Learning-based Fault Diagnosis Methods Research Articles

An explainable one-dimensional convolutional neural networks based fault diagnosis method for building heating, ventilation and air conditioning systems

Due to the frequently changed outdoor weather conditions and indoor requirements, heating, ventilation and air conditioning (HVAC) experiences faulty operations inevitably throughout its lifespan. Therefore, it is important to monitor and diagnose HVAC fault operations. Recently, deep learning methods have attracted more attentions for their guarantee of better diagnosis performance under various system configurations and operating conditions. However, these methods are black-box models which though highly accurate for fault diagnosis but are extremely hard to explain. To overcome the disadvantage of poor interpretability of deep learning black-box models, this study therefore proposes a novel explainable deep learning based fault diagnosis method that is suitable for HVACs. To maintain HVAC operational information and variable locations of all chiller input data samples, proposed method is established with three characteristics: 1) the pooling layer is excluded, 2) the size of convolution filter kernel is set as 1, and 3) use softsign as an activation function. Considering the resulting impacts of HVAC faults on system operating variables, a new Absolute Gradient-weighted Class Activation Mapping (Grad-Absolute-CAM) method is proposed to visualize the fault diagnosis criteria and make the model explainable by providing the fault-discriminative information. The proposed method is validated using fault experimental dataset of a typical building HVAC system (i.e., chiller) from the ASHRAE research project 1043 (RP-1043). The fault diagnosis accuracy is over 98.5% for seven chiller faults. Results indicates that it is capable of interpreting the model work mechanism by activation feature maps and explaining the fault diagnosis criteria by Grad-Absolute-CAM.

Attention Recurrent Autoencoder Hybrid Model for Early Fault Diagnosis of Rotating Machinery

Early fault diagnosis of rotating machinery is crucial in the industry. The network parameters of the traditional deep learning-based fault diagnosis method are optimized only by the single loss function, and the extracted features are usually not the most valuable features of the input signal. This article proposes a novel method, attention recurrent autoencoder (AE) hybrid model classification algorithm, for early fault diagnosis and severity detection of rotating machinery. The AE can learn the most valuable features in an unsupervised way. By adjusting the weight proportion of the two loss functions and optimizing the multibranch network simultaneously, the proposed method enables the network to extract the most valuable features of the input signals. Moreover, the proposed method can take the raw 1-D vibration signal as the input and does not need time-frequency conversion. By introducing long short-term memory networks in AE, the time-dependent features of the data can be extracted effectively. The proposed method was verified by five kinds of gears at different pitting degrees under six load conditions. The results indicate that the method can provide simultaneous accurate fault diagnosis and severity detection for different pitting degrees.

Deep Learning-Based Bearing Fault Diagnosis Method for Embedded Systems.

Bearing elements are vital in induction motors; therefore, early fault detection of rolling-element bearings is essential in machine health monitoring. With the advantage of fault feature representation techniques of time–frequency domain for nonstationary signals and the advent of convolutional neural networks (CNNs), bearing fault diagnosis has achieved high accuracy, even at variable rotational speeds. However, the required computation and memory resources of CNN-based fault diagnosis methods render it difficult to be compatible with embedded systems, which are essential in real industrial platforms because of their portability and low costs. This paper proposes a novel approach for establishing a CNN-based process for bearing fault diagnosis on embedded devices using acoustic emission signals, which reduces the computation costs significantly in classifying the bearing faults. A light state-of-the-art CNN model, MobileNet-v2, is established via pruning to optimize the required system resources. The input image size, which significantly affects the consumption of system resources, is decreased by our proposed signal representation method based on the constant-Q nonstationary Gabor transform and signal decomposition adopting ensemble empirical mode decomposition with a CNN-based method for selecting intrinsic mode functions. According to our experimental results, our proposed method can provide the accuracy for bearing faults classification by up to 99.58% with less computation overhead compared to previous deep learning-based fault diagnosis methods.

Deep Transfer Learning-Based Fault Diagnosis for Gearbox under Complex Working Conditions

In the large amount of available data, information insensitive to faults in historical data interferes in gear fault feature extraction. Furthermore, as most of the fault diagnosis models are learned from offline data collected under single/fixed working condition only, this may cause unsatisfactory performance for complex working conditions (including multiple and unknown working conditions) if not properly dealt with. This paper proposes a transfer learning-based fault diagnosis method of gear faults to reduce the negative effects of the abovementioned problems. In the proposed method, a cohesion evaluation method is applied to select sensitive features to the task with a transfer learning-based sparse autoencoder to transfer the knowledge learnt under single working condition to complex working conditions. The experimental results on wind turbine drivetrain diagnostics simulator show that the proposed method is effective in complex working conditions and the achieved results are better than those of traditional algorithms.

Bearing Fault Diagnosis Based on Multilayer Domain Adaptation

Bearing fault diagnosis plays a vitally important role in practical industrial scenarios. Deep learning-based fault diagnosis methods are usually performed on the hypothesis that the training set and test set obey the same probability distribution, which is hard to satisfy under the actual working conditions. This paper proposes a novel multilayer domain adaptation (MLDA) method, which can diagnose the compound fault and single fault of multiple sizes simultaneously. A special designed residual network for the fault diagnosis task is pretrained to extract domain-invariant features. The multikernel maximum mean discrepancy (MK-MMD) and pseudo-label learning are adopted in multiple layers to take both marginal distributions and conditional distributions into consideration. A total of 12 transfer tasks in the fault diagnosis problem are conducted to verify the performance of MLDA. Through the comparisons of different signal processing methods, different parameter settings, and different models, it is proved that the proposed MLDA model can effectively extract domain-invariant features and achieve satisfying results.

Hierarchical deep convolution neural networks based on transfer learning for transformer rectifier unit fault diagnosis

Convolutional neural networks (CNNs) are able to extract features automatically. Fault identification and location of transformer rectifier units (TRUs) which are widely used as an avionic secondary power supply are significant for system reliability. Based on the analyzation of TRUs, this paper discusses the design process of the developed discrete time-series convolution neural network (DTCNN) and develops a hierarchical method for TRUs fault diagnosis along with a transfer learning-based fault diagnosis method instead of training new models for different TRUs. The DTCNN construction is determined firstly. Then, the performance of HDCNN is validated. On the basis, the conditions of a suitable source dataset for TRU fault diagnosis and the transfer layers from the pre-trained HDCNN are discussed. The comparison with other algorithms under different noise conditions shown that the transfer learning is an effective way to construct the diagnosis network for similar equipment and often can lead to better performance.

Unsupervised Deep Transfer Learning for Fault Diagnosis in Fog Radio Access Networks

The rapid development of the Internet of Things with the requirements of ultrareliability and ultralow latency has imposed huge challenges on the radio access network operation and maintenance. Using artificial intelligence technologies can provide the accurate fault diagnosis rapidly and efficiently, but it is usually hampered by the lack of historical data as well as the certified fault labels. To deal with these challenges, in this article, an unsupervised deep transfer learning-based fault diagnosis method in fog radio access networks is proposed. Specifically, a transfer learning-based density-based spatial clustering of applications with noise method is first utilized to detect and label fault data in each interval by using the core-level information. Then, an unsupervised deep transfer learning method combining a convolutional neural network with a domain adversarial neural network is applied to classify the categories of unlabeled fault data by using cell-level information. The experimental results show that the proposed method can reduce the missed detection rate than the traditional method, and has better fault diagnosis accuracy than the reference methods.

Ensemble sparse supervised model for bearing fault diagnosis in smart manufacturing

Abstract Machinery fault diagnosis is of great significance to improve the reliability of smart manufacturing. Deep learning based fault diagnosis methods have achieved great success. However, the features extracted by different models may vary resulting in ambiguous representation of the data, and even wasted time with manually selecting the optimal hyperparameters. To solve the problems, this paper proposes a new framework named Ensemble Sparse Supervised Model (ESSM), in which a typical deep learning model is treated as two phases of feature learning and model learning. In the feature learning phase, the original data is represented to be a feature matrix as non-redundant as possible by applying sparse filtering. Then, the feature matrix is fed into the model learning phase. Regularization, dropout and rectified linear unit (ReLU) are used in the model's neurons and layers to build a sparse deep neural network. Finally, the output of the sparse deep neural network provides feedback to the first phase to obtain better sparse features. In the proposed method, hyperparameters need to be pre-specified and a python library of talos is employed to finish the process automatically. The proposed method is verified using the bearing data provided by Case Western Reserve University. The result demonstrates that the proposed method can capture the effective pattern of data with the help of sparse constraints and simultaneously provide convenience for the operators with assuring performance.

Fault Diagnosis under Variable Working Conditions Based on STFT and Transfer Deep Residual Network

Fault diagnosis plays a very important role in ensuring the safe and reliable operations of machines. Currently, the deep learning-based fault diagnosis is attracting increasing attention. However, fault diagnosis under variable working conditions has been a significant challenge due to the domain discrepancy problem. This problem is also unavoidable in deep learning-based fault diagnosis methods. This paper contributes to the ongoing investigation by proposing a new approach for the fault diagnosis under variable working conditions based on STFT and transfer deep residual network (TDRN). The STFT was employed to convert vibration signal to time-frequency image as the input of the TDRN. To address the domain discrepancy problem, the TDRN was developed in this paper. Unlike traditional deep convolutional neural network (DCNN) methods, by combining with transfer learning, the TDRN can make a bridge between two different working conditions, thereby using the knowledge learned from a working condition to achieve a high classification accuracy in another working condition. Moreover, since the residual learning is introducing, the TDRN can overcome the problems of training difficulty and performance degradation existing in traditional DCNN methods, thus further improving the classification accuracy. Experiments are conducted on the popular CWRU bearing dataset to validate the effectiveness and superiority of the proposed approach. The results show that the developed TDRN outperforms those methods without transfer learning and/or residual learning in terms of the accuracy and feature learning ability for the fault diagnosis under variable working conditions.

Assessment of characteristics of acoustic emission parameters for valve damage detection under varying compressor speeds

Acoustic emission technique is often employed to detect valve abnormalities. With the development of technology, machine learning-based fault diagnosis methods are prevalent in the nondestructive testing industry as they can automatically detect valve problems without any human intervention. Nevertheless, feeding in all possible input parameters into the learning algorithm without any prior assessment may result in high computational cost and time, while adding to the risk of having false alarms. This study intended to obtain characteristics of acoustic emission signal for various valve conditions and compressor speeds by examining the four most commonly used parameters, namely the acoustic emission root mean square, acoustic emission crest factor, acoustic emission variance, and acoustic emission kurtosis. The study begins with time–frequency analysis of one revolution acoustic emission signal acquired from a faulty suction valve through discrete wavelet transform to obtain the signal characteristics of valve events. To associate signals with valve movements, the reconstructed discrete wavelet transform signals are further segregated into six time segments, and the four acoustic emission parameters are computed from each of the time segments. These parameters are analyzed through statistical analysis namely the two-way analysis of variance, followed by the Tukey test to obtain the best parameter which can differentiate each valve condition clearly at all speeds. The results revealed that acoustic emission root mean square is the best parameter especially in identification of heavy grease valve condition during suction valve opening event while acoustic emission crest factor is capable to detect leaky valve during the suction valve closing event at all speeds. It is believed that effective valve diagnosis strategy can be delivered by referring to the features of parameters and the characteristic valve event timing corresponding to each valve condition and speed.

Deep Learning-Based Partial Domain Adaptation Method on Intelligent Machinery Fault Diagnostics

In the past years, deep learning-based machinery fault diagnosis methods have been successfully developed, and the basic diagnostic problems have been well addressed where the training and testing data are collected under the same operating conditions. When the training and testing data are from different distributions, domain adaptation approaches have been introduced. However, the existing methods generally assume the availability of the target-domain data in all the health conditions during training, which is not in accordance with the real industrial scenarios. This article proposes a deep learning-based fault diagnosis method to address the partial domain adaptation problems, where the unsupervised target-domain training data do not cover the full machine health state label space. The conditional data alignment and unsupervised prediction consistency schemes are proposed to achieve partial domain adaptation. The experimental results on two rotating machinery datasets suggest the proposed method offers a promising tool for this practical industrial problem.

Deep representation clustering-based fault diagnosis method with unsupervised data applied to rotating machinery

Despite the recent advances on intelligent data-driven machinery fault diagnostics, large amounts of high-quality supervised data are mostly required for model training. However, it is usually difficult and expensive to collect sufficient labeled data in real industries, and the difficulty in data preparation significantly hinders the application of the intelligent diagnostic methods. In order to address the data sparsity issue with insufficient labeled data, a deep learning-based fault diagnosis method is proposed in this study, exploring additional unsupervised data which are generally easy for collection. A three-stage training scheme is adopted, i.e. pre-training, representation clustering and enhanced supervised learning. The auto-encoder structure is used for feature extraction, and distance metric learning and k-means clustering method are integrated in the neural network architecture for unsupervised learning. Two rotating machinery datasets are used for validations. The proposed method not only achieves promising diagnostic performance on the semi-supervised learning tasks with few labeled data, but also is well suited for pure unsupervised learning problems. The experimental results suggest the proposed method offers a promising approach on exploiting unsupervised data for fault diagnostics.

Single- and Multi-Fault Diagnosis Using Machine Learning for Variable Frequency Drive-Fed Induction Motors

In this article, an effective machine learning-based fault diagnosis method is developed for induction motors driven by variable frequency drives (VFDs). Two identical 0.25 HP induction motors under healthy, single-, and multifault conditions were tested in the lab with different VFD output frequencies and motor loadings. The stator current and the vibration signals of the motors were recorded simultaneously under steady-state for each test, and both signals are evaluated for their suitability for fault diagnosis. The signal processing technique, discrete wavelet transform, is chosen in this article to extract features for machine learning. Four families of machine learning algorithms in the MATLAB Classification Learner Toolbox, decision trees, support vector machines, k -nearest neighbors, and ensemble, with 20 classifiers are evaluated for their classification accuracy when used for fault diagnosis of induction motors fed by VFDs. To allow fault diagnosis for untested motor operating conditions, the feature calculation formulas are developed through surface fitting using experimental data of a range of tested frequencies and loadings of the motor to provide training data for untested conditions.

Deep Forest-Based Fault Diagnosis Method for Chemical Process

With the rapid expanding of big data in all domains, data-driven and deep learning-based fault diagnosis methods in chemical industry have become a major research topic in recent years. In addition to a deep neural network, deep forest also provides a new idea for deep representation learning and overcomes the shortcomings of a deep neural network such as strong parameter dependence and large training cost. However, the ability of each base classifier is not taken into account in the standard cascade forest, which may lead to its indistinct discrimination. In this paper, a multigrained scanning-based weighted cascade forest (WCForest) is proposed and has been applied to fault diagnosis in chemical processes. In view of the high-dimensional nonlinear data in the process of chemical industry, WCForest first designs a set of relatively suitable windows for the multigrained scan strategy to learn its data representation. Next, considering the fitting quality of each forest classifier, a weighting strategy is proposed to calculate the weight of each forest in the cascade structure without additional calculation cost, so as to improve the overall performance of the model. In order to prove the effectiveness of WCForest, its application has been carried out in the benchmark Tennessee Eastman (TE) process. Experiments demonstrate that WCForest achieves better results than other related approaches across various evaluation metrics.

A deep convolutional neural network approach with information fusion for bearing fault diagnosis under different working conditions

In recent years, deep learning-based fault diagnosis methods have drawn lots of attention. However, for most cases, the success of machine learning-based models relies on the circumstance that training data and testing data are under the same working condition, which is too strict for real implementation cases. Combined with the features of robustness of deep convolutional neural network and vibration signal characteristics, information fusion technology is introduced in this study to enhance the feature representation capability as well as the transferability of diagnosis models. With the basis of multi-sensors and narrow-band decomposition techniques, a convolutional architecture named fusion unit is proposed to extract multi-scale features from different sensors. The proposed method is tested on two data sets and has achieved relatively higher generalization ability when compared with several existing works, which demonstrates the effectiveness of our proposed fusion unit for feature extraction on both source task and target task.

A Deep Learning Based Fault Diagnosis Method With Hyperparameter Optimization by Using Parallel Computing

Bearing fault diagnosis is of great significance to ensure the safe operation of mechanical equipment. This paper proposes an intelligent fault diagnosis method of rolling bearings based on deep belief network (DBN) with hyperparameter optimization by using parallel computing. Different with traditional diagnosis methods that extract the features manually depending on much prior knowledge about signal processing techniques and diagnostic expertise, DBN extracts fault features automatically by machine learning mechanism. Considering the time consuming problem, parallel computing is adopted to the DBN training process by using a Master/Slave mode to improve the training speed so that the global optimization with Genetic Algorithm and higher diagnosis accuracy can be achieved. Finally, the proposed method is verified with the public datasets provided by Case Western Reserve University (CWRU) with various fault depths in different locations and loads of rolling bearings. The results indicate that the proposed method can identify bearing faults under different conditions correctly which significantly enhances the intelligence of fault classification and reduces the time for parameter selection of deep learning models.

Machinery fault diagnosis with imbalanced data using deep generative adversarial networks

Despite the recent advances of intelligent data-driven fault diagnosis methods on rotating machines, balanced training data for different machine health conditions are assumed in most studies. However, the signals in machine faulty states are usually difficult and expensive to collect, resulting in imbalanced training dataset in most cases. That significantly deteriorates the effectiveness of the existing data-driven approaches. This paper proposes a deep learning-based fault diagnosis method to address the imbalanced data problem by explicitly creating additional training data. Generative adversarial networks are firstly used to learn the mapping between the distributions of noise and real machinery temporal vibration data, and additional realistic fake samples can be generated to balance and further expand the available dataset afterwards. Through experiments on two rotating machinery datasets, it is validated that the data-driven methods can significantly benefit from the data augmentation, and the proposed method offers a promising tool on fault diagnosis with imbalanced training data.

Fault diagnosis of rotating machinery with ensemble kernel extreme learning machine based on fused multi-domain features

Accurate and reliable fault diagnosis for rotating machinery, especially under variable working conditions remains a great challenge. Existing deep learning methods which extract features from single domain are insufficient to ensure reliable diagnosis results. In this study, a new deep learning based fault diagnosis method, which extracts features from both time and frequency domains is proposed. Two sets of deep features from multiple domains are fused into intrinsic low-dimensional features by local and global principle component analysis. And a new ensemble kernel extreme learning machine is proposed for fault pattern classification based on the fused features. Extensive experiments on gearbox, rotor and engine rolling bearing show that the proposed method has better diagnosis performance than state-of-the-art methods and is more adaptable to the fluctuation of working conditions.

Machine Learning-Based Fault Diagnosis for Single- and Multi-Faults in Induction Motors Using Measured Stator Currents and Vibration Signals

In this paper, a practical machine learning-based fault diagnosis method is proposed for induction motors using experimental data. Various single- and multi-electrical and/or mechanical faults are applied to two identical induction motors in lab experiments. Stator currents and vibration signals of the motors are measured simultaneously during experiments and are used in developing the fault diagnosis method. Two signal processing techniques, matching pursuit, and discrete wavelet transform, are chosen for feature extraction. Three classification algorithms, support vector machine (SVM), K-nearest neighbors (KNN), and ensemble, with 17 different classifiers offered in MATLAB Classification Learner toolbox are used in the study to evaluate the performance and suitability of different classifiers for induction motor fault diagnosis. It is found that five classifiers (fine Gaussian SVM, fine KNN, weighted KNN, bagged trees, and subspace KNN) can provide near 100% classification accuracy for all faults applied to each motor, but the remaining 12 classifiers do not perform well. A novel curve fitting technique is developed to calculate features for the motors that stator currents or vibration signals under certain loadings are not tested for a particular fault. The proposed fault diagnosis method can accurately detect single- or multi-electrical and mechanical faults in induction motors.

Transfer Learning Based Fault Diagnosis with Missing Data Due to Multi-Rate Sampling.

Deep learning is an effective feature extraction method widely applied in fault diagnosis fields since it can extract fault features potentially involved in multi-sensor data. But different sensors equipped in the system may sample data at different sampling rates, which will inevitably result in a problem that a very small number of samples with a complete structure can be used for deep learning since the input of a deep neural network (DNN) is required to be a structurally complete sample. On the other hand, a large number of samples are required to ensure the efficiency of deep learning based fault diagnosis methods. To solve the problem that a structurally complete sample size is too small, this paper proposes a fault diagnosis framework of missing data based on transfer learning which makes full use of a large number of structurally incomplete samples. By designing suitable transfer learning mechanisms, extra useful fault features can be extracted to improve the accuracy of fault diagnosis based simply on structural complete samples. Thus, online fault diagnosis, as well as an offline learning scheme based on deep learning of multi-rate sampling data, can be developed. The efficiency of the proposed method is demonstrated by utilizing data collected from the QPZZ- II rotating machinery vibration experimental platform system.