On-line Visual Ferrograph Research Articles

An improved YOLOv8-CGA-ASF-DBB method for multi-class wear debris recognition of online visual ferrograph image

Abstract The analysis of wear based on On-Line Visual Ferrograph (OLVF) provides crucial insights for the analysis of wear faults in mechanical equipment.However, online ferrograph analysis has been greatly limited by the low imaging quality and recognition accuracy of particle chains and high hubbles when analyzing lubricant oils in practical applications. To address this issue,this paper proposes an enhanced OLVF wear image detection model based on YOLOv8 and applies it to the multi-class intelligent recognition of ferrograph images .The Cascade Group Attention(CGA) module is introduced enhance the diversity of features and improve computational efficiency. The fusion of attention scale sequence is introduced to achieve precise and rapid recognition of small targets. This Diverse Branch Block(DBB) module is introduced efficiently extracts features without compromising reasoning speed during training. For verification , a test of the bridge transmission box was conducted based on OLVF, and 992 ferrograph images were collected. Experimental results reveal that the improved algorithm achieves an accuracy of 94.53% on the dataset of bridge transmission box ferrograph wear debris images collected through OLVF. This represents a 5.2% increase in recognition accuracy compared to the original algorithm, while maintaining a processing time of only 0.69ms per image. These findings provide compelling evidence for the significant enhancements in both recognition accuracy and processing speed achieved by the improved algorithm, thereby establishing its considerable value for engineering applications.

Real-Time Ferrogram Segmentation of Wear Debris Using Multi-Level Feature Reused Unet.

The real-time monitoring and fault diagnosis of modern machinery and equipment impose higher demands on equipment maintenance, with the extraction of morphological characteristics of wear debris in lubricating oil emerging as a critical approach for real-time monitoring of wear, holding significant importance in the field. The online visual ferrograph (OLVF) technique serves as a representative method in this study. Various semantic segmentation approaches, such as DeepLabV3+, PSPNet, Segformer, Unet, and other models, are employed to process the oil wear particle image for conducting comparative experiments. In order to accurately segment the minute wear debris in oil abrasive images and mitigate the influence of reflection and bubbles, we propose a multi-level feature reused Unet (MFR Unet) that enhances the residual link strategy of Unet for improved identification of tiny wear debris in ferrograms, leading to superior segmentation results.

Short-Term Cross-Sectional Time-Series Wear Prediction by Deep Learning Approaches

Abstract Wear is one of the major causes that affect the performance and reliability of tribo-systems. To mitigate its adverse effects, it is necessary to monitor the wear progress so that preventive maintenance can be timely scheduled. An online visual ferrograph (OLVF) apparatus is used to obtain online measurements of wear particle quantities, and monitor the wearing of a four-ball tribometer under different lubrication conditions, and several popular deep learning algorithms are evaluated for their effectiveness in providing maintenance decisions. The obtained data are converted to the cross-sectional time series (CSTS), for its effectiveness in representing the variation trends of multiple variables, and the data are used as the input to the deep learning algorithms. Experimental results indicate that the CSTS together with the bidirectional long short-term memory (Bi-LSTM) architecture outperforms other tested settings in terms of the mean-squared error (MSE). Increased prediction accuracy is observed for tribological pairs with a stochastically changing coefficient of friction.

Lubricating oil deterioration modeling and remaining useful life prediction based on hidden semi-Markov modeling

Lubricating oil, which carries information about machine’s health condition, is of great importance to the performance of machines in the full life cycle. The main purpose of oil deterioration modeling and its remaining useful life prediction is to determine the exact time that the lubricating oil has degraded and it is no longer able to maintain its functions. Generally, lubricating oil deterioration can be partially detected by condition monitoring based on wear debris analysis, and thus can be categorized into three states. In our paper, vector data, which contain wear debris concentration and carry information about the state of lubricating oil, are obtained by an on-line visual ferrograph sensor from a four-ball tester at regular sampling epochs. The oil’s state process is described by a hidden semi-Markov model, and its sojourn times in each state are assumed to be Erlang distributed. A vector autoregressive method based on time series modeling is presented to obtain residual observations, which are regarded as the observable process of oil information in the hidden semi-Markov model framework. The unknown parameters of the hidden semi-Markov model are then estimated by using expectation-maximization algorithm. Afterward, a Bayesian updating approach is presented to derive the explicit formulas of the conditional reliability and mean residual life. To validate the proposed approach, a real case study of lubricating oil deterioration is demonstrated and a comparison with the hidden Markov model is given to illustrate the effectiveness of the new developed remaining useful life prediction approach for lubricating oil.

A Full Field-of-View Online Visual Ferrograph Debris Detector Based on Reflected Light Microscopic Imaging

For improving the detection accuracy of wear debris visual information for online wear monitoring, a full field-of-view Online Visual Ferrograph (OLVF) debris detector is presented. In current OLVF debris detector, a semi-transparent optical diffusion film is clamped into the gap between the bottom optical quartz window and two magnetic poles as the background of full field-of-view OLVF microscopic imaging, by which the emitted light of circular ring LED array is diffusely reflected. A reflected light microscopic imaging system is developed to form the full field-of-view imaging of OLVF, and a mathematical relationship between its depth of field and the oil-pumping and oil-free axial deviations of CMOS image plane is established, by which the oil chamber height can be determined. An optical focusing mechanism is applied to adjusting the optical magnification and the optical resolution of about 7.2 μm is derived under 2.0× optical magnification. Moreover, a microscopic imaging model is proposed to evaluate the characteristics of full field-of-view OLVF microscopic imaging. The simulation results show that the high-quality microscopic imaging can be obtained near the optical axis of CMOS image plane under 2.0× optical magnification and oil attenuation coefficient of less than 2.0. Two experiments were conducted to validate the microscopic imaging performance and effectiveness of the current detector. The results show that the detector can be applied for capturing the full field-of-view reflected ferrogram of wear debris from different in-use lube oils, and it can also detect debris concentration more accurately for monitoring the fast-changing wear of the machines.

Running-in real-time wear generation under vary working condition based on Gaussian process regression approximation

In real-time lubricant monitoring operation, the wear’s generation inside the flow lubricant is identified as the lubricant’s performance indicator. The online visual ferrograph (OLVF) based on magnetic deposition and image analysis could detect the wear during the running machine process. However, the time gap during OLVF’s data processing makes the corrective actions of abnormal high wear rate are not based on time-sensitive data. Therefore, this study proposes a data-driven approach based on Gaussian process regression approximation to analyze the acquired data and thus create new knowledge on the understanding of real-time wear generation. Moreover, running-in processes with high seasonality (abnormal) data are being investigated to give insight into the initial process’s phenomena. In this study, the combined temperature-controlled four-ball tribometer with an OLVF is employed to acquire debris data inside the flow channel, and the coefficient of friction (COF) represents the wear trend under varying working conditions. Results show that the experimental setup and the approach can be used to investigate running-in real-time wear generation.

A Segmentation Algorithm of Wear Debris Reflected image Based on Watershed and H-minima Transform for On-Line Visual Ferrograph Analysis

Aiming to solve the problems of identifying the over-brightness and over-dark parts in the reflected light wear debris image, this paper proposed a new OLVF (on-line visual ferrgraph) reflected light wear debris segmentation algorithm. If the wear debris image is bright and dark, the morphological black-hat operation is used to eliminate these interferences. Then the H-minima transform is applied to corrected image local extreme value interference and the threshold is obtain through Otsu algorithm. The watershed algorithm is adopted to separate over-bright and over-dark wear debris. Eventually the morphological dilate corrosion of opening and closing reconstruction is applied to filled holes and connected wear debris profile. On the basis of the above-mentioned method, the OLVF reflected light wear debris are accurate shape and continuous profile in the final result. Compared with other wear debris divide algorithms, the proposed method effectively suppresses the influence of reflected light, and better retain the information of wear debris image. At the same time, the signal-to-noise ratio is 12dB higher than others at the maximum, and the root mean square error is reduced by 0.051 at the maximum.

Wear Debris Segmentation of Reflection Ferrograms Using Lightweight Residual U-Net

Extraction of lube oil wear debris morphological features is an important means for real-time monitoring of equipment wear, and online visual ferrograph (OLVF) is one of the representative technologies. In the current OLVF wear monitoring, the transmission ferrogram (TF) is basically relied on, but the more informative reflection ferrogram (RF) has not yet been applied, because its complex surface color distribution and bubble interference make it difficult to segment the RF. Accordingly, a convolutional neural network (CNN) model called lightweight residual U-net (Res-UNet) is constructed in this article. Simultaneously, with both RFs and TFs, an automatic labeling method is proposed to label the RFs and make a training dataset to implement network training. The experimental results demonstrate that the trained network can achieve accurate segmentation of RFs with excellent anti-interference performance. The proposed method lays the foundation for the feature extraction of reflection OLVF ferrograms and provides an alternative image segmentation method for other image wear debris sensors.

Method of microscopic imaging quality evaluation for board lens based on image parameter extraction of pinhole imaging

The modulation transfer function area (MTFA) of pinhole imaging and the MTFA relative variation ratio were used as the image surface parameters to quantitatively evaluate the microscopic imaging quality of the board lens. The light spot images located in different fields of view positions were obtained by using the pinhole imaging method to extract the modulation transfer function (MTF) of microscopic imaging system for board lens. The MTF mathematical model of lens system for board lens was established, and the MTFA in sagittal direction and meridian direction of edge field of view as well as the MTFA relative variation ratio were extracted to measure the microscopic imaging definition and image surface flatness of the measured lens. An experiment which was for measuring the board lens was carried out, and the microscopic imaging performance of the measured lens was quantitatively evaluated by using the proposed model and method. The results show that the Lens 2 can obtain the optimal microscopic imaging quality. The definition value of average power spectrum value (APSV), the gray mean gradient (GMG) and the laplacian summation (LS) were calculated respectively from the RGB bitmap images captured by these measured lens. The results show that the maximum parameter value of the APSV, GMG and LS of Lens 2 bitmap images is equal to 2.720 2, 17.024 4 and 94.921 2, respectively. At the same time, the bitmap images have the highest definition, which is consistent with the image surface parameter evaluation results, indicating that the proposed method used in the quantitative evaluation of microscopic imaging performance of board lens is accurate and effective, which is of great significance and engineering value to improve the online image probe design of the on-line visual ferrograph (OLVF).

Wear Performance of GCr15 Friction Pairs with Effect of Initial Radial Micro-Grooves

Background: Grooves may inevitably occur on the surface of the friction pair caused by severe wear or residual stress, which will play an important role on the reliability of machine parts during operation. Objective: The effect of the micro-grooves perpendicular to sliding direction on the wear performance of the friction pairs should be studied. Method: Micro-grooves can be machined on discs of friction pairs using electrical discharge machining. On-line visual ferrograph method was used to monitor the wear process to research the wear rate changing characteristic. Profilemeter and metallurgical microscope were used to observe the wear scars. Results: Comparing to the non-groove test, i) in one-groove test, wear volume and rate were approximate the same, and the wear scar was smooth, ii) when the grooves more than 4, the test running-in stage will be obviously prolonged, particularly for the test with 8 grooves on the disc, the duration of running-in stage is 4 times than that without grooves on specimen, and the wear rate and volume increase significantly, and then decrease with fluctuation, iii) the abrasive wear can be avoid with the debris stagnating in the groove, however, fatigue wear will significantly emerge. Conclusion: Abrasive wear can be avoided and smooth running-in surfaces can be obtained with proper amount of initial radial micro-grooves.

Segmentation of Online Ferrograph Images with Strong Interference Based on Uniform Discrete Curvelet Transformation.

Through real-time acquisition of the visual characteristics of wear debris in lube oil, an on-line visual ferrograph (OLVF) achieves online monitoring of equipment wear in practice. However, since a large number of bubbles can exist in lube oil and appear as a dynamically changing interference shadow in OLVF ferrograms, traditional algorithms may easily misidentify the interference shadow as wear debris, resulting in a large error in the extracted wear debris characteristic. Based on this possibility, a jam-proof uniform discrete curvelet transformation (UDCT)-based method for the binarization of wear debris images was proposed. Through multiscale analysis of the OLVF ferrograms using UDCT and nonlinear transformation of UDCT coefficients, low-frequency suppression and high-frequency denoising of wear debris images were conducted. Then, the Otsu algorithm was used to achieve binarization of wear debris images under strong interference influence.

A Wear Debris Segmentation Method for Direct Reflection Online Visual Ferrography.

Wear debris in lube oil was observed using a direct reflection online visual ferrograph (OLVF) to monitor the machine running condition and judge wear failure online. The existing research has mainly concentrated on extraction of wear debris concentration and size according to ferrograms under transmitted light. Reports on the segmentation algorithm of the wear debris ferrograms under reflected light are lacking. In this paper, a wear debris segmentation algorithm based on edge detection and contour classification is proposed. The optimal segmentation threshold is obtained by an adaptive canny algorithm, and the contour classification filling method is applied to overcome the problems of excessive brightness or darkness of some wear debris that is often neglected by traditional segmentation algorithms such as the Otsu and Kittler algorithms.

A direct reflection OLVF debris detector based on dark-field imaging

To solve the problems of monitoring wear debris in black oil, a direct reflection online visual ferrograph (OLVF) debris detector is presented. In current OLVF detectors, a reflected light source is used. The emitted light is reflected by wear debris directly instead of passing through the lube oil. Therefore, the transparency of the lube oil ceases to matter. Two experiments were conducted to validate the wear debris imaging feasibility and effectiveness of the newly developed detector. The results show that the visual feature information of the wear debris can be reliably obtained from black oil by this detector, and it can also be used to track the fast-changing wear of tribopairs at different wear stages. To the best of our knowledge, to date there is no other report for solving this issue.

Prediction of wear trend of engines via on-line wear debris monitoring

Abstract On-line wear debris monitoring is a useful technology for real-time machine wear condition monitoring but needs further development. This study, based on previous developments of an on-line visual ferrograph (OLVF), focused on (i) data reconstruction for extracting representative and reliable wear condition related characteristics, and (ii) development of an improved model for on-line wear prediction. Wear monitoring of a diesel engine was performed using this on-line wear debris monitoring system. Experimental results and comparisons between the improved relevance vector machine (RVM) model and other models show that the improved RVM model gives an earlier warning and enhances the prediction accuracy.

Parameter estimation and remaining useful life prediction of lubricating oil with HMM

Lubricating oil plays a vital role in the full life-span performance of the machine. Lubricating oil deterioration, which leads to the attenuation of oil performance and severe wear afterwards, is a slow degrading process, which can be observed by condition monitoring, but the actual degree of the oil degradation is often very difficult to examine. The main purpose of lubricating oil degradation prediction is to estimate the failure time when the oil no longer fulfills its functions. We suppose that the state process evolution of lubricating oil degradation can be modeled using a hidden Markov model (HMM) with three states: healthy state, unhealthy state, and failure state. Only the failure state is observable. While the lubricating oil is in service, vector data that are stochastically related to the deterioration state are obtained through on-line condition monitoring by an OLVF (On-line Visual Ferrograph) sensor at regular sampling epochs. A method of Time Series Analysis (TSA) is applied to the healthy portions of the oil data histories to get the residuals as the observable process containing partial information to fit the hidden Markov model. The unknown parameters of the fitted hidden Markov model are estimated by the Expectation-Maximization (EM) algorithm. The remaining useful life (RUL) of lubricating oil can be evaluated through explicit formulas of the characteristics such as the conditional reliability function (CRF) and mean residual life (MRL) function in terms of the posterior probability.

Modeling and experimental investigations on the relationship between wear debris concentration and wear rate in lubrication systems

The aim of this work is to map the wear debris concentration with wear rate. A wear debris attenuation function is proposed to depict quantitatively the removal of wear debris. The total residual mass of wear debris in lubrication systems can be stated as the convolution of the wear rate and the wear debris attenuation function. Additionally, the dynamic responses of the wear debris concentration under different wear rate excitations are also presented. To simulate and verify how the debris concentration changes, an oil lubrication system test rig is built, and on-line visual ferrograph (OLVF) is used to monitor the debris concentration. Two experimental results indicate that the variation of the wear debris concentration is consistent with the model result.

An oil monitoring method of wear evaluation for engine hot tests

Abnormal wear of a piston ring-cylinder liner pair may happen after 9 min hot tests of internal combustion engines, while the engine performance parameters were within predetermined threshold ranges. Few differences were observed among oil samples from the engines with or without abnormal wear in the spectrometric and Kittiwake Analex PQ analysis. Therefore, a manual confirmation by disassembling the oil pan was often required. In this work, an oil monitoring method for wear evaluation of the engines was proposed. The oil samples were rapidly analyzed on site by on-line visual ferrograph (OLVF). For the abnormal engines, it was found that the index of particle coverage area (IPCA), characterizing the wear debris concentration, was low. Moreover, large debris was rarely observed on OLVF ferrograms, which was consistent with the results obtained from analytical ferrography, and the reason was analyzed and discussed. In addition, an on-site abnormal wear evaluation procedure for the 9 min hot tests was proposed based on a trained Naive Bayes Classifier. As observed from the results of 27 engines, 4 abnormal engines were found. Among one of them, longitudinal scratches were found on the cylinder wall, which were evaluated as abnormal wear by the classifier. This method can cut down the quantity of disassembly inspection and is more efficient.

在线图像可视铁谱成像系统的像面照度均匀性

在线图像可视铁谱成像系统的像面照度均匀性

Prediction on wear of a spur gearbox by on-line wear debris concentration monitoring

The aim of this work is to develop a method for quantifying and predicting gear wear based on on-line wear monitoring. A model of wear debris concentration has been built based on Kragelsky's method with different wear coefficients in corresponding wear stages. A gear test rig with oil-bath lubrication for accelerated wear test was built, and a full-life wear monitoring test was performed by employing an on-line visual ferrograph (OLVF). An index of particle coverage area (IPCA) characterizing wear debris concentration and an OLVF ferrogram were obtained by sampling in-use oil every 2min. The experiment results indicate that the IPCA curve is consistent with the proposed model, and the early-warning signs of abnormalities can also be observed. Additionally, depositing experiments show that appropriate depositing time (usually not less than 30s) is crucial for a valid OLVF sampling. Therefore, it is feasible to predict gear wear by on-line wear debris concentration monitoring.

Correction strategies of debris concentration for engine wear monitoring via online visual ferrograph

Changing or replenishing the oil in an engine influences the accuracy of oil monitoring. An online visual ferrograph (OLVF) was used to monitor engine wear in this study. The variation in the debris concentration in an engine lubrication system was studied, and its equilibrium concentration of debris was determined to correct the data acquired using an OLVF after fresh oil was added to the system. This correction made the monitoring data of OLVF continuous, thereby enabling the raw data collected before and after oil replenishment to be compared. Changing the oil caused an interruption in the raw data. By combining different prediction modes, a comprehensive trend prediction framework was constructed to correct the data acquired after the engine’s oil was changed. Thus, the data acquired before and after the oil change could be analysed and compared. In addition, the engine’s whole wear trends could be observed. This correction strategy could also be used to assess the engine’s health and to predict its degradation trend.