REAL-TIME TOOL CONDITION MONITORING AND SURFACE ROUGHNESS PREDICTION IN SMART MANUFACTURING

Abstract

The increasing demand for high-quality products, reduced production losses, improved machine availability, and adaptive process control has accelerated the adoption of intelligent monitoring technologies in modern manufacturing environments. Tool degradation and surface-quality variation are among the most significant challenges in machining operations because progressive tool wear can increase cutting forces, vibration, temperature, dimensional deviations, energy consumption, and surface irregularities while also creating unexpected production interruptions. Conventional tool inspection methods frequently depend on scheduled manual measurements or offline quality inspection, which may identify deterioration only after defective components have already been produced. This paper proposes a real-time tool condition monitoring and surface roughness prediction framework for smart manufacturing that integrates Industrial Internet of Things sensing, multi-source process data acquisition, edge preprocessing, feature extraction, machine-learning analytics, toolhealth assessment, surface-quality prediction, Digital Twin-based operational context, secure API integration, adaptive decision support, and continuous model lifecycle management. The framework collects heterogeneous machining information including vibration, cutting force, acoustic emission, spindle current, temperature, spindle speed, feed rate, depth of cut, machining duration, tool history, and contextual production variables. Edge-level processing performs signal validation, noise reduction, synchronization, segmentation, and feature generation before relevant observations are transferred to analytical services. The proposed methodology develops continuously updated tool-condition profiles and predicts surface roughness using combined sensor and process information rather than relying on isolated measurements. A Digital Twin layer maintains contextual representations of machine state, tool usage, process settings, predicted wear condition, and expected quality behavior. The decision-support component generates recommendations for tool inspection, parameter adjustment, maintenance intervention, and controlled tool replacement when predicted risks exceed acceptable operational limits. Secure OpenAPI-oriented interfaces enable interoperable communication among sensors, edge gateways, Digital Twin services, analytical models, manufacturing applications, and qualitymanagement platforms. A representative analytical evaluation compares conventional periodic monitoring with the proposed real-time framework using tool-wear detection accuracy, surface roughness prediction quality, defect rate, unplanned downtime, tool utilization, false alarm rate, response time, energy consumption, and overall equipment effectiveness. The results indicate that the proposed framework can substantially improve early wear detection, surface-quality consistency, production continuity, and tool-resource utilization. The study concludes that integrated multi-sensor analytics, contextual Digital Twins, and continuously managed predictive models provide a practical foundation for real-time intelligent machining supervision in smart manufacturing environments.