False Alarm Rate Elimination Schemes For Seismic Event Detection On Smartphones

Accelerometer and gyroscopes equipped smartphones which are widely adopted by humans,can sample seismic waves as the similar way to those of seismometers,potentially become the devices for monitoring earthquakes.However,accelerometers in the mobile phone are off-the-shelf MEMS sensors,so result in poor sensing precisions.At the same time,the smartphone is susceptible to daily activities of the phone holders,consequently leads to very high false alarm rates.In order to solve this problem,this paper focuses on the schemes of effective seismic pick-up algorithm to reduce false alarm rate caused by white noise and the human activities,according to the characteristics of data sampled by smart phone's sensors and the laws of human daily activities.The main work of this article is as follows:Firstly,the strategies of data processing in smartphones are introduced for conquering the shortcomings in the process of data sampling by its sensors,as the data of accelerometers are easily affected by the positions of the mobile phone,and the white Gaussian noise is large,lead to the imprecision of the samling results.According to these characteristics,reasonable data processing methods are proposed.The data coordinate transformation is realized by the rotation matrix,and the data perpendicular to the ground is collected to eliminate the influence of different positions of the mobile phone.And then the Gaussian white noise is removed by the wavelet transform,and we import an eigenfunction to obtain good denoising effect.Secondly,according to the characteristics that mobile phones are easily affected by human activities,a high-precision abnormal event detection algorithm of STA/LTA+DTW is proposed.In order to reduce the impact of human behavior on earthquake anomaly detection and reduce the false alarm rate,this paper analyzes the time series signal characteristics DTW of human daily behavior,and designs a double thresholds detection method.The first layer is a STA/LTA-based vibration sensing algorithm for detecting changes of smartphone's motion.The second layer is an abnormal vibration detection layer based on the DTW calculation,which is used to distinguish abnormal events from daily events.By comparing the results of mew scheme and the traditional method,proves that the proposed algorithm has asignificant improvement in reducing the false alarm rate.Finally,the sheme of precision improvement in smart phone is proposed through data fusion between accelerometers and gyroscopes.Accelerometers in smartphones are different from professional seismometers in that the sensing accuracy is low and will deteriorate over time,so this paper uses multi-sensor data fusion to solve this problem.The rotation azimuth of the mobile phone and the accelerometer value are modeled,and fused,by the extended Kalman filter fusion to realize the calibration of the accelerometer value.After testing the ordinary smartphones by lots of volunteer,the results show the improvement of detection precision and false alarm rate elimination of the proposed schemes.