Embedded IoT solutions for Industry 4.0 manufacturing process digitalization

The fourth industrial revolution has created significant opportunities for intelligent, data-driven manufacturing. However, the adoption of Industry 4.0 technologies remains inconsistent across the industrial landscape, particularly among small and medium-sized enterprises. These companies form a critical part of global manufacturing yet often lack the financial resources, technical expertise, and organizational readiness needed to undertake large-scale digital transformation. Their production environments frequently rely on legacy machinery that still performs reliably but lacks the connectivity and sensor infrastructure required for real-time monitoring and advanced analytics. As a result, small and medium-sized enterprises risk falling behind technologically even as the value of data-driven manufacturing becomes increasingly evident. This thesis investigates how low-cost embedded Industrial Internet of Things solutions can support a gradual and sustainable pathway toward Industry 4.0 adoption. Rather than promoting disruptive modernization or extensive reinvestment in new equipment, the work focuses on retrofitting existing machinery with embedded sensing, data processing, and communication capabilities. The central contribution is the STRIDE4.0 framework, a structured methodology that guides companies through incremental digital transformation. STRIDE4.0 incorporates principles from digital transformation theory, continuous-improvement practices inspired by DMAIC, and test-before-invest prototyping strategies. This combination ensures that each step of the transformation produces meaningful value, building organizational motivation and reducing financial and operational risk. A major component of this thesis focuses on the role of embedded systems in facilitating the early phases of digitalization. Embedded microcontroller-based devices and industrial control platforms represent affordable and flexible tools for acquiring and processing industrial data, supporting edge-computing paradigms, and enabling local decision-making where appropriate. These systems can be selected and configured to meet requirements for robustness, scalability, security, and unobtrusiveness, all of which are essential for deployment in small and medium-sized enterprises. Retrofitting then becomes a practical strategy that extends the operational life of legacy machinery while turning it into a source of actionable data. The practical validation of STRIDE4.0 is carried out through a real-world case study on hydraulic broaching machines used in the production of polypropylene tubes. Broaching is a mechanically demanding process where tool wear significantly affects part dimensional accuracy, tool life, downtime, and scrap rates. Despite its industrial relevance, broaching is traditionally monitored manually or with minimal sensing infrastructure. The machines investigated in this research relied on analog gauges without any capability for recording or analyzing pressure profiles. Digital pressure sensors were installed on the machines and connected to embedded Industrial Internet of Things devices that provided real-time data capture, wireless communication, and cloud-based storage. A complete data pipeline was developed, including firmware modules for sensor management, processing routines for noise reduction and resampling, algorithms for repairing and normalizing pressure traces, and interfaces for visualization and inspection. The system was designed to integrate smoothly into the existing factory workflow and required no interruptions to production. The collected hydraulic pressure data revealed characteristic patterns that correlate with tool condition and process behavior. Based on these observations, automated categorization algorithms were developed to divide each broaching cycle into distinct operational phases, including idle, without-load, ramp-up, plateau, ramp-down, cut, and block conditions. These categories allow a deeper interpretation of pressure signals and support the extraction of features relevant to process health. A key scientific contribution of this thesis is the development of a hybrid predictive-maintenance framework. The approach combines elements of physics-informed reasoning with data-driven analysis and focuses especially on trend-based monitoring. The central metric in this framework is the Wear Index, an interpretable indicator derived from the temporal progression of pressure peaks in the ramp-up phase. The Wear Index captures gradual increases in cutting resistance as the broach teeth degrade. Through the analysis of extensive industrial datasets, the Wear Index was shown to follow stable and interpretable growth patterns, making it a powerful tool for predicting when maintenance should be performed. The predictive-maintenance framework is strengthened by additional techniques such as polynomial trend modeling, anomaly-detection heuristics, and environmental normalization. Environmental factors, including temperature, humidity, and atmospheric pressure, were shown to influence pressure measurements, and compensating for these variations significantly improves predictive robustness. The resulting predictive model is both reliable and interpretable, and it is designed specifically for deployment in manufacturing environments where transparency and trust are essential for operator acceptance. The results of the industrial deployment demonstrate that STRIDE4.0 provides an effective and scalable pathway for small and medium-sized enterprises seeking to adopt Industry 4.0 technologies. The retrofitted broaching machines became transparent, data-generating assets that supported improved decision-making and deeper process understanding. The Wear Index enabled proactive scheduling of tool replacement, reducing unplanned downtime and preventing deviations that would otherwise result in defective parts. Operators and decision-makers gained confidence as early benefits became visible, helping to overcome the financial and motivational barriers that often hinder digital transformation. Beyond the specific broaching application, the thesis highlights the broader relevance of the STRIDE4.0 methodology. Its modular hardware and software components, scalable architecture, and hybrid data-modeling approach make it suitable for a wide range of industrial processes that rely on legacy equipment. The methodology is adaptable, accessible, and grounded in practical considerations that reflect the realities of small and medium-sized industrial environments. Future research directions include the integration of additional sensory modalities such as vibration and acoustic emissions, the use of more advanced machine-learning methods for anomaly detection and time-series forecasting, and the extension of the framework toward full digital-twin capabilities. In summary, this thesis presents a coherent strategy for enabling small and medium-sized enterprises to progress toward Industry 4.0 through the use of embedded IoT systems and hybrid predictive-maintenance techniques. The development of STRIDE4.0, in combination with the retrofitting of broaching machines and the creation of the Wear Index, demonstrates that digital transformation can be practical, affordable, and incrementally achievable. The research bridges the gap between advanced Industrial Internet of Things concepts and the operational constraints of real manufacturing environments, contributing to a more inclusive and sustainable evolution of industrial processes.