system, numerical method has to be implemented. In this study, the equations of motion of co-axial dual rotor system was solved numericallyusing MATLAB programming language. This study compares the result of two methods used: the pseudo-modal method and the direct method, with the aim of getting the most effective method in predicting critical rotational speedby plotting Campbell diagram. It was foundthat the difference in the value of critical speeds between Pseudo Modal and the Direct Method is 0%-9,56% for the first 6 lowest frequencies and 0.09%-3.58% for the first 8 lowest frequencies.

Development of a Linear Acoustic Array for Aero-Acoustic Quantification of Camber-Bladed Vertical Axis Wind Turbine.

Vertical axis wind turbines (VAWT) are a source of renewable energy and are used for both industrial and domestic purposes. The study of noise characteristics of a VAWT is an important performance parameter for the turbine. This study focuses on the development of a linear microphone array and measuring acoustic signals on a cambered five-bladed 45 W VAWT in an anechoic chamber at different tip speed ratios. The sound pressure level spectrum of VAWT shows that tonal noises such as blade passing frequencies dominate at lower frequencies whereas broadband noise corresponds to all audible ranges of frequencies. This study shows that the major portion of noise from the source is dominated by aerodynamic noises generated due to vortex generation and trailing edge serrations. The research also predicts that dynamic stall is evident in the lower Tip speed ratio (TSR) region making smaller TSR values unsuitable for a quiet VAWT. This paper compares the results of linear aeroacoustic array with a 128-MEMS acoustic camera with higher resolution. The study depicts a 3 dB margin between two systems at lower TSR values. The research approves the usage of the 8 mic linear array for small radius rotary machinery considering the results comparison with a NORSONIC camera and its resolution. These observations serve as a basis for noise reduction and blade optimization techniques.

Energy harvesting from a dynamic vibro-impact dielectric elastomer generator subjected to rotational excitations

A vibro-impact (VI) dielectric elastomer generator (DEG) subjected to rotational excitations is studied in this paper for energy harvesting (EH). The VI DEG that has been demonstrated as an advanced vibrational energy harvester is first introduced, and an impact model is proposed to experimentally validate the energy harvesting mechanism of the impact-based DEGs. The VI DEG is then considered to harvest energy from a rotational excitation by being installed onto rotational machinery. The dynamical and electrical responses of the proposed system are fully studied through both theoretical analysis and numerical simulations. It is found that the system presents rich dynamical behaviors under the rotational excitation, and the rotational speed and the distance between two membranes are two key parameters to affect the system’s response and EH performance. Research results show the superiority of the proposed system which can produce a maximal output power as high as 0.88 mW from rotational excitations, and also provide guidelines to study the rotational speed-related system EH performance with given dimensional parameters or optimize the dimensional parameters under a given rotational speed.

Ensemble deep learning with multi-objective optimization for prognosis of rotating machinery

With the emerging of Internet of Things and smart sensing techniques, enormous monitoring data has been collected by prognostics and health management (PHM) systems. Predicting the Remaining useful life (RUL) of mechanical components from monitoring data has always been a challenging task in many industries, yet determining RUL accurately is identified as one of the most demanded outcomes of PHM systems. In this study, an ensemble deep learning with multi-objective optimization (EDL-MO) method is proposed for RUL prediction. A novel ensemble deep learning algorithm for RUL prediction is designed by combining accuracy and diversity. By introducing the diversity, uncorrelated error is produced in each individual iteration, and performance of prediction will be improved by evolving deep networks. The presented EDL-MO employs evolutionary optimization to optimize the two conflicting objectives, that is, diversity and accuracy. To validate the proposed algorithm, bearing run-to-failure experiments were carried out under constant load. The vibration signals are recorded and utilized to predict the RUL by using the proposed EDL-MO method, as well as other existing methods for performance comparison. The effectiveness and superiority of EDL-MO are analyzed, which outperforms the current algorithms in predicting RUL on rotation machineries.

Development of an algorithm for reconstruction of droplet history based on deposition pattern using computational fluid dynamics and convolutional neural network

Liquids are one of the fundamental components for the functionality of a wide range of mechanical equipment (e.g. for lubrication, cooling, hydraulic, etc.). However, they could lead to secondary issues such as corrosion or contamination, caused, for instance, due to leakage of the liquid. As investigating the equipment during their operation is not always possible (e.g. for rotary machinery or in case of high-temperature working conditions), locating the origin of the leakage could be a challenging task, especially if the only traces left behind are a few droplets. In the present work, an algorithm for prediction of the leakage position, i.e. position of the injector, is developed. In order to guarantee intelligence and enhance flexibility, the designed algorithm is powered by a machine learning approach, Convolutional Neural Network. The developed algorithm is based upon using different deposition patterns, calculated by numerical simulations, as the input. The robust algorithm is designed to be so intelligent that it could ideally traceback (predict) the leakage location (i.e. reconstruct the droplet history) by only an image of the deposition.

Fault diagnosis based on the quantification of the fault features in a rotary machine

Fault diagnosis of rotary machinery is essential to fixing defective rotary machines and preventing rotary systems from breaking. Recently, fault diagnosis techniques have moved from traditional methods to artificial intelligence techniques. Many research groups have reported on the development of various artificial intelligence-based classifiers to improve diagnostic performance. In classifier design, selecting the datasets that include standard or fault features is essential to obtaining a high-performance classifier. However, there are few studies on the techniques to evaluate the quality of datasets numerically. In this study, we have developed a fault severity criterion to quantify the faultless and fault features of measured data. Using the proposed criterion, we have determined the dominant direction of representative faults in a rotary machine. The standard and fault data have been obtained, considering the dominant direction of each fault. The intrinsic mode function (IMF) that presents the fault features has been obtained by an empirical mode decomposition and a sensitive IMF selection criterion. Finally, we have designed an accurate and memory-efficient classifier using the extracted data and verified its performance by diagnosing a 7.5 kW servo motor. A conventional support vector machine has been used to verify the effects of the proposed algorithm on the classifier’s performance improvement. The developed classifier demonstrated an increase of performance up to an average of 51.9%, compared with the classifiers using training datasets measured in an arbitrary direction, with the detection rate of 99.9%. The study results suggest that the proposed classifier design technique based on the quantification of the fault features is useful for creating high-quality training datasets, machine learning, and deep learning-based classifiers.

Intelligent Condition-Based Monitoring Techniques for Bearing Fault Diagnosis

In recent years, intelligent condition-based monitoring of rotary machinery systems has become a major research focus of machine fault diagnosis. In condition-based monitoring, it is challenging to form a large-scale well-annotated dataset due to the expense of data acquisition and costly annotation. The generated data have a large number of redundant features which degraded the performance of the machine learning models. To overcome this, we have utilized the advantages of minimum redundancy maximum relevance (mRMR) and transfer learning with a deep learning model. In this work, mRMR is combined with deep learning and deep transfer learning framework to improve the fault diagnostics performance in terms of accuracy and computational complexity. The mRMR reduces the redundant information from data and increases the deep learning performance, whereas transfer learning, reduces a large amount of data dependency for training the model. In the proposed work, two frameworks, i.e., mRMR with deep learning and mRMR with deep transfer learning, have explored and validated on CWRU and IMS rolling element bearings datasets. The analysis shows that the proposed frameworks can obtain better diagnostic accuracy compared to existing methods and can handle the data with a large number of features more quickly.

Comparative performance of prognostics for remaining useful life of bearing

The frequent failure components in rotary machinery are the rolling element bearings as it is important for prognostics of rolling element bearings. In this study, the data-driven technique was used to develop the prognostic models based on particle swarm optimization techniques. These models apply to estimate the rolling element bearing degradation and predict remaining useful life. Initially, the fault features were extracted by processing the vibration signals through wavelet packet decomposition based on the intensifying impulsive characteristics and superiority of features. Health indicators evaluate from fault features. The prognostic models are developed with the following approaches: Gated recurrent neural network, classification and regression tree, and Autoregressive-Moving Average models. Performances of models are verified on the basis of error in prediction. To verify the suggested methodology, an experiment on normal and faulty bearings was conducted using a bearing test rig. Experimental results clarify that the prognostic algorithms predict bearing remaining useful life with significant robustness. Outperformance of the proposed method and conformation of degree of accuracy from results indicate that remaining useful life is conjectured as inference compared with the published literature.

Fault Diagnosis of Rotary Machines Using Deep Convolutional Neural Network with Wide Three Axis Vibration Signal Input.

Fault diagnosis is considered as an essential task in rotary machinery as possibility of an early detection and diagnosis of the faulty condition can save both time and money. This work presents developed and novel technique for deep-learning-based data-driven fault diagnosis for rotary machinery. The proposed technique input raw three axes accelerometer signal as high definition 1D image into deep learning layers which automatically extract signal features, enabling high classification accuracy. Unlike the researches carried out by other researchers, accelerometer data matrix with dimensions 6400 × 1 × 3 is used as input for convolutional neural network training. Since convolutional neural networks can recognize patterns across input matrix, it is expected that wide input matrix containing vibration data should yield good classification performance. Using convolutional neural networks (CNN) trained model, classification in one of the four classes can be performed. Additionally, number of kernels of CNN is optimized using grid search, as preliminary studies show that alternating number of kernels impacts classification results. This study accomplished the effective classification of different rotary machinery states using convolutional artificial neural network for classification of raw three axis accelerometer signal input.