A Method For Detecting Abnormal Environmental Vibration Based On Smartphone Inertial Sensor Sampling

Abnormal vibration in the geographic environment usually indicates the occurrence of events such as geological disasters,and the detection of abnormal vibration events has become an effective means to reduce casualties and property damage.The existing common geological vibration monitoring equipment is usually the traditional vibrometer containing acceleration sensors,which has a high recognition accuracy,but cannot be deployed in a large area because of the high price,large size and other shortcomings.With the rapid development of electronic technology and the comprehensive popularity of smartphones,smartphone acceleration sensors have become new infrastructure for environmental vibration practice monitoring,and the use of electronic devices such as smartphones of common users to achieve the detection of abnormal events has become a new research hotspot.However,the use of smartphones for environmental vibration monitoring usually faces the influence of the daily activities of the users,which leads to deficiencies such as high false alarm rate and low leakage rate of monitoring.This paper focuses on how to distinguish human activities from environmental abnormal vibration events and the accurate pickup of P-wave arrival of abnormal vibration events,so as to provide an effective basis for the accurate monitoring of environmental abnormal vibration events using ordinary smartphones in the future.The main work and contributions of this paper are as follows.First,the conversion between the coordinate system of smartphone sampling data to the geographic coordinate system is realized by using the quadratic method,as the error source about acceleration data,collected by the smartphone,is generated due to the user's action state.Furthermore,owing to the device error caused by the brand and the different manufacturing process of the built-in acceleration sensor of the smartphone,a multi-Attitude Model is constructed to systematically correct the error of the sampling value of the acceleration sensor of the mobile phone.In the sampling process,median filtering is used to denoise the signal and extract sample features to address the problem of noisy smartphone sensor sampling,and the experiments demonstrate that the proposed method has good denoising effect.Subsequently,according to the problem that smartphones are vulnerable to human action by mobile phone users,Support Vector Machine,which is widely used in smartphone recognition and classification,are used to identify human action events and abnormal vibration events.The simulation generates data sets to train the support vector machine model and get the optimal classification label.Finally,the smartphone sampled data after the above processing is accurately picked up by P-wave arrival.The basic principles of traditional STA/LTA method and AIC method are analyzed first,and an improved STA/LTA-AIC synthesis method based on wavelet packet decomposition is proposed.The experimental results show that the P-wave pick-up method based on wavelet packet decomposition is better than three traditional contrast pick-up methods,which verifies the possibility of abnormal vibration event warning based on smart phone.