Middleware architecture application for predictive maintenance

In parallel with the development of technology, the concepts of Industrial Internet of Things (IIoT) and Intermediate layer come to the fore in industrial systems that digitize. Thanks to the development of IIoT, many benefits are provided in the industry, such as increasing efficiency, saving time and labor, reducing error rates, improving quality and providing easy management. Middleware architectures are needed to meet industrial needs and provide high-efficiency environments. In addition, predictive maintenance activities can be carried out thanks to the interpretation of the data collected from the sensors used in the working environments and the anomaly detection performed on the machinery and equipment. The ability to detect anomaly in advance ensures that maintenance needs can be met with low cost and high reliability. Thus, the systems used are continuously monitored and largely made sustainable. In our study, a data pipeline system is designed using open source Fiware middleware components, Orion Context Broker, OPC-UA IoT Agent and Draco. In this designed system, Modbus RTU communication protocol is used between V20 driver and Plc, which are hardware components, and OPC-UA communication protocol is used between Plc broadcasting OPC-UA server and middleware. The current and frequency data of the NGSI data model, which is read by the Fiware component OPC-UA IoT Agent, are transferred to the middle layer through the test setup created by programming the V20 motor driver, touch panel screen and Plc hardware components one by one. The data is forwarded to the PostgreSQL database and stored in the PostgreSQL database by undergoing the necessary transformation processes with the middleware component Draco. Finally, the flow data stored in the database is simplified by using feature selection methods and an anomaly detection application is carried out using eight different machine learning algorithms.