Equipment Sub-system Extraction and its Application in Predictive Maintenance

We developed a method for extracting sub-systems for predictive maintenance. The method processes sensor data obtained from target equipment to extract several sub-systems that contain correlated variables. First, the latent variables are generated by analyzing correlation coefficients. Next, the method divides the variables (latent variables and sensors) into different groups by applying a hierarchical clustering method. Each group represents a sub-system that has the variables strongly correlated with each other. The application of sub-systems in predictive maintenance helps users to detect anomalies earlier and more accurately than the conventional methods, leading to better maintenance and productivity. The effectiveness of this method was evaluated by using sensor data obtained from compact electrical generators. We also describe possible ways of visualizing the sub-systems.