Incorporating deep learning data analytics techniques in the optimisation of capacitated planned maintenance

Abstract

Manufacturing systems must be supported by the availability of materials, a streamlined production process and a prepared production line to achieve the production target. In a mass customization manufacturing system, the number of machines required for customization is relatively small. Conse-quently, maintenance on critical machines will impact this manufacturing system the most. Two types of maintenance strategies are implemented: corrective and preventive maintenance. The corrective maintenance requires more resources since the time and cost to repair the breakdown machine will be higher due to fatal failure. For the management to consider preventive maintenance while the binding machines are still operational, it must be equipped with a deep analysis demonstrating that fewer resources will be required. This paper discusses two deep analyses: accurate prediction of the binding machines' breakdown based on Mean Time Between Failure (MTBF) data using a deep learning data analytics technique and optimizing the maintenance total cost in the available capacitated time. The findings and results of this paper show that the proposed deep learning data analytics technique can increase the MTBF prediction accuracy by up to 66.12% and reduce the total maintenance cost by up to 4% compared with the original model.