Automated Configuration of Multi-Spindle Drilling Gear Machines by Sub-Domain Knowledge Processing and Adaptive Algorithms

A software concept and its realisation based on heuristic knowledge and pattern identification techniques for automated design of a multi-spindle drilling gear machine used in furniture production process is presented. The aim is to find an optimised design of the target-machine, this means to find a machine design with minimised number of drills and with the antagonistic goal to provide a fast production of the boards by minimising production-cycles per board. The design experience of a human expert was transferred to a design tool using his heuristic knowledge in combination with special developed pattern detection and recognition algorithms. Known and interpretable patterns are identified and used as information for a pre-design of the machine. The feasibility to manufacture each board is reached by analysing each single board to recognise known patterns for which drills are already equipped on the gears and the detection of new, uninterpretable patterns for which free spindle places can be equipped with suitable drills.