Development of Advanced Predictive Maintenance System

Unplanned or unnecessary maintenance of the equipment and instruments leads to waste of money and time in production or manufacturing plants. Current predictive maintenance techniques make use of offline data and basic prognostics tools that provide insufficiently accurate predictions. Moreover, existing predictive maintenance systems developed by external vendors are only accessible by large firms due to their high price. This project was initiated to develop an online, open-source and relatively accurate predictive maintenance system that employs Autoregressive Moving Average (ARMA) statistical prognostics method for small-sized companies that aim to supply products with the least waste of time and money in the process.
In this project, the proposed system was applied on 5 cases including current, voltage, active power, cold air temperature and discharge pressure of a process plant.