Wear Prediction Model Research Articles

Performance evaluation for tool wear prediction based on Bi-directional, Encoder–Decoder and Hybrid Long Short-Term Memory models

PurposeExcessive tool wear is responsible for damage or breakage of the tool, workpiece, or machining center. Thus, it is crucial to examine tool conditions during the machining process to improve its useful functional life and the surface quality of the final product. AI-based tool wear prediction techniques have proven to be effective in estimating the Remaining Useful Life (RUL) of the cutting tool. However, the model prediction needs improvement in terms of accuracy.Design/methodology/approachThis paper represents a methodology of fusing a feature selection technique along with state-of-the-art deep learning models. The authors have used NASA milling data sets along with vibration signals for tool wear prediction and performance analysis in 15 different fault scenarios. Multiple steps are used for the feature selection and ranking. Different Long Short-Term Memory (LSTM) approaches are used to improve the overall prediction accuracy of the model for tool wear prediction. LSTM models' performance is evaluated using R-square, Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) parameters.FindingsThe R-square accuracy of the hybrid model is consistently high and has low MAE, MAPE and RMSE values. The average R-square score values for LSTM, Bidirection, Encoder–Decoder and Hybrid LSTM are 80.43, 84.74, 94.20 and 97.85%, respectively, and corresponding average MAPE values are 23.46, 22.200, 9.5739 and 6.2124%. The hybrid model shows high accuracy as compared to the remaining LSTM models.Originality/valueThe low variance, Spearman Correlation Coefficient and Random Forest Regression methods are used to select the most significant feature vectors for training the miscellaneous LSTM model versions and highlight the best approach. The selected features pass to different LSTM models like Bidirectional, Encoder–Decoder and Hybrid LSTM for tool wear prediction. The Hybrid LSTM approach shows a significant improvement in tool wear prediction.

Prediction of machine tool wear using the weighted k-nearest neighbor method and Neural network model with acceleration response

In the manufacturing industry, with the decrease in the working population, there is a demand for intelligent manufacturing processes (smart factory). When focusing on cutting, predicting and detecting tool wear during machining is an important technical issue for autonomy. However, the tool wear of cutting tools varies from machining to machining even under the same machining conditions, with the same type of tool, and with the same material and shape of workpiece. Therefore, even if a representative wear prediction model is constructed to predict tool wear, the prediction accuracy is low and not practical. In this research, we will verify whether the amount of tool wear can be predicted by a neural network, which is a type of machine learning, using the vibration data obtained during cutting and the past. The accuracy of the proposed method is also verified by comparing the wear volume prediction by one representative neural network with the wear volume prediction by the combination of the proposed multiple neural network.

Indirect tool wear measurement and prediction using multi-sensor data fusion and neural network during machining

The quality produced of the machined workpiece is affected by various machining parameters including the tool wear. The condition of tool can be monitored by analyzing various machining responses through sensors. The use of multiple sensors allows for a comparison of data and information gathered from various sources, assisting in deciding the state of the tool and the workpiece. In current work, multiple sensors are attached to a lathe during the machining of AISI 4140. This paper presents a multi-sensor data fusion method for flank wear measurement and prediction using various parameters such as vibration, power, temperature, force, and surface roughness. Dry machining is performed by using PVD coated tool insert, and the flank wear is analyzed. Experiments are performed with a full factorial experimental design L27 orthogonal array. Taguchi approach has been utilized to evaluate the effect of machining parameters on tool wear. Also, the tool wear is predicted through an artificial neural network (ANN) model, and the results are compared with manually measured data. The results showcased the effectiveness of the proposed tool wear prediction model with the least measurement error. The work presents a novel neural network approach for the accurate prediction of tool wear and hence eliminates the necessity of manual measurement of tool wear and leading towards the increased productivity and quality of the work part.

Study on Wear Prediction of Shield Disc Cutter in Hard Rock and Its Application

It is important to enable the prediction of disc cutter wear during shield tunnelling through hard rock because such wear is associated with project delays and increased costs. Based on a theoretical analysis of Archard's wear mechanism and Euler's rotation theorem, a displacement equation of rock-breaking point A on the disc cutter ring is studied, and then a new wear prediction model is established combined with the theoretical analysis of the disc cutter force and wear test. The cutter wear prediction model is verified by using the data of two cases, and the results reveal that the errors between measured and predicted values are less than 25%. The research results also show that abrasive wear is the main wear mechanism (approximately 88%). In addition, the influence of fatigue wear is very small, accounting for only approximately 2%, which can be ignored. Finally, the maximum tunneling distance of disc cutters under different installation radii and penetration is studied using the model, and the results can be used as references for optimizing shield tunneling parameters and predicting the disc cutter replacement time.

Tool wear prediction method based on symmetrized dot pattern and multi-covariance Gaussian process regression

Cutting tool plays a critical role in modern manufacturing system and tool wear prediction has a great effect on product cost and quality. Many efforts have been devoted to developing tool wear prediction models. The previous studies mainly focus on utilizing a single model during the whole prediction process named global model and thus ignore the local characteristics of different wear stages. However, paying no attention to the impact of wear stage division may lead to mismatch the distribution characteristics of datasets and overlapping of feature information, failing to get the desired accuracy. To achieve more accurate prediction results, this paper proposes a wear stage division-based tool wear prediction method (WSDTWP) based on the improved symmetrized dot pattern (ISDP) and multi-covariance Gaussian process regression (MCGPR). Firstly, according to the varying trend of tool wear value, tool wear process is divided into three stages, including initial wear stage, moderate wear stage and severe wear stage. Then, SDP technique reconstructs the original signals visually and the main SDP parameters are adaptively selected by developing an evaluation model. Further, to deal with the issue of high-dimension and small-size datasets in initial and severe wear stages, MCGPR is developed and hyperparameters are optimized by particle swarm optimization for accurately predicting tool wear. According to the varying prediction requirements, different approaches are assigned into different wear stages to achieve better performance. Finally, three cutting tests are conducted to validate the effectiveness of the proposed approach. The experimental results indicate that WSDTWP is more accurate than existing methods, which provides new theoretical and practical support for identifying tool working condition and predicting tool wear.

Abrasive tool wear prediction based on an improved hybrid difference grey wolf algorithm for optimizing SVM

In the era of intelligent manufacturing, it is necessary to monitor the wear condition of abrasive tools in real time to prevent the deterioration of workpiece quality due to tool breakage and wear. A wear prediction model of abrasive tools based on an improved hybrid differential grey wolf optimization algorithm for optimizing support vector machine (IHDGWO-SVM) is proposed based on the grinding of zirconia ceramic holes by sintered diamond grind bits. The features of the force and vibration signals were extracted by time domain, frequency domain and wavelet analysis. The wear experimental results showed that the prediction accuracy of the IHDGWO-SVM model was 92%, which was significantly higher than the prediction accuracy of 68%, 80% and 72% of SVM, GWO-SVM and DE-SVM. The new IHDGWO-SVM model provides a theoretical and practical method for the on-line wear monitoring of abrasive tools during grinding of NMBM.

Sediment wear prediction model of ZG06Cr13Ni4Mo turbine guide vane in sediment-laden hydropower station

The guide vane of a hydraulic turbine in any sediment-laden hydropower station is one of the components most seriously affected by sediment abrasion. Damage to a guide vane can significantly impact stable operation and energy characteristics of the unit, and it is thus essential to address and effectively manage this problem. In this study, the k-ε solid–liquid two-phase turbulence model and sample algorithm were used to numerically simulate the sand-water flow through the entire passage of a hydraulic turbine and sand samples were subsequently collected from the hydropower station to examine the sediment abrasion damage to turbine’s guide vane, which was made of ZG06Cr13Ni4Mo. Thereafter, calculation and test results were used to establish a prediction model for sediment abrasion of hydraulic turbine guide vane. These research findings could provide guidance for improved hydraulic turbine design and could thus contribute to the optimized operation of sediment-laden hydropower stations.

Wear behavior of tool flank in the side milling of Ti6Al4V: An analytical model and experimental validation

Due to the poor machinability of Ti6Al4V material, the cutting tool can easily suffer flank wear during the process of high-speed side milling, which reduces the tool life as well as the surface integrity of workpiece. Further, an effective method for predicting the flank wear of end mill during side milling of Ti6Al4V is lacking in the existing literature, which makes it difficult to improve the productivity of the overall process. To this end, in this study, a flank wear prediction model is constructed based on three main mechanisms: abrasive wear, adhesive wear, and diffusive wear. Subsequently, a normal stress model and temperature field model of wear land on the flank of end mill are established. Finally, these two models are incorporated in the flank wear model to obtain the variation rate of wear land width, which is regarded as a criterion to evaluate the reliability of the proposed flank wear prediction model of side mill. The prediction results are found to be in excellent agreement with the experimental results, which verifies the high prediction accuracy of the proposed model. Overall, this model can serve as a useful theoretical basis for the rational selection of tool geometry and cutting parameters.

Development and validation of a wear prediction model for railway applications including track flexibility

The simultaneous prediction of wheel surface wear is one of the most challenging tasks in the railway field, which requires the knowledge of vehicle–track dynamics, rolling contact mechanics, and other tribological phenomena. In this paper, a new model for predicting wheel profile wear is presented to study in detail the characteristics of wear in the railway. The model is able to consider the effect of track flexibility on wheel wear evolution. The proposed modelling approach includes a vehicle–track coupled model for vehicle dynamical simulation, the local contact model for wheel–rail interaction, and an energetic approach for surface wear evaluation. The whole model has been validated by the comparison with experimental data coming from Chinese high-speed railway lines. The results show that the new model assures high computational efficiency as well as reaches a good prediction accuracy. The proposed model allows the investigation of the role played by different track parameters related to track flexibility on the wheel and rail wear evolution. The influence of the fastener and railpad stiffness on the wheel wear evolution is presented, using the proposed model.

Experimental validation of a fast wheel wear prediction model

Wheel-rail wear prediction is a fundamental matter in order to study problems such as wheel lifespan prediction and evolution of vehicle dynamic characteristics with time. However, one of the principal drawbacks of the existing methodologies for calculating the wear evolution is the computational cost. The developed methodology is based on (1) representing the whole route in a reduced number of characteristic points and (2) substituting dynamic simulations by quasi-static ones, in order to reduce computational cost significantly. This paper presents its integration into the complete simulation process for predicting wheel wear evolution and its validation by comparing simulated results with in-service wheel profile measurements. The generality of the method has been verified by applying it to two different operational cases with very different track characteristics. For the validation, the evolution of the worn depth in the tread has been compared according to the travelled distance, as well as the morphology of the worn depth (worn area). Promising results have been obtained in the prediction of wheel wear evolution: in both cases, the simulated curve approaches the measurement curve satisfactorily. Therefore, it can be concluded the validation of the developed method adequately.

Effect of off-sized balls on contact stiffness and stress and analysis of the wear prediction model of linear rolling guideways

The characteristics between the rolling balls and raceways are the key to study a linear rolling guideway (LRG). In this paper, the contact stresses of an LRG with off-sized balls incorporating the variation of the contact angle are given by the established LRG joint model. Moreover, the effect of the location, number, and the deviation degree of the off-sized balls on the stress distribution are studied. In addition, the contact stress distribution between the balls and raceway for different arrangement cases of the off-sized balls are analyzed. The random arrangement case can improve the stiffness and service life of the LRG. Based on the Archard wear theory, the wear prediction model of the LRG is established and the displacements and angular displacements of the slider caused by wear in reciprocating motion are obtained. The effectiveness of the contact stiffness and wear prediction model of the LRG is verified by simulations and analysis.

Hybrid adaptive model to optimise components replacement strategy: A case study of railway brake blocks failure analysis

In this paper, we propose a novel hybrid adaptive model (HAM) that integrates Gaussian mixture probabilistic machine learning (ML), Weibull time-to-failure feature, and value of information (VOI) techniques for complex engineering failure analysis. The objective is to establish an optimum components replacement intervention strategy for composite brake blocks of railway rolling stocks to better curtail failures and possible accidents. The HAM considers brake blocks as rudimentary systems for emergency brake application to prevent catastrophic rail accidents under nonlinear and dynamic environmental conditions. A Gaussian mixture regression technique with a rational quadratic kernel function is used to develop a predictive wear model that applies wear measurement as training data. The Weibull feature estimates the threshold maintenance replacement strategy pertaining to other premature failure modes before the brake blocks’ legal scrapping wear limit is reached. Finally, the VOI feature establishes the net loss in terms of cost for the proposed replacement options to guide optimum component replacement selection strategy under organisational resource constraints. Based on operational data obtained from several brake blocks of the London Underground Trains fleet, the proposed HAM failure analysis technique provided a better balance between safety and cost-effectiveness compared to other popular ML approaches.

Development of an ANN model for prediction of tool wear in turning EN9 and EN24 steel alloy

An imperative requirement of a modern machining system is to detect tool wear while machining to maintain the surface quality of the product. Vibration signatures emanating during machining with a single point cutting tool have proven to be good indicators for the tool’s health. The current research undertaken utilizes vibration signatures while turning EN9 and EN24 steel alloy to predict tool life using Artificial Neural Network (ANN). During initial meager experimentation, tool acceleration during machining was recorded, and the width of the flank wear at the end of each run was measured using Tool Makers Microscope. The recorded experimental data is utilized to develop the neural network with the variation of operating parameters and corresponding tool vibration with measured tool flank wear. The endeavor undertaken for the development of ANN flank wear prediction model was effective with a regression coefficient of 0.9964. The proposed methodology of indirect measurement of tool wear is efficient, economical for the machining industry to predict tool life, which in turn avoids catastrophic tool failure.

A Review on Mechanisms and Testing of Wear in Slurry Pumps, Pipeline Circuits, and Hydraulic Turbines

AbstractThis paper presents an overview about particulate wear that occurs in various components of slurry pumps, pipeline systems, and hydraulic turbines due to the mechanical action caused by the flow of solid–liquid mixtures. Three most common wear problems, namely, erosion, erosion–corrosion, and abrasion occur in different industries like thermal and hydro power plants, mining, chemical, and marine industries. Therefore, the efficiency of these industries and power plants highly depends on the wear damages. So, it becomes necessary to govern the wear phenomenon. In this paper, the various properties of particles and target are discussed on which the particulate wear depends. Present overview explains the experimental methods of measuring the particulate wear at in situ and ex situ conditions by using the different types of testers, rigs, and pilot plant test loops. Moreover, the empirical and analytical correlations or models for the prediction of particulate wear in pumps, piping circuits, and hydraulic turbines are also discussed in present literature review. By studying the all possible advantages and disadvantages, the gaps in knowledge of wear prediction methods and models are highlighted. At the end, a “think-model” for the prediction as well as reduction of wear in the various components is presented on the basis of different experimental and computational fluid dynamics (CFD) based simulation work. Further investigations can be carried out to develop the more accurate and flexible models that can be used to predict the particulate wear in a wide range of applications.

COMPARATIVE ANALYSIS OF METRO WHEEL WEAR PREDICTION BASED ON TWO TRACK MODELING METHODS

The track has an important influence on wheel wear. A metro wheel wear prediction model is established, which includes a vehicle dynamic model, a local contact model based on Hertzian theory and on FASTSIM algorithm, a wear model based on the Tγ/A-wear rate function, a smoothing and updating strategy, and a track model. It is compared for the differences in modeling complexity, in wheel wear prediction results and, in the calculation efficiency between the equivalent track method and the actual track method. The Tγ/A-wear rate function is corrected by the measured wheel wear, and the wear prediction model is verified. The results show that, compared with the equivalent track method, the actual track method is more complex, but it can better simulate the actual track. There is little difference between the two wear prediction results, and both of them are in a good agreement with the measurement results. The calculated efficiency of the equivalent track model is 3.7 times that of the actual track model. It is suggested to use the equivalent track method for modeling, which can give the consideration for calculation accuracy and efficiency.

Prediction of Abrasive and Impact Wear Due to Multi-Shaped Particles in a Centrifugal Pump via CFD-DEM Coupling Method

Since solid particles suspended in the fluid can cause wear in centrifugal pumps, intensive attention has been focused on the numerical prediction for the wear of flow parts in centrifugal pumps. However, most numerical studies have focused on only one wear model and a sphere particle model. The impact of particle shape on the wear of flow parts in centrifugal pumps is under-studied, particularly considering abrasive and impact wear simultaneously. In this work, the Computational Fluid Dynamics (CFD)-Discrete Element Method (DEM) coupling method with an abrasive and impact wear prediction model was adopted to study the wear characteristics of a centrifugal pump. Moreover, four regular polyhedron particles and a sphere particle with the same equivalent diameter but different sphericity were mainly analyzed. The results demonstrate that more particles move closer to the blade pressure side in the impeller passage, and particles tend to cluster in specific areas within the volute as sphericity increases. The volute suffers the principal wear erosion no matter what the shapes of particles and wear model are. Both the impact and abrasive wear within the impeller occur primarily on the blade leading edge. The pump’s overall impact wear rate decreases first and then increases with particle sphericity rising, while the pump’s overall abrasive wear rate grows steadily.

Multi-frequency-band deep CNN model for tool wear prediction

A reliable data-driven tool condition monitoring system is more and more promising for cutting down on machine downtime and economic losses. However, traditional methods are not able to address machining big data because of low model generalizability and laborious feature extraction by hand. In this paper, a novel deep learning model, named multi-frequency-band deep convolution neural network (MFB-DCNN), is proposed to handle machining big data and to monitor tool condition. First, samples are enlarged and a three-layer wavelet package decomposition is applied to obtain wavelet coefficients in different frequency bands. Then, the multi-frequency-band feature extraction structure based on a deep convolution neural network structure is introduced and utilized for sensitive feature extraction from these coefficients. The extracted features are fed into full connection layers to predict tool wear conditions. After this, milling experiments are conducted for signal acquisition and model construction. A series of hyperparameter selection experiments is designed for optimization of the proposed MFB-DCNN model. Finally, the prediction performance of typical models is evaluated and compared with that of the proposed model. The results show that the proposed model has outstanding generalizability and higher prediction performance, and a well designed structure can remedy the absence of complicated feature engineering.

Automatic Identification of Tool Wear Based on Thermography and a Convolutional Neural Network during the Turning Process.

This article presents a control system for a cutting tool condition supervision, which recognises tool wear automatically during turning. We used an infrared camera for process control, which—unlike common cameras—captures the thermographic state, in addition to the visual state of the process. Despite challenging environmental conditions (e.g., hot chips) we protected the camera and placed it right up to the cutting knife, so that machining could be observed closely. During the experiment constant cutting conditions were set for the dry machining of workpiece (low alloy carbon steel 1.7225 or 42CrMo4). To build a dataset of over 9000 images, we machined on a lathe with tool inserts of different wear levels. Using a convolutional neural network (CNN), we developed a model for tool wear and tool damage prediction. It determines the state of a cutting tool automatically (none, low, medium, high wear level), based on thermographic process data. The accuracy of classification was 99.55%, which affirms the adequacy of the proposed method. Such a system enables immediate action in the case of cutting tool wear or breakage, regardless of the operator’s knowledge and competence.

Effect of Blade Angles on the Shovel Muck Capacity and Wear Characteristics for TBM Scraper

During tunnel boring machine (TBM) excavation, scrapers play the role in shoveling the broken rock fragments, directly contacting with them, which may result in seriously wear and failure, causing high economic loss. Therefore, it is an urgent problem to study the wear mechanism of scraper and how to reduce the blade wear. In this study, the wear prediction model of scraper was established based on working mechanism and wear mechanism analysis. The factors influence on wear characteristic were obtained. Then, the numerical model of TBM scraper shoveling the muck was established using discrete element method. The shovel capacity and wear characteristic of scraper were studied. The structural parameters of blade were optimal designed, comprehensively considering the shovel capacity and wear characteristics. The results indicate that abrasive wear is the main type of scraper wear. The main influencing factors of scraper wear are blade angles. When the front angle of the blade is 25° and the back angle is 8°, the wear performance of the scraper would be improved. The optimal results were successfully applied to a tunneling project and the actual data in the engineering construction of the scraper during the construction were analyzed to verify the structural design method of TBM scraper.

A wheel wear prediction model of non-Hertzian wheel-rail contact considering wheelset yaw: Comparison between simulated and field test results

The standard model-based approach for predicting wheel wear consists of four main sub-modules: a macro vehicle-track model, a normal wheel-rail contact model, a tangential wheel-rail contact model, and a wear model. In terms of the normal contact model, the few existing studies using non-Hertzian methods to calculate wheel wear rarely consider wheelset yaw, although it is an unavoidable phenomenon when a train is running. Aiming at this issue, a non-Hertzian normal wheel-rail contact model considering wheelset yaw is proposed based on ANALYN and Sun's approximate expression (i.e., ANALYN-YAW). In terms of the wear model, the currently available models were developed for specific wheel/rail materials and under fixed conditions. To extend these models to a new situation, an automatic adjustment strategy based on Kriging surrogate model and particle swarm optimization algorithm (KSM-PSO) is introduced to adjust the wear rate of the wear model developed by the University of Sheffield (USFD model). Finally, an ANALYN-YAW-FaStrip-adjusted USFD-based wheel wear prediction model is proposed, and a comparison of results from simulation and field tests is presented and shows that the predicted result of the ANALYN-YAW-FaStrip-adjusted USFD model is more plausible than that of the ANALYN-FaStrip-adjusted USFD model and the Hertzian-FaStrip-adjusted USFD model.