Why is machine tool diagnostics and prognostics difficult?

Real time asset monitoring, diagnostics and prognostics are complex tasks in any piece of equipment. In case of machine tools prognostics it becomes even more challenging. In industrial equipment we monitor the condition, analyze the data, trend and predict the failure of the equipment. The goal is to accurately estimate the eminent failure (act before it fails) and maximize the uptime and output. In machine tools this is not enough. A machine tool transforms the raw material into the finished part. The machine tool is like the mother of the tons of children products. The goal is to not only minimize the downtime but to produce quality parts. The health of the machine plays a vital role in quality of a component in terms of dimensional tolerance and surface quality. Thus correlating the machine health with the effect it has on the quality and mapping it with the requirements is the key in machine tool prognostics. The aim is to build machine health monitoring system from a machine-centric to a part quality-centered point of view. The second issue lies in the proprietary manufacturing hardware and software, which makes it difficult to capture real-time data in the required format and granularity. This has resulted in a customized health management for each system. The third issue is the lack of methodology to evaluate single component health information, detect imminent failure, and diagnose the type of fault in case of multiple-failure mode. There is a need for a decision-making and information fusion methodology to correlate single component health information with critical product quality, as well as to generate a unified machine tool health indicator to describe the machine’s overall condition.