【作者】 王雪梅；

【导师】 吴鑫；

【作者基本信息】 四川师范大学 ， 安全工程（专业学位）， 2022， 硕士

【摘要】 我国地质类型丰富且较为复杂,部分地区滑坡灾害严重,各类新型技术开始用于边坡监测,但主要是监测地表或浅部范围的位移和变形场,对边坡内部变形却难以掌握。边坡失稳是一段时期内不稳定性影响因素的积累结果,包括微裂隙的产生和扩展,微观尺度破裂总是伴随着不同频域或时域的弹性波产生。因此本文基于声发射信号模式识别对土体剪切速率进行预测,为进一步提高滑坡预测精度和提供早期预警。本文的主要研究内容如下:(1)通过开展砂砾填充的有源波导杆在不同加载速率下(2、5、10、20mm/min)的三点弯曲试验,分析加载速率与声发射信号参数之间的内在联系。试验结果表明:幅值随着加载速率的增加逐渐增大,b值(表示低幅值事件占高幅值事件的比例)逐渐减小,且声发射事件数与振幅呈负相关。振铃计数随着时间的增加呈线性增加,振铃计数的增长速率随加载速率的增大呈幂函数增加;随着阈值的不断增大,振铃计数减少的速率越来越快。能量随着时间的增加呈指数函数增加,能量增加的速率随加载速率的增大越来越快。且随着力的增加,振铃计数与能量增加的速率变化规律为:2<5<10<20mm/min。在加载前期主频幅值相对稳定,加载后期主频幅值明显增加;加载速率越大,主频幅值增加的速率也越快;且高频信号的主频幅值相对更小,主频幅值在低频与中低频段(20～200k Hz)上的变化更明显。(2)通过开展在不同含水率下(18%～24%)和不同剪切速率下(5、10、20mm/min)的土体剪切试验,分析土体含水率和剪切速率与声发射信号参数之间联系。试验结果表明:力、幅值、振铃计数和能量都随着剪切时间的增加逐渐增大;累计振铃计数与累计能量随着剪切时间的增加呈线性增长,且剪切速率越大,其线性增长的速率越快,b值与之相反,随着剪切时间增加逐渐减小。在同一剪切速率下,随着土体含水率逐渐增大,土体强度逐渐降低,力、振铃计数和能量随之减小。含水率越小,振铃计数、能量增加的速率越快,b值减小的速率越快。振铃计数和能量随含水率的增加呈数幂函数减小,b值则相反,随含水率的增加呈数幂函数增加。(3)基于BP神经网络分别对三点弯曲的加载速率和土体的剪切速率与力进行预测,将振铃计数、能量、b值、幅值、含水率等参数的变化作为预测特征。预测结果如下:三点弯曲加载速率真实值与预测值的误差在[-1.5,1.5],剪切速率的误差在[-1,1],剪切力的误差在[-0.1,0.1],平均绝对误差、均方误差、均方根误差的值都小于1,相关性系数R都约等于1。因此通过识别声发射信号参数的特征来预测边坡滑移速度是可行的。更多还原

【Abstract】 The geological types in my country are rich and complex,and landslide disasters are serious in some areas.Various new technologies have begun to be used for slope monitoring,but they are mainly used to monitor the displacement and deformation fields based on the surface or shallow range,and it is difficult to grasp the internal deformation of the slope.Slope instability is the accumulation result of instability influencing factors in a period of time,including the generation and expansion of micro-cracks.Micro-scale ruptures are always accompanied by elastic waves in different frequency or time domains.Therefore,this paper predicts the soil shear rate based on the pattern recognition of acoustic emission signals,in order to further improve the prediction accuracy of landslides and provide early warning.The main research contents are as follows:(1)Three-point bending test of the active waveguide rod filled with gravel at different loading rates(2,5,10,20 mm/min)was carried out.Analyze the intrinsic relationship between acoustic emission signal parameters and loading rate.The experimental results show that the amplitude increases gradually with the increase of the loading rate,and the b value(representing the proportion of low-amplitude events to high-amplitude events)gradually decreases.The number of acoustic emission events is negatively correlated with the amplitude.The ringing count increases linearly with time,and the growth rate of the ringing count increases as a power function with the loading rate.As the threshold increases,the ring count decreases faster and faster.The energy increases exponentially with time,and the rate of energy increase becomes faster and faster as the loading rate increases.With the increase of the force,the change rule of the rate of ringing count and energy increase is: 2<5<10<20mm/min.In the early stage of loading,the main frequency amplitude is relatively stable,and the main frequency amplitude increases significantly in the later stage of loading.The greater the loading rate,the faster the main frequency amplitude increases.The main frequency amplitude of the high frequency signal is relatively smaller,and the main frequency amplitude changes more obviously in the low frequency and mid-low frequency bands(20～200k Hz).(2)By carrying out soil shear tests at different moisture contents(18%～24%)and different shear rates(5,10,20mm/min).Analyze the relationship between soil moisture content and shear rate and acoustic emission signal parameters.The test results show that the force,amplitude,ringing count and energy all increase gradually with the increase of shearing time;the cumulative ringing count and cumulative energy increase linearly with the increase of shearing time.The larger the shear rate,the faster the linear growth rate,and the b-value is opposite to that,and gradually decreases with the increase of shear time.At the same shear rate,as the soil moisture content gradually increases,the soil strength gradually decreases,and the force,ringing count and energy decrease accordingly.The smaller the moisture content,the faster the rate of ringing count,energy increase,and the faster the rate of b-value decrease.The ringing count and energy decreased with the increase of water content as a power function,while the b value was the opposite,and increased with the increase of the water content as a power function.(3)Based on BP neural network,the loading rate of three-point bending and the shear rate and force of soil are predicted respectively.Changes in parameters such as ringing count,energy,b-value,amplitude,and moisture content were used as prediction features.The prediction results are as follows: the error between the actual value of the loading rate and the predicted value is [-1.5,1.5],the error of the soil shear rate is [-1,1],and the error of the soil shear force is [-0.1,0.1].The values of mean absolute error,mean square error,and root mean square error are all less than 1,and the correlation coefficient R is approximately equal to 1.Therefore,it is feasible to predict the slope slip rate by identifying the characteristic parameters of the acoustic emission signal.更多还原