Stabilizing Manufacturing Lines of Customized Building Mechanical Components under Uncertain Machinery Deterioration

In customized building mechanical component production, more frequent quality issues and maintenance activities caused by unexpected machine failures interrupt production and result in waste and delays. Integrating predictive maintenance strategies into manufacturing operations is necessary to solve this problem. However, under the uncertainties of machinery deterioration and uncertain productivity, existing literature lacks a quantifiable index for precise evaluation for measuring the stability of joint maintenance and production schedules. A significant challenge exacerbating this gap is the scarcity of production and maintenance data to infer these deterioration models accurately. As a result, current methods struggle to balance production goals with production machine maintenance needs, often leading to sub-optimal operational performance and increased machine downtime. This paper presents a joint maintenance and production scheduling framework that integrates data extraction and process mining to infer manufacturing line deterioration models. This framework generates stable maintenance coordination plans, considering uncertain customized products' production schedules and machinery deterioration rates. It establishes the needed predictive maintenance using the deterioration models estimated from limited production and maintenance data. The production and predictive maintenance activities can be thus scheduled simultaneously to stabilize productivity and quality. The authors validated the proposed framework using data from a ventilation duct manufacturing line. The results illustrate that this framework can use limited production and maintenance data to produce reliable machine deterioration models that support stable joint maintenance and production scheduling.