Predictive Maintenance In The IoT Era
Published 2019 · Computer Science
Predictive maintenance in the Internet of Things (IoT) era can be summarized as a maintenance methodology that brings together the power of machine learning and streaming sensor data to maintain machines before they fail, optimize resources, and thereby reduce unplanned downtime. This chapter introduces the fundamental concepts of a predictive maintenance program and its applicability to machines via the explosion of IoT. It analyses machine learning methodologies as they apply to predictive maintenance, their challenges, best practices, and risks. Preventive maintenance is typically scheduled using a bathtub curve. A bathtub curve indicates the probability of failure of components, thereby illustrating the life and reliability of the component population. The IoT refers to a network of interconnected objects or things. Industrial systems today are getting more complex via instrumentation of sensors that continuously monitor machine and environment parameters. Machine‐learning techniques can be broadly classified as supervised and unsupervised.