Classification and detection of cavitation, particle contamination and oil starvation in journal bearing through machine learning approach using acoustic emission signals

Poddar, Surojit; Tandon, N.

doi:10.1177/1350650121991316

Cited by 14 publications

(8 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…According to the Euclidean distance d between each sample, the distance between all samples and the cluster center is defined as the objective function J . The formulas are as follows 29,30 :

d (x_{i}, x_{j}) = {(‖‖, x_{i} goodbreak- x_{j})}^{2},

J = \sum_{i = 1}^{N} \sum_{k = 1}^{K} r_{italicik} {(‖‖, x_{i} goodbreak- μ_{k})}^{2},

where,

r_{italicik} \in \{0, 1\}

, i = 1,2…, N , k = 1, 2, …. For K , if

r_{italicik} = 1

, it means that sample

x_{i}

belongs to the k clusters, and if

j \neq k

r_{italicik} = 0

, the sample

x_{i}

belongs to only one cluster.…”

Section: Model and Ae Database Constructionmentioning

confidence: 99%

Effects of thermal aging on damage evolution behavior of glass fiber reinforced composites with multilayer graphene by acoustic emission

Wang

Zhou

Guo

et al. 2022

J of Applied Polymer Sci

View full text Add to dashboard Cite

The thermal aging behavior of glass fiber reinforced composites has attracted much attention because of its application in elevated temperatures environments. In this study, to improve the thermal aging resistance performance of composites, multilayer graphene was embedded in glass fiber reinforced composites and the damage evolution behaviors of the composites after thermal aging treatments were investigated. The mapping relationship between damage modes and acoustic emission parameters is established by principal component analysis and k-means clustering methods, then the acoustic emission data were trained by K-Nearest Neighbor algorithms to obtain a damage modes identification model with positive predictive values above 90%. The results showed that more matrix cracking signals are captured (or matrix cracking accumulation acoustic emission (AE) energy is abruptly raised) at the early stage of loading after 32 days of thermal aging. With the addition of multilayer graphene, the fiber/matrix debonding accumulation AE energy is significantly less than the matrix cracking accumulation AE energy, which is more obvious after thermal aging. Damage initiation and extension obtained by the acoustic emission events help to find the correlation between aging time and interior interface damage mechanism.

show abstract

“…According to the Euclidean distance d between each sample, the distance between all samples and the cluster center is defined as the objective function J . The formulas are as follows 29,30 :

d (x_{i}, x_{j}) = {(‖‖, x_{i} goodbreak- x_{j})}^{2},

J = \sum_{i = 1}^{N} \sum_{k = 1}^{K} r_{italicik} {(‖‖, x_{i} goodbreak- μ_{k})}^{2},

where,

r_{italicik} \in \{0, 1\}

, i = 1,2…, N , k = 1, 2, …. For K , if

r_{italicik} = 1

, it means that sample

x_{i}

belongs to the k clusters, and if

j \neq k

r_{italicik} = 0

, the sample

x_{i}

belongs to only one cluster.…”

Section: Model and Ae Database Constructionmentioning

confidence: 99%

Effects of thermal aging on damage evolution behavior of glass fiber reinforced composites with multilayer graphene by acoustic emission

Wang

Zhou

Guo

et al. 2022

J of Applied Polymer Sci

View full text Add to dashboard Cite

show abstract

“…The modeling of acoustic emission of rolling element bearings was proposed by Patil et al [21], which presented a mechanistic model of rolling bearing acoustic emission generation. Various studies [22][23][24] combined acoustic emission signals with machine learning methods for rolling bearing fault diagnosis detection. However, the large size of acoustic emission signal data means that it is difficult to process directly [25], so the features of acoustic emission signals need to be extracted and combined with suitable methods to achieve fault diagnosis [26].…”

Section: Introductionmentioning

confidence: 99%

An adaptive selective ensemble algorithm for fault classification

Zhang

Liu

et al. 2023

Meas. Sci. Technol.

View full text Add to dashboard Cite

Aiming at the problem of improving the classification performance of single Extreme Learning Machine (ELM) for fault diagnosis, an adaptive selective ensemble classification algorithm is proposed based on the idea of ensemble learning of ELM,. and it can achieve better classification performance than a single model, and the advantages of proposed algorithm is that the selected ELM subnetworks are selected adaptively and the ensemble weights of selected ELM subncetworks is calculated adaptively, and it can avoid the complex process of manually selecting subnetworks and calculating ensemble weights. Firstly, a regularized error weight adjustment ELM is constructed by introducing regularization parameters and error weight adjustment matrix to the standard ELM, where the error weight adjustment matrix is obtained by the method of adaptive gaussian kernel density estimation. Then, the discrepancy subnetworks are constructed by six different activation functions, and ensemble weights of subnetworks are obtained adaptively according to normalized ratio of mean and variance of subnetwork F-scores to achieve ensemble of subnetworks. Finally, the adaptive selective ensemble classification algorithm is validated with UCI dataset and experimental acoustic emission signals of gearbox faults, the results show that the adaptive selective ensemble method can improve the stability and accuracy of classification algorithm, and the classification accuary of experimental acoustic emission was achieved at 0.9773.

show abstract

“…Rossopoulos and Papadopoulos 55 proposed a methodology for predictive analytics that employs ML algorithms to assess the performance condition of marine journal bearings, focusing on factors such as maximum pressure, minimum film thickness, Sommerfeld number, load, and shaft speed. In their study, Poddar and Tandon 56 utilized ML techniques and acoustic emission signals to classify and detect cavitation, particulate contamination, and oil deficiency in journal bearings. Mohammed et al 57 employed ML techniques to estimate the FC of Su-8 and its composite coatings.…”

Section: Introductionmentioning

confidence: 99%

Prediction of friction coefficient and torque in self-lubricating polymer radial bearings produced by additive manufacturing: A machine learning approach

Baş,

Karabacak

2023

Proceedings of the Institution of Mechanical Engineers, Part J:

View full text Add to dashboard Cite

Additive manufacturing is a rapidly developing technology that enables the production of complex parts with intricate geometries. Self-lubricating radial bearings are one of the machine elements that can be produced using additive manufacturing. In this research, we present a machine learning-based approach to model the friction coefficient and friction torque in self-lubricating radial bearings manufactured by additive manufacturing using polyether ether ketone (PEEK), polylactic acid (PLA), acrylonitrile butadiene styrene (ABS), and nylon. The proposed approach includes different machine learning models (artificial neural networks, support vector machines, regression trees, linear regression models) that utilize experimental data to predict the coefficient of friction and friction torque of different types of radial bearings. Experimental data were obtained by performing tribological tests on self-lubricating radial bearings under various operating conditions. The results reveal that the machine learning models are successful in predicting the friction coefficient and friction torque in self-lubricating radial bearings with high accuracy. The approach can be utilized to optimize the design and performance of self-lubricating radial bearings manufactured using additive manufacturing.

show abstract

Classification and detection of cavitation, particle contamination and oil starvation in journal bearing through machine learning approach using acoustic emission signals

Cited by 14 publications

References 20 publications

Effects of thermal aging on damage evolution behavior of glass fiber reinforced composites with multilayer graphene by acoustic emission

Effects of thermal aging on damage evolution behavior of glass fiber reinforced composites with multilayer graphene by acoustic emission

An adaptive selective ensemble algorithm for fault classification

Prediction of friction coefficient and torque in self-lubricating polymer radial bearings produced by additive manufacturing: A machine learning approach

Contact Info

Product

Resources

About