The use of artificial intelligence for anomaly detection in industrial production lines

The collection of real-world data in industrial automation systems can be challenging due to factors such as variable environmental conditions, accuracy of sensors, high cost and complexity of systems. These challenges affect the efficiency and reliability of the data collection process. Therefore, data collection and processing techniques need to be continuously improved. Large volumes of data and their analysis require data processing and storage capacity. Overcoming these challenges is an important step in the development of industrial automation systems by strengthening data-driven decision-making processes. Artificial Intelligence (AI) is revolutionizing many industries today, transforming business processes, decision-making mechanisms, and user experiences. AI, which finds a wide range of applications from health to education, from finance to the automotive industry, is critical in increasing efficiency, optimizing decision-making processes and solving complex problems. This rapid progress creates a continuous wave of innovation and development in society and industry. In this study, a simulator is developed to reduce the cost of sensor installations required for the collection of real-world data. Sound, vibration and temperature data from the real environment are compared with the data generated from proposed simulation architecture using 19 unsupervised deep learning algorithms. Both the verification of the proposed simulation architecture and the analysis of sound, vibration and temperature data are examined, and the results are evaluated.