Construction of Machine Tool Calibration Plans Using Domain-Independent Automated Planning

The evolution in precision manufacturing has resulted in the requirement to produce and maintain more accurate machine tools. This new requirement coupled with desire to reduce machine tool downtime places emphasis on the calibration procedure during which the machine's capabilities are assessed. Machine tool downtime can be as much as $120 per hour and is significant for manufacturers because the machine will be unavailable for manufacturing use, therefore wasting the manufacturer's time and potentially increasing lead-times for clients. In addition to machine tool downtime, the uncertainty of measurement, due to the schedule of the calibration plan, has significant implications on tolerance conformance, resulting in an increased possibility of false acceptance and rejection of machined parts.

Currently calibrations are planned based on expert knowledge and there are no intelligent tools aiding to produce optimal calibration plans. This thesis describes a method of intelligently constructing calibration plans, optimising to reduce machine tool downtime and the estimated uncertainty of measurement due to the plan schedule. This resulted in the production of a novel, extensible domain model that encodes the decision making capabilities of a subject expert. Encoding the knowledge in PDDL2 requires the discretization of non-linear resources, such as continuous temperature change.

Empirical analysis has shown that when this model is used alongside state-of-the-art automated planning tools, it is possible to achieve a reduction in machine tool downtime greater than 10% (12:30 to 11:18) over expert generated plans. In addition, the estimated uncertainty due to the schedule of the plan can be reduced by 59% (48 µm to 20 µm). Further experiments on a PC architecture investigate the trade-o� when optimising calibration plans for both time and the uncertainty of measurement. These experiments
demonstrated that it is possible to optimise both metrics reaching a compromise that is on average 5% worse that the best-known solution for each individual metric. Additional experiments using a High Performance Computing architecture show that on average optimality of calibration plans can be improved by 4%; a potential saving of 30 minutes for a single machine and 10 hours for a company with 20 machines tools. This could incur a financial saving in excess of $1200 saving.