Fuzzy Logic-Enhanced Expert System for Real Time Anomaly Detection in CNC Machines

Computer Numerical Control (CNC) machines play a pivotal role in modern precision manufacturing, where real-time monitoring is essential to prevent catastrophic failures and minimize downtime. This study proposes a Fuzzy Logic-Enhanced Expert System (FLEES) for real-time anomaly detection in CNC machines, leveraging linguistic rule inference fused with physical constraints and data-driven optimization. The system processes 14 distinct sensory inputs, including spindle vibration, cutting torque, thermal gradients, and acoustic emissions, gathered from 120 hours of high-frequency CNC machine operation under varying load conditions. Fuzzification maps raw sensor signals to 42 linguistic variables using Gaussian and trapezoidal membership functions. A total of 96 fuzzy rules were formulated based on expert knowledge and refined via Particle Swarm Optimization (PSO) guided by energy consistency and classification loss minimization. Experiments conducted on a benchmark CNC dataset show that FLEES achieves 96.7% anomaly classification accuracy, with 95.3% sensitivity and 97.9% specificity, outperforming existing methods, including SVM (91.2%) and LSTM (94.1%). Moreover, the system maintains a real-time response under 180 milliseconds per inference cycle. These results confirm that integrating fuzzy reasoning with physics-informed optimization enhances reliability and interpretability for real-time fault diagnostics in smart manufacturing.