| With the rapid development of earthquake related work in China,the seismic event signals obtained in various places are increasing.Earthquakes continue to threaten the lives and property of people all over the world.Seismic signal classification-that is,the classification of signal source type,different kinds of earthquake caused by the vibration signal response time,energy,are not the same.Classification is difficult,traditional seismic monitoring system can only analyze the low-dimensional characteristics of waveform,so as to identify and classify seismic signals.With the rise of artificial intelligence,big data and other emerging technologies,it is increasingly important to adopt the latest information technology to replace traditional methods and develop efficient,high-precision and universal event classification algorithms in the era of seismic big data.Therefore,a seismic event classification system based on wavelet packet decomposition and deep learning is proposed to solve these problems.This paper reviews and analyzes a large number of domestic and foreign literatures and finds some deficiencies.The existing seismic monitoring system has a comprehensive structure and complete functions,including data reception,data query and time pickup,etc.but few functions such as event classification are involved.This paper proposes a seismic event classification system based on deep learning,which includes system management module,data processing module,pickup module and event classification module.Aiming at the problem of seismic event classification,a deep learning model based on CNN-RNN based on wavelet packet decomposition is proposed.Feature extraction uses convolutional neural network input to the recurrent neural network to extract its time feature.Experimental results show that this method has good precision and anti-noise performance performance of the two algorithms are improved to a certain extent compared with the traditional ones,which provides a certain practical significance for the study of seismic event classification.The seismic event classification system is developed based on CS architecture.On this basis,Python,My SQL and other related languages are used as the main development languages.In addition to the classification function,it also contains other functions such as filtering and pickup on time.In reality,different stations have different sampling points for different seismic events.In the existing classification algorithms,they basically intercept fixed points.In this paper,The same sampling points are realized by wavelet packet decomposition and reconstruction under the premise of keeping seismic event characteristics.According to the requirements and detailed design,The function points of each module are realized and tested to ensure that the The The system has achieved the expected design goals and met the needs of users,thus ensuring that the subsequent development and maintenance of the system is more efficient. |
