Abstract
Fiber-optic seismic sensor has become an emerging effective tool for microseismic monitoring, which is of importance for oil and gas production improvement. When processing the fiber-optic data, the recognition of microseismic events is the first crucial step, which is rarely reported due to the lack of such data. In this paper, based on the features of central data distribution, skewness, kurtosis, energy entropy, etc., the classification accuracy of microseismic (MS) events obtained by fiber-optic MS monitoring system and electronic one is compared using machine learning algorithm (e. g. SVM and KNN) for the first time. The results show that fiber-optic data has higher classification accuracy than electronic data when using the feature of central data distribution owing to the high signal-to-noise ratio of fiber-optic data. However, when choosing the features of energy entropy, zero-crossing rate and the energy proportion of specific frequency band, electronic data has higher classification accuracy than fiber-optic data benefiting from the longer events duration and the lower frequency components of electronic data. When using skewness and kurtosis, the classification performance in fiber-optic data is almost consistent with the electronic one. Moreover, the results indicate that the characteristics of MS signal itself have a greater impact on the discrimination ability of MS events than the applied machine learning classification algorithm.
Highlights
MICROSEISMIC(MS) monitoring has a pivotal role in oil and gas industry, in which arrays of geophone are deployed in downhole to monitor the seismic wave generated by the fracturing well during the fracturing process
This paper compares the case of MS events recognition between fiber-optic MS monitoring system (FMMS) and electronic microseismic monitoring system (EMMS) using support vector machine (SVM) and KNN, respectively
The experiment results show that different monitoring systems demonstrate different characteristics in time domain and frequency domain about the MS events
Summary
MICROSEISMIC(MS) monitoring has a pivotal role in oil and gas industry, in which arrays of geophone are deployed in downhole to monitor the seismic wave generated by the fracturing well during the fracturing process. Machine learning method to recognize MS events from noise has been explored [4-6]. Pingan Peng et al proposed an automatic classification method based on a deep learning approach for MS records classifying in underground mines. The event type was correctly determined by the trained convolutional neural network (CNN) classifier with an accuracy of 98.2% [8]. Guangdong Song et al converted raw data into two-dimensional image format by STransform and used CNN to recognize the MS signal. Yumei Kang et al used the deep belief network model to identify MS events and blasting signals generated in construction processes, which took the advantage of the strong learning and feature extraction abilities of such models and achieved a classification accuracy of 94.4% [10]. Ruisheng Jia et al presented permutation entropy and SVM to detect low signal-to-noise ratio (SNR) MS events more efficiently [11]
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
