| Our country's topography is varied and complicated, vast mountainous areas, variety of crisscross which leads lot of geological disasters. Landslide disaster is particularly serious. There are eight thousand two hundred and twenty-four geological disasters in twenty-eight provinces last year, the number of landslide disaster is 5616. Unevenly stressing inside the mountain leads to redistribute stress, accompany with releasing elastic wave which is called acoustic emission. It includes a lot of information, such as the location of the acoustic emission source, intensity, etc. Acoustic emission technology has the advantages of dynamic and real-time detection and is used for nondestructive testing of materials and structures,the research of landslide earlywarning which is based on acoustic emission technology becomes to a hotspot. The research of the characteristics and mechanism has great significance to landslide earlywarning and correct understanding of the landslide process. The main research content and result are as follows:First, a hardware platform is established which is used to collect AE signal. Choose the series of FalconTM's 4193 microphone to gather acoustic emission signals. Transmission acquired signals by ZigBee which has mature wireless transmission technology, STM32F407 as master controller. The location method is sensitive to the position coordinates and time synchronization of the monitor. Use GPS to achieve the location of the monitor and complete time synchronization. By doing this we can get the more precisely location.Second, analysis the signals. Refer to a large number of literature, analysis of signal, noise characteristics and transmission mechanism. We find that the noise is composed by environmental noise, system noise, mechanical noise and other acoustic emission signals. According to the characteristics of signal and noise wavelet is proposed to de-noise. ICA is used to acquire independent signal by dealing with signals. Wavelet and EMD are used to de-noised the independent signal. The results showed that a moderate demixing matrix can isolate the noise and signal from multiple nodes monitoring signals. The signal with global best features which is reconstitution by different wavelet basis function of wavelet threshold de-noising method. We can get a better de-noised effect by choosing an appropriate decomposition layer and global MOMENT. The Direct-Threshold and part of the refactoring de-noised method get acoustic emission signal local best features.The last, this paper complete the positioning of landslide by acoustic emission signal which is collected by many monitor nodes. Based on the analysis of TDOA, area location, centroid location and intelligent location's theory. Comparing the several kinds of localization algorithm in landslides in the applications of AE source location advantages and disadvantages. TDOA can get a precisely position coordinates under the precondition of monitoring the time synchronization, coordinates and under the premise of signal propagation velocity is known. Area location by the sequence of the signal, it can't locate a precisely position coordinates. Centroid location has a Low demands on the time synchronization, using signal strength and choosing the right weight value we can get an accurate sound source location coordinates. BP and RBF both of them are smart positioning methods, can overcome the shortage of TDOA and centroid localization. Simulation results show that we can realize accurate localization of sound source by choosing arrival time, signal range, signal energy, node location coordinates as the inputs of neural network model. |
PDF Full Text Request