RESUMEN

Fault detection, classification, and location prediction are crucial for maintaining the stability and reliability of modern power systems, reducing economic losses, and enhancing system protection sensitivity. This paper presents a novel Hierarchical Deep Learning Approach (HDLA) for accurate and efficient fault diagnosis in transmission lines. HDLA leverages two-stage transformer-based classification and regression models to perform Fault Detection (FD), Fault Type Classification (FTC), and Fault Location Prediction (FLP) directly from synchronized raw three-phase current and voltage samples. By bypassing the need for feature extraction, HDLA significantly reduces computational complexity while achieving superior performance compared to existing deep learning methods. The efficacy of HDLA is validated on a comprehensive dataset encompassing various fault scenarios with diverse types, locations, resistances, inception angles, and noise levels. The results demonstrate significant improvements in accuracy, recall, precision, and F1-score metrics for classification, and Mean Absolute Errors (MAEs) and Root Mean Square Errors (RMSEs) for prediction, showcasing the effectiveness of HDLA for real-time fault diagnosis in power systems.

RESUMEN

The periodical impulses caused by localized defects of components are the vital characteristic information for fault detection and diagnosis of rotating machines. In recent years, multitudinous spectrum analysis-based signal processing methods have been developed and authenticated as the powerful tools for excavating fault-related repetitive transients from the measured complex signals. Nonetheless, in practice, their applications can be severely confined by the constraints of limited system signal availability and incomplete information extraction under intricate noise interferences. To tackle the aforementioned issues, this paper proposes a periodic-modulation-oriented noise resistant correlation (PMONRC) method for target period detection and fault diagnosis of rotating machinery. Firstly, the envelope of raw signal is obtained via a novel sequential procedure of signal element-wise squaring, spectral Gini index-guided adaptive low-pass filtering, and signal element-wise square root computation, to highlight the modulated wave component that is more likely to be related to the potential fault-induced periods. Subsequently, a series of sub-signals, which can encode the fault-related repetitive information and enhance noise resistance, are constructed utilizing the envelope signal. Based upon the envelope signal and the obtained sub-signals, a weighted envelope noise resistant correlation function can be derived with the assistance of the L-moment ratio-based indicator and Sigmoid transformation. Finally, the specific fault type of the rotating machinery can be identified and affirmed accordingly. The proposed PMONRC method, which is nonparametric and completely adaptive to the signal being processed itself, overcomes the deficiencies of spectral analysis-based approaches, and is applicable for the engineering circumstances of system signal limitation and low signal-to-noise ratio (SNR), possessing immense practical merit. Both simulation analyses and experimental validations profoundly demonstrate that the proposed method is superior to other existing state-of-the-art time-domain correlation methods. Moreover, as an attempt as well as exemplar to apply this method, the PMONRC-based incipient fault diagnostic results of rolling bearing data from the well-known experimental platform PRONOSTIA are presented and discussed as well, to further elucidate the effectiveness and practical engineering significance of the proposed method.

RESUMEN

This study introduces a fault diagnosis algorithm based on particle filtering for open-cycle liquid-propellant rocket engines (LPREs). The algorithm serves as a model-based method for the startup process, accounting for more than 30% of engine failures. Similar to the previous fault detection and diagnosis (FDD) algorithm for the startup process, the algorithm in this study is composed of a nonlinear filter to generate residuals, a residual analysis, and a multiple-model (MM) approach to detect and diagnose faults from the residuals. In contrast to the previous study, this study makes use of the modified cumulative sum (CUSUM) algorithm, widely used in change-detection monitoring, and a particle filter (PF), which is theoretically the most accurate nonlinear filter. The algorithm is confirmed numerically using the CUSUM and MM methods. Subsequently, the FDD algorithm is compared with an algorithm from a previous study using a Monte Carlo simulation. Through a comparative analysis of algorithmic performance, this study demonstrates that the current PF-based FDD algorithm outperforms the algorithm based on other nonlinear filters.

RESUMEN

Fault diagnosis plays a pivotal role in identifying the root causes of a fault. Current fault diagnosis methods encounter the shortcomings being unable to assess the fault amplitude or having low efficiency for batch fermentation process. In order to solve the above problems, this paper proposes a fault detection model named convolutional neural network based on variational autoencoder (CNN-VAE) and a fault diagnosis based on counterfactual inference (FDCI). To begin with, quality-related process variables are selected using mutual information (MI). Next, a two-dimensional moving window is used to obtain input sequences from the process data. Then, two statistics from the latent and residual domains of the CNN-VAE model are constructed for fault detection. Additionally, once a fault occurs, FDCI is used to locate the root cause of a fault. Finally, a simulation process and a real-world L. plantarum batch fermentation process are provided to demonstrate the effectiveness of the proposed approache.

RESUMEN

The automation of Fault Detection and Diagnosis (FDD) is a central task for many industries today. A myriad of methods are in use, although the most recent leading contenders are data-driven approaches and especially Machine Learning (ML) methods. ML algorithms fall into two main categories: supervised and unsupervised methods, depending on whether or not the instances are labeled with the expected outputs. However, a new approach called Semi-Supervised Learning (SSL) has recently emerged that uses a few labeled instances together with other unlabeled instances for the training process. This new approach can significantly improve the accuracy of conventional ML models for industrial environments where labeled data are scarce. SSL has been tested as a promising solution over the past few years for several FDD problems, although there have been no systemic reviews of this sort of approach up until the present review. In this study, an attempt to organize the existing literature on SSL for FDD using the taxonomy of van Engelen & Hoos is reported. The most and the least frequently used SSL algorithms are identified and considered in terms of different fault detection tasks and their most common dataset structure. Moreover, a set of best practices are proposed in the conclusions of this work for implementation under real industrial conditions, so as to avoid some of the most common faults.

RESUMEN

In the past decade, machine learning techniques have seen wide industrial applications for design of data-based process monitoring systems with an aim to improve industrial productivity. An efficient process monitoring system for wastewater treatment process (WWTP) ensures increased efficiency and effluents meeting stringent emission norms. Benchmark simulation model No. 1 (BSM1) provides a simulation platform to researchers for developing efficient data-based process monitoring, quality monitoring, and process control systems for WWTPs. The present article presents a review of all research works reporting applications of various machine learning techniques for sensor and process fault detection of BSM1. The review focuses on process monitoring of biological wastewater treatment process, which uses a series of aerobic and anaerobic reactions followed by secondary settling process. Detailed information on various parameters monitored, different machine learning techniques explored, and results obtained by different researchers are presented in tabular and graphical format. In the review, it was observed that principal component analysis (PCA) and its variants account for the maximum number of research works for process monitoring in WWTPs and there are very few applications of recently developed deep learning techniques. Following the review and analysis, various future scopes of research (such as techniques yet to be explored or improvement of results for a particular fault) are also presented. These information will assist prospective researchers working on BSM1 to take forward the research.

Asunto(s)

RESUMEN

Most methodologies for fault detection and diagnosis in prognostics and health management (PHM) systems use machine learning (ML) or deep learning (DL), in which either some features are extracted beforehand (in the case of typical ML approaches) or the filters are used to extract features autonomously (in the case of DL) to perform the critical classification task. In particular, in the fault detection and diagnosis of industrial robots where the primary sources of information are electric current, vibration, or acoustic emissions signals that are rich in information in both the temporal and frequency domains, techniques capable of extracting meaningful information from non-stationary frequency-domain signals with the ability to map the signals into their constituent components with compressed information are required. This has the potential to minimise the complexity and size of traditional ML- and DL-based frameworks. The deep scattering spectrum (DSS) is one of the approaches that use the Wavelet Transform (WT) analogy for separating and extracting information embedded in a signal's various temporal and frequency domains. Therefore, the primary focus of this work is the investigation of the efficacy and applicability of the DSS's feature domain relative to fault detection and diagnosis for the mechanical components of industrial robots. For this, multiple industrial robots with distinct mechanical faults were studied. Data were collected from these robots under different fault conditions and an approach was developed for classifying the faults using DSS's low-variance features extracted from input signals. The presented approach was implemented on the practical test benches and demonstrated satisfactory performance in fault detection and diagnosis for simple and complex classification problems with a classification accuracy of 99.7% and 88.1%, respectively. The results suggest that, similarly to other ML techniques, the DSS offers significant potential in addressing fault classification challenges, especially for cases where the data are in the form of signals.

Asunto(s)

RESUMEN

In recent years, explainable artificial intelligence (XAI) techniques have been developed to improve the explainability, trust and transparency of machine learning models. This work presents a method that explains the outputs of an air-handling unit (AHU) faults classifier using a modified XAI technique, such that non-AI expert end-users who require justification for the diagnosis output can easily understand the reasoning behind the decision. The method operates as follows: First, an XGBoost algorithm is used to detect and classify potential faults in the heating and cooling coil valves, sensors, and the heat recovery of an air-handling unit. Second, an XAI-based SHAP technique is used to provide explanations, with a focus on the end-users, who are HVAC engineers. Then, relevant features are chosen based on user-selected feature sets and features with high attribution scores. Finally, a sliding window system is used to visualize the short history of these relevant features and provide explanations for the diagnosed faults in the observed time period. This study aimed to provide information not only about what occurs at the time of fault appearance, but also about how the fault occurred. Finally, the resulting explanations are evaluated by seven HVAC expert engineers. The proposed approach is validated using real data collected from a shopping mall.

Asunto(s)

RESUMEN

The necessity of vehicle fault detection and diagnosis (VFDD) is one of the main goals and demands of the Internet of Vehicles (IoV) in autonomous applications. This paper integrates various machine learning algorithms, which are applied to the failure prediction and warning of various types of vehicles, such as the vehicle transmission system, abnormal engine operation, and tire condition prediction. This paper first discusses the three main AI algorithms, such as supervised learning, unsupervised learning, and reinforcement learning, and compares the advantages and disadvantages of each algorithm in the application of system prediction. In the second part, we summarize which artificial intelligence algorithm architectures are suitable for each system failure condition. According to the fault status of different vehicles, it is necessary to carry out the evaluation of the digital filtering process. At the same time, it is necessary to preconstruct its model analysis and adjust the parameter attributes, types, and number of samples of various vehicle prediction models according to the analysis results, followed by optimization to obtain various vehicle models. Finally, through a cross-comparison and sorting, the artificial intelligence failure prediction models can be obtained, which can correspond to the failure status of a certain car model and a certain system, thereby realizing a most appropriate AI model for a specific application.

RESUMEN

Bearings are vital components of rotating machines that are prone to unexpected faults. Therefore, bearing fault diagnosis and condition monitoring are essential for reducing operational costs and downtime in numerous industries. In various production conditions, bearings can be operated under a range of loads and speeds, which causes different vibration patterns associated with each fault type. Normal data are ample as systems usually work in desired conditions. On the other hand, fault data are rare, and in many conditions, there are no data recorded for the fault classes. Accessing fault data is crucial for developing data-driven fault diagnosis tools that can improve both the performance and safety of operations. To this end, a novel algorithm based on conditional generative adversarial networks (CGANs) was introduced. Trained on the normal and fault data on actual fault conditions, this algorithm generates fault data from normal data of target conditions. The proposed method was validated on a real-world bearing dataset, and fault data were generated for different conditions. Several state-of-the-art classifiers and visualization models were implemented to evaluate the quality of the synthesized data. The results demonstrate the efficacy of the proposed algorithm.

RESUMEN

The board channel is a connection between a data acquisition system and the sensors of a plant. A flawed channel will bring poor-quality data or faulty data that may cause an incorrect strategy. In this paper, a data-driven approach is proposed to detect the status of the enclosed board channel based on an error time series obtained from multiple excitation signals and internal register values. The critical faulty data, contrary to the known healthy data, are constructed by using a null matrix with maximum projection and are labelled as training examples together with healthy data. Finally, the status of the enclosed board channel is validated by a well-trained probabilistic neural network. The experimental results demonstrate the effectiveness of the proposed method.

Asunto(s)

RESUMEN

This paper introduces a novel robust adaptive fault detection and diagnosis (FDD) observer design approach for a class of nonlinear systems with parametric uncertainty, unknown system fault and time-varying internal delays. The conditions for the existence of the proposed FDD are obtained based on the well-known Linear Matrix Inequalities (LMI) technique. Using Lyapunov stability theory, the adaptation laws for updating the observer weights and unknown faults estimation are derived based on which the convergence of the state estimation error to zero and asymptotic stability of the error dynamics are proven. Toward this, a new structural algorithm for FDD observer design is also derived based on LMIs. The performance of the proposed method is also investigated while applying to some industrial systems. Simulation results illustrate superior performance of the proposed method for the systems subject to time-varying unknown delays on states, uncertainty in nonlinear system modeling and unknown system faults.

Asunto(s)

RESUMEN

Heating, ventilation, and air conditioning (HVAC) systems are a popular research topic because buildings' energy is mostly used for heating and/or cooling. These systems heavily rely on sensory measurements and typically make an integral part of the smart building concept. As such, they require the implementation of fault detection and diagnosis (FDD) methodologies, which should assist users in maintaining comfort while consuming minimal energy. Despite the fact that FDD approaches are a well-researched subject, not just for improving the operation of HVAC systems but also for a wider range of systems in industrial processes, there is a lack of application in commercial buildings due to their complexity and low transferability. The aim of this review paper is to present and systematize cutting-edge FDD methodologies, encompassing approaches and special techniques that can be applied in HVAC systems, as well as to provide best-practice heuristics for researchers and solution developers in this domain. While the literature analysis targets the FDD perspective, the main focus is put on the data-driven approach, which covers commonly used models and data pre-processing techniques in the field. Data-driven techniques and FDD solutions based on them, which are most commonly used in recent HVAC research, form the backbone of our study, while alternative FDD approaches are also presented and classified to properly contextualize and round out the review.

Asunto(s)

RESUMEN

Incipient faults in running gear systems corrupt the overall performance of high-speed trains, increasing the necessity of fault detection and diagnosis whose purpose is to maintain the safe and stable operation of high-speed trains. For this purpose, a novel data-driven method, that utilizes Hellinger distance and slow feature analysis, is proposed in this study. By integrating Hellinger distance into slow feature analysis, a new test statistic is defined for detecting incipient faults in running gear systems. Furthermore, the hidden Markov method is developed for performing reliable fault diagnosis tasks. The salient strengths of the proposed method lie in its satisfactory fault detectability on the one hand and the considerable robustness against high-level noises on the other hand. Finally, the effectiveness of the proposed method is verified through a numerical example and a running gear system of high-speed trains under actual working conditions.

RESUMEN

Industrial applications of fault detection and diagnosis face great challenges as they require not only accurate identification of faulty statuses but also the effective understandability of the results. In this paper, a two-step robust and understandable fault detection and diagnosis framework is developed to address this challenge by exploiting denoising sparse autoencoder and smooth integrated gradients. Specifically, denoising sparse autoencoder(DSAE) is utilized to detect faults in the first step. DSAE is more robust to noise corruption and has better generalization performance compared to the existing autoencoder-based methods. In the second step, smooth integrated gradients(SIG) is used to diagnose the root-cause variables of the faults detected. Smooth integrated gradients can provide a denoising effect on the feature importance. The proposed framework is evaluated through an application to the Tennessee Eastman process. As proved in the experiments, the presented DSAE-SIG method not only achieves higher diagnosis accuracy but also successfully identifies the potential root-cause variables of process disturbances.

RESUMEN

The malfunctioning of the heating, ventilating, and air conditioning (HVAC) system is considered to be one of the main challenges in modern buildings. Due to the complexity of the building management system (BMS) with operational data input from a large number of sensors used in HVAC system, the faults can be very difficult to detect in the early stage. While numerous fault detection and diagnosis (FDD) methods with the use of statistical modeling and machine learning have revealed prominent results in recent years, early detection remains a challenging task since many current approaches are unfeasible for diagnosing some HVAC faults and have accuracy performance issues. In view of this, this study presents a novel hybrid FDD approach by combining random forest (RF) and support vector machine (SVM) classifiers for the application of FDD for the HVAC system. Experimental results demonstrate that our proposed hybrid random forest-support vector machine (HRF-SVM) outperforms other methods with higher prediction accuracy (98%), despite that the fault symptoms were insignificant. Furthermore, the proposed framework can reduce the significant number of sensors required and work well with the small number of faulty training data samples available in real-world applications.

Asunto(s)

RESUMEN

To satisfy the market, competition in the industrial sector aims for productivity and safety in industrial plant control systems. The appearance of a fault can compromise the system's proper functioning process. Therefore, Fault Detection and Diagnosis (FDD) methods contribute to avoiding any undesired events, as there are techniques and methods that study the detection, isolation, identification and, consequently, fault diagnosis. In this work, a new methodology that uses faults emulation to obtain parameters similar to the Development and Application of Methods for Diagnosis of Actuators in Industrial Control Systems (DAMADICS) benchmark model will be developed. This methodology uses previous information from tests on sensors with and without faults to detect and classify the situation of the plant and, in the presence of faults, perform the diagnosis through a process of elimination in a hierarchical manner. In this way, the definition of residue signature is used as well as the creation of a decision tree. The whole process is carried out incorporating FDD techniques, through the Non-Linear Auto-Regressive Neural Network Model With Exogenous Inputs (NARX), in the diagnosis of the behavioral prediction of the signals to generate the residual values. Then, it is applied to the construction of the decision tree based on the most significant residue of a certain signal, enabling the process of acquisition and formation of the signature matrix. With the procedures in this article, it is possible to demonstrate a practical and systematic method of how to emulate faults for control valves and the possibility of carrying out an analysis of the data to acquire signatures of the fault behavior. Finally, simulations resulting from the most sensitized variables for the production of residuals that is generated by neural networks are presented, which are used to obtain signatures and isolate the flaws. The process proves to be efficient in computational time and makes it easy to present a fault diagnosis strategy that can be reproduced in other processes.

RESUMEN

In prognostics and health management (PHM), the majority of fault detection and diagnosis is performed by adopting segregated methodology, where electrical faults are detected using motor current signature analysis (MCSA), while mechanical faults are detected using vibration, acoustic emission, or ferrography analysis. This leads to more complicated methods for overall fault detection and diagnosis. Additionally, the involvement of several types of data makes system management difficult, thus increasing computational cost in real-time. Aiming to resolve that, this work proposes the use of the embedded electrical current signals of the control unit (MCSA) as an approach to detect and diagnose mechanical faults. The proposed fault detection and diagnosis method use the discrete wavelet transform (DWT) to analyze the electric motor current signals in the time-frequency domain. The technique decomposes current signals into wavelets, and extracts distinguishing features to perform machine learning (ML) based classification. To achieve an acceptable level of classification accuracy for ML-based classifiers, this work extends to presenting a methodology to extract, select, and infuse several types of features from the decomposed wavelets of the original current signals, based on wavelet characteristics and statistical analysis. The mechanical faults under study are related to the rotate vector (RV) reducer mechanically coupled to electric motors of the industrial robot Hyundai Robot YS080 developed by Hyundai Robotics Co. The proposed approach was implemented in real-time and showed satisfying results in fault detection and diagnosis for the RV reducer, with a classification accuracy of 96.7%.

RESUMEN

Process monitoring at industrial sites contributes to system stability by detecting and diagnosing unexpected changes in a system. Today, as the infrastructure of industrial sites is advancing because of the development of communication technology, vast amounts of data are generated, and the importance of a way to effectively monitor such data in order to diagnose a system is increasing daily. Because a method based on a deep neural network can effectively extract information from a large amount of data, methods have been proposed to monitor processes using such networks to detect system faults and abnormalities. Neural-network-based process monitoring is effective in detecting faults, but has difficulty in diagnosing because of the limitations of the black-box model. Therefore, in this paper we propose a process-monitoring framework that can detect and diagnose faults. The proposed method uses a class activation map that results from diagnosis of faults and abnormalities, and verifies the diagnosis by post-processing the class activation map. This improves the detection of faults and abnormalities and generates a class activation map that provides a more verified diagnosis to the end user. In order to evaluate the performance of the proposed method, we did a simulation using publicly available industrial motor datasets. In addition, after establishing a system that can apply the proposed method to actual manufacturing companies that produce sapphire nozzles, we carried out a case study on whether fault detection and diagnosis were possible.

RESUMEN

Weld bead geometry features (WBGFs) such as the bead width, height, area, and center of gravity are the common factors for weighing welding quality control. The effective modeling of these WBGFs contributes to implementing timely decision making of welding process parameters to improve welding quality and enhance automatic levels. In this work, a dynamic modeling method of WBGFs is presented based on machine vision and learning in multipass gas metal arc welding (GMAW) with typical joints. A laser vision sensing system is used to detect weld seam profiles (WSPs) during the GMAW process. A novel WSP extraction method is proposed using scale-invariant feature transform and machine learning. The feature points of the extracted WSP, namely the boundary points of the weld beads, are identified with slope mutation detection and number supervision. In order to stabilize the modeling process, a fault detection and diagnosis method is implemented with cubic exponential smoothing, and the diagnostic accuracy is within 1.50 pixels. A linear interpolation method is presented to implement sub pixel discrimination of the weld bead before modeling WBGFs. With the effective feature points and the extracted WSP, a scheme of modeling the area, center of gravity, and all-position width and height of the weld bead is presented. Experimental results show that the proposed method in this work adapts to the variable features of the weld beads in thick plate GMAW with T-joints and butt/lap joints. This work can provide more evidence to control the weld formation in a thick plate GMAW in real time.