data transformation and attribute subset selection: do they help make differences in software failure prediction?

IEEE, IEEE Comp Soc, Tech Council Software Engn

An online failure prediction system for private IaaS platforms

The size and complexity of cloud environments make them prone to failures. The traditional approach to achieve a high dependability for these systems relies on constant monitoring. However, this method is purely reactive. A more proactive approach is provided by online failure prediction (OFP) techniques. In this paper, we describe a OFP system for private IaaS platforms, currently under development, that combines di_erent types of data input, including monitoring information, event logs, and failure data. In addition, this system operates at both the physical and virtual planes of the cloud, taking into account the relationships between nodes and failure propagation mechanisms that are unique to cloud environments.

System failure prediction through rare-events elastic-net logistic regression

Predicting failures in a distributed system based on previous events through logistic regression is a standard approach in literature. This technique is not reliable, though, in two situations: in the prediction of rare events, which do not appear in enough proportion for the algorithm to capture, and in environments where there are too many variables, as logistic regression tends to overfit on this situations; while manually selecting a subset of variables to create the model is error- prone. On this paper, we solve an industrial research case that presented this situation with a combination of elastic net logistic regression, a method that allows us to automatically select useful variables, a process of cross-validation on top of it and the application of a rare events prediction technique to reduce computation time. This process provides two layers of cross- validation that automatically obtain the optimal model complexity and the optimal mode l parameters values, while ensuring even rare events will be correctly predicted with a low amount of training instances. We tested this method against real industrial data, obtaining a total of 60 out of 80 possible models with a 90% average model accuracy.

Stress based prediction of formability and failure in incremental sheet forming

A strain-based forming limit criterion is widely used in sheet-metal forming industry to predict necking. However, this criterion is usually valid when the strain path is linear throughout the deformation process [1]. Strain path in incremental sheet forming is often found to be severely nonlinear throughout the deformation history. Therefore, the practice of using a strain-based forming limit criterion often leads to erroneous assessments of formability and failure prediction. On the other hands, stress-based forming limit is insensitive against any changes in the strain path and hence it is first used to model the necking limit in incremental sheet forming. The stress-based forming limit is also combined with the fracture limit based on maximum shear stress criterion to show necking and fracture together. A derivation for a general mapping method from strain-based FLC to stress-based FLC using a non-quadratic yield function has been made. Simulation model is evaluated for a single point incremental forming using AA 6022-T43, and checked the accuracy against experiments. By using the path-independent necking and fracture limits, it is able to explain the deformation mechanism successfully in incremental sheet forming. The proposed model has given a good scientific basis for the development of ISF under nonlinear strain path and its usability over conventional sheet forming process as well.

Stress-based predictions of formability and failure in incremental sheet forming

 This research investigates the deformation mechanism in incremental sheet forming (ISF) with relation to necking and failure. A strain-based forming limit criterion is widely used in sheet-metal forming industry to predict necking. However, this criterion is strictly valid only when the strain path is linear throughout the deformation process. Where the strain path in ISF is often found to be severely nonlinear throughout the deformation history. Therefore, the practice of using a strain-based forming limit criterion often leads to erroneous assessments of formability and failure prediction. On the other hands, stress-based forming limit is insensitive against any changes in the strain path and hence it is used to model the necking and fracture limits. Simulation model is evaluated for a single point incremental forming using AA 6022-T4E32 and checked the accuracy against experiments.

Investigation of surface ratchetting due to rail/wheel contact

This project advances the knowledge of rail wear and crack formation due to rail/wheel contact in Australian heavy-haul railway lines. This comprehensive study utilised numerous techniques including: simulation using a twin-disk test-rig, scanning electron microscope particle analysis and finite element modeling for material failure prediction. Through this work, new material failure models have been developed which may be used to predict the lifetime and reliability of materials undergoing severe contact conditions.

A review of recent developments in parametric based acoustic emission techniques applied to concrete structures

This article presents a review of recent developments in parametric based acoustic emission (AE) techniques applied to concrete structures. It recapitulates the significant milestones achieved by previous researchers including various methods and models developed in AE testing of concrete structures. The aim is to provide an overview of the specific features of parametric based AE techniques of concrete structures carried out over the years. Emphasis is given to traditional parameter-based AE techniques applied to concrete structures. A significant amount of research on AE techniques applied to concrete structures has already been published and considerable attention has been given to those publications. Some recent studies such as AE energy analysis and b-value analysis used to assess damage of concrete bridge beams have also been discussed. The formation of fracture process zone and the AE energy released during the fracture process in concrete beam specimens have been summarised. A large body of experimental data on AE characteristics of concrete has accumulated over the last three decades. This review of parametric based AE techniques applied to concrete structures may be helpful to the concerned researchers and engineers to better understand the failure mechanism of concrete and evolve more useful methods and approaches for diagnostic inspection of structural elements and failure prediction/prevention of concrete structures.

Evolution-induced catastrophe and its predictability

Both earthquake prediction and failure prediction of disordered brittle media are difficult and complicated problems and they might have something in common. In order to search for clues for earthquake prediction, the common features of failure in a simple nonlinear dynamical model resembling disordered brittle media are examined. It is found that the failure manifests evolution-induced catastrophe (EIC), i.e., the abrupt transition from globally stable (GS) accumulation of damage to catastrophic failure. A distinct feature is the significant uncertainty of catastrophe, called sample-specificity. Consequently, it is impossible to make a deterministic prediction macroscopically. This is similar to the question of predictability of earthquakes. However, our model shows that strong stress fluctuations may be an immediate precursor of catastrophic failure statistically. This might provide clues for earthquake forecasting.