Extremal Optimisation Approach to Component Placement in Blood Analysis Equipment

Abstract

This reports present an initial study on generative mechatronic design of equipment for blood analysis where the samples and chemicals are forwarded in thin single millimeter vessels. The system of vessels in the equipment transfers the fluids to different stations where chemical reactions and studies are performed. One of the stations is an optical inspection that requires controllable lighting conditions using an array of LEDs of different types.

The focus is on the generative design of the placement and configuration of the LEDs. The placement of the LEDs has been taken as a studying case for the method of Extremal Optimisation (EO) approach to mechatronic design. This method forms an opposing strategy to methods like genetic algorithms and simulated annealing. This is because it discriminate the individual parts or components of the configuration that underperform in a particular aspect instead of the more classical strategy of favouring good configurations from global measures. The presented study also relates to the class of many-objective optimisation methods (MaOP) and originates from the concept of self-organised criticality (SOC). The characteristics of avalanche barrier crossings in the parameter search space is inherited from such systems.

The test case used for the evaluation places occupying circles onto a quarter ring domain representing LEDs and circuit board. The fluid vessels are represented by lit up small domains that are also approximated by a circular disc. Some conclusion upon the methods capability to form a valid solution are made. A framework for describing a set of local flaw-improvement rules, called D2FI is introduced.