Securing Production Engineering: Data Science and Cybersecurity in Product Development

Abstract

The convergence of data science and cybersecurity is reshaping the landscape of modern production engineering, where digital tools and interconnected systems are central to product development. This study investigates how the integration of machine learning-driven anomaly detection and cybersecurity frameworks can enhance the resilience, efficiency, and security of production environments. Using a multi-phase methodology that includes unsupervised learning models, vulnerability analysis, and statistical evaluation, the research evaluates security risks across CAD systems, IoT gateways, PLCs, and cloud-based platforms. Isolation Forest and Autoencoder models were tested across three production lines, with Isolation Forest demonstrating superior detection performance (F1-score > 0.96) and lower latency. Penetration testing identified critical vulnerabilities, particularly in control systems and network gateways, while correlation and hypothesis testing revealed significant relationships between vulnerability severity, incident frequency, and operational downtime. The post-integration period showed a marked decline in anomaly incidents, validating the effectiveness of embedded security protocols. This study concludes that combining data science with cybersecurity within the production lifecycle not only protects intellectual property and system integrity but also enables scalable, secure innovation. The findings advocate for a DevSecOps-oriented approach in engineering, ensuring that security is treated as a core component of intelligent, data-driven manufacturing.