Analysis of the structure of vibration signals for tool wear detection

Sánchez, Francisco Javier Alonso; Salgado, David Rodríguez

doi:10.1016/j.ymssp.2007.09.012

Cited by 172 publications

(62 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…It appeared that only the RMS and variance of the detrended signals showed a monotonic behavior with tool wear, which meant that the information in the vibration signals about flank wear was mostly contained in the high-frequency components. Later, they extended their technique by applying cluster analysis to group the SSA decomposition of the vibration signals [101]. This time, they found that only the RMS and standard deviation of the medium and high frequency signals of the longitudinal vibration and the RMS and standard deviation of the high-frequency components of the transverse vibration showed a monotonic behavior with tool wear.…”

Section: Signal Pre-processingmentioning

confidence: 99%

Advanced monitoring of machining operations

et al. 2010

View full text Add to dashboard Cite

Section: Signal Pre-processingmentioning

confidence: 99%

Advanced monitoring of machining operations

et al. 2010

View full text Add to dashboard Cite

“…Of the 81 cutting conditions, 57 data points (cutting conditions) are used for training, 12 cutting conditions are allotted for validation, and another 12 data points (cutting conditions) for testing. After training the data using LM back propagation [21][22][23] algorithm, based upon the minimum value of MSE, the optimum network is arrived at. The different trials carried out and the MSE values are shown in Table 3.…”

Section: Artificial Neural Network Based Tool Wear Estimation On Dry mentioning

confidence: 99%

“…Since the relationship between features from sensors and tool wear are nonlinear, the regression equation may not hold well. The artificial neural networks (ANN) using a mapping technique between the input and output are extensively employed [21][22][23] whenever the relation is nonlinear. The selection of input parameters, hidden layer, and inner error depend upon the cutting process in ANN.…”

mentioning

confidence: 99%

Artificial neural network based tool wear estimation on dry hard turning processes of AISI4140 steel using coated carbide tool

Rajeev¹,

Dinakaran²,

Singh³

2017

Bulletin of the Polish Academy of Sciences Technical Sciences

View full text Add to dashboard Cite

Abstract. Nowadays, finishing operation in hardened steel parts which have wide industrial applications is done by hard turning. Cubic boron nitride (CBN) inserts, which are expensive, are used for hard turning. The cheaper coated carbide tool is seen as a substitute for CBN inserts in the hardness range (45-55 HRC). However, tool wear in a coated carbide tool during hard turning is a significant factor that influences the tolerance of machined surface. An online tool wear estimation system is essential for maintaining the surface quality and minimizing the manufacturing cost. In this investigation, the cutting tool wear estimation using artificial neural network (ANN) is proposed. AISI4140 steel hardened to 47 HRC is used as a work piece and a coated carbide tool is the cutting tool. Experimentation is based on full factorial design (FFD) as per design of experiments. The variations in cutting forces and vibrations are measured during the experimentation. Based on the process parameters and measured parameters an ANN-based tool wear estimator is developed. The wear outputs from the ANN model are then tested. It was observed that as the model using ANN provided quite satisfactory results, and that it can be used for online tool wear estimation.Key words: hard turning, coated carbide, cutting force, vibration, ANN.Artificial neural network based tool wear estimation on dry hard turning processes of AISI4140 steel using coated carbide tool cutting forces, tool dynamometers are widely used. Another prominent feature used in TCM is the vibration signal. The amplitude of the vibration signal in the dynamic frequency band of the tool holder's natural frequency along the z direction is more profound to wear, and it can be considered a feature for TCM [15][16][17].Since tool wear is a complex phenomenon, the signal information from a single sensor is inadequate to predict the wear accurately. Hence, it is advisable to employ multiple sensors. The highlight of the multi-sensor system is the abundance of information available, which can be used for decision making. Many researchers have used a combination of force and vibration signals, as well as acoustic emission signals, to monitor tool wear and roughness [18][19][20].The estimation of tool wear from the sensor signals is performed by developing a mathematical model from the experimental data, referred to as the regression equation. Since the relationship between features from sensors and tool wear are nonlinear, the regression equation may not hold well. The artificial neural networks (ANN) using a mapping technique between the input and output are extensively employed [21][22][23] whenever the relation is nonlinear. The selection of input parameters, hidden layer, and inner error depend upon the cutting process in ANN.The main research factor in hard turning is the estimation wear in coated carbide cutting tool which can be used as a replacement for expensive CBN tools. Even though many investigations have been carried out on wear estimation and TCM, the research wor...

show abstract

“…The amplitude of vibration components decreases with the increase of cutting speed, and increases with the increase of feed rate and depth of cut. Alonso, F. J. and Salgado, D. R. [4], developed a reliable tool condition monitoring system (TCMS) for industrial application. They employed singular spectrum analysis (SSA) and cluster analysis for analysis of the tool vibration signals.SSA was non-parametric technique, of time series analysis that decomposes the acquired tool vibration signals and Cluster analysis was used to group the SSA decomposition in order to obtain several independent components in the frequency domain and that are apply to feed forward back-propagation (FFBP) neural network to determine the tool flank wear.…”

Section: Introductionmentioning

confidence: 99%

Experimental Investigation to Minimize Resultant Vibration Signal in CNC Turing Operation of Hard AISI M2 Tool Steel

Pawar¹

2017

IJRASET

View full text Add to dashboard Cite

Analysis of the structure of vibration signals for tool wear detection

Cited by 172 publications

References 33 publications

Advanced monitoring of machining operations

Advanced monitoring of machining operations

Artificial neural network based tool wear estimation on dry hard turning processes of AISI4140 steel using coated carbide tool

Experimental Investigation to Minimize Resultant Vibration Signal in CNC Turing Operation of Hard AISI M2 Tool Steel

Contact Info

Product

Resources

About