Industrial instruments arranged with sensors and computers inside an interconnected network allows for improvements in productivity and efficiency of operations. In recent years, this revolution has been most commonly referred to as the Industrial Internet of Things (IIoT). By collecting vast amounts of data on processes, they can be analyzed and used to train machine learning algorithms to classify conditions, detect anomalies, and forecast the future. The ability to digest, reason, and make inferences about very high-dimensional data allow for longer-term more optimal strategies. After deployment, these machine learning models respond quickly, allowing decisions to be made over finer time intervals– this opens up a new world of tactics that lead to more optimal business results.Some domains of industry that this has impacted most in recent years are agriculture, energy, and manufacturing, where even fractionally small improvements have significant economic value.Automatic sensor readings may indicate the health of a machine asset. This realization has led to an increased desire to perform maintenance based on condition rather than solely based on time since the last scheduled event. Condition-based maintenance leads to significant financial savings. One way to achieve this is by building a data-driven model that can predict the remaining useful life (RUL) of machines based on sensor readings.

To optimize the maintenance schedule, we built a system of machine learning models. The system may be queried from on-premise computers or from a remote server, depending on the clients particular performance and privacy considerations. Our solution: