Analysis of multiphase induction machines with winding faults

The paper shows how the techniques of generalized harmonic analysis may be used to simulate the steady-state behavior of a multiphase cage induction motor with any form of open-circuit or short-circuit fault in the stator winding. The analytical model is verified using a four-pole machine with a 48-slot stator. Each coil of the stator winding of this machine is brought out to a patchboard that enables the stator to be configured for single-phase, two-phase, three-phase, four-phase, six-phase, or 12-phase excitation. Experimental results are compared with computer predictions for a six-phase machine with both open-circuit and short-circuit faults.     

Risk Management and Design of Critical Bridges for Terrorist Attacks

In the aftermath of the September 11th tragedies, the vulnerability of the United States' transportation infrastructure to terrorist attack has gained national attention. In light of this vulnerability, various governmental agencies are looking into ways to improve the design of structures to better withstand extreme loadings. Until recently, little attention has been given to bridges which are critical to our economy and transportation network. This paper summarizes the results of ongoing research to investigate economical, unobtrusive, and effective methods to mitigate the risk of terrorist attacks against critical bridges. It outlines a recommended plan to reduce these threats through proven risk management techniques, lists possible cost-effective security measures, discusses blast effects on bridges, and provides structural design and retrofit guidelines. It also discusses ongoing research oriented towards the development of a performance-based design methodology. In using proper risk management techniques, transportation managers and bridge engineers can mitigate the risk of terrorist attacks against critical bridges to an acceptable level, while ensuring efficient use of limited resources.     

Online learning via congregational gradient descent

We propose and analyse a populational version of stepwise gradient descent suitable for a wide range of learning problems. The algorithm is motivated by genetic algorithms which update a population of solutions rather than just a single representative as is typical for gradient descent. This modification of traditional gradient descent (as used, for example, in the backpropogation algorithm) avoids getting trapped in local minima. We use an averaging analysis of the algorithm to relate its behaviour to an associated ordinary differential equation. We derive a result concerning how long one has to wait in order that, with a given high probability, the algorithm is within a certain neighbourhood of the global minimum. We also analyse the effect of different population sizes. An example is presented which corroborates our theory very well.This work was supported by the Australian Research Council.     

Estimation of the inter-bar resistance of a cast cage rotor

This paper describes an experimental investigation of bar-to-bar resistance. It addresses the interpretation of experimental data, and the use of that data to develop a means of estimating bar-to-bar resistance at the design stage.     

A Dynamic Interaction Between Machine Learning and the Philosophy of Science

The relationship between machine learning and the philosophy of science can be classed as a dynamic interaction: a mutually beneficial connection between two autonomous fields that changes direction over time. I discuss the nature of this interaction and give a case study highlighting interactions between research on Bayesian networks in machine learning and research on causality and probability in the philosophy of science.     

Development and evaluation of an intervention to reduce rip current related beach drowning

The objective of this research was to evaluate a campaign to improve beachgoer recognition of calm-looking rip currents, known to contribute to surf drowning. Posters, postcards, and brochures conveying the message “Don’t get sucked in by the rip” were distributed in an intervention area. Beachgoers were interviewed in this and a similar control area one year before and immediately after the intervention (respective response rates: 69.9% and 82.3%), Consenting respondents were sent follow-up questionnaires after approximately 6 months and 55% responded. In the intervention area, 28.8% of post-intervention, and 57.2% of follow-up respondents, had seen our campaign. At post-intervention, intervention respondents demonstrated improvement (relative to baseline) in intentions to swim away from a calm-looking rip, ability and confidence in identifying a rip, intention never to swim at unpatrolled beaches, and responses to being caught in a rip, compared to the control respondents. Similar improvements were observed post-intervention for respondents in the intervention area who had seen our campaign (relative to those who had not), and at 6 month follow-up for intervention respondents (relative to control respondents). The relatively brief print-based campaign was effective in warning beachgoers about calm-looking rips.     

Using Social Network Analysis to Visualize the Social-Media Networks of Community Groups: Two Case Studies from Sydney

Community participation in planning is generally considered crucial for the delivery of positive outcomes; however, the network structures that can be created by community groups that use social media and participate in the network are not widely understood. This paper explores the use of social media, specifically Twitter, by two community groups in Sydney. In the context of this study, community groups are self-created and organized groups of citizens that form to oppose a proposal to amend planning controls for a specific site. Employing the research technique of Social Network Analysis (SNA), this paper seeks to visualize community group social media networks, as well as understand who is connected and who is participating within the networks. For the two community groups investigated, it was found that they do not attract large numbers of friends and followers on Twitter and key stakeholders play a passive listening role in the networks.