Combined analysis of acoustic emission and vibration signals in monitoring tool wear, surface quality and chip formation when turning SCM440 steel using MQL

Abstract

With modern production, Minimum Quantity Lubricant (MQL) technology has emerged as an alternative to conventional liquid cooling. The MQLs is an environmentally friendly lubricant method with low cost while meeting the requirements of machining conditions. In this study, the experimental and analytical results show that the obtained acoustic emission (AE) and vibration signal components can effectively monitor various circumstances in the SCM440 steel turning process with MQL, such as surface quality and chip formation as cutting tool conditions. The AE signals showed a significant response to the tool wear processes. In contrast, the vibration signal showed an excellent ability to reflect the surface roughness during turning with MQL. The chip formation process through the cutting mode parameters (cutting speed, feed and depth of cut) was detected through analysis amplitude of the vibration components Ax, Ay and Az and the AE signal. Finally, Gaussian process regression and adaptive neuro-fuzzy inference systems (GPR-ANFIS) algorithms were combined to predict the surface quality and tool wear parameters of the MQL turning process. Tool condition monitoring devices assist the operator in monitoring tool wear and surface quality limits, stopping the machine in case of imminent tool breakage or lower surface quality. With the unique combination of AE and vibration analysis model and the training and testing samples established by the experimental data, the corresponding average prediction accuracy is 97.57 %. The highest prediction error is not more than 3.8 %, with a confidence percentage of 98 %. The proposed model can be used in industry to predict surface roughness and wear of the tools directly during turning

Methodology for Tool Wear Detection in CNC Machines Based on Fusion Flux Current of Motor and Image Workpieces

“…Current had a low contribution to predictions in their experiments. Similarly, the results reported by Hoang et al [18] indicated the ability of AE and vibrations to predict wear and roughness in the machining of SCM440 steel, combining a Gaussian process regression and adaptive neuro-fuzzy inference system (GPR-ANFIS) algorithms to process RMS values of the signals, achieving an average prediction accuracy of 97.57% in online monitoring with the proposed methodology.…”

“…Several literature reviews have identified the main techniques, tools and trends for processing [21][22][23][24]. Firstly, signals are processed directly in the time domain [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18] or techniques or transforms are used for their analyses in the frequency [3,5,[18][19][20] or time-frequency [19,20] domains. From here, several techniques are used to obtain features that allow the analysis to be carried out in a better way, such as the use of statistical indicators [3][4][5][6]8,14,[18][19][20], time or time-frequency transforms for direct feature extraction [3][4][5]19,20] and, in some cases, methods for the selection of the most appropriate features or dimensionality reduction such as heuristic techniques [5,6,22], linear discriminant analysis (LDA) [19] or principal component analysis (PCA).…”

“…From here, several techniques are used to obtain features that allow the analysis to be carried out in a better way, such as the use of statistical indicators [3][4][5][6]8,14,[18][19][20], time or time-frequency transforms for direct feature extraction [3][4][5]19,20] and, in some cases, methods for the selection of the most appropriate features or dimensionality reduction such as heuristic techniques [5,6,22], linear discriminant analysis (LDA) [19] or principal component analysis (PCA). Subsequently, classification or decision-making techniques tend to use intelligent systems such as different types of neural networks [8,13,14,16,19,25], support vector machines (SVM) [21][22][23][24], hidden Markov models (HMM) [21][22][23][24], fuzzy systems [8,9,17,18] and DL systems [4,11,14,15].…”

“…The use of intelligent techniques is of great value when processing large amounts of information generated by fused sensors, allowing a more adequate manipulation and interpretation of the data. Another point to note is the implementation of the methodologies, since they can be performed online, allowing process monitoring [3][4][5][7][8][9][10][11][12][14][15][16][17][18][19], or offline once the machining of a part has been carried out [6,13,20,26]. This is an important aspect, since in methodologies that use vision systems or images, the machining process must be stopped for the acquisition of images of the tool or they may be taken after the process; however, this depends on the needs and planning of the process.…”