Пошук академічних текстів

Інформація × Реєстраційний номер 2125U004243, Матеріали видань та локальних репозитаріїв Категорія Стаття Назва роботи Enhanced Super SCADA architecture with integrated predictive analytics for spiral steel pipe manufacturing Автор Bakhtiyarov B.Jabiyeva A.Khudaverdiyeva M.Bakhtiyarov B.Jabiyeva A.Khudaverdiyeva M. Дата публікації 01-01-2025 Постачальник інформації Сумський державний університет Першоджерело https://essuir.sumdu.edu.ua/handle/123456789/101036 Видання Sumy State University Опис The article proposes an enhanced Super SCADA architecture with internal predictive analytics, specifically designed for the spiral steel pipe manufacturing industry. Modern industrial environments require intelligent supervisory systems that can handle massive, heterogeneous sensor data streams in real-time. The overall aim of this research is to enhance the quality, stability, and production predictability of production processes by implementing enhanced anomaly detection and data-driven decision mechanisms integrated into a comprehensive supervisory control platform. The proposed architecture combines Apache Kafka for high-throughput message brokering and Apache Spark Streaming for real-time analytics, guaranteeing continuous data streams and low-latency event processing. A synthetic supervisory control and data acquisition (SCADA) dataset of 100,000 sensor records was created to mimic a real industry environment scenario, including 5 % labeled anomalies such as temperature spikes, motor current anomalies, and sensor drift. Anomaly marking was performed based on the operational source and duration of the fault to obtain realistic benchmarking conditions. For model evaluation, the predictive layer was compared with a traditional baseline, which included a rule-based threshold detector and a statistical z-score model, achieving 82 % and 85 % accuracy, respectively. On the other hand, the proposed Super SCADA model achieved an anomaly-detection accuracy of 93 % and reduced the average detection latency to 2.3 s, representing a 1 % improvement in performance compared to traditional SCADA topologies. Added a self-optimization feedback loop, increasing system reliability and learning potential. Overall, the research contributes to the development of an autonomous, scalable, and intelligent supervisory control paradigm in line with Industry 5.0 principles. The resulting architecture is suitable for modular implementation, easy interfacing with manufacturing execution systems (MES) / enterprise resource planning (ERP) systems, and online process management, providing an essential step towards implementing next-generation smart manufacturing environments. Додано в НРАТ 2026-04-17 Закрити