A framework for real-time fault detection and diagnosis using temporal data

Successful real-time sensor-based fault detection and diagnosis in large and complex systems is seldom achieved by operators. The lack of an effective method for handling temporal data is one of several key problems in this area. A methodology is introduced which advantageously uses temporal data in performing fault diagnosis in a subsystem of a Navy ship propulsion system. The methodology is embedded in a computer program designed to be used as a decision aid to assist the operator. It utilizes machine learning, is able to cope with uncertainty at several levels, and works in real-time. Program performance data is presented and analysed. The approach illustrates how relatively simple existing techniques can be assembled into more powerful real-time diagnostic tools.     

A self-tuning adaptive trend extraction method for process monitoring and diagnosis

Trend analysis is an efficient tool for process monitoring and diagnosis. However, the performances of a trend-based diagnosis system depend on the reliability of the trends extracted from the signals. One challenge in trend analysis is to design algorithms able to adapt themselves to the varying conditions of background noise and artefacts occurring non-deterministically on a same signal. Moreover, while long term trends such as decreasing/increasing have been extensively studied other subtle changes such as slow drifts and step-like transients have received little attention. In this paper, an adaptive on-line trend-extraction method is presented. It extends a former algorithm based on a linear segmentation to filter the signal and extract trends. In this version, the tuning parameters are not set to a fixed value for a given signal but can self-adapt on-line according to an estimation of the noise variance. An increasing or decreasing trend is detected if the variations on the signal are significantly higher than the level of the background noise. An initialisation phase is proposed to automatically set the initial values of the parameters, making the algorithm a self-tuned algorithm with minimal user intervention.The method was evaluated on a set of simulated data with various levels of background noise. It was also applied on real physiological data recorded from babies hospitalised in a Neonate Intensive Care Unit. It showed improved performances compared to the non adaptive algorithm, whatever the level of noise corrupting the data.     

综合性SDG故障诊断架构

基于模型的SDG(Signed DiGraph,符号有向图)故障诊断方法因其具有完备性好、推理深度高等优点在过程工业安全工程中具有十分重要的意义,已成为安全工程中的1种关键技术。本文在以前研究的基础上,提出了1个综合性SDG故障诊断架构,以期能够实现在生产过程中及时发现故障并判明故障源。该综合性故障诊断架构按模型、推理和应用3个层次搭建,以传统定性SDG及概率SDG理论为基础,包含了从模型建立到故障诊断推理,从定性SDG方法到结合统计监控的SDG方法再到概率SDG方法等一系列实施方案。该综合性SDG故障诊断架构由于引入了多元统计监控模型,使得在系统没有表现出明显的故障征兆时就能够及时敏感地检测到异常变化,进而触发SDG及PSDG推理来实现对故障源的查找,并给出各故障源发生故障的概率值,以指导使用者按照概率值的大小顺序采取处理措施。以某石化公司的气体分馏装置为实际背景,利用该装置实时数据库中的实际生产工艺数据对该综合性诊断架构进行了实例验证,其故障诊断结果与实际发生的故障相吻合,证明了该综合性故障诊断架构的有效性。     

A self-learning and self-optimizing framework for the fault diagnosis knowledge base in a workshop

The knowledge base is an essential part of the fault diagnosis system, which is crucial to the performance of fault recognition. As the intelligence of the fault diagnosis system has made persistent advance, the increasing demands for diversity and dynamic update have posed challenges to the knowledge base. In this paper, a framework for the fault diagnosis knowledge base is proposed to address the challenges mentioned above. Firstly, a dynamic clustering model is designed using the proposed semi-supervised multi-spatial manifold clustering method to recognize attribute clusters and aggregate new types. When new types are added to this model, it is constantly updated to achieve the automatic evolution of the knowledge base for the diversity of fault. Then, a knowledge evolution model is established by the generative adversarial network algorithm to achieve self-learning and self-optimizing capabilities of the knowledge base. This method learns the distribution of knowledge elements and generates new knowledge elements to optimize the clustering model. Finally, a series of comparative experiments are carried out on bearing datasets to verify the validity of the mentioned framework and models. The comparison results indicate that the proposed method has better performance in fault diagnosis. This research can not only update the knowledge base, but also provide a feasible approach for designing an autonomous knowledge base with self-optimizing and self-learning capabilities.     

设备故障诊断中的特征提取

以某航天器姿控系统为研究对象,对故障的原始特征进行特征提取和选择,以使所建立的故障标准模式由少数几个新特征给予有效的表达,较好地实现了主要故障模式的分离,为提高设备故障诊断能力和故障定位能力奠定了基础。同时,通过考察权系数,可判断出主分量中各个参数的贡献率,这将有利于指导设计阶段的测点选择。     

A formal framework for on-line software version change

The usual way of installing a new version of a software system is to shut down the running program and then install the new version. This necessitates a sometimes unacceptable delay during which service is denied to the users of the software. An online software replacement system replaces parts of the software while it is in execution, thus eliminating the shutdown. While a number of implementations of online version change systems have been described in the literature, little investigation has been done on its theoretical aspects. We describe a formal framework for studying online software version change. We give a general definition of validity of an online change, show that it is in general undecidable and then develop sufficient conditions for ensuring validity for a procedural language     

A software architecture framework for on-line option pricing

The problem of growing computational complexity in the finance industry demands manageable, high-speed and real-time solutions in solving complex mathematical problems such as option pricing. In current option trading scenarios, determining a fair price for options “any time” and “anywhere” has become vital yet difficult computational problem. In this study, we have designed, implemented, and deployed an architecture for pricing options on-line using a hand-held device that is J2ME-based Mobile computing-enabled and is assisted by web mining tools. In our architecture, the client is a MIDP user interface, and the back end servlet runs on a standalone server bound to a known port address. In addition, the server uses table-mining techniques to mine real-time data from reliable web sources upon the mobile trader’s directive. The server performs all computations required for pricing options since mobile devices have limited battery power, low bandwidth, and low memory. We have parallelized and implemented various computational techniques such as binomial lattice and finite differencing. To the best of our knowledge, this is one of the first studies that facilitate the mobile-enabled-trader to compute the price of an option in ubiquitous fashion. This architecture aims at providing the trader with various computational techniques to avail (to provide results from approximate to accurate results) while on-the-go and to make important and effective trading decisions using the results that will ensure higher returns on investments in options.

A fault tolerance framework for cooperative robotic manipulators

The problem of fault tolerance in cooperative manipulators rigidly connected to an undeformable load is addressed in this paper. Four categories of faults are considered: free-swinging joint faults (FSJFs), locked joint faults (LJFs), incorrectly measured joint position faults (JPFs), and incorrectly measured joint velocity faults (JVFs). Free-swinging and locked joint faults are detected via artificial neural networks (ANNs). Incorrectly measured joint position and velocity faults are detected by considering the kinematic constraints of the cooperative system. When a fault is detected, the control system is reconfigured according to the nature of the isolated fault and the task is resumed to the largest extent possible. The fault tolerance framework is applied to an actual system composed of two cooperative robotic manipulators. The results presented demonstrate the feasibility and performance of the methodology.     

Use of multiple models and qualitative knowledge for on-line moving horizon disturbance estimation and fault diagnosis

An integrated fault detection, fault isolation, and parameter estimation technique is presented in this paper. Process model parameters are treated as disturbances that dynamically affect the process outputs. A moving horizon estimation technique minimizes the error between process and model measurements over a finite horizon by calculating model parameter values across the estimation horizon. To implement qualitative process knowledge, this minimization is constrained such that only a limited number of different faults (parameters) may change during a specific horizon window. Multiple linear models are used to capture nonlinear process characteristics such as asymmetric response, variable dynamics, and changing gains. Problems of solution multiplicity and computational time are addressed. Results from a nonlinear chemical reactor simulation are presented.     

A rough set-based fault ranking prototype system for fault diagnosis

Fault diagnosis is a complex and difficult problem that concerns effective decision-making. Carrying out timely system diagnosis whenever a fault symptom is detected would help to reduce system down time and improve the overall productivity. Due to the knowledge and experience intensive nature of fault diagnosis, the diagnostic result very much depends on the preference of the decision makers on the hidden relations between possible faults and the presented symptom. In other words, fault diagnosis is to rank the possible faults accordingly to give the engineer a practical priority to carry out the maintenance work in an efficient and orderly manner. This paper presents a rough set-based prototype system that aims at ranking the possible faults for fault diagnosis. The novel approach engages rough theory as a knowledge extraction tool to work on the past diagnostic records, which is registered in a pair-wise comparison table. It attempts to extract a set of minimal diagnostic rules encoding the preference pattern of decision-making by domain experts. By means of the knowledge acquired, the ordering of possible faults for failure symptom can then be determined. The prototype system also incorporates a self-learning ability to accumulate the diagnostic knowledge. A case study is used to illustrate the functionality of the developed prototype. Result shows that the ranking outcome of the possible faults is reasonable and sensible.     

A new integrated on-line fuzzy clustering and segmentation methodology with adaptive PCA approach for process monitoring and fault detection and diagnosis

A new on-line fuzzy clustering-based algorithm is developed using integration of an adaptive principal component analysis approach with a weighted fuzzy C-means (WFCM) methodology for process fault detection and diagnosis (FDD) applications. The proposed algorithm is based on the segmentation of measured multivariate time series process data through a sliding window scheme being realized in a bottom-up cluster merging approach to enable detection of probable changes embedded in their hidden structure. The method recursively maintain updated PCA models and their corresponding fuzzy membership functions based on the most recent arrival of each independent chunk of process data. The extracted chunk features are then retained in the memory to be merged using a new on-line fuzzy C-means methodology before incoming of the following chunks of data. A new formula is then presented for cluster merging improvement by incorporating an on-line weight to address the issue of cluster’s weight updating in the on-line WFCM methodology. The cluster merging mechanism is coordinated by a compatibility criterion, utilizing both similarities of the adapted clusters-based PCA models and their center closeness. The proposed algorithm has been evaluated on an artificial case study and Tennessee Eastman benchmark process plant. The observed performances demonstrate promising capabilities of the proposed algorithm to successfully detect and diagnose the introduced fault scenarios.     

一种基于PCNN的电力设备故障区域提取方法

针对红外自动监控电力设备是否存在故障问题,结合脉冲耦合神经网络(PCNN)同步点火特性,提出一种基于PCNN的红外图像感兴趣区域提取方法。首先针对原始的动态阈值振荡问题,采用神经元点火信息构建新的动态阈值,并建立连接系数与点火区域信息之间的内在关系,从而使得神经元自适应地发生点火。为了进一步确保每一次迭代中所捕获的神经元与点火区域的相似性,在模型框架内融合了一种聚类规则,进而有效更新动态阈值,并给出了停止迭代的方法。实验表明,该提取区域方法性能优于传统阈值、normalized cuts以及经典PCNN模型等方法。     

一种新的故障检测与诊断方法及其应用研究

给出了在故障检测与诊断中采用经验模式分解与希尔伯特变换相结合的方法。经验模式分解不同于小波变换、KL变换、奇异值分解(SVD)等信号分解方法,它把数据序列分解为能够表示数据中嵌入的不同振荡的本征模式函数。首先介绍方法的原理与特点,然后将该方法用于齿轮故障的探测与诊断,结果显示,这种方法能准确识别出裂缝故障的特征频率。     

矿用干式变压器绝缘在线监测及故障诊断预警系统研制

针对传统的矿用干式变压器监测方法采用热电偶检测干式变压器内部温度而存在监测量单一,不能有效反映干式变压器绝缘实时状态的问题,提出了一种矿用干式变压器绝缘在线监测及故障诊断预警系统的设计方案,介绍了系统整体结构和故障诊断方法,建立了干式变压器匝间绝缘故障类型库。现场调试结果表明,该系统运行稳定,可实现对矿用干式变压器的远程实时监测和故障诊断预警功能。     

A framework for automatic static and dynamic video thumbnail extraction

Video thumbnails enable users to see quick snapshots of video collections. To display the video thumbnails, the first frame or a frame selected by using simple low level features in each video clip has been set to the default thumbnail for the sake of computational efficiency and implementation simplicity. However, such methods often fail to represent the gist of the clip. To overcome this limitation, we present a new framework for both static and dynamic video thumbnail extraction. First, we formulate energy functions using the features which incorporate mid-level information to obtain superior thumbnailing. Since it is considered that frames whose layouts are similar to others in the clip are relevant in video thumbnail extraction, scene layouts are also considered in computing overall energy. For dynamic thumbnail generation, a time slot is determined by finding the duration showing the minimum energy. Experimental results show that the proposed method achieves comparable performance on a variety of challenging videos, and the subjective evaluation demonstrates the effectiveness of our method.