A Fuzzy C-Means With Support Vector Regression for Calibration and Fault Detection in Modern Instruments

From the past few decades, Calibration and Fault Detection in modern instruments are crucial for ensuring accuracy, reliability, and safety by configuring instruments to provide accurate measurements and identifying anomalies or malfunctions. These processes prevent accidents, reduce downtime, and increase productivity, but also have drawbacks, including being time-consuming and labor-intensive, requiring specialized expertise and equipment, and being expensive, while also being susceptible to environmental factors and requiring frequent recalibration. To overcome these problems, a Fuzzy C-Means with Support Vector Regression (FCM-SVR) is proposed for detecting calibration and fault in modern instruments using Predictive Electrical Fault Analysis (PEFA) dataset. The collected data is undergone preprocessing using Label Encoder (LE) to normalize data without reducing the dimension. Then, Empirical Mode Decomposition and Sample Entropy are considered for feature extraction for robust fault detection and to reduce false alarm. At last, FCM-SVR is employed as classifier for enhancing computational efficiency, flexibility, and overall performance. From the results, it clearly shows that the proposed FCM-SVR obtained higher values in the context of precision, recall, F1 score and accuracy when compared to existing method K-Nearest Neighbor (KNN).