Tool Wear Experiments Research Articles

Tool remaining useful life prediction using deep transfer reinforcement learning based on long short-term memory networks

Tool wear and faults will affect the quality of machined workpiece and damage the continuity of manufacturing. The accurate prediction of remaining useful life (RUL) is significant to guarantee the processing quality and improve the productivity of automatic system. At present, the most commonly used methods for tool RUL prediction are trained by history fault data. However, when researching on new types of tools or processing high value parts, fault datasets are difficult to acquire, which leads to RUL prediction a challenge under limited fault data. To overcome the shortcomings of above prediction methods, a deep transfer reinforcement learning (DTRL) network based on long short-term memory (LSTM) network is presented in this paper. Local features are extracted from consecutive sensor data to track the tool states, and the trained network size can be dynamically adjusted by controlling time sequence length. Then in DTRL network, LSTM network is employed to construct the value function approximation for smoothly processing temporal information and mining long-term dependencies. On this basis, a novel strategy of Q-function update and transfer is presented to transfer the deep reinforcement learning (DRL) network trained by historical fault data to a new tool for RUL prediction. Finally, tool wear experiments are performed to validate effectiveness of the DTRL model. The prediction results demonstrate that the proposed method has high accuracy and generalization for similar tools and cutting conditions.

Tool wear analysis of cutting Ti-5553 with uncoated carbide tool under liquid nitrogen cooling condition using tool wear maps

Tool wear experiments involving Ti-5553, a newly developed titanium superalloy, were carried through cryogenic machining with liquid nitrogen. Tool wear modes under various cutting parameters were observed and recorded, and the tool wear mechanism was analyzed. The tool wear map for the rake face was constructed to analyze tool wear qualitatively. Using the experimental results, the tool wear map for the flank face was developed to analyze tool wear quantitatively using the three-dimensional isosurface and response surface methodologies. This new method of combining the tool wear maps of the rake and flank faces provides a quick and useful reference for selecting suitable cutting parameters.

New Tool Wear Estimation Method of the Milling Process Based on Multisensor Blind Source Separation

Timely and effective identification and monitoring of tool wear is important for the milling process. However, traditional methods of tool wear estimation have run into difficulties due to under small samples with less prior knowledge. This article addresses this issue by employing a multisensor tool wear estimation method based on blind source separation technology. Stationary subspace analysis (SSA) technology is applied to transform multisensor signals to stationary and nonstationary sources without prior information of signals. Ten dimensionless time-frequency indices of the nonstationary signal are extracted to train least squares support vector regression (LS-SVR) to obtain a tool wear estimation model for small samples. The analysis and comparison of one benchmark tool wear dataset and tool wear experiments verify the feasibility and effectiveness of the proposed method and outperform other two current methods.

A GAPSO-Enhanced Extreme Learning Machine Method for Tool Wear Estimation in Milling Processes Based on Vibration Signals

Tools are the most vulnerable components in milling processes conducted using numerical control milling machines, and their wear condition directly influences work-product quality and operational safety. As such, tool wear estimation is an essential component of NC milling operations. This study addresses this issue by proposing an extreme learning machine (ELM) method enhanced by a hybrid genetic algorithm and particle swarm optimization (GAPSO) approach for conducting tool wear estimation based on workpiece vibration signals. Here, a few feature parameters in the time, frequency, and time–frequency (Ensemble empirical mode decomposition, EEMD) domains of the workpiece vibration signals are extracted as the input of the ELM model. Then, the initialized weights and thresholds of the ELM model are optimized based on the GAPSO approach with training dataset. Finally, tool wear is estimated using the optimized ELM model with testing dataset. The effectiveness of the proposed method is verified by its application to vibration signals collected from two milling tool wear experiments (an open-access benchmark dataset and a milling tool wear experiment) by comparison to the ELM, GA-ELM, and PSO-ELM methods. The results indicate that the estimation accuracy and optimization efficiency of the proposed method outperforms that of other three methods.

Physics-guided logistic classification for tool life modeling and process parameter optimization in machining

This paper describes a physics-guided logistic classification method for tool life modeling and process parameter optimization in machining. Tool life is modeled using a classification method since the exact tool life cannot be measured in a typical production environment where tool wear can only be directly measured when the tool is replaced. In this study, laboratory tool wear experiments are used to simulate tool wear data normally collected during part production. Two states are defined: tool not worn (class 0) and tool worn (class 1). The non-linear reduction in tool life with cutting speed is modeled by applying a logarithmic transformation to the inputs for the logistic classification model. A method for interpretability of the logistic model coefficients is provided by comparison with the empirical Taylor tool life model. The method is validated using tool wear experiments for milling. Results show that the physics-guided logistic classification method can predict tool life using limited datasets. A method for pre-process optimization of machining parameters using a probabilistic machining cost model is presented. The proposed method offers a robust and practical approach to tool life modeling and process parameter optimization in a production environment.

Experimental Study on the Influence of Tool Wear on the Cutting Process of Ti6Al4V

The tool wear is a big problem when machining titanium alloy. If the tool wear law is not mastered well, the machining quality will not be guaranteed. In this paper, the theoretical model of the tool is established, which can understand the spatial dimension model of the tool. At the same time, the tool wear experiment of the most commonly used tool for titanium alloy machining in the manufacturing industry is carried out, and the influence of the tool cutting wear on the cutting process is studied. The experimental analysis shows that the tool produces the chip extrusion stripe when it is worn in the initial stage and normal stage, which is perpendicular to the chip in the outflow direction, the fringe protrusions are parallel, and the contact surface of the chips is relatively flat. When the tool reaches the stage of severe wear, the protrusion stripes perpendicular to the outflow direction of the chips are fluffy, and the stripes are twisted. With the continuous cutting wear time of the tool becoming longer, the cutting force, the residual stress of the machined surface layer and the cutting temperature obtained in each stage gradually increase; with the tool reaching the stage of severe wear, the protrusion stripes are fluffy, and the stripes are twisted The tool wear time gradually increases, and the cutting surface roughness obtained by continuous tool wear decreases first and then increases; the surface quality of the tool in the initial wear stage is relatively obvious and the edge burr is large, while in the normal wear stage, the surface quality of the tool is gradually gentle, but when the tool reaches the severe wear stage, the surface quality of the tool begins to be processed Sharp drop and thick edge burr.

Theoretical study on frequency spectrum characteristics of surface profiles generated in micro-end-milling process

Micro-milling is a flexible and promising fabrication process. Based on the analysis of the surface formation process in micro-end milling, the theoretical surface model was established to investigate the influence of seven identified factors, i.e., minimum cut thickness (MCT), elastic recovery, plastic side flow, radial and axial runout, tool wear (corner radius and unbalanced wear coefficient), and vibration on the frequency spectrum characteristics. The increase of MCT coefficient and plastic side flow height could increase the amplitude of the sft and 2sft frequency, while the scenario for the elastic recovery coefficient witnessed a reverse trend. The amplitude of the 0.5sft frequency decreased with the increase of the radial runout offset, while the amplitude of sft frequency increased as the radial runout offset increased; the scenario for the influence of the radial runout angle was opposite. The increase of the axial runout offset could lead to the increase of the amplitude of both 0.5sft and sft frequency. The increase of corner radius could lead to a slight reduction of the amplitude of the 0.5sft and sft frequency. The occurrence of the vibration with high frequency leaded to a low spatial frequency. The corresponding tool wear experiment was conducted, and the frequency-spectrum characteristics of the micro-milled surface were compared with the simulated results to validate the established profile models. The paper successfully correlated the relative importance of the identified individual factors to the corresponding frequency spectrum characteristics, and it had potential applications in predicting and controlling the spatial content of machined surfaces.

Logistic classification for tool life modeling in machining

This paper describes the application of logistic classification for tool life modeling and prediction in an industrial setting using shop floor data. Tool life is treated as a classification problem since tool wear can only be measured at the time of tool replacement in a production environment. Laboratory tool wear experiments are used to simulate shop floor wear data by two states: not worn (class 0); and worn (class 1). To incorporate non-linearity in logistic classification, a log-transformation of input features is performed. The logistic classification approach, results, and interpretability of the logistic model are presented.

Tool wear mechanism and prediction in milling TC18 titanium alloy using deep learning

Rapid tool wear from milling TC18 (Ti-5Al-5Mo-5V-1Cr-1Fe) leads to increased surface deterioration and manufacturing costs. Here, a real-life tool wear experiment was introduced, and the three stages of tool wear were analyzed in detail according to the tool wear micro-topography and chemical elements. In the initial and normal stage, the tool wear was slow because of the protection of the adhesive titanium layer and dense alumina film. Diffusion wear and oxidation wear occurred until the sever wear stage. Based on the above wear mechanism determination, after acquiring the real time cutting force, the tool wear prediction models were established using a convolutional bi-directional long short-term memory networks (CNN + BILSTM) and a convolutional bi-directional gated recurrent unit (CNN + BIGRU). The results show that the errors of the predicted minimum values are all within 8%, demonstrating that the deep learning method offers a new and promising approach for in monitoring tool wear on-line.

Effect of machinability, microstructure and hardness of deep cryogenic treatment in hard turning of AISI D2 steel with ceramic cutting

This study examined the hard turning of AISI D2 cold work tool steel subjected to deep cryogenic processing and tempering and investigated the effects on surface roughness and tool wear. In addition, the effects of the deep cryogenic processes on mechanical properties (macro and micro hardness) and microstructure were investigated. Three groups of test samples were evaluated: conventional heat treatment (CHT), deep cryogenic treatment (DCT-36) and deep cryogenic treatment with tempering (DCTT-36). The samples in the first group were subjected to only CHT to 62 HRc hardness. The second group (DCT-36) underwent processing for 36h at −145°C after conventional heat treatment. The latter group (DCTT-36) had been subjected to both conventional heat treatment and deep cryogenic treatment followed by 2h of tempering at 200°C. In the experiments, Al2O3 + TiC matrix-based untreated mixed alumina ceramic (AB30) and Al2O3 + TiC matrix-based TiN-coated ceramic (AB2010) cutting tools were used. The artificial intelligence method known as artificial neural networks (ANNs) was used to estimate the surface roughness based on cutting speed, cutting tool, workpiece, depth of cut and feed rate. For the artificial neural network modeling, the standard back-propagation algorithm was found to be the optimum choice for training the model. Three different cutting speeds (50, 100 and 150m/min), three different feed rates (0.08, 0.16 and 0.24mm/rev) and three different cutting depths (0.25, 0.50 and 0.75mm) were selected. Tool wear experiments were carried out at a cutting speed of 150m/min, a feed rate of 0.08mm/rev and a cutting depth of 0.6mm. As a result of the experiments, the best results for both surface roughness and tool wear were obtained with the DCTT-36 sample. When cutting tools were compared, the best results for surface roughness and tool wear were obtained with the coated ceramic tool (AB2010). The macroscopic and micro hardness values were highest for the DCT-36. From the microstructural point of view, the DCTT-36 sample showed the best results with homogeneous and thinner secondary carbide formations.

Influence of vibration in cutting on tool flank wear: Fundamental study by conducting a cutting experiment with forced vibration in the depth-of-cut direction

In this research, we developed special vibration tools, in which a simple harmonic vibration is generated with controlled frequency and amplitude, and experimentally clarified the influence of vibration in cutting on the tool flank wear. The results of the tool wear experiment show that the wear increasing rate (increment in the flank wear width with respect to the cutting length) increases depending on the vibration speed, whereas wear is not affected by the frequency if the vibration speed is the same. Moreover, the observations reveal that the tool flank face interferes with the workpiece as the vibration speed increases, thereby increasing the abrasive wear.

Using the temperature method for the prediction of tool life in sustainable production

The tool workpiece-type thermocouple is practical, inexpensive and provides an indication of temperature trends in the tool-chip-workpiece interface. These temperatures also correlate with tool wear. Using this sensor system, a novel method for predicting functional lifetime in tool parts has been developed. In this paper, an extended Taylor's functional relationship was determined based on measurements of the tool work temperature. This method allows prediction of tool life based on a tool-work thermocouple temperature with a few traditional tool-wear experiments. To confirm the reliability of the proposed temperature-based method of tool-life prediction, the traditional tool-life testing methods are applied using standard experimental procedures for exploring tool-to-surface responses.

Study on the tool wear and its effect of PCD tool in micro milling of tungsten carbide

Tungsten carbide is the most promising material to produce glass molds due its unique properties. This paper conducts an tool wear experiment on PCD micro end mill in micro milling of tungsten carbide. The tool wear characteristics are observed and tool wear mechanisms are analyzed. The results show that PCD tool wear is concentrated on the tool tip, and a triangular wear belt is formed on the tool bottom surface. The involved tool wear mechanisms mainly include the adhesive wear, micro chipping and abrasive wear. After PCD tool is worn, the cutting forces is increasing, some tungsten carbide is removed with brittle fracture, then micro defects are generated on the milled surface. Tool tip shape of the worn PCD tool is copied to the milled groove and makes it a reverse cone-shape.

High-Efficiency Milling of Steam Turbine Blade

Conventional methods often use ball end mills with a small diameter to finish machining of a steam turbine blade to satisfy accuracy requirements by using a small pick feed value. Thus, the cutting length increases, resulting in increased wear and a lower milling efficiency. Therefore, a new method using a tilt-taper end mill is proposed. This paper presents the validity of the proposed method used for milling planes by comparing the ball and square end mills through tool wear experiments. Factors including removal degree, surface roughness, tool wear, and machined surfaces are investigated with respect to the plane model. The experimental results show that tilt end mill can retard the tool wear remarkably to obtain a steady surface profile, and the maximum surface roughness value, using the tilt-taper end mill, is less than 6μm until process completion.

Machinery Degradation Inspection and Maintenance Using a Cost-Optimal Non-Fixed Periodic Strategy

Optimization of inspection and maintenance strategies based on machinery degradation could improve the operation safety and manufacturing efficiency. In this paper, a novel non-fixed periodic inspection strategy will be proposed and applied in precognitive maintenance to monitor and repair a multistage degradation process. The overall cost function of each degradation stage can be individually constructed based on its existing cost factors, and optimal solutions of the above-applied inspection intervals will be obtained by minimizing the corresponding cost function. Due to the high breakdown risk in the final stage, a warning threshold is tailored to divide it into two parts, where two additional inspection intervals are applied, accordingly. The effectiveness of our cost-optimal non-fixed periodic inspection strategy is testified for the inspection and maintenance of an industrial tool wear experiment in a high-speed milling machine, which can save 37% of the total maintenance cost without accuracy loss of the wear estimation and prediction, compared with the conventional periodic inspection strategy.

Effects of tool axis offset in turn-milling process

The orthogonal turn-milling process draws increasing attention parallel to the developments in multi-tasking machine tools. However, the effects of process parameters used in orthogonal turn-milling are not fully understood. The basic objective of this study is to describe one of the most important and unique cutting parameters which is the tool axis offset, and determine its effects on process outputs such as cutting force, surface quality and tool wear. In order to achieve this objective, the effect of the tool axis offset is defined based upon geometrical relationships which are used to develop a mechanistic force model. Surface quality including the tool axis offset effect is examined by both analytical and experimental methods. Additionally, tool wear experiments were conducted to obtain the tool life under different settings for the tool axis offset. It is found that the process outcomes such as cutting forces, surface quality and tool life can be improved by taking into account the tool axis offset in orthogonal turn-milling operation. As a valuable conclusion, the proposed models and experimental results can be used to select a proper tool axis offset value while performing orthogonal turn-milling process in order to improve the process outcome.

Gamma process with recursive MLE for wear PDF prediction in precognitive maintenance under aperiodic monitoring

Effective degradation prediction under aperiodic monitoring can assist to save inspection costs and schedule maintenance for the realistic manufacturing systems. In this paper, an extended time-/condition-based framework will be proposed to predict the Probability Density Function (PDF) of unobservable degradation under aperiodic condition monitoring. Furthering our earlier work of indirect degradation estimation via a mixed time-/condition-based approach, a stage-based Gamma process is designed and implemented for the prediction of degradation PDF where a recursive Maximum Likelihood Estimation (MLE) algorithm deduced from the conventional MLE algorithm is able to update the modeling parameters based on each aperiodic monitoring interval. The effectiveness of our extended framework is tested on the tool wear experiments in an industrial high speed computer numerical control milling machine, which can achieve an acceptable prediction performance with the average error of predicted bounds less than 15.0% as well as the average accuracy more than 94.3%.

Milling Force Modeling of Worn Tool and Tool Flank Wear Recognition in End Milling

The wear state of a cutting tool is an important factor which affects machining quality. Therefore, monitoring tool wear is extremely essential to ensure workpiece quality and improve tool life. This paper models the milling forces of a worn tool and proposes a recognition method of milling tool wear state based on the influence relationships between the milling force features and tool wear. In the milling force model, the friction effect force and the shearing force are treated separately, and the friction stress distribution on tool flank is described. Then the force model is calibrated and verified through experiments. In the tool wear recognition method, the relationship between the milling force feature vector and tool wear is investigated. On this basis, the tool flank wear recognition method is proposed. A tool wear experiment is performed using superalloy material. In the experiment, the recognition results are expressed in confidence intervals which can represent the recognized tool wear more effectively and accurately. Finally, the scheme of tool flank wear online monitoring is proposed.

Study on Tool Wear in Green Milling Ni-Based Superalloy

In order to solve the difficult machining problem of Ni-based superalloy, the water vapor and ionized air was applied as coolant and lubricant in milling process. The tool wear experiments on water vapor, ionized air, wet and dry cutting with carbide tool YG6A machining Ni-based superalloy GH4169 was carried out. The result showed that tool wear was quickly on milling GH4169 with carbide tool YG6A, and increased with the cutting speed.The flank wear land was uneven and the boundary wear was serious. During the ionized air condition the flank wear was more slowly than wet and dry cutting condition, the values of flank wear were reduced about 15 and 10 percent respectively. In addition, the result showed that The milling force and machined surface roughness increased with the flank wear.

Cost optimization and experimental design in milling using surrogate models and value of information

Tool wear experiments are necessary for tool life model development and cost optimization. However, tool wear experiments are costly and time-consuming, thereby limiting the number of tool life evaluations as a function of machining parameters. This paper describes an approach for experimental design for tool life testing and machining cost optimization in milling using surrogate modeling and the value of information method. The machining cost in milling was modeled using a kriging surrogate. The machining cost function was evaluated using experimental tool life results and was considered to be the “true” function. The value of information, expressed as expected reduction in machining cost optimum after experiment, was used to identify optimal test parameters for tool life experiments. The procedure to calculate value of information is demonstrated using a one-dimensional cost function. Results show a convergence to the true optimum in two and three dimensions. The value of information criterion balances local search by exploiting where the cost function is minimum and global search by exploring where the prediction uncertainty is high. The method performs better than the traditional statistical design of experiments such as Taguchi orthogonal arrays, central composite design, and factorial designs, especially in three or more dimensions.