Automatic fault detection in seismic data using Gaussian process regression

Noori, Maryam; Hassani, Hossein; Javaherian, Abdolrahim; Amindavar, Hamidreza; Torabi, Siyavash

doi:10.1016/j.jappgeo.2019.02.018

Cited by 22 publications

(9 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The basic method is to apply a Gaussian distribution model. 5–8 The method demonstrates good results with a small amount of input data. Another approach to the problem of anomaly detection is the application of autoencoders (AE).…”

Section: Introductionmentioning

confidence: 94%

Fault diagnosis systems for rotating machines operating with fluid-film bearings

Stebakov

Kornaev

Popov

et al. 2022

Proceedings of the Institution of Mechanical Engineers, Part J:

View full text Add to dashboard Cite

The paper deals with the application of deep learning methods to rotating machines fault diagnosis. The main challenge is to design a fault diagnosis system connected with multisensory measurement system that will be sensitive and accurate enough in detecting weak changes in rotating machines. The experimental part of the research presents the test rig and results of high-speed multisensory measurements. Six states of a rotating machine, including a normal one and five states with loosened mounting bolts and small unbalancing of the shaft, are under study. The application of deep network architectures including multilayer perceptron, convolutional neural networks, residual networks, autoencoders and their combination was estimated. The deep learning methods allowed to identify the most informative sensors, then solve the anomaly detection and the multiclass classification problems. An autoencoder based on ResNet architecture demonstrated the best result in anomaly detection. The accuracy of the proposed network is up to 100% while the accuracy of an expert is up to 65%. A one-dimensional convolutional neural network combined with a multilayer perceptron that contains a pretrained encoder demonstrated the best result in multiclass classification. The detailed fault detection accuracy with the determination of the specific fault is 83.3%. The combinations of known deep network architectures and application of the proposed approach of pretraining of the encoders together with using a block of inputs for one prediction demonstrated high efficiency.

show abstract

Section: Introductionmentioning

confidence: 94%

Fault diagnosis systems for rotating machines operating with fluid-film bearings

Stebakov

Kornaev

Popov

et al. 2022

Proceedings of the Institution of Mechanical Engineers, Part J:

View full text Add to dashboard Cite

show abstract

“…Not only can it be stated that GPR could represent a valuable pre-trenching tool for paleo-seismic surveys, but also that the technique has shown great potential in tracing fault segments along strikes, assessing variations in their displacements, and, hence, determining their chronology. Within this domain, geometric and morphological attributes have been widely developed to highlight the features of fault zones that are hardly visible in standard GPR profiles [239,240].…”

Section: Gpr Applications In Geologymentioning

confidence: 99%

From Its Core to the Niche: Insights from GPR Applications

2022

View full text Add to dashboard Cite

Thanks to its non-destructive, high-resolution imaging possibilities and its sensitivity to both conductive and dielectric subsurface structures, Ground-Penetrating Radar (GPR) has become a widely recognized near-surface geophysical tool, routinely adopted in a wide variety of disciplines. Since its first development almost 100 years ago, the domain in which the methodology has been successfully deployed has significantly expanded from ice sounding and environmental studies to precision agriculture and infrastructure monitoring. While such expansion has been clearly supported by the evolution of technology and electronics, the operating principles have always secured GPR a predominant position among alternative inspection approaches. The aim of this contribution is to provide a large-scale survey of the current areas where GPR has emerged as a valuable prospection methodology, highlighting the reasons for such prominence and, at the same time, to suggest where and how it could be enhanced even more.

show abstract

“…(1) Regression algorithm GPR (Mahmoodzadeh et al 2021;Noori et al 2019) has good adaptability and strong generalization ability to address high-dimensional, small-sample, nonlinear, and complex problems. Compared with neural network and SVM, this method has the advantages of easy implementation and adaptive acquisition of superparameters.…”

Section: Selection Of Prediction Algorithmmentioning

confidence: 99%

Cooperative prediction method of gas emission from mining face based on feature selection and machine learning

Zhou

Lin

Jin

et al. 2022

Int J Coal Sci Technol

View full text Add to dashboard Cite

Collaborative prediction model of gas emission quantity was built by feature selection and supervised machine learning algorithm to improve the scientific and accurate prediction of gas emission quantity in the mining face. The collaborative prediction model was screened by precision evaluation index. Samples were pretreated by data standardization, and 20 characteristic parameter combinations for gas emission quantity prediction were determined through 4 kinds of feature selection methods. A total of 160 collaborative prediction models of gas emission quantity were constructed by using 8 kinds of classical supervised machine learning algorithm and 20 characteristic parameter combinations. Determination coefficient, normalized mean square error, mean absolute percentage error range, Hill coefficient, mean absolute error, and the mean relative error indicators were used to verify and evaluate the performance of the collaborative forecasting model. As such, the high prediction accuracy of three kinds of machine learning algorithms and seven kinds of characteristic parameter combinations were screened out, and seven optimized collaborative forecasting models were finally determined. Results show that the judgement coefficients, normalized mean square error, mean absolute percentage error, and Hill inequality coefficient of the 7 optimized collaborative prediction models are 0.969–0.999, 0.001–0.050, 0.004–0.057, and 0.002–0.037, respectively. The determination coefficient of the final prediction sequence, the normalized mean square error, the mean absolute percentage error, the Hill inequality coefficient, the absolute error, and the mean relative error are 0.998%, 0.003%, 0.022%, 0.010%, 0.080%, and 2.200%, respectively. The multi-parameter, multi-algorithm, multi-combination, and multi-judgement index prediction model has high accuracy and certain universality that can provide a new idea for the accurate prediction of gas emission quantity.

show abstract

Automatic fault detection in seismic data using Gaussian process regression

Cited by 22 publications

References 23 publications

Fault diagnosis systems for rotating machines operating with fluid-film bearings

Fault diagnosis systems for rotating machines operating with fluid-film bearings

From Its Core to the Niche: Insights from GPR Applications

Cooperative prediction method of gas emission from mining face based on feature selection and machine learning

Contact Info

Product

Resources

About