Tool Condition Monitoring In Milling Research Articles

On the Stability and Homogeneous Ensemble of Feature Selection for Predictive Maintenance: A Classification Application for Tool Condition Monitoring in Milling.

Feature selection (FS) represents an essential step for many machine learning-based predictive maintenance (PdM) applications, including various industrial processes, components, and monitoring tasks. The selected features not only serve as inputs to the learning models but also can influence further decisions and analysis, e.g., sensor selection and understandability of the PdM system. Hence, before deploying the PdM system, it is crucial to examine the reproducibility and robustness of the selected features under variations in the input data. This is particularly critical for real-world datasets with a low sample-to-dimension ratio (SDR). However, to the best of our knowledge, stability of the FS methods under data variations has not been considered yet in the field of PdM. This paper addresses this issue with an application to tool condition monitoring in milling, where classifiers based on support vector machines and random forest were employed. We used a five-fold cross-validation to evaluate three popular filter-based FS methods, namely Fisher score, minimum redundancy maximum relevance (mRMR), and ReliefF, in terms of both stability and macro-F1. Further, for each method, we investigated the impact of the homogeneous FS ensemble on both performance indicators. To gain broad insights, we used four (2:2) milling datasets obtained from our experiments and NASA's repository, which differ in the operating conditions, sensors, SDR, number of classes, etc. For each dataset, the study was conducted for two individual sensors and their fusion. Among the conclusions: (1) Different FS methods can yield comparable macro-F1 yet considerably different FS stability values. (2) Fisher score (single and/or ensemble) is superior in most of the cases. (3) mRMR's stability is overall the lowest, the most variable over different settings (e.g., sensor(s), subset cardinality), and the one that benefits the most from the ensemble.

Deep Learning Algorithms for Tool Condition Monitoring in Milling: A Review

The 4th Industrial Revolution (Industry 4.0) necessitates implementing the prognostics and health management (PHM) practices in manufacturing processes. The traditional machine learning approach has well assisted the PHM practices within the same data distributions. However, when a high noise environment, versatile operating conditions, and cross-domain machining is considered, it still lacks key steps of generalizing unknown tool faults. In an attempt to address PHM practices under such domains, a generic Deep Learning-based scheme is gaining significant attention. In this paper, an inclusive review is presented in order to provide an insight into the application of DL in tool condition monitoring (TCM), particularly in milling. Commonly used DL algorithms and their applications toward TCM are initially discussed and number of illustrative DL models applied for TCM is presented. Later, emergent DL themes & their computational techniques are summarized with an intention to provide framework for domain generalization. Finally, challenges in further exploration and futuristic trends in TCM are discussed.

Vibration-based tool condition monitoring in milling of Ti-6Al-4V using an optimization model of GM(1,N) and SVM

Titanium alloys are the difficult to cut metals due to their low thermal conductivity and chemical affinity with tool material. Since the tool vibration is a replica of tool wear and surface roughness, the present study has proposed a methodology for estimating tool wear and surface roughness based on tool vibration for milling of Ti-6Al-4V alloy using cemented carbide mill cutter. Experiments are conducted at optimum levels of cutting speed, feed per tooth, and depth of cut, and experimental results for the tool vibration, tool wear, and surface roughness are collected until the flank wear reached 0.3 mm (ISO3685:1993). In the next stage, an optimization model of grey prediction GM(1,N) system and support vector machine (SVM) are used and estimated tool wear and surface roughness related to tool vibration. The predicted values of tool wear and surface roughness are compared with the experimental results. The optimization model of GM(1,N) predicted the tool wear and surface roughness with an average error of 3.03% and as 0.7% respectively while the SVM predicted with an average error of 7.67% and 4.45%, respectively.

Sound singularity analysis for milling tool condition monitoring towards sustainable manufacturing

Manufacturing plays an important role since they are among the largest energy consumers in modern societies. With the enhancement of environmental protection and a severe shortage of energy and resources globally, sustainable manufacturing technology has been recognized as an important future trend of manufacturing industries. Tool wear is inevitable in manufacturing and affects the surface quality and geometric tolerance significantly. Therefore, a robust and efficient Tool Condition Monitoring (TCM) system is needed to maximize tool life, ensure work-piece quality and benefit the cost control of manufacturers. Even though lots of tool condition monitoring systems have been established using various sensors, there is still an urgent demand for a low-cost and simple setup system. This article presents a sound singularity analysis approach for TCM in milling; this approach has never been previously employed for milling tools. A de-noising algorithm based on Wavelet Transform Modulus Maxima (WTMM) estimation was proposed to eliminate noises and preserve singularities. Then a wavelet basis selection method was established for optimal sound singularity analysis. The singularity was estimated by Holder Exponents (HE). A full tool life-cycle milling experiments were conducted to obtain the sound signals. The mutual information method was employed to rank HE features. The Means, Standard deviations, Minima of estimated HEs and Quantities of singular points are found most correlated to tool conditions. Then a Support Vector Machine (SVM) model trained by these features for TCM has been proposed, achieving classification accuracy of 85%. Finally, the manufacturing sustainability of the TCM approach was evaluated by tracking and improving the usages of ten same type cutters in the CNC manufacturing plant. Experimental results indicate that this approach is efficient and capable of providing effective guidance on tool replacement, and can enhance the manufacturing sustainability.

Tool condition monitoring in milling using a force singularity analysis approach

Tool condition monitoring (TCM) is extremely important to ensure production efficiency and workpiece quality. It is crucial to extract and select suitable features from raw signals to improve the robustness and feasibility of TCM systems. This paper introduces a cutting force singularity analysis approach for TCM in milling, which correlates tool wear states with force waveform variations. The Holder exponents (HEs) were selected as the index of singularities. HEs are calculated by wavelet transform modulus maxima (WTMM). The raw signal is de-noised based on WTMM estimation, which can effectively preserve singularities compared with traditional filters. Fisher’s discriminant ratio (FDR) is employed to rank the discriminant capability of statistical features of HEs. It is found that means of HEs and quantities of singular points estimated from feed forces Fx show the strongest class-discriminant ability. Then, these features were chosen as training samples to propose a TCM approach based on the support vector machine (SVM). Experimental results indicate that this approach provides reliable and effective advice for tool change.

Singularity Analysis of Cutting Force and Vibration for Tool Condition Monitoring in Milling

Tool wear is inevitable in manufacturing and affects the surface quality and geometric tolerance significantly. A robust and efficient tool condition monitoring (TCM) system is needed to maximize tool life, ensure work-piece quality, and benefit the cost control of manufacturers. This paper presents a systematic singularity analysis approach of cutting force and vibrations for feature extraction of TCM in milling. The singularity of sensory signals is estimated by Holder Exponents (HE), which are determined by wavelet transform modulus maxima (WTMM). A comprehensive wavelet basis selection approach is proposed to choose the appropriate wavelet basis for different sensory signals. A de-noising algorithm based on WTMMs' estimation was used as a pre-processing technique to improve noise reduction and preserve the singularities. The mutual information method was employed to rank HE features. The effectiveness of the singularity analysis approach is validated through the Support Vector Machine (SVM) models trained by these ranked features. The estimating results of case studies confirm the efficacy and efficiency of the proposed approach.

A Multisensor Fusion Method for Tool Condition Monitoring in Milling

Tool fault diagnosis in numerical control (NC) machines plays a significant role in ensuring manufacturing quality. Tool condition monitoring (TCM) based on multisensors can provide more information related to tool condition, but it can also increase the risk that effective information is overwhelmed by redundant information. Thus, the method of obtaining the most effective feature information from multisensor signals is currently a hot topic. However, most of the current feature selection methods take into account the correlation between the feature parameters and the tool state and do not analyze the influence of feature parameters on prediction accuracy. In this paper, a multisensor global feature extraction method for TCM in the milling process is researched. Several statistical parameters in the time, frequency, and time–frequency (Wavelet packet transform) domains of multiple sensors are selected as an alternative parameter set. The monitoring model is executed by a Kernel-based extreme learning Machine (KELM), and a modified genetic algorithm (GA) is applied in order to search the optimal parameter combinations in a two-objective optimization model to achieve the highest prediction precision. The experimental results show that the proposed method outperforms the Pearson’s correlation coefficient (PCC) based, minimal redundancy and maximal relevance (mRMR) based, and Principal component analysis (PCA)-based feature selection methods.

Audio-Based Tool Condition Monitoring in Milling of the Workpiece Material With the Hardness Variation Using Support Vector Machines and Convolutional Neural Networks

Machining industry has been evolving toward implementation of automation into the processes for higher productivity and efficiency. Although many studies have been conducted in the past to develop intelligent monitoring systems in various application scenarios of machining processes, most of them just focused on cutting tools without considering the influence of the nonuniform hardness of workpiece material. This study develops a compact, reliable, and cost-effective tool condition monitoring (TCM) system to detect the cutting tool wear in machining of the workpiece material with hardness variation. The generated audible sound signals during the machining process are analyzed by state-of-the-art artificial intelligent techniques, support vector machine (SVM) and convolutional neural network (CNN), to predict the tool wear and the hardness variation of the workpiece. A four-level classification model is developed for the system to detect the tool wear condition based on the width of the flank wear land and the hardness variation of the workpiece. This study also involves the comparative analysis between two employed artificial intelligent techniques to evaluate the performance of the model in prediction. The proposed TCM system has shown a high prediction accuracy in detecting the tool wear from the audible sound into the proposed multiclassification wear level in end milling of the nonuniform hardened workpiece.

Benchmarking of Pattern Recognition Techniques for Online Tool Wear Detection

Pattern recognition techniques have been implemented in real-time tool condition monitoring (TCM) systems to improve their robustness and reliability. The performance and accuracy of these techniques vary depending on their algorithm and the dataset properties. This research benchmarks six pattern recognition techniques to optimize the learning effort, classification accuracy and calculation time for TCM in milling of Al-Alloys using spindle-drive feedback. The techniques were tested using a generalized dataset where the tool condition has a dominant effect over the cutting conditions. The analysis demonstrated the high capability of the linear discriminant analysis technique compared to other techniques.

Sensor fusion for online tool condition monitoring in milling

The objective of this paper is to combine a direct sensor (vision) and an indirect sensor (force) to create an intelligent integrated tool condition monitoring (TCM) system for online monitoring of flank wear and breakage in milling, using the complementary strengths of the two types of sensors. For flank wear, images of the tool are captured and processed in-cycle using successive moving-image analysis. Two features of the cutting force, which closely indicate flank wear, are extracted in-process and appropriately pre-processed. A self-organizing map (SOM) network is trained in a batch mode after each cutting pass, using the two features derived from the cutting force, and measured wear values obtained by interpolating the vision-based measurement. The trained SOM network is applied to the succeeding machining pass to estimate the flank wear in-process. The in-cycle and in-process procedures are employed alternatively for the online monitoring of the flank wear. To detect breakage, two features in time domain derived from cutting force are used, and the thresholds for them are determined dynamically. Again, vision is used to verify any breakage identified in-process through the cutting force monitoring. Experimental results show that this sensor fusion scheme is feasible and effective for the implementation of online tool condition monitoring in milling, and is independent of the cutting conditions used.

Tool condition monitoring in end milling using a torque-based sensorized toolholder

The automation of metal-cutting processes needs robust sensing systems and monitoring methods, able to detect the tool condition on-line. In milling operations, powerful tool-monitoring solutions are based on cutting force measurement, performed by using sensorized toolholders. This paper deals with a particular application of this method, consisting of acquiring the cutting torque signal in end milling by means of a commercial low-cost rotating dynamometer, usually applied in other machining operations. The dynamometer is directly integrated into a toolholder body, properly modified in order to obtain the needed specifications. The proposed solution is not invasive and, therefore, avoids a strong reduction in mechanical properties of the tool and the toolholder set. Preliminary cutting tests were performed in order to detect if the acquired signal can be successfully used to recognize on-line the status of the cutting tool. On-line threshold-based monitoring strategies were also proposed, and significant indicators were considered for tool condition monitoring in milling.

Tool condition monitoring in milling based on cutting forces by a neural network

Automated machining systems require reliable online monitoring processes. The application of a multilayered neural network for tool condition monitoring in face milling is introduced and evaluated against cutting force data. The work uses the back-propagation algorithm for training neural network of 5 2 10 2 2 architecture. An artificial neural network was used for feature selection in order to estimate flank wear ( Vb ) and surface roughness ( Ra ) during the milling operation. The relationship of cutting parameters with Vb and Ra was established. The sensor selection using statistical methods based on the experimental data helps in determining the average effect of each factor on the performance of the neural network model. This model, including cutting speed, feed rate, depth of cut and two cutting force components (feed force and vertical Z -axis force), presents a close estimation of Vb and Ra . Therefore, the neural network with parallel computation ability provides a possibility for setting up intelligent sensor systems.