Predictive maintenance project implementation based on data-driven & data mining

Today's global economy is experiencing a sharper slowdown than ever before. With a higher risk of deterioration or disappearance of manufacturing companies than usual. As a result, manufacturers must face this situation by adopting strong strategies, including equipment management and cost optimization in connection. This is why predictive maintenance is an important pillar in achieving these objectives. Except that predictive maintenance requires a budget and time for the proper implementation, such investment can only be accepted by investors if it is sure that the expected results will contribute effectively to the reduction of maintenance costs. To do this, it is necessary to assess the risks that could impact this project's success, mainly the data reliability and the machine learning model performance. The aptitude to predict the need for maintenance of a system in a perfect way at a specific time is one of the main challenges in this scope. This paper proposes a methodology for data management in the case of predictive maintenance project implementation. It starts by introducing the project study phase for cost & benefit evaluation based on data-driven. Then it presents the predictive concept based on data mining & machine learning tools for optimal model building, as well for the project performance follow up a monitoring approach is proposed based on the continuous improvement concept.