Spindle Vibration Signal Research Articles

Vibration signal acquisition and extraction for cutter wear monitoring

Abstract The cutter wear state is key in manufacturing. During the machining, the vibration signal of the tool spindle will change with the increase of cutter wear, which provides an effective method for cutter wear monitoring. However, due to the popularity of machining centers, the machining trajectory of a cutter is extremely complex in practical production. The complexity includes a single process containing multiple work steps, a step containing random and long non-cutting time, etc., making it difficult to acquire and extract the specific step signal as the effective signal for cutter wear monitoring. Therefore, this paper proposes a system that can synchronously acquire machine parameters and tool spindle vibration signals. Then, two algorithms—the cutting signal recognition algorithm and the step signal similarity detection algorithm—are proposed. The former algorithm is based on sliding window energy, which can accurately identify the instant the cutter begins to cut the workpiece. The step signal similarity detection algorithm based on the Edit Distance on Real (EDR) sequence ensures the correctness of the step extraction. The experimental results show that the system has efficient acquisition of vibration signals and correct step signal extraction, providing a data basis for cutter wear monitoring.

Study on tool wear state recognition algorithm based on spindle vibration signals collected by homemade tool condition monitoring ring

With a view to further realising the intelligence of tool condition monitoring (TCM) and to address the high cost and low stiffness problems of existing smart tool holders, this study firstly proposed a tool condition monitoring ring, which can be mounted on the tool shank and is easily disassembled and reused. This ring acquires the triaxial vibration signals of the spindle. This study also firstly applied manifold learning to TCM model for data dimensionality reduction and proposed a weighted random forest based on K-nearest neighbour (KWRF). A tool wear state recognition algorithm based on wavelet threshold de-noising (WTD), variational mode decomposition (VMD), manifold learning and KWRF was proposed. Four sets of full life cycle milling experiments were carried out on 45 steel with cemented carbide milling inserts. The results showed that the recognition accuracy of the proposed model reached 98.74%, which verified the effectiveness of the proposed model.

Monitoring Built-Up Edge, Chipping, Thermal Cracking, and Plastic Deformation of Milling Cutter Inserts through Spindle Vibration Signals

Condition monitoring provides insights into the type of damage occurring in the cutting tool during machining to facilitate its timely maintenance or replacement. By detecting and analyzing machining consequences (vibrations, chatter, noise, power consumption, spindle load, etc.), correlating them with different tool conditions enables real-time monitoring and the automated detection of tool failures. Machine learning (ML) plays a vital role in making tool condition monitoring (TCM) frameworks intelligent, and most research is geared toward classifying various types of tool wear. However, monitoring built-up edges, chipping, thermal cracking, and plastic deformation of milling cutter inserts are challenging and need careful consideration. To effectively monitor these phenomena, spindle vibrations can narrate the corresponding dynamic behavior of tool conditions and therefore have been investigated in this research. The acquired vibration data are then analyzed using histogram features and trained through the Partial C4.5 (PART) classifier to extract meaningful recommendations related to the milling cutter inserts condition.

Research on feature extraction method of spindle vibration detection of weak signals for rapier loom fault diagnosis in strong noise background

Fault diagnosis of rapier loom is an inevitable requirement to meet the demand of intelligent manufacturing. Facing the strong noise interference caused by complex working environment, accurate and reliable vibration signal detection of blade loom spindle is the key to realize the rapier loom fault diagnosis. This paper proposes a method to extract the spindle vibration signal of the rapier loom by Adaptive Piecewise Hybrid Stochastic Resonance (APHSR) after the Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN). Firstly, ICEEMDAN is used to pre-process the weak vibration signal containing noise, decompose the signal into multiple IMF components and display the high and low frequency signal characteristics of the original signal. Then, the energy density method and the correlation coefficient method are used to remove high and low noise, respectively, to filter the optimal IMF components, and then the signal containing valid information is reconstructed. Finally, the reconstructed signal is input to APHSR for noise-assisted enhancement after scale transformation to restore the faint vibration signal feature frequencies and achieve effective feature extraction. Through the simulation experiment and the engineering fault experiment analysis, comparing ICEEMDAN-APHSR with CEEMDAN-SR, ICEEMDAN-SR, CEEMDAN-APHSR methods. The difference between the spectrum amplitude, the spectrum amplitude and the maximum noise and the maximum signal to noise ratio (SNR) of the fault feature frequency of the rapier loom spindle bearing increased by 3.3668 dB,1.7205 dB,2.3952 dB, respectively. The results show that ICEEMDAN-APHSR method can accurately extract the fault feature frequency of the spindle bearing of rapier loom, and effectively solves the problem of extracting the weak vibration signal feature of rapier loom in the background of strong noise. This method is beneficial to the future research of rapier loom fault diagnosis, and is of great significance to promote the maintenance of loom equipment and production safety and quality.

Real-time tool breakage monitoring based on dimensionless indicators under time-varying cutting conditions

Tool breakage occurs randomly during machining operations, which induces more severe impacts on the quality of components compared to progressive tool wear. It is widely acknowledged that the unpredictable changes in cutting conditions will cause fluctuations in the signal amplitude and thus generate false alarms. This study introduced a novel method for tool breakage monitoring based on dimensionless indicators under time-varying cutting conditions. The amplitude ratio (AR) and the energy ratio (ER) were proposed according to the power spectrum of the spindle vibration signal, which represents the change of amplitude and the energy distribution, respectively. The AR and ER are normalized and integrated into a unified indicator for real-time breakage monitoring. The floating monitoring threshold is designed based on the Gaussian distribution. Moreover, the material removal rate (MRR) is selected as a secondary indicator to accurately identify tool breakage based on determining the amplitude fluctuation caused by cutting conditions or teeth breakage. The effectiveness of the proposed method for tool breakage monitoring has been verified under the constant, time-varying, and entry/exit cutting conditions. The results show that the proposed indicators have higher sensitivity than the traditional root mean square (RMS) features and eliminate false alarms during condition change transients. This research provides a potential solution for tool breakage monitoring under complex cutting conditions.

Bearing Fault Diagnosis Using a Vector-Based Convolutional Fuzzy Neural Network

The spindle of a machine tool plays a key role in machining because the wear of a spindle might result in inaccurate production and decreased productivity. To understand the condition of a machine tool, a vector-based convolutional fuzzy neural network (vector-CFNN) was developed in this study to diagnose faults from signals. The developed vector-CFNN mainly comprises a feature extraction part and a classification part. The feature extraction phase encompasses the use of convolutional layers and pooling layers, while the classification phase is facilitated through the deployment of a fuzzy neural network. The fusion layer plays an important role by being placed between the feature extraction and classification parts. It combines the characteristics and then passes the feature information to the classification part to improve the model’s performance. The developed vector-CFNN was experimentally evaluated against existing fusion methods; vector-CFNN required fewer parameters and achieved the highest average accuracy (99.84%) in fault diagnosis relative to conventional neural networks, fuzzy neural networks, and convolutional neural networks. Moreover, vector-CFNN achieved superior fault diagnosis using spindle vibration signals and required fewer parameters relative to its counterparts, indicating its feasibility for online spindle vibration monitoring.

Mechanical vibration monitoring system for electrocardiogram machine based on Hilbert‐Huang transformations

The monitoring of health and the technologies that are related to it are an exciting area of research. The paper proposes a mechanical manufacturing vibration monitoring system that is based on Hilbert-Huang transformation (HHT) feature extraction to monitor the running state of the spindle of a mechanical numerical control (NC) machine tool of an electrocardiogram (ECG) machine. Real-time monitoring of the time–frequency characteristic quantity of the spindle vibration signal for ECG signals has been made possible due to the online empirical mode decomposition (EMD) method, which is used to obtain the time–frequency characteristic quantity of the spindle vibration signal based on HHT. The experiment shows that the frequency doubling characteristic components in the time–frequency distribution are obvious in the time interval without copper rod contact, but they disappear in the time interval during which copper rods are in contact (0.3 1.1 s, 3 4s in the figure). It has been demonstrated that the system is capable of not only accurately monitoring the characteristic quantity in the frequency domain of the vibration signal produced by the NC machine tool spindle, but also of successfully implementing the monitoring of the time–frequency characteristic quantity in real time.

Effects of wheel spindle vibration on surface formation in wafer self-rotational grinding process

As a universal machining method, wafer self-rotational grinding is applied to obtain a planar surface with high processing accuracy. However, the wheel spindle vibration significantly deteriorates the wafer flatness but is rarely studied. This paper primarily focuses on the impacts caused by wheel spindle vibration on the surface topography of ground wafer. Firstly, a mathematical model is proposed to analyze the dynamic behaviors of wheel spindle, involving radial, axial and tilting movements. The dynamic response and frequency characteristics of vibration-induced errors are presented. Then, a surface formation model is developed, taking grinding kinematics, grain size randomness, vibration-induced error and overlapping effect into account. Two kinds of patterns, i.e., waviness along circumferential direction and waviness along mark direction, are observed from the simulated wafer surface. Moreover, their wavelengths are calculated to explore the effect of grinding parameters and vibration frequency on waviness features. Finally, grinding tests are performed under different machining conditions, as well as the spindle vibration signal and wafer topography are measured. The obtained waviness characteristics are consistent well with those in simulation. In addition, the variation of surface roughness under different vibration parameters is further revealed. This work is expected to provide new insight into the machined error prediction and wafer flatness improvement considering the dynamic performance of wheel spindle.

A novel adversarial domain adaptation transfer learning method for tool wear state prediction

Tool wear prediction is of critical importance to maintain the desired part quality and improve productivity in machining. The traditional tool wear prediction based on deep learning mostly considers the same type of tools under the same working conditions. It assumes that the collected data obey the same distribution and that the training data labels are sufficient, which has significant limitations in practical machining applications. In this paper, a novel adversarial domain adaptation transfer learning was proposed to predict the tool wear state of end milling tools under different working conditions, including the laboratory and actual industrial machining conditions. Firstly, the dual-path deep residual shrinkage network was used to extract the tool wear multiscale sensitive features from the spindle vibration signals. Then, a balance parameter was added to the traditional adversarial domain adaptation model, which can dynamically and quantitatively evaluate the relative importance of marginal and conditional distribution. Thus, the alignment of the source and target tool feature space was realized by dynamically learning domain invariant representations. Finally, the proposed method was verified on an 8 mm and 2 mm tool wear states prediction. Compared with different transfer learning methods, the superiority of the proposed dynamic adversarial domain adaptation method was proved.

Modeling and tagging of time sequence signals in the milling process based on an improved hidden semi-Markov model

Vibration signals are widely used in the field of tool wear, tool residual life prediction and health monitoring of mechanical equipment. However, the current data-driven research methods mostly rely on high-value and high-density labeled data to establish relevant models and algorithms. Therefore, it is of great significance to solve the problem of automatic tagging of data, realize automatic signal interception, and enhance the value density of manufacturing process data.The Hidden semi-Markov model (HSMM) can describe the real spatial statistical characteristics of random models through observable data. As HSMM does not need the real labels of the signal, it can reduce tagging work to improve the marking efficiency. In this paper, an improved HSMM was proposed to model and tag the spindle vibration signals in the milling process. First, the Mel frequency cepstral coefficients (MFCCs) were extracted as observation sequences from the collected spindle vibration signals, and the dimension of the original features was reduced by linear discriminant analysis (LDA). Subsequently, a signal automatic tagging model based on HSMM was developed, in which the state duration can be explicitly modeled. Finally, the evaluation of the proposed methodology was carried out in the laboratory and real industry machining. The experimental results confirmed the effectiveness and robustness of the proposed model.

Effects of wheel spindle error motion on surface generation in grinding

In ultra-precision grinding, error motions of the aerostatic bearing wheel spindle lead to unexpected variations in the actual depth of cut, thereby directly affecting machining accuracy. However, few studies have focused on dynamic characteristics of the wheel spindle error motion and its effects on surface generation. In this paper, the ground surface generation mechanism under the wheel spindle error motion is explored by establishing a grinding kinematics-based surface topography model integrated with a wheel spindle dynamics model considering the wheel topography and the overlapping effect. Grinding experiments are performed to identify the theoretical results through spindle vibration signals and machined surface topographies. The results show that the spindle error motions include the drift of the axis average line which produces the surface form error, and the spindle vibration relative to the axis average line that contributes to the waviness circumferentially. It is also found that the increasing wheel speed leads to the decreasing waviness amplitude at a small radial distance of the workpiece while the trend of the waviness amplitude with the wheel speed is the same as that of the spindle vibration amplitude at a large radial distance. Furthermore, based on the ground surface simulation, the grinding mark pattern formation mechanism is found to include three types, identified by the machined surfaces. A novel and convenient method, where the ratio of the grinding contact width to the cross feed distance per workpiece revolution and the reduced fraction of rotational speed ratio are used, is proposed to predict the grinding mark patterns and spatial frequencies, and validated by the ground surface simulation and experimental results.

Condition monitoring of deep-hole drilling process based on improved empirical wavelet de-noising and high multiple frequency components of rotation frequency

The machining condition monitoring is an important task to improve product quality and avoid economic losses in deep-hole drilling. Out-of-roundness-tolerance is one of major defects in deep-hole drilling process. In this work, the spindle vibration signals generated in deep-hole drilling process are studied to seek the characteristics which can reveal the roundness error change. Considering the influence of background noise on monitoring effect and the sensitivity of high multiple frequency components of spindle vibration signal towards the roundness error change, an improved empirical wavelet de-noising method is proposed to remove the noise of the spindle vibration signal. The high multiple frequency components of the rotation frequency are extracted from the de-noised spindle vibration signal. Firstly, the spindle vibration signal is decomposed by empirical wavelet transform into a set of wavelet coefficients, and then an improved thresholding function is proposed to process the Fourier spectrum of wavelet coefficients for removing the noise from the wavelet coefficients. Secondly, the wavelet coefficients are reconstructed based on the de-noised wavelet coefficients, and the wavelet coefficients which contain high multiple components of rotation frequency are selected. Finally, the energy entropy of the selected wavelet coefficients is calculated as a roundness error monitoring indicator to detect the roundness error of deep hole. The proposed method is verified with well-designed deep-hole drilling tests. The test results show that the monitoring indicator can effectively reflect the roundness error change in deep-hole drilling process.

Improved empirical wavelet denoising algorithm with application to whirling detection in deep hole drilling process

This paper investigates the empirical wavelet denoising algorithm, to extract the high multiple frequency signals of rotation frequency from the spindle vibration signal in deep hole drilling. With the characteristic of adaptive frequency partition, empirical wavelet transform is introduced to decompose the vibration signal. Considering the influence of background noise on the vibration signal, an improved threshold denoising method is proposed to remove the noise of the spindle vibration signal. Thus, an improved empirical wavelet denoising algorithm is proposed for the high multiple frequency signals of rotation frequency extraction. The improved empirical wavelet denoising algorithm is combined with energy entropy to detect whirling in deep hole drilling process.

Deep Stacked Auto-Encoder Network Based Tool Wear Monitoring in the Face Milling Process

Tool wear identification plays an important role in improving product quality and productivity in the manufacturing industry. The actual tool wear status with input cutting parameters may cause different levels of spindle vibration during the machining process. This research proposes an architecture comprising a deep learning network (DLN) to identify the actual wear state of machining tool. Firstly, data on spindle vibration signals are obtained from an acceleration sensor. The data are then pre-processed using the fast Fourier transform (FFT) method to reveal the relevant outstanding features in the frequency domain. Finally, the DLN is constructed based on stacked auto-encoders (SAE) and softmax, which is trained with the input data on the vibration features of the respective tool wear state. This DLN architecture is then used to identify the actual wear statuses of machining tool. The experimental results from the collected data show that the proposed DLN architecture is capable of identifying actual tool wear with high accuracy.

Spindle vibration signal extraction method based on improved all phase Fast Fourier Transform

In the modern processing technology, the machine tool plays an irreplaceable role as the main processing tool, and its machining accuracy directly affects the processing quality of the part. Because the spindle vibration caused by the rotor mass imbalance will seriously affect the machining accuracy of the machine tool, it is necessary to dynamically balance the spindle. The key step of dynamic balancing is to accurately extract the characteristics of the vibration signal, the phase of the vibration signal of the spindle was extracted by the all phase fast Fourier transform, the amplitude of vibration signal was extracted by cross correlation analysis, the results are applied to the influence coefficient method to realize the dynamic balance of spindle vibration The results show that the vibration signal feature extraction method combined with the results of the two extraction methods has high precision and stability.

Combining translation-invariant wavelet frames and convolutional neural network for intelligent tool wear state identification

On-machine monitoring of tool wear in machining processes has found its importance to reduce equipment downtime and reduce tooling costs. As the tool wears out gradually, the contact state of the cutting edge and the workpiece changes, which has a significant influence on the vibration state of the spindle. The performance of traditional intelligent fault diagnosis methods depend on feature extraction of dynamic signals, which requires expert knowledge and human labor. Recently, deep learning algorithms have been applied widely in machine health monitoring. In this paper, we present a novel intelligent technique for tool wear state recognition using machine spindle vibration signals. The proposed technique combines derived wavelet frames (DWFs) and convolutional neural network (CNN). Constructed based on dual tree wavelets, DWF are equipped with merits of centralized multiresolution and nearly translation-invariance. In this method, DWFs are employed to decompose the original signal into frequency bands of different bandwidths and different center frequencies, which are more pronounced as the tool wears. Further, the reconstructed sub-signals are stacked into a 2-D signal matrix to match the structure of 2-D CNN while retaining more dynamic information. The 2-D convolutional neural network is utilized to automatically recognize features from the multiscale 2-D signal matrix. End-milling experiments were performed on a S45C steel workpiece at different machining parameters. The experiment results of the recognition for tool wear state show the feasibility and effectiveness of the proposed method.

Investigation assembly state of spindle bearing based on improved maximum correlated kurtosis deconvolution and support vector machine

Aiming at the skew and misalignment between the inner and outer rings of the spindle bearing during the installation process, a new method for detecting the spindle bearing assembly state is proposed combined an improved maximum correlated kurtosis deconvolution (IMCKD) and support vector machine (SVM). The vibration and impulse signals induced by bearing assembly skew or dislocation are quite weak. In this paper, the collected vibration signals were preprocessed by IMCKD algorithm with optimized parameters between the normal and skewed state of the outer rings at first, which aims to deconvolve periodic impulse characteristic and enhance the periodic impulse component from a new assembly spindle vibration signal. Then, wavelet packet decomposition method was adopted to extract the characteristic energy value of different frequency bands, and the eigenvector of the assembly state of the spindle bearing was calculated. Furthermore, the support vector machine classification method is implemented to analyze the eigenvectors under different assembly states. Finally, the proposed method was verified by experimental results. The results show that the IMCKD and SVM methods can effectively detect the bearing assembly skew state of the spindle with an accuracy of more than 95 %.

Early Mechanical Failure Prediction Technology of CNC Machining Centers Motorized Spindle

CNC machining center spindle early mechanical failure has hidden and complex features. In this paper, PeakVue technology early detection of potential and subtle abnormal spindle vibration signal, to solve the problem of weak signal difficult to detect. In order to quickly and accurately determine the motorized spindle potential mechanical failure, this paper PSO global search capability and fast optimization speed and fuzzy neural network(FNN) fault-tolerant ability, self-adaptable characteristics of a method is proposed. The method combines fuzzy logic, RBF neural networks and FNN, spindle in order to predict an early mechanical failure of CNC machining center. In order to optimize the parameters of fuzzy neural network structure, velocity updating formula and inertia weight of PSO has been improved, so as to establish an improved PSO optimized fuzzy neural network CNC machining center spindle early failure prediction method. The results show that: the method of early mechanical failure of the spindle has higher prediction accuracy and generalization ability stronger.

Application of Mathematical Morphology Filter in Non-Stationary Signal of the Spindle Dynamic Balance Regulator Online Adjustment Process

The signal which is generated by the dynamic balance regulator for unbalanced spindle in the process of adjustment, having strong nonstationarity. The traditional signal processing methods could not deal with these signals, but also could not express the processing of signal changes. Aiming at the problem, a method combining mathematical morphology filter and the short-time Fourier transform is applied in the non-stationary signal processing in this paper. Using mathematical morphology filter could reduce the noise of spindle vibration signal, and improve the SNR. The amplitude of spindle vibration signal could be got by means of short-time Fourier transform method. Utilized the method proposed in this paper, non-stationary signal of the spindle dynamic balance regulator online adjustment process has been analysed, the results of the experiment show the feasibility of the method.

An experimental technique to detect tool–workpiece contact in micromilling

There is a need for simple, accurate, and low-cost techniques to detect tool–workpiece contact (or tool touch-off) in micromilling operations. This paper presents a method that is based on monitoring changes in the power spectral characteristics of the spindle vibration signal. The accuracy of this contact detection method is evaluated under different conditions by measuring the overshoot of the tool into the workpiece surface. Specifically, the effects of tool geometry, workpiece surface roughness and hardness, tool wear, step size, and contact detection threshold on the overshoot are analyzed through experiments carried out on a 3-axis micromilling machine. The results show that the method is capable of sub-micron contact detection accuracy depending on the workpiece hardness, roughness, and contact detection threshold.