Data Realtime Sensing And Adaptive Processing Methods For Wireless Microseismic Monitoring

With the advantages of long-distance,three-dimensional and realtime capacity,microseismic monitoring technology has become one of the most effective disaster prevention and early warning methods in underground engineerings,such as tunneling and mining.Most existing monitoring systems still use wired acquisition schemes,resulting in poor system maintainability,portability,and scalability,which is not conducive to the flexible and efficient development of large-scale monitoring.Therefore,the wireless system has become an inevitable development trend and a hot topic in current research and application of microseismic monitoring technology.At present,wireless monitoring mainly faces the challenges of a large amount,high redundancy,low quality and diverse characteristics of microseismic data,and fragmentation between data acquisition and processing systems,embodied in the following three aspects:(1)The collection,transmission,and processing of massive data will bring colossal consumption of energy,bandwidth,hashrate,and other resources,leading to significant adverse impact on the realtime performance and reliability of the monitoring system;(2)Under the combined effects of environmental noise interference and propagation medium characteristics,the monitoring data have significant low signal-to-noise ratio and characteristic diversity,resulting in poor accuracy and robustness of data analysis;(3)The software and hardware systems of data acquisition and processing are not integrated enough to support the implementation of improvements in data sensing effectively.In view of the above problems,this dissertation focuses on improving the realtime capacity,accuracy,and reliability of wireless microseismic monitoring,from the perspective of data sensing and processing methods and hardware and software fusion design,by taking the core monitoring process of data acquisition,transmission,and processing analysis as the research context.The main research contents are as follows:(1)For the realtime sensing problem of massive monitoring data,traditional methods assume that the data obeys specific distribution or structural characteristics,resulting in poor signal feature expression and anti-noise performance.Thus,they cannot guarantee stable and reliable realtime data sensing.To solve that problem,this dissertation proposes a noise adaptation realtime sensing method for massive monitoring data.Based on the signal multiscale features and the noise non-stationarity and feature diversity,this method combines the feature learning principle from the convolutional neural network and the noise adaptive estimation capacity from wavelet packet decomposition,significantly improving event identification performance.The accuracy and robustness of event detection are driven by the realtime sensing of effective event signal acquisition and transmission.Experiment results show that the proposed method can effectively reduce the total amount of data in the monitoring system by about 80% via dropping irrelevant noisy records,thereby significantly improving the realtime performance and reliability of the system.(2)For the realtime sensing problem of the large number of signals containing event information when the microseismic activity is frequent,existing data compression methods have high complexity and poor robustness,making it challenging to ensure realtime data sensing strictly.To solve that problem,this dissertation proposes a realtime sensing method for effective microseismic signals based on adaptive compressed sensing.According to the feature diversity of microseismic signals and the influence of noise interference on data compression,it first completed the over-complete dictionary construction,and then studied the fast signal compression and transmission strategy guided by random sampling,and the adaptive reconstruction algorithm directed by residual energy attenuation.Experiment results show that this method can reduce the amount of effective event signal transmission load by more than 80% with a very low time cost,thereby further improving the realtime performance and reliability of the system.(3)For the problem of signal processing under the background of poor quality and diverse characteristics of monitoring data,traditional methods rely on artificial parameters or/and data models,resulting in low data-adaptability and unable to meet the requirements of accurate and robust data analysis.In view of this,this paper focuses on the three problems of noise suppression,first arrival time identification and waveform picking in the processing of microseismic monitoring data,and proposes a noise suppression method based on time-time-frequency domain joint fuzzy analysis,and an adaptive first arrival time recognition and waveform picking method based on multiscale feature fusion analysis.Among them,the noise suppression method introduces the fuzzy analysis idea into the identification and extraction of effective signal elements in the original time domain and time-frequency domain at the same time,significantly improving the signal component retention and noise removal effect.The first arrival time identification and waveform picking method combines adaptive fuzzy clustering,the ability of short-time window features to accurately reflect vibration information,and the long-time window features to express the whole vibration process,to carry out the rough picking and calibration of waveforms step by step,thereby significantly improving the accuracy and integrity of waveform picking.The research shows that the noise suppression method has better signal quality improvement performance than the traditional algorithms.At the same time,the first arrival time identification and waveform picking method also have significantly better accuracy,stability,and data adaptation than the conventional method.(4)In view of the impact of the split relationship between data acquisition and processing on the realtime performance and reliability of the monitoring system,this dissertation focuses on edge-cloud collaboration,software and hardware integration design,and low power consumption and reliability design of acquisition instruments.According to the actual needs of the monitoring work,a novel wireless data acquisition instrument was developed.On that basis,this research proposes a cloud collaborative wireless realtime data sensing and processing system with the effective integration of the data sensing and adaptive processing methods.The results of laboratory tests and field monitoring experiments show that,compared with traditional monitoring systems,the proposed wireless realtime data sensing and adaptive processing system can collect and analyze microseismic activities efficiently,effectively,stably and reliably meet the requirements of actual microseismic activities.Monitor specific requirements for convenient deployment,realtime sensing,and long-term stable work.