Product quality driven auto-prognostics: Low-cost digital solution for SMEs

Abstract

Setting out existing prognostics solutions in small and medium enterprises (SMEs) is accompanied by challenges. These include employing expensive sensors, acquisition systems; and attending geometric limitations. Additionally, these solutions call for a specialist to take on feature engineering, machine learning algorithm selection, etc. Presented in this paper is a low-cost digital solution (intelligently integrate cost-cutting off-the-shelf technologies) for SMEs via product quality driven auto-prognostics. First, we develop upon existing solutions by addressing their drawbacks viz. cost, geometric limitations via a new product quality-centered condition monitoring strategy. Every SME must investigate the quality of their products, and therefore the authors believe this to be a low-cost solution. Next, the proposed solution integrates automated machine learning via Auto-WEKA, an off-the-shelf open-source technology. Lastly, the practical advantages of the proposed solution over the existing sensor-based solution were investigated via a case study. Results depict that this low-cost prognostics solution is vital for maintenance planning in SMEs.