Predictive Maintenance Model for Pneumatic Systems Supported by Industry 4.0 and Edge Computing
Pneumatic systems are actuators that convert compressed air from the atmosphere into pressure energy and use it for mechanical action. The movements referred to here are mechanical movements, such as linear or circular movements. Pneumatic systems have been used in the manufacturing industry since the 1950s. Today, we see that pneumatic installations are also being constructed alongside electrical and water lines in factories that produce goods. This situation clearly demonstrates that pneumatic systems have become an indispensable and fundamental necessity in production processes. When evaluating pneumatic systems from an Industry 4.0 perspective, it becomes apparent that a great deal of data can be collected from the environments in which pneumatic systems are used to increase efficiency. Although there are different approaches to managing and analyzing the data obtained from these sources, cloud computing and edge computing are the two most prominent approaches. Today, it is common to see these two approaches used together. Data obtained from the source via Edge Computing is analyzed at the source and used to obtain quick results, while Cloud Computing is used only for the transfer and analysis of important data. This reduces unnecessary time and performance losses in the system, making it more efficient. In this study, the Industry 4.0 concept's Edge Computing approach was applied to a real-life problem. The problem addressed was predictive maintenance in pneumatic systems. This problem was studied on the fully automated bread processing and stacking line of Brotmas, a company based in Konya that manufactures baking machines. Data was collected from the company on key points in the system, including the detection of leaks in the air piping, compressor oil level control, manometer pressure value control, compressor belt tension control, and inverter-based compressor motor health control. The collected data was compared with new information to prevent potential failures. As a result of this study, it was observed that error notifications caused by compressor shutdown, pressure drop, maintenance deficiencies, and pneumatic piston failures decreased by 67%.