Users of high availability (HA) computing require systems that run continuously, with little or no downtime. Modern PCs address HA needs by monitoring operating system parameters such as voltage, temperature, and hard drive status in order to anticipate possible system failure. However, one modality for PC monitoring that has been underutilized is sound. The application described here uses wavelet theory to analyze sounds produced by PC hard drives during standard operation. When twenty-nine hard drives were tested with the application and the results compared with the drives' Self-Monitoring, Analysis, and Reporting Technology (S.M.A.R.T.) data, the binomial distribution's low p-value of 0.012 indicated better than chance agreement. While the concurrence between the two systems shows that sound is an effective tool in detecting hardware failures, the disagreements between the systems show that the application can complement S.M.A.R.T. in an HA system.

Unscheduled aircraft maintenance causes a lot problems and costs for aircraft operators. This is due to the fact that aircraft cause significant costs if flights have to be delayed or canceled and because spares are not always available at any place and sometimes have to be shipped across the world. Reducing the number of unscheduled maintenance is thus a great costs factor for aircraft operators. This thesis describes three methods for aircraft health monitoring and prediction; one method for system monitoring, one method for forecasting of time series and one method that combines the two other methods for one complete monitoring and prediction process. Together the three methods allow the forecasting of possible failures. The two base methods use decision trees for decision making in the processes and genetic optimization to improve the performance of the decision trees and to reduce the need for human interaction. Decision trees have the advantage that the generated code can be fast and easily processed, they can be altered by human experts without much work and they are readable by humans. The human readability and modification of the results is especially important to include special knowledge and to remove errors, which the automated code generation produced.

Submitted by Geandra Rodrigues (geandrar@gmail.com) on 2018-03-27T17:53:34Z No. of bitstreams: 1 nielsonsoares.pdf: 4342256 bytes, checksum: bcb08c0e8cbff9d4ed4b60f92f7715b2 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2018-03-27T18:03:27Z (GMT) No. of bitstreams: 1 nielsonsoares.pdf: 4342256 bytes, checksum: bcb08c0e8cbff9d4ed4b60f92f7715b2 (MD5) / Made available in DSpace on 2018-03-27T18:03:27Z (GMT). No. of bitstreams: 1 nielsonsoares.pdf: 4342256 bytes, checksum: bcb08c0e8cbff9d4ed4b60f92f7715b2 (MD5) Previous issue date: 2018-02-27 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / As máquinas de chave são equipamentos eletromecânicos de grande importância em uma malha ferroviária. A ocorrência de falhas nesses equipamentos pode ocasionar interrupções das ferrovias e acarretar potenciais prejuízos econômicos. Assim, um diagnóstico precoce dessas falhas pode representar uma redução de custos e um aumento de produtividade. Essa dissertação tem como objetivo propor um modelo preditivo, baseado em técnicas de inteligência computacional, para a solução desse problema. A metodologia aplicada compreende o uso de técnicas de extração e seleção de características baseada em testes de hipóteses e modelos de aprendizado de máquina não supervisionado. O modelo proposto foi testado em uma base de dados disponibilizada por uma empresa ferroviária brasileira e se mostrou eficiente ao constatar como críticas as operações realizadas próximas à operação classificada como falha. / Railroad switch machines are important electromechanical equipment in a railway network, and the occurrence of failures in such equipment can cause railroad interruptions and lead to potential economic losses. Thus, an early diagnosis of these failures can represent a reduction of costs and an increase in productivity. This dissertation aims to propose a predictive model, based on computational intelligence techniques, to solve this problem. The applied methodology includes the use of features extraction and selection techniques based on hypothesis tests and unsupervised machine learning models. The proposed model was tested in a database made available by a Brazilian railway company and proved to be efficient when considering as critical the operations performed close to the operation classified as failure.

Following the growth of high performance computing systems (HPC) in size and complexity, and the advent of faster and more complex Exascale systems, failures became the norm rather than the exception. Hence, the protection mechanisms need to be improved. The most de facto mechanisms such as checkpoint/restart or redundancy may also fail to support the continuous operation of future HPC systems in the presence of failures. Failure prediction is a new protection approach that is beneficial for HPC systems with a short mean time between failure. The failure prediction mechanism extends the existing protection mechanisms via the dynamic adjustment of the protection level. This work provides a prototype to analyze and predict system behavior using statistical analysis to pave the path toward resilience in HPC systems. The proposed anomaly detection method is noise-tolerant by design and produces accurate results with as little as 30 minutes of historical data. Machine learning models complement the main approach and further improve the accuracy of failure predictions up to 85%. The fully automatic unsupervised behavior analysis approach, proposed in this work, is a novel solution to protect future extreme-scale systems against failures.:1 Introduction 1.1 Background and Statement of the Problem 1.2 Purpose and Significance of the Study 1.3 Jam–e Jam: A System Behavior Analyzer 2 Review of the Literature 2.1 Syslog Analysis 2.2 Users and Systems Privacy 2.3 Failure Detection and Prediction 2.3.1 Failure Correlation 2.3.2 Anomaly Detection 2.3.3 Prediction Methods 2.3.4 Prediction Accuracy and Lead Time 3 Data Collection and Preparation 3.1 Taurus HPC Cluster 3.2 Monitoring Data 3.2.1 Data Collection 3.2.2 Taurus System Log Dataset 3.3 Data Preparation 3.3.1 Users and Systems Privacy 3.3.2 Storage and Size Reduction 3.3.3 Automation and Improvements 3.3.4 Data Discretization and Noise Mitigation 3.3.5 Cleansed Taurus System Log Dataset 3.4 Marking Potential Failures 4 Failure Prediction 4.1 Null Hypothesis 4.2 Failure Correlation 4.2.1 Node Vicinities 4.2.2 Impact of Vicinities 4.3 Anomaly Detection 4.3.1 Statistical Analysis (frequency) 4.3.2 Pattern Detection (order) 4.3.3 Machine Learning 4.4 Adaptive resilience 5 Results 5.1 Taurus System Logs 5.2 System-wide Failure Patterns 5.3 Failure Correlations 5.4 Taurus Failures Statistics 5.5 Jam-e Jam Prototype 5.6 Summary and Discussion 6 Conclusion and Future Works Bibliography List of Figures List of Tables Appendix A Neural Network Models Appendix B External Tools Appendix C Structure of Failure Metadata Databse Appendix D Reproducibility Appendix E Publicly Available HPC Monitoring Datasets Appendix F Glossary Appendix G Acronyms

With the increased focus on reducing carbon emissions in today’s society, several industries have to overcome new challenges, where especially the automotive industry is under a lot of scrutiny to deliver improved and more environmentally friendly products. To meet the demands from customers and optimize vehicles aerodynamically, new cars often contain complex body geometries, together with advanced materials that are introduced to reduce the total vehicle weight. With the introduction of the complex body components and advanced materials,one area in the automotive industry that has to overcome these challenges is manufacturing engineering, and in particular the departments working with the sheet metal forming process. In this process complex body component geometries can lead to non-linear strain paths and stretch bending load cases, and newly introduced advanced materials can be prone to exhibit behaviour of edge cracks not observed in conventional sheet metals. This thesis takes it onset in the challenges seen in industry today with predicting failure of the three complex load cases: Non-Linear Strain Paths, Stretch-Bending,and Edge Cracks. Through Finite Element simulation attempts are made to accurately predict failure caused by aforementioned load cases in industrial components or experimental setups in an effort to develop post-processing methods that are applicable to all cases.

Node downtime and failed jobs in a computing cluster translate into wasted resources and user dissatisfaction. Therefore understanding why nodes and jobs fail in HPC clusters is essential. This paper provides analyses of node and job failures in two university-wide computing clusters at two Tier I US research universities. We analyzed approximately 3.0M job execution data of System A and 2.2M of System B with data sources coming from accounting logs, resource usage for all primary local and remote resources (memory, IO, network), and node failure data. We observe different kinds of correlations of failures with resource usages and propose a job failure prediction model to trigger event-driven checkpointing and avoid wasted work. We provide generalizable insights for cluster management to improve reliability, such as, for some execution environments local contention dominates, while for others system-wide contention dominates.

The objective of the proposed research is to develop statistical algorithms for controlling failure trends through targeted maintenance of at-risk components. The at-risk components are identified via chronological history and diagnostic data, if available. Utility systems include many thousands (possibly millions) of components with many of them having already exceeded their design lives. Unfortunately, neither the budget nor manufacturing resources exist to allow for the immediate replacement of all these components. On the other hand, the utility cannot tolerate a decrease in reliability or the associated increased costs. To combat this problem, an overall maintenance model has been developed that utilizes all the available historical information (failure rates and population sizes) and diagnostic tools (real-time conditions of each component) to generate a maintenance plan. This plan must be capable of delivering the needed reliability improvements while remaining economical. It consists of three facets each of which addresses one of the critical asset management issues: * Failure Prediction Facet - Statistical algorithm for predicting future failure trends and estimating required numbers of corrective actions to alter these failure trends to desirable levels. Provides planning guidance and expected future performance of the system. * Diagnostic Facet - Development of diagnostic data and techniques for assessing the accuracy and validity of that data. Provides the true effectiveness of the different diagnostic tools that are available. * Economics Facet - Stochastic model of economic benefits that may be obtained from diagnostic directed maintenance programs. Provides the cost model that may be used for budgeting purposes. These facets function together to generate a diagnostic directed maintenance plan whose goal is to provide the best available guidance for maximizing the gains in reliability for the budgetary limits utility engineers must operate within.

This study is concerned with developing a model to identify small-medium U.K. companies at risk of financial failure up to five years in advance. The importance of small companies in an economy, the impact of their failures, and the lack of failure research with respect to . this population, provided justification for this study. The research was undertaken in two stages. The first stage included a detailed description and discussion of the nature and role of small business in the UK economy, heir relevance, problems and Government involvement in this sector, together with literature review and assessment of past research relevant to this study. The second stage was involved with construction of the models using multiple discriminant analysis, applied to published accountancy data for two groups of failed and nonfailed companies. The later stage was performed in three parts : (1) evaluating five discriminant models for each of five years prior to failure; (2) testing the performance of each of the .five models over time on data not used . in their construction; (3) testing the discriminant models on a validation sample. The purpose was to establish the "best" discriminant model. "Best" was determined according to classification ability of the model and interpretation of variables. Finally a model comprising seven financial ratios measuring four aspects of a company's financial profile, such as profitability, gearing, capital turnover and liquidity was chosen. The model has shown to be a valid tool for predicting companies' health up to five years in advance.

In this thesis, an evaluation of the existing control system for ABB: s HVDC Classic Links is performed in order to investigate whether a possible improvement to commutation failure prediction is possible and to be recommended. The thesis starts with a theoretical approach to the complexity of consequences of increasing the extinction angle (γ) in order to prevent commutation failure in inverter operation, which is later confirmed through using the simulation software PSCAD to evaluate coherence between simulation results and theory. Dynamic power studies are performed through simulations in the electromagnetic time domain transient tool PSCAD in order to establish a possible improvement to the existing commutation failure prediction today used in ABB control systems for HVDC applications.

This doctoral thesis focuses on the problems of IGBT failure prediction in pulse converters using measurable changes of selected parameters (so--called trending variables) being influenced of transistor degradation during aging. Firstly the state--of--the--art in this field is presented in the dizertation. The description of designed and constructed automated measurement stand follows, enabling monitoring and recording of switching processes during accelerated aging. Further the problems of high--bandwidth measurement of electrical quantities during IGBT switching are described. Especially the problems of current sensing are analyzed and the most suitable sensor is selected. The data recorded using the developed apparatus served to identify potential trending variables allowing the failure prediction. Here the dependence of trending variables on aging and on parasitic influences (current, temperature, voltage) had to be distinguished. Finally the evaluation of trending variables is performed. Their insignificant sensitivity on accelerated aging is shown which complicates their practical implementation for the purpose of failure prediction.