Tool and Workpiece Condition Classification Using Empirical Mode Decomposition (EMD) with Hilbert–Huang Transform (HHT) of Vibration Signals and Machine Learning Models

Existing studies have attempted to determine the tool chipping condition using indirect method of data capture and intelligent analysis techniques considering machine parameters, conditions signal processing techniques. Due obstructive nature machining operation, however, it is daunting use capturing intelligently as well that workpiece. This study aimed apply some advanced vibration signals captured experimentally during operation for decision making workpiece conditions. Vibration were turning operations while four (4) classes tools, based on their flank wear. The first pre-processed decomposed Empirical Mode Decomposition (EMD) method. Hilbert–Huang transform (HHT) was applied resulting IMFs obtained compute feature vectors used classify A total 12 features, consisting instantaneous properties such energy, frequencies, amplitudes, training classification To optimize process, selection performed a genetic algorithm (GA) reduce number features from 4 classification. trained with neural network scaled conjugate gradient (SCG) algorithm. result showed model error 0.102. Two other learning models, support vector (SVM) K-Nearest Neighbors (KNN), also implemented classifying conditions, vector, most accurately predicted tool. avoid bias misclassification errors, k-fold cross-validation technique ‘k’ taken 5 10. computed inputs train both SVM KNN models machining. loss each evaluated plotted review performance. average overall 0.5031 observed 5-fold cross-validation, whereas 0.0318 cross-validation. 0.5009 10-fold when selected by (GA), 0.0343 model. optimal performance all training, better implemented. losses be less in compared SCG. results developed 10 times than predicting process.