On-line tool wear estimation in CNC turning operations using fuzzy neural network model

In recent past, several neural network models which employ cutting forces and AErms or their derivatives for estimation as well as classification of flank wear have been developed. However, a significant variation in mean cutting forces and AErms at the start of cutting operation for similar new tools can result in estimation and classification error. In order to deal with this problem, a new on-line fuzzy neural network (FNN) model is presented in this paper. This model has four parts. The first part of the model is developed to classify tool wear by using fuzzy logic. The second part of this model is designed for normalizing the inputs for the next part. The third part consisting of modified least-square backpropagation neural network is built to estimate flank and crater wear. The development of forth part was done in order to adjust the results of the third part. Several basic and derived parameters including forces, AErms, skew and kurtosis of force bands, as well as the total energy of forces were employed as inputs in order to enhance the accuracy of tool wear prediction. The experimental results indicate that the proposed on-line FNN model has a high accuracy for estimating progressive flank and crater wear with small computational time.     

On-line monitoring of tool wear in turning using a neural network

Tool wear has long been identified as the most undesirable characteristic of the machining operations. Flank wear, in particular directly affects the workpiece dimensions and the surface quality. A reliable and sensitive technique for monitoring the tool wear without interrupting the process, is crucial in realization of the modern manufacturing concepts like unmanned machining centres, adaptive control optimization, etc. In this work an optoelectronic sensor is used in conjunction with a multilayered Neural Network for predicting the flank wear on the cutting tool. The gap sensing system consists of a bifurcated optical fibre, a laser source and a photodiode circuit. The output of the photodiode circuit is amplified and converted to the digital form using an A/D converter. The digitized sensor signal along with the cutting parameters form the inputs to a three layered, feed forward, fully connected Neural Network. The Neural Network, trained off-line using a backpropagation algorithm and the experimental data, is used to predict the flank wear. A geometrical relation is also used to correlate the flank wear on the cutting tool with the change in the workpiece dimension. The values predicted using the Neural Network and those calculated using the geometrical relation are compared with the actual values measured using a tool maker's microscope. Results showed the ability of the Neural Network to accurately predict the flank wear.     

A brief review: acoustic emission method for tool wear monitoring during turning

Research during the past several years has established the effectiveness of acoustic emission (AE)-based sensing methodologies for machine condition analysis and process monitoring. AE has been proposed and evaluated for a variety of sensing tasks as well as for use as a technique for quantitative studies of manufacturing processes. This paper reviews briefly the research on AE sensing of tool wear condition in turning. The main contents included are:

1. The AE generation in metal cutting processes, AE signal classification, and AE signal correction.

2. AE signal processing with various methodologies, including time series analysis, FFT, wavelet transform, etc.

3. Estimation of tool wear condition, including pattern classification, GMDH methodology, fuzzy classifier, neural network, and sensor and data fusion.

A review of AE-based tool wear monitoring in turning is an important step for improving and developing new tool wear monitoring methodology.     

One-step-ahead prediction of flank wear using cutting force

Tool wear prediction plays an important role in the tip geometry compensation during precision machining. The purpose of this study is to develop a reliable method to predict flank wear during a turning process. The force ratio and increment values are applied to predict one-step-ahead flank wear. The results of this paper show that using force ratios, flank wear can be predicted to within 8 and 11.9%, and also using force increment, flank wear can be predicted to within 10.3% of the actual wear for various turning conditions.     

Gradual wear monitoring of turning inserts using wavelet analysis of ultrasound waves

Discrete wavelet transforms of ultrasound waves is used to measure the gradual wear of carbide inserts during turning operations. Ultrasound waves, propagating at a nominal frequency of 10 MHz, were pulsed into the cutting tools towards the cutting edge at a burst frequency of 10 KHz. The reflected waves off the mark, nose and flank surfaces were digitized at a sampling rate of 100 MHz. Daubechies Quadrature Mirror Filter pair was used to decompose ultrasound signals into frequency packets using a tree structure.Normalized signals in each level of decomposition were used to search for a neural network architecture that correlates the ultrasound measurements to the wear level on the tool. A three-layer Multi-Layer Perceptron architecture yielded the best correlation (95.9%) using the wave packets from the fourth level of decomposition with frequencies 3.75–4.375 and 5.625–6.875 MHz.     

An experimental investigation on monitoring of crater wear in turning using ultrasonic technique

Tool life prediction and tool change strategies are now based on most conservative estimates of tool life from past tool wear data. Hence usually tools are underutilized. In an unmanned factory, this has the effect of increased frequency of the tool changes and therefore increased cost. An ultrasound online monitoring of crater wear of the uncoated carbide insert during the turning operation is presented. The method relies on inducing ultrasound waves in the tool, which are reflected by side flank surface. The amount of reflected energy is correlated with crater wear depth. Various ultrasonic parameters are considered for defining the crater wear and individual contribution of each parameter is analyzed. The ultrasonic parameters, amplitude, pulse width and root mean square (RMS) of the signal are used to quantify the crater depth and width. The power spectrum analysis of received signals shows the importance of frequency components in defining the tool wear. In the presented work, the normalizing of signals are carried out by insert hole, which is provided for clamping. This signal is not influenced by the wear but affected by other factors like tool material variation, improper couplant, temperature, etc. The response of the wear signal is normalized to the response of hole signal by mathematical division. A new approach adaptive neuro-fuzzy inference system (ANFIS) for monitoring of crater in carbide insert is presented. This improves the system accuracy and eliminates the limitation in statistical modeling that was presented in previous studies.     

Tool wear measurement in turning using force ratio

The aim of this work was to develop a reliable method to predict flank wear during the turning process. The present work developed a mathematical model for on-line monitoring of tool wear in a turning process. Force signals are highly sensitive carriers of information about the machining process and, hence, they are the best alternatives for monitoring tool wear. In the present work, determination of tool wear has been achieved by using force signals. The relationship between flank wear and the ratio of force components was established on the basis of data obtained from a series of experiments. Measurement of the ratio between the feed force and the cutting force components (F_f/F_c) has been found to provide a practical method for an in-process approach to the quantification of tool wear. A series of experiments was conducted to study the effects of tool wear as well as other cutting parameters on the cutting force signals, and to establish a relationship between the force signals, tool wear and other cutting parameters. The flank wear and the ratio of forces at different working conditions were collected experimentally to develop a mathematical model for predicting flank wear. The model was verified by comparing the experimental values with the predicted values. The relationship was then used for determination of tool flank wear.     

Role of temperature and surface finish in predicting tool wear using neural network and design of experiments

The present work focuses on the two of the techniques, namely design of experiments and the neural network for predicting tool wear. In the present work, flank wear, surface finish and cutting zone temperature were taken as response (output) variables measured during turning and cutting speed, feed and depth of cut were taken as input parameters. Predictions for all the three response variables were obtained with the help of empirical relation between different responses and input variables using design of experiments (DOE) and also through neural network (NN) program. Predicted values of the responses by both techniques, i.e. DOE and NN were compared with the experimental values and their closeness with the experimental values was determined. Relationship between the surface roughness and the flank wear and also between the temperature and the flank wear were found out for indirect measurement of the flank wear through surface roughness and cutting zone temperature.     

Prediction of flank wear by using back propagation neural network modeling when cutting hardened H-13 steel with chamfered and honed CBN tools

Productivity and quality in the finish turning of hardened steels can be improved by utilizing predicted performance of the cutting tools. This paper combines predictive machining approach with neural network modeling of tool flank wear in order to estimate performance of chamfered and honed Cubic Boron Nitride (CBN) tools for a variety of cutting conditions. Experimental work has been performed in orthogonal cutting of hardened H-13 type tool steel using CBN tools. At the selected cutting conditions the forces have been measured using a piezoelectric dynamometer and data acquisition system. Simultaneously flank wear at the cutting edge has been monitored by using a tool makers microscope. The experimental force and wear data were utilized to train the developed simulation environment based on back propagation neural network modeling. A trained neural network system was used in predicting flank wear for various different cutting conditions. The developed prediction system was found to be capable of accurate tool wear classification for the range it had been trained.     

Monitoring of flank wear of coated tools in high speed machining with a neural network ART2

A monitoring system for classifying the levels of the tool flank wear of coated tools into some categories has been developed using an unsupervised and self-organizing artificial neural network, ART2. The input pattern used for the ART2 was an array of normalized mean wavelet coefficients of the feed force, which was affected by not only the flank wear but also the severe crater wear observed in high speed machining. The outputs of ART2 were classified into four or five categories of wear levels: the incipient stage, one or two intermediate stages, final stage and hazardous stage. For two apparently different series of input data obtained under the same cutting conditions, which are often experienced in the experiment, the ART2 neural network showed very similar classification of tool wear levels from the beginning to the end of cutting. Further study proved that this monitoring system detected the excessive wear in the hazardous stage for different cutting speeds 5–7 m/s and different feed rates 0.10–0.20 mm/rev.     

Real time implementation of on-line tool condition monitoring in turning

This paper describes a real-time tool condition monitoring system for turning operations. The system uses a combination of static and dynamic neural networks with off-line and on-line training and cutting force components are used as diagnostic signals. The system is capable of monitoring several wear components simultaneously. The wear estimation system has been implemented experimentally to evaluate its suitability for use in shop floor conditions. The tests were performed in real time with different cutting conditions. The experimental results showed that the system was successful in predicting three wear components in real time. However, the accuracy of the wear prediction was not the same for all three wear components. The crater wear predictions were less accurate partly because of the opposing effects of crater and flank wear components on cutting force components.     

Tool wear detection with fuzzy classification and wavelet fuzzy neural network

In the paper, a new method of tool wear detection with cutting conditions and detected signals is presented, which includes the model of wavelet fuzzy neural network with acoustic emission (AE) and the model of fuzzy classification with motor current. The results of tool wear estimated by cutting conditions and detected signals (spindle motor current, feed motor current and AE) are fused by fuzzy inference. Experimental results show that the method of tool wear detection is reliable and practical.     

Predictive modeling of surface roughness and tool wear in hard turning using regression and neural networks

In machining of parts, surface quality is one of the most specified customer requirements. Major indication of surface quality on machined parts is surface roughness. Finish hard turning using Cubic Boron Nitride (CBN) tools allows manufacturers to simplify their processes and still achieve the desired surface roughness. There are various machining parameters have an effect on the surface roughness, but those effects have not been adequately quantified. In order for manufacturers to maximize their gains from utilizing finish hard turning, accurate predictive models for surface roughness and tool wear must be constructed. This paper utilizes neural network modeling to predict surface roughness and tool flank wear over the machining time for variety of cutting conditions in finish hard turning. Regression models are also developed in order to capture process specific parameters. A set of sparse experimental data for finish turning of hardened AISI 52100 steel obtained from literature and the experimental data obtained from performed experiments in finish turning of hardened AISI H-13 steel have been utilized. The data sets from measured surface roughness and tool flank wear were employed to train the neural network models. Trained neural network models were used in predicting surface roughness and tool flank wear for other cutting conditions. A comparison of neural network models with regression models is also carried out. Predictive neural network models are found to be capable of better predictions for surface roughness and tool flank wear within the range that they had been trained.Predictive neural network modeling is also extended to predict tool wear and surface roughness patterns seen in finish hard turning processes. Decrease in the feed rate resulted in better surface roughness but slightly faster tool wear development, and increasing cutting speed resulted in significant increase in tool wear development but resulted in better surface roughness. Increase in the workpiece hardness resulted in better surface roughness but higher tool wear. Overall, CBN inserts with honed edge geometry performed better both in terms of surface roughness and tool wear development.     

Analytical model for tool wear monitoring in turning operations using ultrasound waves

This paper presents an analytical model to monitor the gradual wear of cutting tools, on-line, during turning operations using ultrasound waves. Ultrasound waves at a frequency of 10 MHz were pulsed continuously inside several cutting tools, towards their cutting edge. The change in tool geometry, due to gradual wear, has been related, in a mathematical form, to the change in the acoustic behavior of ultrasound waves inside the body of the cutting tools. Physical laws governing the propagation and reflection of ultrasound waves along with geometrical analysis of the wear area were used in deriving the mathematical model. The experimental setup and model evaluation is based on a previously published research work by the author, which presented an empirical model showing a corresponding change in the ultrasound behavior with tool gradual wear. The current work emphasizes the previous findings and presents the relation between the acoustic behavior of ultrasound waves and the progressive tool gradual wear in a mathematical form that can be easily used in machine control operations.     

Topography of the flank wear surface

In many applications, topography represents the main external features of a surface. This paper describes the topography of the flank wear surface and also presents the relationship between the maximum flank wear and the topography parameters (roughness parameters) of the flank wear surface during the turning operation. A modern CNC lathe machine (Okuma LH35-N) was used for the machine turning operation. Three-dimensional surface roughness parameters of the flank wear surface were measured by a surface texture instrument (from Talysurf series) using surface topography software (Talymap). Based on the resulting experimental data, it is found that as the flank wear increases, the roughness parameters (sR_a, sR_q, and sR_t) on the flank surface increase significantly. The greater the roughness value of the flank wear surface, the higher the friction of the tool on the workpiece and the greater the heat generation that will occur, thus ultimately causing tool failure. On the other hand, positive skewness (sR_sk) indicates the presence of a small number of spikes on the flank surface of the cutting tool, which could quickly wear off during the machining process.     

Data fusion neural network for tool condition monitoring in CNC milling machining

Several data fusion methods are addressed in this research to integrate the detected data for the neural network applications of on-line monitoring of the tool condition in CNC milling machining. One dynamometer and one accelerometer were used in the experiments. The collected signals were pre-processed to extract the feature elements for the purpose of effectively monitoring the tool wear condition. Different data fusion methods were adopted to integrate the obtained feature elements before they were applied into the learning procedure of the neural networks. The training-efficiency and test-performance of the data fusion methods were then analyzed. The convergence speed and the test error were recorded and used to represent the training efficiency and test performance of the different data fusion methods. From an analysis of the results of the calculations based on the experimental data, it was found that the performance of the monitoring system could be significantly improved with suitable selection of the data fusion method.     

A computer algorithm for flank and crater wear estimation in CNC turning operations

In order to increase the productivity of turning processes, several attempts have been made in the recent past for tool wear estimation and classification in turning operations. The tool flank and crater wear can be predicted by a number of models including statistical, pattern recognition, quantitative and neural network models. In this paper, a computer algorithm of new quantitative models for flank and crater wear estimation is presented. First, a quantitative model based on a correlation between increases in feed and radial forces and the average width of flank wear is developed. Then another model which relates acoustic emission (AE_rms) in the turning operation with the flank and crater wear developed on the tool is presented. The flank wear estimated by the first model is then employed in the second model to predict the crater wear on the tool insert. The influence of flank and crater wear on AE_rms generated during the turning operation has also been investigated. Additionally, chip-flow direction and tool–chip rake face interfacing area are also examined. The experimental results indicate that the computer program developed, based on the algorithm mentioned above, has a high accuracy for estimation of tool flank wear.     

Modeling of tool wear in drilling by statistical analysis and artificial neural network

The useful life of a cutting tool and its operating conditions largely control the economics of the machining operations. Hence, it is imperative that the condition of the cutting tool, particularly some indication as to when it requires changing, to be monitored. The drilling operation is frequently used as a preliminary step for many operations like boring, reaming and tapping, however, the operation itself is complex and demanding.

Back propagation neural networks were used for detection of drill wear. The neural network consisted of three layers input, hidden and output. Drill size, feed, spindle speed, torque, machining time and thrust force are given as inputs to the ANN and the flank wear was estimated. Drilling experiments with 8 mm drill size were performed by changing the cutting speed and feed at two different levels. The number of neurons in the hidden layer were selected from 1, 2, 3, …, 20. The learning rate was selected as 0.01 and no smoothing factor was used. The estimated values of tool wear were obtained by statistical analysis and by various neural network structures. Comparative analysis has been done between statistical analysis, neural network structures and the actual values of tool wear obtained by experimentation.     

Tool wear estimation using resource allocation network

This paper presents tool wear estimation in face milling operations using the resource allocation network (RAN). Acoustic emission (AE) signals, surface roughness parameters and cutting conditions (cutting speed, feed) have been used to formulate input patterns. The performance of RAN has been compared with the multi-layer perceptron (MLP) trained using back-propagation (BP) algorithm, and the results are presented.