Data-driven design of monitoring and diagnosis systems for dynamic processes: A review of subspace technique based schemes and some recent results

In this paper, the development of data-driven design of process monitoring and fault diagnosis (PM-FD) systems is reviewed and some recent results are presented. A major objective of this work is to sketch a process input–output data based framework of designing PM-FD systems for dynamic processes. The main focus of our study is on the data-driven design of observer-based PM-FD systems, which are, thanks to their high robustness and real-time ability, suitable for industrial applications.     

Smart anomaly detection in sensor systems: A multi-perspective review

Anomaly detection is concerned with identifying data patterns that deviate remarkably from the expected behavior. This is an important research problem, due to its broad set of application domains, from data analysis to e-health, cybersecurity, predictive maintenance, fault prevention, and industrial automation. Herein, we review state-of-the-art methods that may be employed to detect anomalies in the specific area of sensor systems, which poses hard challenges in terms of information fusion, data volumes, data speed, and network/energy efficiency, to mention but the most pressing ones. In this context, anomaly detection is a particularly hard problem, given the need to find computing-energy-accuracy trade-offs in a constrained environment. We taxonomize methods ranging from conventional techniques (statistical methods, time-series analysis, signal processing, etc.) to data-driven techniques (supervised learning, reinforcement learning, deep learning, etc.). We also look at the impact that different architectural environments (Cloud, Fog, Edge) can have on the sensors ecosystem. The review points to the most promising intelligent-sensing methods, and pinpoints a set of interesting open issues and challenges.     

A spiking neural network-based approach to bearing fault diagnosis

Fault diagnosis, with the aim of accurately identifying the presence of various faults as early as possible so at to provide effective information for maintenance planning, has been extensively concerned in advanced manufacturing systems. With the increase of the amount of condition monitoring data, fault diagnosis methods have gradually shifted from the model-based paradigm to data-driven paradigm. Intelligent fault diagnosis approaches which can automatically mine useful information from a huge amount of raw data are becoming promising ways to identify faults of manufacturing systems in the context of massive data. In this paper, the Spiking Neural Network (SNN), as the third generation neural network, is tailored as an intelligent fault diagnosis tool for bearings in rotating machinery. Compared to the perceptron and the back propagation neural network (BPNN) which are respectively the first and second generations of neural networks. SNN, which introduces the concept of time into its operating model can more closely mimic natural neural networks and possesses high bionic characteristics. In the proposed SNN-based approach to bearing fault diagnosis, features extracted from raw vibration signals through the local mean decomposition (LMD) are encoded into spikes to train an SNN with the improved tempotron learning rule. The performance of the proposed method is examined by the CWRU and MFPT datasets, and the experimental results show that the method can achieve a promising accuracy in bearing fault diagnosis.     

Gaussian Discriminative Analysis aided GAN for imbalanced big data augmentation and fault classification

With data in industrial processes being larger in scale and easier to access, data-driven technologies have become more prevalent in process monitoring. Fault classification is an indispensable part of process monitoring, while machine learning is an effective tool for fault classification. In most practical cases, however, the number of fault data is far smaller than normal data, and this imbalance of dataset would lead to the significant decline in performance of common classifier learning algorithms. To this issue, we propose a data augmentation method, which is based on Generative Adversarial Networks(GAN) and aided by Gaussian Discriminant Analysis(GDA), for enhancement of fault classification accuracy. To validate the effectiveness of this method for imbalanced fault classification, on toy data and the Tennessee Eastman (TE) benchmark process, common oversampling method and the basic GAN are compared to our method, with different classification algorithms. Besides, proposed method is deployed and parallelly trained on Tensorflow platform, which is suitable for applications like data augmentation and imbalanced fault classification in industrial big data environments.     

Augmented Industrial Data-Driven Modeling Under the Curse of Dimensionality

The curse of dimensionality refers to the problem of increased sparsity and computational complexity when dealing with high-dimensional data. In recent years, the types and variables of industrial data have increased significantly, making data-driven models more challenging to develop. To address this problem, data augmentation technology has been introduced as an effective tool to solve the sparsity problem of high-dimensional industrial data. This paper systematically explores and discusses the necessity, feasibility, and effectiveness of augmented industrial data-driven modeling in the context of the curse of dimensionality and virtual big data. Then, the process of data augmentation modeling is analyzed, and the concept of data boosting augmentation is proposed. The data boosting augmentation involves designing the reliability weight and actual-virtual weight functions, and developing a double weighted partial least squares model to optimize the three stages of data generation, data fusion, and modeling. This approach significantly improves the interpretability, effectiveness, and practicality of data augmentation in the industrial modeling. Finally, the proposed method is verified using practical examples of fault diagnosis systems and virtual measurement systems in the industry. The results demonstrate the effectiveness of the proposed approach in improving the accuracy and robustness of data-driven models, making them more suitable for real-world industrial applications.      

A data-driven two-stage maintenance framework for degradation prediction in semiconductor manufacturing industries

To reduce the production costs and breakdown risks in industrial manufacturing systems, condition-based maintenance has been actively pursued for prediction of equipment degradation and optimization of maintenance schedules. In this paper, a two-stage maintenance framework using data-driven techniques under two training types will be developed to predict the degradation status in industrial applications. The proposed framework consists of three main blocks, namely, Primary Maintenance Block (PMB), Secondary Maintenance Block (SMB), and degradation status determination block. As the popular methods with deterministic training, back-propagation Neural Network (NN) and evolvable NN are employed in PMB for the degradation prediction. Another two data-driven methods with probabilistic training, namely, restricted Boltzmann machine and deep belief network are applied in SMB as the backup of PMB to model non-stationary processes with the complicated underlying characteristics. Finally, the multiple regression forecasting is adopted in both blocks to check prediction accuracies. The effectiveness of our proposed two-stage maintenance framework is testified with extensive computation and experimental studies on an industrial case of the wafer fabrication plant in semiconductor manufactories, achieving up to 74.1% in testing accuracies for equipment degradation prediction.     

基于卷积神经网络的发动机故障预测方法

近年来,卷积神经网络(CNN)等深度学习方法的发展为发动机故障诊断和预测带来了新的思路。CNN具有局部连接、权值共享、池化操作以及多层结构等特点,能够有效提取局部特征,降低网络的训练难度,使CNN具有很强的学习能力和特征表达能力。开展了深度卷积神经网络故障预测方法研究,实现了面向发动机气路故障预测算法架构。利用基于发动机试验仿真数据对该方法进行了验证,并与其他几种常见的基于数据驱动的预测方法进行了比较,验证结果表明本文提出的基于卷积神经网络的预测方法具有较好的可行性和效果,可作为开展发动机PHM技术研究的参考。     

数字孪生驱动的火箭控制系统故障诊断研究综述

在第四次工业革命中,智能制造成为各国工业发展的重点方向,数字孪生技术作为一项新兴技术,能够有效实现物理信息的融合;将其应用于火箭控制系统的故障诊断和健康管理,能够进一步提高故障诊断的事前准确性,提升火箭发射的可靠性;本文对数字孪生技术在航天领域的研究现状进行归纳整理;首先梳理了NASA的数字孪生目标,国内领域按照设计、生产、支持服务阶段对数字孪生应用进行分类;其次,按照故障树、专家系统、神经网络、数据驱动的方法阐述控制系统故障诊断的研究现状;在介绍数字孪生驱动的健康管理方法的基础上,提出数字孪生驱动的火箭控制系统的故障诊断方法;详细介绍其基本组成框架,分析关键技术及应用难点,并提出数字孪生健康管理平台的基本流程;该方法预期实现火箭控制系统的事前诊断和维修策略的制定     

深度学习在故障诊断领域中的研究现状与挑战

现代工业系统已呈现出向大型化、复杂化的方向发展,使得针对工业系统的故障诊断方法遇到一系列的技术难题.近年来,深度学习(deep learning)在特征提取与模式识别方面显示出独特的优势与潜力,将深度学习应用于解决复杂工业系统故障诊断的研究已初现端倪.为此,首先介绍几种典型的基于深度学习方法实现工业系统故障诊断方法;然后对基于深度学习实现故障诊断的主要思想和建模方法进行描述;最后总结和讨论了复杂工业系统故障的特点,并探讨了深度学习在实现复杂工业系统故障诊断方面所面临的挑战,展望了未来值得继续研究的方向.     

功能交联条件下飞机混合增强故障诊断方法

由于飞机系统结构复杂,功能交联程度高,传统以机内测试（BIT）方法为基础的机载综合诊断方法受到机载测试点设置限制,故障检测、隔离能力无法完全满足部队使用维护需求。近年来,数据驱动方法发展迅速,并被广泛运用到故障诊断中,其中概率神经网络（PNN）凭借其结构简单、计算速度快、运算精度高和分类结果好等优势,非常适合故障诊断与分类问题。但同时,数据驱动方法由于样本不足,工程应用存在困难,不能完全替代传统方法。为提升系统故障检测率、隔离率,降低虚警率,通过应用PNN处理飞机健康状态数据进行故障诊断,并结合BIT中的诊断信息,利用二者间的互补性提出了一种混合增强飞机系统故障诊断方法。详细介绍了该融合方法的基本原理、融合层次、融合规则等。最后,通过某型号飞机分系统诊断实例表明,该方法检测、隔离能力有显著提升,能够满足功能交联条件下飞机故障诊断需求。     

智能装备故障预测与健康管理系统研究

针对智能装备预测性维护存在的智能化和网络化程度不高、物理模型建模困难等 问题,研究了数据驱动的智能装备远程故障预测与健康管理系统(PHM)的实施框架、关键技术 和系统开发方法。具体阐述了数据驱动 PHM 系统的运行模式,在此基础上分析了 PHM 系统的 软件架构和关键技术,首先利用 EEMD 对原始信号进行降噪和重构,将重构后的信号作为输入 建立基于 RBF 神经网络的故障诊断模型;然后采用动态神经网络建立基于时间序列的故障预测 模型,并建立基于故障阈值的故障报警机制;最后利用混合编程和网络化开发技术开发了数据 驱动的远程 PHM 系统。实际应用结果表明,该系统能以较高效率完成故障诊断、故障预测等 核心功能,具有良好的实用性。     

Condition monitoring and fault diagnostics for hydropower plants

This paper presents a condition monitoring and fault diagnostics (CMFD) system for hydropower plants (HPP). CMFD is based on the concept of industrial product-service systems (IPS2), in which the customer, turbine supplier, and maintenance service provider are the IPS2 stakeholders. The proposed CMFD consists of signal acquisition, data transfer to the virtual diagnostics center (VDC) and fault diagnostics. A support vector machine (SVM) classifier has been used for fault diagnostics. CMFD has been implemented on an HPP with three Kaplan units. A signal acquisition system for CMFD consists of data acquisition from a unit control system and a supplementary system for high-frequency data acquisition. The implemented SVM method exhibits high training accuracy and thus enables adequate fault diagnostics. The data are analyzed in the VDC, which allows all stakeholders access to diagnostic information from anywhere at any time. Based on this information, the service providers can establish condition-based maintenance and offer operational support. Furthermore, through the VDC, cooperation between the stakeholders can be achieved; thus, better maintenance scheduling is possible, which will be reflected in higher system availability.     

偏最小二乘线性模型及其非线性动态扩展模型综述

偏最小二乘(Partial least square,PLS)是一种基于数据驱动可以处理多个因变量对多个自变量的回归建模方法,因其具有提取质量相关信息的特性,在质量相关复杂工业过程监控中得到广泛的应用,成为近几十年复杂工业过程故障检测和诊断领域的研究热点.对此,介绍线性、非线性、动态PLS模型及其故障检测技术.首先,介绍标准PLS模型,在此基础上对传统PLS模型进行细化分并指出其优缺点,针对标准PLS存在的两个问题以及工业过程数据的两种极端情况,从数据预处理类、多空间类和分块类三方面梳理线性PLS模型的发展和改进历程;其次,将非线性PLS模型扩展方法分为两类,重点介绍核函数非线性PLS模型的研究现状;再次,指出动态扩展方法的两种基本思路,对PLS动态模型进行分类,阐明动态特性的成因,从本质上揭示两种动态扩展方法的原理,按照分类综述动态PLS模型的发展现状;最后,指出该领域亟需解决的问题和未来研究方向.     

从结构推断到根因识别——工业过程故障根因诊断研究综述

由于现代工业过程的复杂结构,变量间普遍存在紧密耦合,故障往往在变量间广泛传播,为过程运维带来挑战.针对该问题,工业根因诊断(industrial root cause diagnosis, IRCD)技术应运而生,其从异常变量中确定故障根因,便于针对性故障处理. IRCD包含两个主要步骤:结构推断和根因识别.前者建立变量间的信息传递结构;后者根据传递结构定位根因.然而,现有IRCD综述多侧重于结构推断,未对根因识别步骤进行调研,且未建立起各类IRCD模型与过程特性间的系统关联.为此,从结构推断和根因识别两个层级展开IRCD的研究综述.首先,依据推断准则的异同,归纳4类经典结构推断模型;其次,考虑到过程的高维度、非线性、非平稳性质以及机理知识的效用,对结构推断模型的变种及适用场景进行梳理;随后,对根因识别方法进行归类,包括纯数据驱动、知识与数据融合驱动的范式,涵盖6类典型方法,并分析它们的优势与不足;最后,讨论IRCD技术中存在的挑战,并给出未来研究方向,为后续研究提供参考.     

Rebooting data-driven soft-sensors in process industries: A review of kernel methods

Soft-sensors usually assist in dealing with the unavailability of hardware sensors in process industries, thus allowing for less fault occurrence and better control performance. However, nonlinear, non-stationary, ill-data, auto-correlated and co-correlated behaviors in industrial data always make general data-driven methods inadequate, thus resorting to kernel-based methods provide a necessary alternative. This paper gives a systematic review of various state-of-the-art kernel-based methods with applications for data pre-processing, sample selection, variable selection, model construction and reliability analysis of soft-sensors. An integrated review of various kernel-based soft-sensor modeling methods is attempted, including on-line, multi-output, small-data-driven, multi-step-ahead and semi-supervised applications. The discussion is further to provide an overview of achieving hard-to-measure variable prediction, fault detection and advanced control of process industries. Finally, data-driven soft-sensors with kernel methods perspectives on potential challenges and opportunities have been highlighted for future explorations in the process industrial communities.     

Robust PLS approach for KPI-related prediction and diagnosis against outliers and missing data

In practical industrial applications, the key performance indicator (KPI)-related prediction and diagnosis are quite important for the product quality and economic benefits. To meet these requirements, many advanced prediction and monitoring approaches have been developed which can be classified into model-based or data-driven techniques. Among these approaches, partial least squares (PLS) is one of the most popular data-driven methods due to its simplicity and easy implementation in large-scale industrial process. As PLS is totally based on the measured process data, the characteristics of the process data are critical for the success of PLS. Outliers and missing values are two common characteristics of the measured data which can severely affect the effectiveness of PLS. To ensure the applicability of PLS in practical industrial applications, this paper introduces a robust version of PLS to deal with outliers and missing values, simultaneously. The effectiveness of the proposed method is finally demonstrated by the application results of the KPI-related prediction and diagnosis on an industrial benchmark of Tennessee Eastman process.     

Model-Based Prognostic Techniques Applied to a Suspension System

Conventional maintenance strategies, such as corrective and preventive maintenance, are not adequate to fulfill the needs of expensive and high availability transportation and industrial systems. A new strategy based on forecasting system degradation through a prognostic process is required. The recent advances in model-based design technology have realized significant time savings in product development cycle. These advances facilitate the integration of model-based diagnosis and prognosis of systems, leading to condition-based maintenance and increased availability of systems. With an accurate simulation model of a system, diagnostics and prognostics can be synthesized concurrently with system design. In this paper, we develop an integrated prognostic process based on data collected from model-based simulations under nominal and degraded conditions. Prognostic models are constructed based on different random load conditions (modes). An Interacting Multiple Model (IMM) is used to track the hidden damage. Remaining-life prediction is performed by mixing mode-based life predictions via time-averaged mode probabilities. The solution has the potential to be applicable to a variety of systems, ranging from automobiles to aerospace systems.      

Digital twin enhanced fault prediction for the autoclave with insufficient data

Since any faulty operations could directly affect the composite property, making early prognosis is particularly crucial for complex equipment. At present, data-driven approach has been typically used for fault prediction. However, for part of complex equipment, it is difficult to access reliable and sufficient data to train the fault prediction model. To address this issue, this paper takes autoclave as an example. A Digital Twin (DT) model containing multiple dimensions for the autoclave is firstly constructed and verified. Then the characteristics of autoclave under different conditions are analyzed and presented with specific parameters. The data in normal and faulty conditions are simulated by using the DT model. Both the simulated data and extracted historical data are applied to enhance fault prediction. A convolutional neural network for fault prediction will be trained with the generated data which matches the feature of the autoclave in faulty conditions. The effectiveness of the proposed method is verified through result analysis.     

Vibration signal-based early fault prognosis: Status quo and applications

To implement Prognostics and Health Management (PHM) for industrial systems, it is paramount to conduct early fault prognosis on the systems to ensure the stability and reliability during their entire lifecycles. Investigations on early fault prognosis have been actively carried out, but there is a lack of systematic analysis and summary of the developed methods. To bridge the gap, in this paper, the relevant methods are comprehensively reviewed from the aspects of signal processing and fault identification. Furthermore, the applications of the methods are systematically described. In the end, to further facilitate researchers and practitioners, statistical and comparative analysis of the reviewed methods are given, and future development directions are outlined.     

发动机故障信息综合研究平台构建技术研究

针对目前多型号多批次的无人机发动机数据存储较为分散、缺少专门的分析手段辅助设计人员快速分析发动机状态的问题,借鉴国内外较为成熟的故障诊断算法和平台构建技术,设计了发动机故障信息综合研究平台。平台内集成了飞行数据状态识别、故障诊断、分析数据库建立等多项关键技术,结合不同的飞行工况分类结果,从长期和短期的角度分别采取基于递归结构辨识(Recursive Structural Identification, RESID)的起动状态故障检测、基于自回归滑动平均(Auto Regressive Moving Average, ARMA)模型的稳态故障检测、基于参数趋势分析的故障诊断和基于深度学习的智能故障识别等不同的诊断方法,并采用飞行报告的形式实现了从数据文件输入至飞行状态分析的全流程计算和结果展示,有效地实现了发动机使用维护的数据支持和技术保障,具有一定的工业应用价值。