Multiple expert classification: a new methodology for parallel decision fusion

Abstract. A new parallel hybrid decision fusion methodology is proposed. It is demonstrated that existing parallel multiple expert decision combination approaches can be divided into two broad categories based on the implicit decision emphasis implemented. The first category consists of methods implementing computationally intensive decision frameworks incorporating a priori information about the target task domain and the reliability of the participating experts, while the second category encompasses approaches implementing group consensus without assigning any importance to the reliability of the experts and ignoring other contextual information. The methodology proposed in this paper is a hybridisation of these two approaches and has shown significant performance enhancements in terms of higher overall recognition rates along with lower substitution rates. Detailed analysis using two different databases supports this claim. Received January 19, 1999 / Revised March 20, 2000     

Dynamic problems of shape optimal design for shallow curved plates

The problem of optimal structural design of shallow thin-walled elements such as curved rectangular plates are formulated and solved for dynamic conditions. The distribution of the initial curvature of shallow plates in a nonstrained state is taken as the control function. Dynamic compliance is considered as the minimized performance functional. Optimality conditions are derived for the distributed parameter system considered and applied for the construction of the analytical solution. The rigorous analysis of extremum conditions and behavioural equations shows that the initial optimization problem is decomposed into several problems of classical structural analysis, which can be successfully solved analytically. Some optimal designs obtained for rectangular plates under stretching and bending, and a plate lying on an elastic foundation and subjected to lateral forces are presented. Received: November 27, 1998     

Optimal structural topologies with transmissible loads

In optimal topological design of structures one obtains the configuration of optimal structures when the design domain, the displacement boundary conditions and the applied loads are specified. In the optimal structure one often notices a marked difference between the main bearing structure and the load transfer zones. The latter are composed of relatively light elements the exact nature of which is not always very distinct. The main purpose of this paper is to allow the main bearing part of the structure to emerge. Moreover the actual location of the load along its line of action is not always a design requirement. In order to include this relaxed condition regarding the loading position the concept of transmissible or sliding forces is introduced in topological design of structures. A transmissible force is a force of given magnitude and direction which can be applied at any point along the line of action of the force. The optimization formulation is similar to standard topological design procedure in addition to the condition of transmissability of the forces. It is shown that this condition reduces to an equal displacement constraint along the line of action of the forces. The method is illustrated by typical structural examples. It is observed that this numerical method produces indeed crisp images of the main structural components, unblurred by the secondary load transfer elements. It is also indicated that many results are often replicas of Prager structures which were previously obtained by analytical methods. Received March 3, 1999     

Maximization of eigenvalues using topology optimization

Topology optimization is used to optimize the eigenvalues of plates. The results are intended especially for MicroElectroMechanical Systems (MEMS) but can be seen as more general. The problem is not formulated as a case of reinforcement of an existing structure, so there is a problem related to localized modes in low density areas. The topology optimization problem is formulated using the SIMP method. Special attention is paid to a numerical method for removing localized eigenmodes in low density areas. The method is applied to numerical examples of maximizing the first eigenfrequency. One example is a practical MEMS application; a probe used in an Atomic Force Microscope (AFM). For the AFM probe the optimization is complicated by a constraint on the stiffness and constraints on higher order eigenvalues. Received June 10, 1999     

The New Frontier Issues for Cognition Technology

Cognitive engineering has developed enormously over the last fifteen years. Yet, despite many excellent research projects and publications, its full potential has not been embraced into mainstream system design. This paper will examine the reasons for this failure and argue that the problem is not simply inertia or lack of education. There are strong organisational influences that cause resistance to this particular approach. The discipline itself has characteristics that make it fragile in the modern corporate structure. In addition, the cognitive engineers themselves are not blameless in the equation. They appear to have done exactly what they criticise the engineering community for doing: they have packaged their product in a manner that is not ‘user friendly’ to its target population, not structured to suit its application, and not output in the format required. Suggestions will be made to rectify the situation: a list of actions is proposed for practising cognitive engineers to make their product more likely to enjoy widespread uptake.     

Ride comfort optimization of a city bus

Ride comfort optimization of a city bus is considered in this article, using stiffness and related damping of the power unit mounts as design variables. The response of the bus, due to road irregularities, is calculated in the time domain by using a finite element (FE) model describing the complete vehicle. The body of the bus is condensed, using a substructuring technique, in order to reduce the simulation times. Operational deflection shapes (ODS) have been used to compare measured and calculated vibrations and some corrections of the FE model have been done. The response, i.e. acceleration, at three locations in the bus has been frequency-weighted according to the comfort standard ISO 2631:1997. The resulting root mean square (RMS) values of these responses are then used in the objective functions in the optimization procedure. The computational results show that it is possible to increase the ride comfort. However, the improvement of the ride comfort is probably too small to have a substantial subjective impact on the passenger in the bus. The results show that the objective function, regarding the overall ride comfort in the bus, has local optima. Received October 10, 1999