Feature Extraction and Analysis of Microseismic Signal Based on Convolutional Neural Network

Seismic detection technology has been widely used in safety detection of engineering construction abroad. Although it has just started in the field of engineering in our country, its role is becoming more and more important. Through computer technology, micro-seismic detection can provide accurate data for the construction safety detection of large-scale projects, which has important practical significance for the rapid and effective identification of micro-seismic signals. Based on this, the purpose of this article is to study the feature extraction and classification of microseismic signals based on neural games. This article first summarizes the development status of microseismic monitoring technology. Using traditional convolutional neural networks for analysis, a multi-scale feature fusion network is proposed on the basis of convolutional neural networks and big data, the multi-scale feature fusion network is used to research and analyze microseismic feature extraction and classification. This article systematically explains The principle of microseismic signal acquisition and the construction of multi-scale feature fusion network. And use big data, comparative analysis method, observation method and other research methods to study the theme of this article. Experimental research shows that the db7 wavelet base has little effect on the Megatron signal.

Experimental Investigations on Charge Induction and Microseismic Characteristics of the Coal and Rock under Different Loading Rates

The research on charge induction and microseismic characteristics of coal and rock under different loading rates is of great significance for rockburst prediction. In this study, the coal and sandstone samples from the No. 11 mine of Pingdingshan Coal Mine are prepared. The charge induction and microseismic synchronous comprehensive monitoring system is built. The uniaxial compression tests of coal and sandstone samples under the different loading rates are conducted. The charge induction and microseismic signal characteristics in the deformation and fracture process of the coal and rock under the different loading rates are studied. The results show that, with the increase of loading rate, the compressive strength of the coal and rock samples increases and the time from the peak stress to instability failure becomes shorter. At the same loading rate, the softening failure stage time of coal is longer than that of sandstone. With the increase of loading rate, the duration of charge-induced signal and microseismic signal is longer and the events’ number and amplitude of charge signal and microseismic signal increase in the deformation and fracture process of the coal and rock. Before the instability failure, the charge-induced signal and microseismic signal have both synchronous and asynchronous signals, and the amplitude of charge-induced and microseismic signals in each channel is different, which is related to the distance from the position of each sensor to the fracture point of the sample. During the instability failure, the charge induction and microseismic signals of each channel are generated synchronously, and the signal amplitude reaches the maximum values of 50 pC and 6 × 10−3 m/s at the same time. With the increase of specimen stress, the dominant frequency of microseismic signals first increases and then decreases, while the amplitude of dominant frequency increases synchronously. The dominant frequency amplitude of microseismic signals is the largest in instability failure. With the increase of loading rate, the spectrum amplitude of microseismic signals changes little in the compaction stage, but the spectrum amplitude increases in other stages. At the same loading speed, the events’ number of the microseismic signal of coal samples after peak stress is more than that of sandstone samples, and the signal amplitude is also larger. However, the spectrum distribution range of microseismic signals of coal samples is wider than that of sandstone samples, and the spectrum amplitude of coal samples is lower than that of sandstone. With the increase of loading rate, the time of the first generation of high-amplitude signals is advanced, and the stress of specimen becomes smaller when the first generation of high-amplitude signals occurs. With the increase of loading rate, the duration of microseismic and charge signal is longer, and the mean square amplitude of charge signal is larger.

Machine Learning Based Identification of Microseismic Signals Using Characteristic Parameters

Microseismic monitoring system is one of the effective means to monitor ground stress in deep mines. The accuracy and speed of microseismic signal identification directly affect the stability analysis in rock engineering. At present, manual identification, which heavily relies on manual experience, is widely used to classify microseismic events and blasts in the mines. To realize intelligent and accurate identification of microseismic events and blasts, a microseismic signal identification system based on machine learning was established in this work. The discrimination of microseismic events and blasts was established based on the machine learning framework. The microseismic monitoring data was used to optimize the parameters and validate the classification methods. Subsequently, ten machine learning algorithms were used as the preliminary algorithms of the learning layer, including the Decision Tree, Random Forest, Logistic Regression, SVM, KNN, GBDT, Naive Bayes, Bagging, AdaBoost, and MLP. Then, training set and test set, accounting for 50% of each data set, were prospectively examined, and the ACC, PPV, SEN, NPV, SPE, FAR and ROC curves were used as evaluation indexes. Finally, the performances of these machine learning algorithms in microseismic signal identification were evaluated with cross-validation methods. The results showed that the Logistic Regression classifier had the best performance in parameter identification, and the accuracy of cross-validation can reach more than 0.95. Random Forest, Decision Tree, and Naive Bayes also performed well in this data set. There were some differences in the accuracy of different classifiers in the training set, test set, and all data sets. To improve the accuracy of signal identification, the database of microseismic events and blasts should be expanded, to avoid the inaccurate data distribution caused by the small training set. Artificial intelligence identification methods, including Random Forest, Logistic Regression, Decision Tree, Naive Bayes, and AdaBoost algorithms, were applied to signal identification of the microseismic monitoring system in mines, and the identification results were consistent with the actual situation. In this way, the confusion caused by manual classification between microseismic events and blasts based on the characteristics of waveform signals is solved, and the required source parameters are easily obtained, which can ensure the accuracy and timeliness of microseismic events and blasts identification.

Multi-Classification of Complex Microseismic Waveforms Using Convolutional Neural Network: A Case Study in Tunnel Engineering

Due to the complexity of the various waveforms of microseismic data, there are high requirements on the automatic multi-classification of such data; an accurate classification is conducive for further signal processing and stability analysis of surrounding rock masses. In this study, a microseismic multi-classification (MMC) model is proposed based on the short time Fourier transform (STFT) technology and convolutional neural network (CNN). The real and imaginary parts of the coefficients of microseismic data are inputted to the proposed model to generate three classes of targets. Compared with existing methods, the MMC has an optimal performance in multi-classification of microseismic data in terms of Precision, Recall, and F1-score, even when the waveform of a microseismic signal is similar to that of some special noise. Moreover, semisynthetic data constructed by clean microseismic data and noise are used to prove the low sensitivity of the MMC to noise. Microseismic data recorded under different geological conditions are also tested to prove the generality of the model, and a microseismic signal with Mw ≥ 0.2 can be detected with a high accuracy. The proposed method has great potential to be extended to the study of exploration seismology and earthquakes.

Microseismic Signal Classification Based on Artificial Neural Networks

The classification of multichannel microseismic waveform is essential for real-time monitoring and hazard prediction. The accuracy and efficiency could not be guaranteed by manual identification. Thus, based on 37310 waveform data of Junde Coal Mine, eight features of statistics, spectrum, and waveform were extracted to generate a complete data set. An automatic classification algorithm based on artificial neural networks (ANNs) has been proposed. The model presented an excellent performance in identifying three preclassified signals in the test set. Operated with two hidden layers and the Logistic activation function, the multiclass area under the receiver operating characteristic curve (AUC) reached 98.6%.

Integrated processing method for microseismic signal based on deep neural network

Summary Denoising and onset time picking of signals are essential before extracting source information from collected seismic/microseismic data. We proposed an advanced deep dual-tasking network (DDTN) that integrates these two procedures sequentially to achieve the optimal performance. Two homo-structured encoder-decoder networks with specially designed structures and parameters are connected in series for handling the denoising and detection of microseismic signals. Based on the similarity of data types, the output of the denoising network will be imported into the detection network to obtain labels for the signal duration. The procedures of denoising and duration detection can be completed in an integrated way, where the denoised signals can improve the accuracy of onset time picking. Results show that the method has a good performance for the denoising of microseismic signals that contain various types and intensities of noise. Compared with existing methods, DDTN removes the noise with a minor waveform distortion. It is ideal for recovering the microseismic signal while maintaining a good capacity for onset time picking when the signal-to-noise ratio is low. Based on that, the second network can detect a more accurate duration of microseismic signals and thus obtain more accurate onset time. The method has great potential to be extended to the study of exploration seismology and earthquakes.