Visualisation methods for supporting the exploration of high dimensional problem spaces in engineering design
Ian Packham, Susan L. Denham · 2004
A visualisation system designed to help users understand and evaluate engineering design problems is described. The data is generated by a genetic algorithm that supplies broad high performing clusters but the quality of these regions of the search space need to be evaluated for robustness and sensitivity by the engineer. A novel clustering technique based on kernel density estimation identifies the clusters in terms of the design variables. Clustering can be performed in alternative coordinate systems such as the principal components that reveal the 'natural' clusters in the data. A number of high dimensional visualisation techniques are included to help understand the data; views are linked by the colour of the clusters. The user is encouraged to search for new data in different parts of the design space. Evaluation of regions for robust engineering design is supported through filtering and a novel 'negative' search mechanism that redefines clusters and confirms whether they satisfy tolerances specified by the user.