Improved Process Monitoring Using Nonlinear Principal Component Models

This paper presents two new approaches for use in complete process monitoring. The firstconcerns the identification of nonlinear principal component models. This involves the application of linear
principal component analysis (PCA), prior to the identification of a modified autoassociative neural network (AAN) as the required nonlinear PCA (NLPCA) model. The benefits are that (i) the number of the reduced set of linear principal components (PCs) is smaller than the number of recorded process variables, and (ii) the set of PCs is better conditioned as redundant information is removed. The result is a new set of input data for a modified neural representation, referred to as a T2T network. The T2T NLPCA model is then used for complete process monitoring, involving fault detection, identification and isolation. The second approach introduces a new variable reconstruction algorithm, developed from the T2T NLPCA model. Variable reconstruction can enhance the findings of the contribution charts still widely used in industry by reconstructing the outputs from faulty sensors to produce more accurate fault isolation. These ideas are illustrated using recorded industrial data relating to developing cracks in an industrial glass melter process. A comparison of linear and nonlinear models, together with the combined use of contribution charts and variable reconstruction, is presented.

Improved principal component monitoring using the local approach

This paper shows that current multivariate statistical monitoring technology may not detect incipient changes in the variable covariance structure nor changes in the geometry of the underlying variable decomposition. To overcome these deficiencies, the local approach is incorporated into the multivariate statistical monitoring framework to define two new univariate statistics for fault detection. Fault isolation is achieved by constructing a fault diagnosis chart which reveals changes in the covariance structure resulting from the presence of a fault. A theoretical analysis is presented and the proposed monitoring approach is exemplified using application studies involving recorded data from two complex industrial processes. © 2007 Elsevier Ltd. All rights reserved.

Aplicação de redes neuronais artificiais à deteção e isolamento de falhas em processos industriais

Trabalho Final de Mestrado para obtenção do grau de Mestre em Engenharia Mecânica

A proposal for the diagnosis of uncertain dynamic systems based on interval models

The performance of a model-based diagnosis system could be affected by several uncertainty sources, such as,model errors,uncertainty in measurements, and disturbances. This uncertainty can be handled by mean of interval models.The aim of this thesis is to propose a methodology for fault detection, isolation and identification based on interval models. The methodology includes some algorithms to obtain in an automatic way the symbolic expression of the residual generators enhancing the structural isolability of the faults, in order to design the fault detection tests. These algorithms are based on the structural model of the system. The stages of fault detection, isolation, and identification are stated as constraint satisfaction problems in continuous domains and solved by means of interval based consistency techniques. The qualitative fault isolation is enhanced by a reasoning in which the signs of the symptoms are derived from analytical redundancy relations or bond graph models of the system. An initial and empirical analysis regarding the differences between interval-based and statistical-based techniques is presented in this thesis. The performance and efficiency of the contributions are illustrated through several application examples, covering different levels of complexity.

Formalization of an architectural model for exception handling coordination based on CA action concepts

Architectures based on Coordinated Atomic action (CA action) concepts have been used to build concurrent fault-tolerant systems. This conceptual model combines concurrent exception handling with action nesting to provide a general mechanism for both enclosing interactions among system components and coordinating forward error recovery measures. This article presents an architectural model to guide the formal specification of concurrent fault-tolerant systems. This architecture provides built-in Communicating Sequential Processes (CSPs) and predefined channels to coordinate exception handling of the user-defined components. Hence some safety properties concerning action scoping and concurrent exception handling can be proved by using the FDR (Failure Divergence Refinement) verification tool. As a result, a formal and general architecture supporting software fault tolerance is ready to be used and proved as users define components with normal and exceptional behaviors. (C) 2010 Elsevier B.V. All rights reserved.

Network reconfiguration and loss allocation in a deregulated environment of distribution systems

Low flexibility and reliability in the operation of radial distribution networks make those systems be constructed with extra equipment as sectionalising switches in order to reconfigure the network, so the operation quality of the network can be improved. Thus, sectionalising switches are used for fault isolation and for configuration management (reconfiguration). Moreover, distribution systems are being impacted by the increasing insertion of distributed generators. Hence, distributed generation became one of the relevant parameters in the evaluation of systems reconfiguration. Distributed generation may affect distribution networks operation in various ways, causing noticeable impacts depending on its location. Thus, the loss allocation problem becomes more important considering the possibility of open access to the distribution networks. In this work, a graphic simulator for distribution networks with reconfiguration and loss allocation functions, is presented. Reconfiguration problem is solved through a heuristic methodology, using a robust power flow algorithm based on the current summation backward-forward technique, considering distributed generation. Four different loss allocation methods (Zbus, Direct Loss Coefficient, Substitution and Marginal Loss Coefficient) are implemented and compared. Results for a 32-bus medium voltage distribution network, are presented and discussed.

Reconfiguração de sistemas de distribuição operando em vários níveis de demanda através de uma meta-heurística de busca em vizinhança variável

Pós-graduação em Engenharia Elétrica - FEIS

Implementación de un entorno de mantenimiento centralizado para el diagnóstico y aislamiento de fallos en plataformas de sistemas modulares

En la actualidad gran parte de las industrias utilizan o desarrollan plataformas, las cuales integran un número cada vez más elevado de sistemas complejos. El mantenimiento centralizado permite optimizar el mantenimiento de estas plataformas, por medio de la integración de un sistema encargado de gestionar el mantenimiento de todos los sistemas de la plataforma. Este Trabajo Fin de Máster (TFM) desarrolla el concepto de mantenimiento centralizado para sistemas complejos, aplicable a plataformas formadas por sistemas modulares. Está basado en la creciente demanda de las diferentes industrias en las que se utilizan este tipo de plataformas, como por ejemplo la industria aeronáutica, del ferrocarril y del automóvil. Para ello este TFM analiza el Estado del Arte de los sistemas de mantenimiento centralizados en diferentes industrias, además desarrolla los diferentes tipos de arquitecturas de sistemas, las técnicas de mantenimiento aplicables, así como los sistemas y técnicas de mantenimiento basados en funciones de monitorización y auto diagnóstico denominadas Built-In-Test Equipment (BITE). Adicionalmente, este TFM incluye el desarrollo e implementación de un modelo de un Entorno de Mantenimiento Centralizado en LabVIEW. Este entorno está formado por el modelo de un Sistema Patrón, así como el modelo del Sistema de Mantenimiento Centralizado y la interfaces entre ellos. El modelo del Sistema de Mantenimiento Centralizado integra diferentes funciones para el diagnóstico y aislamiento de los fallos. Así mismo, incluye una función para el análisis estadístico de los datos de fallos almacenados por el propio sistema, con el objetivo de proporcionar capacidades de mantenimiento predictivo a los sistemas del entorno. Para la implementación del modelo del Entorno de Mantenimiento Centralizado se han utilizado recursos de comunicaciones vía TCP/IP, modelización y almacenamiento de datos en ficheros XML y generación automática de informes en HTML. ABSTRACT. Currently several industries are developing or are making use of multi system platforms. These platforms are composed by many complex systems. The centralized maintenance allows the maintenance optimization, integrating a maintenance management system. This system is in charge of managing the maintenance dialog with the different and multiple platforms. This Master Final Project (TFM) develops the centralized maintenance concept for platforms integrated by modular and complex systems. This TFM is based on the demand of the industry that uses or develops multi system platforms, as aeronautic, railway, and automotive industries. In this way, this TFM covers and analyzes several aspects of the centralized maintenance systems like the State of the Art, for several industries. Besides this work develops different systems architecture types, maintenance techniques, and techniques and systems based on Built-in-test Equipment functions. Additionally, this TFM includes a LabVIEW Centralized System Environment model. This model is composed by a Standard System, the Centralized Maintenance System and the corresponding interfaces. Several diagnostic and fault isolation functions are integrated on the Centralized Maintenance Systems, as well a statistic analysis function, that provides with predictive maintenance capacity, based on the failure data stored by the system. Among others, the following resources have been used for the Centralized System Environment model development: TCP/IP communications, XML file data modelization and storing, and also automatic HTML reports generation.

Entorno de supervisión para vehículos aéreos no tripulados

La Universidad Politécnica de Madrid está investigando en el campo de la robótica inteligente, concretamente con el empleo de vehículos aéreos no tripulados (UAV). El objetivo final que se persigue con las investigaciones en este campo es el desarrollo de sistemas capaces de operar de forma más autónoma en un amplio espectro de situaciones. Dentro de este marco, este trabajo fin de grado se centra en el desarrollo de un sistema de supervisión para UAVs que persigue facilitar la monitorización de la ejecución de los procesos y facilitar la inclusión de procedimientos para incrementar la tolerancia a los fallos software. A lo largo de esta memoria se ofrece una revisión del estado del arte en el ámbito de la robótica, haciendo especial hincapié en la robótica inteligente con los métodos de desarrollo existentes y la definición de los distintos marcos de clasificación de la autonomía. También se ofrece una vista a las distintas técnicas existentes para lograr una mayor tolerancia a los fallos software, de entre las que han sido seleccionadas varias de ellas en la realización de este trabajo. Finalmente se describe el sistema de supervisión desarrollado, explicando primero el sistema desde un punto de vista funcional para más adelante adentrarse en la solución técnica elaborada. ---ABSTRACT--- The Universidad Politécnica de Madrid is currently handling several investigations regarding AI robotics, some of them are actually directing their efforts into the use of unmanned aerial vehicles (UAV). The goal in the long term for this investigations is the accomplishment of systems capable of operating autonomously, regardless of the situation the robot is place at. From this perspective, this final degree project focuses on de design and development of a supervision system for UAV’s, which function is to ease the monitoring of executing processes and the inclusion of fault tolerant procedures. During the development of this document a state of the art revision is offered, in which a thorough description through development methods and autonomy definitions for AI robotics is made. It is also offered a look around the different existing techniques for achieving a greater software fault tolerance, from which some of them were chosen for the development of this project. Finally the developed supervision system is described, first from a pure functional perspective of what the system should do and latter with a description of the actual technical solutions developed for this system.

Peer review of a formal verification/design proof methodology : summary of a sub-working-group meeting /

Bibliography: p. 48.

Predicting fault-prone software modules with rank sum classification

The detection and correction of defects remains among the most time consuming and expensive aspects of software development. Extensive automated testing and code inspections may mitigate their effect, but some code fragments are necessarily more likely to be faulty than others, and automated identification of fault prone modules helps to focus testing and inspections, thus limiting wasted effort and potentially improving detection rates. However, software metrics data is often extremely noisy, with enormous imbalances in the size of the positive and negative classes. In this work, we present a new approach to predictive modelling of fault proneness in software modules, introducing a new feature representation to overcome some of these issues. This rank sum representation offers improved or at worst comparable performance to earlier approaches for standard data sets, and readily allows the user to choose an appropriate trade-off between precision and recall to optimise inspection effort to suit different testing environments. The method is evaluated using the NASA Metrics Data Program (MDP) data sets, and performance is compared with existing studies based on the Support Vector Machine (SVM) and Naïve Bayes (NB) Classifiers, and with our own comprehensive evaluation of these methods.

A sensor fault detection and isolation method in interior permanent-magnet synchronous motor drives based on an extended Kalman filter

Interior permanent-magnet synchronous motors (IPMSMs) become attractive candidates in modern hybrid electric vehicles and industrial applications. Usually, to obtain good control performance, the electric drives of this kind of motor require one position, one dc link, and at least two current sensors. Failure of any of these sensors might lead to degraded system performance or even instability. As such, sensor fault resilient control becomes a very important issue in modern drive systems. This paper proposes a novel sensor fault detection and isolation algorithm based on an extended Kalman filter. It is robust to system random noise and efficient in real-time implementation. Moreover, the proposed algorithm is compact and can detect and isolate all the sensor faults for IPMSM drives. Thorough theoretical analysis is provided, and the effectiveness of the proposed approach is proven by extensive experimental results.

Actuator fault detection and isolation in linear systems

A scheme for the detection and isolation of actuator faults in linear systems is proposed. A bank of unknown input observers is constructed to generate residual signals which will deviate in characteristic ways in the presence of actuator faults. Residual signals are unaffected by the unknown inputs acting on the system and this decreases the false alarm and miss probabilities. The results are illustrated through a simulation study of actuator fault detection and isolation in a pilot plant doubleeffect evaporator.

Predicting fault inflow in iterative software development processes: an industrial evaluation

This paper addresses the need for accurate predictions on the fault inflow, i.e. the number of faults found in the consecutive project weeks, in highly iterative processes. In such processes, in contrast to waterfall-like processes, fault repair and development of new features run almost in parallel. Given accurate predictions on fault inflow, managers could dynamically re-allocate resources between these different tasks in a more adequate way. Furthermore, managers could react with process improvements when the expected fault inflow is higher than desired. This study suggests software reliability growth models (SRGMs) for predicting fault inflow. Originally developed for traditional processes, the performance of these models in highly iterative processes is investigated. Additionally, a simple linear model is developed and compared to the SRGMs. The paper provides results from applying these models on fault data from three different industrial projects. One of the key findings of this study is that some SRGMs are applicable for predicting fault inflow in highly iterative processes. Moreover, the results show that the simple linear model represents a valid alternative to the SRGMs, as it provides reasonably accurate predictions and performs better in many cases.

Hybrid fault detection and isolation in an heater bank of an HVAC system

In this work a hybrid technique that includes probabilistic and optimization based methods is presented. The method is applied, both in simulation and by means of real-time experiments, to the heating unit of a Heating, Ventilation Air Conditioning (HVAC) system. It is shown that the addition of the probabilistic approach improves the fault diagnosis accuracy.