Failure Detection in PLC Controlled Industrial Machines Using Machine Learning

Abstract

PLCs are the backbone of modern industrial automation, providing precise and reliable control over manufacturing processes, ensuring efficiency, safety, and cost-effectiveness. The use of PLCs has revolutionized the industrial sector, enabling companies to optimize their production processes, reduce downtime, increase productivity, and ultimately, stay ahead of the competition. A range of factors such as electrical and mechanical issues, environmental factors, software errors, and human error can cause failure in machines controlled by PLCs, underlining the critical importance of regular maintenance, proper testing, and thorough validation of PLC programming to prevent costly downtime and ensure the smooth operation of industrial processes. Identifying errors in these machines can be challenging due to the complex nature of industrial processes, the potential for multiple causes of failure, and the intricate programming involved in PLCs, underscoring the importance of having experienced professionals with specialized knowledge in PLCs and industrial automation to diagnose and resolve issues quickly and effectively. This work presents the development of a machine learning algorithm to identify operation sequence failures in machines controlled by PLCs. The system achieved an accuracy of 90% in detecting abnormal actuator activation or non-activation, registering an alarm condition and providing the exact input or output address of the error. The results indicate that the system has the potential to reduce downtime and the number of stops in industrial operations, showcasing the potential of machine learning techniques in industrial applications.