Fig 3 - uploaded by Juan Carlos Quiroz
Content may be subject to copyright.
Source publication
This paper proposes a new approach to diagnose broken rotor bar failure in a line start-permanent magnet synchronous motor (LS-PMSM) using random forests. The transient current signal during the motor startup was acquired from a healthy motor and a faulty motor with a broken rotor bar fault. We extracted 13 statistical time domain features from the...
Contexts in source publication
Context 1
... squirrel-cage bars in LS-PMSM develop the startup performance during motor run up by enabling the rotor to have direct-on-line movement. When the load on the motor is unbalanced or the rotation speed is fluctuated, the squirrel-cage bars lessen the counter-rotating fields of the air gap, which otherwise would lead to significant losses [29]. Fig. 3 depicts the cross-section of one pole of a three-phase, four pole LS-PMSM. The rotor bar produces the starting torque as a result of the induction current in the bars establishing an electromagnetic field, which interacts with the rotatory field and subsequently pulls the motor toward the synchronism. Despite the induced current, the ...
Context 2
... squirrel-cage bars in LS-PMSM develop the startup performance during motor run up by enabling the rotor to have direct-on-line movement. When the load on the motor is unbalanced or the rotation speed is fluctuated, the squirrel-cage bars lessen the counter-rotating fields of the air gap, which otherwise would lead to significant losses [29]. Fig. 3 depicts the cross-section of one pole of a three-phase, four pole LS-PMSM. The rotor bar produces the starting torque as a result of the induction current in the bars establishing an electromagnetic field, which interacts with the rotatory field and subsequently pulls the motor toward the synchronism. Despite the induced current, the ...
Context 3
... squirrel-cage bars in LS-PMSM develop the startup performance during motor run up by enabling the rotor to have direct-on-line movement. When the load on the motor is unbalanced or the rotation speed is fluctuated, the squirrel-cage bars lessen the counter-rotating fields of the air gap, which otherwise would lead to significant losses [29]. Fig. 3 depicts the cross-section of one pole of a three-phase, four pole LS-PMSM. The rotor bar produces the starting torque as a result of the induction current in the bars establishing an electromagnetic field, which interacts with the rotatory field and subsequently pulls the motor toward the synchronism. Despite the induced current, the ...
Similar publications
Very high resolution canopy cover maps of spekboom are required to assist with the restoration of degraded habitat in the Little Karoo, a large semi-arid region in South Africa. Variations in habitat and level of degradation, in addition to radiometric variations in the imagery, make spekboom mapping at a regional scale a challenging problem. In th...
Objective: Algorithms to predict shock outcome based on ventricular fibrillation (VF) waveform features are potentially useful tool to optimize defibrillation strategy (immediate defibrillation versus cardiopulmonary resuscitation). Researchers have investigated numerous predictive features and classification methods using single VF feature and/or...
Screening shows impact on cancer mortality rate by decreasing the number of advanced cancers with poor diagnosis, while cancer treatment works through decreasing the case-fatality rate. The prediction of breast cancer survivability has been a challenging research problem for many researchers. The objective of this research work is to propose a Nove...
The present strategies for the prediction of price and load may be difficult to deal with huge amount of load and price data. To resolve the problem, three modules are incorporated within the model. Firstly, the fusion of Decision Tree (DT) and Random Forest (RF) are used for feature selection and to remove the redundancy among feature. Secondly, R...
Shape recognition has been a popular application of machine learning, where each shape is defined as a class for training classifiers that recognize the shapes of new instances. Since training of classifiers is essentially achieved through learning from features, it is crucial to extract and select a set of relevant features that can effectively di...
Citations
... Signal processing methods mainly include wavelet transform (WT) [2,3], local mean decomposition (LMD) [4], empirical modal decomposition (EMD) [5], and short-time Fourier transform (STFT) [6,7]. The main shallow machine learning methods are support vector machine (SVM) [8,9], k-nearest neighbor (KNN) [10], and random forest (RF) [11]. Zheng et al. [12] implemented fault classification by calculating the composite multiscale fuzzy entropy (CMFE) of vibration signals based on ensemble support vector machine (ESVM). ...
Fault diagnosis is one of the important applications of edge computing in the Industrial Internet of Things (IIoT). To address the issue that traditional fault diagnosis methods often struggle to effectively extract fault features, this paper proposes a novel rolling bearing fault diagnosis method that integrates Gramian Angular Field (GAF), Convolutional Neural Network (CNN), and Vision Transformer (ViT). First, GAF is used to convert one-dimensional vibration signals from sensors into two-dimensional images, effectively retaining the fault features of the vibration signal. Then, the CNN branch is used to extract the local features of the image, which are combined with the global features extracted by the ViT branch to diagnose the bearing fault. The effectiveness of this method is validated with two datasets. Experimental results show that the proposed method achieves average accuracies of 99.79% and 99.63% on the CWRU and XJTU-SY rolling bearing fault datasets, respectively. Compared with several widely used fault diagnosis methods, the proposed method achieves higher accuracy for different fault classifications, providing reliable technical support for performing complex fault diagnosis on edge devices.
... Applying machine learning to detect and diagnose faults from data is simple and does not require explicit models. Shallow neural networks mainly include logistic regression [102], support vector machines (SVM) [103], random forest [104], the k-nearest neighbor algorithm [105], and naive Bayes [106], which can adaptively learn features without necessitating the creation of a precise mathematical model, thereby reducing the uncertainty and intricacy associated with human involvement. Nevertheless, conventional shallow neural networks have limitations, including gradient vanishing, overfitting, susceptibility to local minima, and the requirement for substantial prior information, all of which diminish the efficacy of fault diagnosis. ...
This article provides a comprehensive overview of state-of-the-art techniques for detecting and diagnosing stator winding inter-turn short faults (ITSFs) in permanent-magnet synchronous motors (PMSMs) for electric vehicles (EVs). The review focuses on the following three main categories of diagnostic approaches: motor model-based, signal processing-based, and artificial intelligence (AI)-based fault detection and diagnosis methods. Motor model-based methods utilize motor state estimation and motor parameter estimation as the primary strategies for ITSF diagnosis. Signal processing-based techniques extract fault signatures from motor measured data across time, frequency, or time-frequency domains. In contrast, AI-based methods automatically extract higher-order fault signatures from large volumes of preprocessed data, thereby enhancing the effectiveness of fault diagnosis. The strengths and limitations of each approach are thoroughly examined, providing valuable insights into the advancements in ITSF detection and diagnosis techniques for PMSMs in EV applications. The emphasis is placed on the application of signal processing methods and deep learning techniques in the diagnosis of ITSF in PMSMs in EV applications.
... The third layer includes typical core AI algorithms used in industry [5], such as linear regression (e.g., for material usage optimization), logistic regression (e.g., for simplifying tomography system), support vector machine (SVM) (e.g., for fault diagnosis [30]), Naive Bayes (e.g., for intrusion detection [31]), decision tree (e.g., for quality disturbance detection [32]), K-nearest neighbor (e.g., for traffic flow prediction [33]), K-means (e.g., for wind power prediction [34]), random forest (e. g., for machinery fault diagnosis [35]), dimensionality reduction (e.g., for wind turbine state monitoring [36]), and neural network (e.g., for flexible joint manipulator control [37]). These algorithms can be applied to various industrial scenarios. ...
With the continuous development of human-centric, resilient, and sustainable manufacturing towards Industry 5.0, Artificial Intelligence (AI) has gradually unveiled new opportunities for additional functionalities, new features, and tendencies in the industrial landscape. On the other hand, the technology-driven Industry 4.0 paradigm is still in full swing. However, there exist many unreasonable designs, configurations, and implementations of Industrial Artificial Intelligence (IndAI) in practice before achieving either Industry 4.0 or Industry 5.0 vision, and a significant gap between the individualized requirement and actual implementation result still exists. To provide insights for designing appropriate models and algorithms in the upgrading process of the industry, this perspective article classifies IndAI by rating the intelligence levels and presents four principles of implementing IndAI. Three significant opportunities of IndAI, namely, collaborative intelligence, self-learning intelligence, and crowd intelligence, towards Industry 5.0 vision are identified to promote the transition from a technology-driven initiative in Industry 4.0 to the coexistence and interplay of Industry 4.0 and a value-oriented proposition in Industry 5.0. Then, pathways for implementing IndAI towards Industry 5.0 together with key empowering techniques are discussed. Social barriers, technology challenges, and future research directions of IndAI are concluded, respectively. We believe that our effort can lay a foundation for unlocking the power of IndAI in futuristic Industry 5.0 research and engineering practice.
... Monferrer et al. [16] used the energy consumption of the spindle as an indicator for the real-time monitoring of the CFRP drilling process and detected multiple types of anomalies in the drilling process. Selvaraj et al. [17] and Quiroz et al. [18] collected and extracted energy consumption data features from systems operating in different states. They then constructed detection models using these features to differentiate between normal and abnormal operation modes of the systems, validating the effectiveness of using energy consumption data as the monitoring parameter. ...
Various abnormal scenarios might occur during the shield tunneling process, which have an impact on construction efficiency and safety. Existing research on shield tunneling construction anomaly detection typically designs models based on the characteristics of a specific anomaly, so the scenarios of anomalies that can be detected are limited. Therefore, the research objective of this article is to establish an accurate anomaly detection model with generalization and identification capabilities on multiple types of abnormal scenarios. Inspired by energy dissipation theory, this paper innovatively detects various anomalies in the shield tunneling process from the perspective of energy consumption and designs the AD_SI model (Anomaly Detection and Scenario Identification model of shield tunneling) based on machine learning. The AD_SI model first monitors the shield machine’s energy consumption status based on the VAE-LSTM (Variational Autoencoder–Long Short-Term Memory) algorithm with a dynamic threshold, thereby detecting abnormal sections. Secondly, the AD_SI model uses the correlation of construction parameters to represent different known scenarios and further clarifies scenarios of the abnormal sections, thus achieving anomaly identification. The application of the AD_SI model in a shield tunneling construction project demonstrates its capability to accurately detect and identify different anomalies, with a recall value exceeding 0.9 and F1 exceeding 0.8, thereby providing guidance for accurately detecting multiple types anomaly scenarios in practical applications.
... The new classification result is obtained by judging the selection result of each class by the decision tree. The subset of elements used to decide the ideal node splits approves weaker elements' representation in RF [24]. The Low correlation between trees in RF is executed by randomization of bootstrap sampling. ...
Inter-turn short circuit (ITSC) is a frequent fault of interior permanent magnet synchronous motors (IPMSM). If ITSC faults are not promptly monitored, it may result in secondary faults or even cause extensive damage to the entire motor. To enhance the reliability of IPMSMs, this paper introduces a fault diagnosis method specifically designed for identifying ITSC faults in IPMSMs. The sparse coefficients of phase current and torque are solved by clustering shrinkage stage orthogonal matching tracking (CcStOMP) in the greedy tracking algorithm. The CcStOMP algorithm can extract multiple target atoms at one time, which greatly improves the iterative efficiency. The multiple features are utilized as input parameters for constructing the random forest classifier. The constructed random forest model is used to diagnose ITSC faults with the results showing that the random forest model has a diagnostic accuracy of 98.61% using all features, and the diagnostic accuracy of selecting three of the most important features is still as high as 97.91%. The random forest classification model has excellent robustness that maintains high classification accuracy despite the reduction of feature vectors, which is a great advantage compared to other classification algorithms. The combination of greedy tracing and the random forest is not only a fast diagnostic model but also a model with good generalisation and anti-interference capability. This non-invasive method is applicable to monitoring and detecting failures in industrial PMSMs.
... The detection and the classification of ISCF were estimated with 88%-99% accuracy. In Ref. [32], random forests (RF) were used in ML algorithms to detect broken rotor failure in PMSM. In the time domain, 13 statistical features obtained from the transient stator current were used as input data for RF and the RF algorithm estimated whether the motor was healthy or the broken rotor faulty with 98.4% accuracy [32]. ...
... In Ref. [32], random forests (RF) were used in ML algorithms to detect broken rotor failure in PMSM. In the time domain, 13 statistical features obtained from the transient stator current were used as input data for RF and the RF algorithm estimated whether the motor was healthy or the broken rotor faulty with 98.4% accuracy [32]. In Ref. [33], the algorithms ANN, SVM, DT, RF and KNN are used to detect ISCF in induction motors with different turn faults. ...
This article presents a novel approach based on the electromechanical torque signal for the inter-turn short-circuit fault (ISCF) detection and the ISCF severity estimation in permanent magnet synchronous motors (PMSMs). The electromechanical torque data have been obtained experimentally in the healthy condition and in three various states of the ISCF at various load rates and at various operating speeds. To extract the features to be used in the ISCF diagnosis, the fast Fourier transform (FFT) implemented to the torque signal. The torque’s second and fourth harmonics were found to be new turn fault features that could be used for ISCF diagnosis. These features were used to train and test the classification algorithms. Four classification algorithms were used to detect ISCF and determine the severity of ISCF: decision trees (DT), artificial neural networks (ANN), K-nearest neighbor (KNN) and support vector machines (SVM). Classification accuracies of 100%, 99.30%, 97.91% and 95.48% were achieved by the ANN, SVM, KNN and DT classifiers, respectively. High accuracy ISCF detection and high accuracy ISCF severity estimation were performed using the developed diagnostic method based on the torque signal.
... The development of AI-based motor systems has effectively made use of ML algorithms like k-Nearest Neighbor (k-NN) (Casimir et al., 2006), Support Vector Machines (SVM) Martínez-Morales et al., 2018), Artificial Neural Networks (ANN) (Han et al., 2006;Su and Chong, 2007;Sun et al., 2016), Bayesian Classifier (BC) (Baraldi et al., 2015), Decision Trees (DT) , Random Forest (RF) (Quiroz et al., 2018), and Seep Learning (DL) (Janssens et al., 2016). Recently, DL and ML have been used for recognition systems by motor researchers (Lei et al., 2016;Cheon et al., 2015;Hausmann et al., 2021;Caramiaux et al., 2020;Elsrogy et al., 2013). ...
... The new classification result is obtained by judging the selection result of each class by the decision tree. The subset of elements used to decide the ideal node splits approves weaker elements' representation in RF [36]. The low correlation between trees in RF is executed by randomization of bootstrap sampling. ...
Permanent magnet synchronous motors (PMSMs) are extensively utilized in production and manufacturing fields due to their wide speed range, high output torque, fast speed response, small size and light weight. PMSMs are susceptible to inter-turn short circuit faults, demagnetization faults, bearing faults, and other faults arising from irregular vibrations and frequent start–brake cycles. While fault diagnosis for PMSMs offers an effective means to enhance operational efficiency, the multi-sensor information fusion is often overlooked. In industrial production processes, the collected data inevitably suffers from noise contamination, which can adversely impact diagnostic outcomes. To enhance the robustness of diagnostic methods in noisy environments and mitigate the risk of overfitting, a PMSM fault diagnosis method based on image features of multi-sensor fusion is proposed. Firstly, the vibration acceleration signals of the PMSM at different positions were acquired. Then, the newly designed multi-signal Gramian Angular Difference Fields (MGADF) method combines sensor signals from three different installation locations into a single image. Next, the multi-texture features are fused to extract the features of the image. Various machine models are compared in the fault feature learning and classification, and the results show that the proposed diagnostic method has good diagnostic accuracy and robustness, with an average diagnostic accuracy of 99.54% and a standard deviation of accuracy of 0.19. It has excellent performance even in noisy environments. The method is non-invasive and can be extended and applied to the condition monitoring and diagnosis of industrial motors.
... One way to put explainability first is to prioritize more explainable methods over less explainable ones. An example of this for PdM is Quiroz et al. (2018) who use machine learning to detect faults in rotor bars. They trained a random forest algorithm on 13 features and realized that the accuracy of the prediction was not different from a model trained on the two most influential features. ...
... This shows that simplicity can sometimes be as efficient as complexity and can drastically improve explainability. Quiroz et al. (2018) additionally compared the performance of a random forest algorithm with other methods such as decision trees, logistic regression and linear ridge using these two influential features. While the accuracy of the random forest algorithm was of 99.6%, the accuracy of the logistic regression was 93.8% and that of the linear ridge 93.8%. ...
... On two features, both the linear ridge and the logistic regression are more comprehensible than the random forest algorithm. Indeed, while LR and LR are deterministic models, the random forest used by Quiroz et al. (2018) contains 100 random decision trees, therefore making it quasi-impossible to trace back a decision. This illustrates that striving for more explainability can sometimes come at the cost of performance. ...
Green growth has emerged as a dominant narrative within environmental policy, holding out the prospect that technological progress will enable an absolute decoupling between economic growth and natural resources usage. To achieve such decoupling, digital technologies are considered particularly promising.
The degrowth movement criticizes the green growth narrative and argues based on an increasing amount of evidence that we need an urgent break out of the growth paradigm. However, the movement’s vision on technology remains undefined, meanwhile influential actors such as industrialised governments dedicate great efforts to push their own imaginary of technological developments forward. This thesis therefore aims at contributing to the degrowth discourse on technology by outlining a degrowth perspective on Artificial Intelligence (AI), specifically machine learning (ML), a digital technology at the center of promises of environmental benefits, strong political support and extremely high investments and research interests. The thesis focuses on the question: could AI be appropriate to a degrowth context? If so, how?
The first step of this thesis (the growth perspective) shows that (1) AI has the capacity to accelerate economic growth and that (2) governments and large consultancy firms show high interests in AI actually driving growth. This thesis therefore argues that a degrowth perspective on AI should oppose this technology as long as it acts as a growth-accelerator, while recognizing that most probably only a change in economic paradigm could divert AI from its growth-acceleration effects.
In the second step (the conviviality perspective), Vetter’s Matrix of Convivial Technologies is applied to machine learning, which leads to the identification of three aspects of ML which strongly limit its conviviality: (1) its high complexity, (2) its environmental impacts and (3) the size of the infrastructure it needs. This thesis argues that if despite the above limitations to conviviality, machine learning were to still be considered by the degrowth movement, then it should at least satisfy the following two conditions to be considered appropriate: (1) it should have no global destructive consequences, and (2) it should be carefully assessed in its local context of application by the affected people, while striving for conviviality.
... Studies have used various signal-processing techniques to extract the relevant features from the measured signals. Typical signal-processing techniques for extracting diagnostic features include domain analyses such as for time [9], frequency [10], and time-frequency [11], which are commonly used as signal-processing methodologies. For instance, in [9], 13 time-domain features were extracted to detect the broken rotor of a PMSM using the stator current signal under various load conditions. ...
... The solution of the Lagrangian function is obtained through optimisation. The minimisation problem given in (8) is solved in the spectral domain within VMD to calculate modes, as shown in (9). ...
