Wear Prediction Research Articles

Machine Learning based Tool Wear Prediction from Variability of Acoustic Sound Emission Signals

Machine Learning based Tool Wear Prediction from Variability of Acoustic Sound Emission Signals

Wavelet Integrated CNN With Dynamic Frequency Aggregation for High-Speed Train Wheel Wear Prediction

Wavelet Integrated CNN With Dynamic Frequency Aggregation for High-Speed Train Wheel Wear Prediction

<b>Cage Wear Prediction for Traction Motor Bearings of Railway Vehicles Based on Measurement of Contact Force between Rolling Element and Cage</b>

<b>Cage Wear Prediction for Traction Motor Bearings of Railway Vehicles Based on Measurement of Contact Force between Rolling Element and Cage</b>

Physics-Informed Neural Networks With Weighted Losses by Uncertainty Evaluation for Accurate and Stable Prediction of Manufacturing Systems.

The state prediction of key components in manufacturing systems tends to be risk-sensitive tasks, where prediction accuracy and stability are the two key indicators. The physics-informed neural networks (PINNs), which integrate the advantages of both data-driven models and physics models, are deemed as an effective approach and research trends for stable prediction; however, the potential advantages of PINN are limited for the situations with inaccurate physics models or noisy data, where the balancing of the weights of the data-driven model and physics model is very important for improving the performance of PINN, and it is also a challenge urgently to be addressed. This article proposed a kind of PINN with weighted losses (PNNN-WLs) by uncertainty evaluation for accurate and stable prediction of manufacturing systems, where a novel weight allocation strategy based on uncertainty evaluation by quantifying the variance of prediction errors is proposed, and an improved PINN framework is established for accurate and stable prediction. The proposed approach is verified with open datasets on tool wear prediction, and experimental results show that the prediction accuracy and stability could be obviously improved over existing methods.

Wear prediction study considering TBM hob slip

Aiming at the problem of serious tool wear and untimely tool change affecting the construction efficiency when TBM is digging in microfractured tuff strata, the hob wear prediction study is carried out based on engineering examples. By analysing the rock-breaking mechanism of hob wear, a linear wear rate model and a life prediction model considering hob slip are derived based on the CSM model and the abrasive wear theory; the wear rate and the wear amount at the tool change point are further calculated by the prediction model in comparison with the measured value, and the furthest digging distance of the hob is calculated by the life prediction model. The results show that: the relative error between the measured average wear rate and the calculated linear wear rate considering the hob slip is 5.3%; the life prediction model predicts that the longest digging distance of the positive hob is 857 rings and the shortest distance is 198 rings. The results of the study are of guiding significance for the prediction of tool wear and the selection of the timing of opening and changing the hob in TBM tunneling.

Tool wear prediction based on kernel principal component analysis and least square support vector machine

Abstract To accurately predict the amount of tool wear in the machining process, a monitoring model of tool wear based on multi-sensor information feature fusion is proposed. First, by collecting the cutting force, vibration, and acoustic emission signals of the tool during the whole life cycle, the multi-domain characteristics of the signal are extracted; then, kernel principal component analysis is used to reduce the dimensionality of the extracted data, and the principal components whose cumulative contribution ratio exceeds 85% are obtained. The redundant features with little correlation with tool wear were removed from the feature vectors to generate the fusion features. Finally, the fusion features are input into the least squares support vector machine model optimized by particle swarm algorithm for regression prediction of tool wear. The non-linear mapping relationship between the physical signal and the tool wear is discovered, which effectively realizes the prediction of the tool wear. Compared with the existing tool wear prediction methods, the method proposed has higher prediction accuracy.

Real-Time Prediction of Disc Cutter Wear in Low-Abrasive Rocks: Integrating Physico-Mechanical Properties and Signal Processing Features Through Machine Learning Methods

Real-Time Prediction of Disc Cutter Wear in Low-Abrasive Rocks: Integrating Physico-Mechanical Properties and Signal Processing Features Through Machine Learning Methods

Research on dynamic and wear prediction models of compound planetary transmission systems

Under high speeds and heavy loads, gear tooth surfaces will be worn, which will lead to accidents. The current wear prediction model is not comprehensive, but the wear prediction model that considers the temperature field and dynamic response can provide more accurate gear wear fault early warning than previous methods. Therefore, according to the Archard model, the discrete wear prediction model of the tooth surface is deduced in this paper, and the wear evolution trend of the tooth surface is predicted and analyzed in each operation stage. The tooth surface temperature field was simulated by the finite-element method, and the tooth surface wear prediction model was improved considering its influence on materials. In addition, the dynamic model of the compound planetary transmission system was established, and the wear prediction model was improved again by using its nonlinear meshing force. Finally, the tooth wear prediction model under the interaction of the dynamic response and temperature field was obtained. A wear measurement method based on duplicate adhesive film was proposed. The wear evolution trend along the tooth length direction and the tooth width direction were measured, and the accuracy of the wear prediction model was verified. The results show that the wear depth along the tooth width direction is larger in the middle part and smaller on both sides under the influence of temperature. After the interference of the dynamic response, the wear depth along the tooth rectangle fluctuates, which is very similar to the results of the theoretical prediction model. It provides a theoretical basis for the early warning of gear health status in practical engineering.

Data-driven prediction of tool wear using Bayesian regularized artificial neural networks

The prediction of wear in cutting tools is pivotal for boosting productivity and reducing manufacturing costs. Although current data-driven models in machine learning and deep learning have advanced predictive capabilities, they often lack generality and demand substantial data training. This paper presents a novel approach using Bayesian Regularized Artificial Neural Networks (BRANNs) to precisely forecast wear in milling tools. Unlike conventional machine learning models, BRANNs merge the strengths of artificial neural networks (ANNs) and Bayesian regularization, yielding a more robust and generalized predictive model. We utilized three open-access datasets from the literature alongside an in-house dataset generated by our milling setup. Initially, we assessed the model’s predictive ability by training and testing it against individual open-access datasets. We investigated the impact of input features, training data size, hidden units, training algorithms, and transfer functions on the model’s predictive capability. Subsequently, we trained the model using three open-access datasets and tested it against our in-house data. Our findings demonstrate that the developed model surpasses existing state-of-the-art models in accuracy and transferability

Velocity-Incorporated Wear Model of Rolling Guide Shoe Material Selection

To ensure an accurate selection of rolling guide shoe materials, an analysis of the intricate relationship between linear speed and wear is imperative. Finite element simulations and experimental measurements are employed to evaluate four distinct types of materials: polyurethane, rubber, polytetrafluoroethylene (PTFE), and nylon. The speed-index of each material is measured, serving as a preparation for subsequent analysis. Furthermore, the velocity-wear factor is determined, providing insights into the resilience and durability of the material across varying speeds. Additionally, a wear model tailored specifically for viscoelastic bodies is explored, which is pivotal in understanding the wear mechanisms within the material. Leveraging this model, wear predictions are made under higher speed conditions, facilitating the choice of material for rolling guide shoes. To validate the accuracy of the model, the predicted degree of wear is compared with experimental data, ensuring its alignment with both theoretical principles and real-world performance. This comprehensive analysis has verified the effectiveness of the model in the selection of materials under high-speed conditions, thereby offering confidence in its reliability and ensuring optimal performance.

Research on tool remaining useful life prediction algorithm based on machine learning

Tool wear during machining significantly impacts workpiece quality and productivity, making continuous monitoring and accurate prediction essential. In this context, the present study develops an efficient tool wear prediction system to enhance production reliability and reduce tool costs. It is worth noting that conventional methods, including support vector regression, autoencoders, attention mechanisms, CNNs, and RNNs, have limitations in feature extraction and efficiency. Aiming at resolving these limitations, a multiscale convolutional neural network (MDCNN)-based algorithm is proposed for predicting the remaining life of milling cutters. The algorithm uses preprocessing techniques like wavelet transform and principal component analysis for noise reduction and feature extraction. It then extracts temporal data features using convolutional layers of different scales and employs a self-attention mechanism for feature encoding. Validation on the PHM2010 milling cutter wear dataset with 10-fold cross-validation demonstrates that the MDCNN model achieves a wear prediction accuracy of 97%, a recall rate of 98%, and an F1 score of 97%. The MDCNN model effectively processes multi-band data and captures complex temporal features, confirming its efficiency and accuracy in predicting milling cutter wear and remaining service life.

Prediction of micro wear depth between engineering polymers

In the polymer industry, changes in the wear area that affect surface aesthetics have emerged as a significant issue in applications where surface appearance is crucial. Even minute changes in surface height can be visible to the naked eye, necessitating the need for micro-scale wear prediction of polymers. While the Archard model has been verified for decades for metal-metal or polymer-metal contact, predicting wear between polymer-polymer contacts remains challenging due to limitations in wear measurement and wear coefficient determination. In this study, we propose a method for determining the wear coefficient for polymer-polymer contact and validate the prediction model with a component wear test on polymer parts. Wear depth, an essential indicator of wear is measured after the sliding wear test of the polycarbonate (PC) pin and Acrylonitrile butadiene styrene (ABS) copolymer plate. The wear prediction model with the obtained wear coefficient exhibits 99.7 % accuracy in comparison to the test wear depth. A component wear test is performed on engineering polymer parts, and wear prediction under test conditions is made using the proposed model. The measured and predicted wear depth of the component demonstrates a difference smaller than 5 μm, which is within the range of no change in scratch visibility. This study ensures the accurate wear prediction of engineering polymer components with affordable computational efficiency. The ability to predict polymer-polymer wear, previously unattainable, will enable several applications in the polymer industry, such as component design, scratch visibility prediction, and quality assurance.

Research on tool wear prediction for milling high strength steel based on DenseNet-ResNet-GRU

Research on tool wear prediction for milling high strength steel based on DenseNet-ResNet-GRU

Analysis and prediction of uneven wear in metro pantograph-rigid catenary system

The non-elastic electrical contact between pantograph strip and wire, via a pantograph-rigid catenary system, results in much wear of metro strip. Every month, hundreds of thousands of strips are consumed in China due to the approximately 9900 kilometers of metro distance. Numerous instances of abnormal strip damage have been observed at the site, likely attributable to the uneven wear of strip. However, the cause of uneven wear on metro strip remains unclear to date. Hence, a wear model was developed based on the thermal equilibrium and wear mechanism, to investigate the uneven wear of strip in metro system. Various parameters, including contact force, electrical current, velocity, and copper content of the strip have been studied to explore their relationship with strip wear. Different laying structures, such as sine, half-wave, and folded line, exhibited varying degrees of uneven wear on the strip surface, with maximum depths observed at 4.20 mm per 104 km, 3.68 mm per 104 km and 1.02 mm per 104 km, respectively. The contact force holds the dominant position in metro wear, followed by current. Additionally, the wear rate also demonstrated sensitivity to the copper content of the strip.

Enhancing high-entropy alloy performance: Predictive modelling of wear rates with machine learning

High-entropy alloys (HEAs) are a category of material with exceptional mechanical, thermal, and chemical properties, making them suitable for a wide range of applications in industries such as aerospace, automotive, and energy storage. The wear rate is a crucial property in determining the reliability and performance of HEAs. It is crucial to accurately predict wear rates when optimizing material for selection and design. This study uses Machine Learning (ML) techniques to create predictive models for wear rate prediction in HEAs. Five regression models, including Linear Regression (LR), Support Vector Regression (SVR), Bayesian Ridge Regression (BRR), Multilayer Perceptron (MLP), and Huber Regression (HR) are built and evaluated. These models are built using a dataset featuring elemental compositions, applied load, sliding distance, sliding velocity, and wear rates of HEAs. The results show that the MLP regressor outperforms other algorithms in terms of accuracy and model fit when predicting wear rates. For phase prediction of non-existing HEAs, Recurrent Neural Networks (RNNs) and Extreme Gradient Boosting (XGBoost) models are employed. Using the output from these phase predictions as input variables, the wear rate of these non-existing HEAs is predicted at different conditions. The prediction provides a large dataset with phase and wear prediction of non-existing HEAs, which can be experimentally designed later for specific applications. The present study highlights the potential of ML techniques in developing materials science and engineering by allowing the design of HEAs with tailored properties for specific applications.

A hybrid tool wear prediction model based on JDA

PurposeAiming at solving the problems of low prediction accuracy and poor generalization caused by the difference in tool wear data distribution and the fixation of single global model parameters, a hybrid prediction modeling method for tool wear based on joint distribution adaptation (JDA) is proposed.Design/methodology/approachFirstly, JDA is exploited to adapt the data features with different data distributions. Then, the adapted data features are identified by the KNN classifier. Finally, according to the tool state classification results, different regression prediction models are assigned to different wear stages to complete the whole tool wear prediction task.FindingsThe results of milling experiments show that the maximum prediction accuracy of this method is 95.13%, and it has good recognition accuracy and generalization performance. Through the application of the tool wear hybrid prediction modeling method, the prediction accuracy and generalization performance of the model are improved and the tool monitoring is realized.Originality/valueThe research results can provide solutions and a theoretical basis for the application of tool wear monitoring technology in practical industrial applications.

Improving milling tool wear prediction through a hybrid NCA-SMA-GRU deep learning model

Milling tool wear, a ubiquitous challenge in industrial automation and manufacturing, leads to diminished equipment utilization, escalating costs, and a decline in product quality. The prediction of tool wear is a complex and challenging task, as it involves numerous variables. This paper introduces a pioneering hybrid approach, the NCA-SMA-GRU model. It involves the hybridization of three major components and is designed to enhance the precision and expedite the process of tool wear prediction. The NCA component is adept at filtering and retaining the most relevant features associated with milling tool wear from the raw signals, and it also improves the model’s interpretability. Subsequently, SMA optimizes the GRU network’s hyperparameters, including the initial learning rate, hidden layer neurons, network training iterations, and the L2 regularization factor, to identify an optimal combination that bolsters predictive performance. The modeling steps and the development of fitness function are explained in detail. The model’s efficacy is rigorously evaluated using data from the 2010 High Speed CNC Machine Tool Health Prediction Contest (PHM 2010), which encompasses wear data from both single and multiple cutters. The performance is compared using metrics such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), R-Squared (R2), and computational time. To assess the ability and characteristics of the proposed approach, other popular hybrid models are constructed for comparative analysis. The results demonstrate that the proposed model addresses the limitations of traditional prediction methods and provides insights into the development of deep learning and optimization algorithms for tool wear prediction.

Prediction of Brake Pad Wear of Trucks Transporting Oversize Loads Based on the Number of Drivers’ Braking and the Load Level of the Trucks—Multiple Regression Models

Brake pad wear forecasting, due to its complex nature, is very difficult to describe using engineering formulas. Therefore, the aim of this publication is to create high-quality brake pad wear forecasts based on three stochastic quantitative models based on multiple regression models (linear model, inverted linear model, and power model). The matrix of explanatory variables was extracted from the controllers of 29 vehicles: A—the driver’s style of using the brake pedal specified on a 4-point scale and B—the number of vehicle load ranges specified on a 5-point scale. Methodology: A matrix of explanatory variables was obtained over a 2-year period from trucks carrying oversize loads via OBD2 socket. The trucks operated under similar operating conditions. The created models were verified in terms of their fit to the source data and by analyzing the residuals of the models. It should be emphasized that only the linear model met all the required criteria. The inverted linear and power-law models were rejected. Results: The verified linear model is characterized by very small MAPE errors. The model was validated on 4 trucks and the brake pad wear prediction errors ranged from −0.39% to 7.03%.

Learning More with Less Data in Manufacturing: The Case of Turning Tool Wear Assessment through Active and Transfer Learning

Monitoring tool wear is key for the optimization of manufacturing processes. To achieve this, machine learning (ML) has provided mechanisms that work adequately on setups that measure the cutting force of a tool through the use of force sensors. However, given the increased focus on sustainability, i.e., in the context of reducing complexity, time and energy consumption required to train ML algorithms on large datasets dictate the use of smaller samples for training. Herein, the concepts of active learning (AL) and transfer learning (TL) are simultaneously studied concerning their ability to meet the aforementioned objective. A method is presented which utilizes AL for training ML models with less data and then it utilizes TL to further reduce the need for training data when ML models are transferred from one industrial case to another. The method is tested and verified upon an industrially relevant scenario to estimate the tool wear during the turning process of two manufacturing companies. The results indicated that through the application of the AL and TL methodologies, in both companies, it was possible to achieve high accuracy during the training of the final model (1 and 0.93 for manufacturing companies B and A, respectively). Additionally, reproducibility of the results has been tested to strengthen the outcomes of this study, resulting in a small standard deviation of 0.031 in the performance metrics used to evaluate the models. Thus, the novelty presented in this paper is the presentation of a straightforward approach to apply AL and TL in the context of tool wear classification to reduce the dependency on large amounts of high-quality data. The results show that the synergetic combination of AL with TL can reduce the need for data required for training ML models for tool wear prediction.

Domain adaptation neural network based prediction of real-time drill bit tooth wear

Drill bit wear reduces drilling efficiency and escalates operational risks. The drill bit tooth wear grading is a measure of lost tooth height. Accurate monitoring of drill bit tooth wear grade is crucial to adjust drilling parameters and plan timely bit replacement, significantly enhancing drilling efficiency and safety. Current real-time monitoring methods for drill bit wear exhibit limitations, particularly in accounting for variations in the geological formation and drilling parameters, resulting in poor model stability. This paper proposes an intelligent modeling approach that utilizes a Domain Adaptation Neural Network (DANN) to establish an ideal drilling rate of penetration (ROP) prediction model. By incorporating measured ROP, wear coefficients are calculated, and a relationship between wear coefficient and tooth wear grade is established, facilitating real-time monitoring of drill bit tooth wear grade. Test results on ten drill bits demonstrate that the DANN ideal ROP prediction model, leveraging data from new drill bits, enhances the model's adaptability to varying working conditions for new drill bits, thereby improving the accuracy of ideal ROP predictions. We establish relationship between calculated wear coefficients and tooth wear grades. The results indicate that the average error in inner wear grade is 0.6 (7.5%) and the average error in outer wear grade is 1.3 (16%) on an 8-point scale which demonstrates the model's effectiveness in monitoring drill bit tooth wear.