Research On Safety Monitoring Technology Of Moisture-bearing Geotechnical Engineering Based On Acoustic Emission

The infrastructure construction has increased rapidly with the economic development of China since twentieth century.Large construction scale brought challenges to the stability and safety.Health monitoring of geotechnical engineering has become a significant issue related to life and property.Moisture can reduce the compressive and shear strength of rock mass.The structural stability would be weakened,which would cause the safety accidents such as mine collapse and dam breakage.The moisture has become an important factor in engineering accidents due to the softening effect on rock structure.For guaranteeing the engineering and lives,it is extremely significant to monitor the moisture content and assess the damage level as well as give warnings in advance.In this study,the influence of moisture content on the damage and AE(Acoustic Emission),the modeling of monitoring data,the moisture content recognition and the damage assessment would be conducted based on the acoustic emission technology.The main content of this thesis are as follows:The influence of moisture content on the damage level and acoustic emission was explored.Time and frequency characteristics of rock acoustic emission signal under different moisture contents were researched using time domain observation,frequency spectrum analysis,high-order spectrum analysis,parameter analysis and line sampling analysis methods.The repression effect of moisture on acoustic emission was demonstrated from numerical simulation and macroscopic experiment.For describing the correlation between moisture and damage levels,the CT scanning was utilized on the damage crack of sandstone samples with different moisture contents.The damage degree was quantified using binary pixel analysis.Connection domain scanning and image thinning algorithm were used to obtain the morphology information of rock crack.The CT images of samples with different moisture contents were quantified using comprehensive damage index.The promoting effect of moisture to damage level can be demonstrated from the quantified analysis.The modelling methods of monitoring data were researched.First,the memory and forgetting mechanism of brain were simulated.The storage areas of monitoring system were divided into short-time memory area and long-time memory area.Noise data can be filtered and the effective can be recorded under the control of threshold.Then,a signal musicalized method was proposed to store the monitoring data sequentially.Damage signal envelop was sampled by musical lines,so that the imperceptible damage signal can be mapped to audio based on the musicalized damage indices.Multi-types damage signals were acquired,and the damage recognition experiment was conducted through LSTM algorithm.The experiment result shows that this method has a good performance on signal feature retention.The signal musicalized method and Kalman algorithm were combined to recover the failure data.The Failure data was fitted and rebuild based on the extreme value before failure and the data of nearby sensor.A twostep acoustic emission parameters selection method was proposed based on the fluctuate law of parameter under environment and the clustering analysis.The algorithms comparison shown that though this method has the large time consumption,the selected parameters perform better in pattern recognition.For monitoring the rock moisture content in real time,a moisture content identification method was proposed based on the fuzzy mathematic theory.The tolerance of parameters was calculated to obtain the importance order and the judgement matrix.The adaptive weight vector was constructed based on the statistical law of acoustic emission.The output of Softmax function on each moisture content was used as a row.Membership matrix was obtained through the combination of rows vector.Moisture content was identified through the fuzzy calculation between the adaptive weight vector and the membership matrix.A sandstone-sandy soil structure was designed.The pressure distribution during the seepage in this structure was simulated by finite element simulation.The mechanical structure of sensor installation and acoustic emission data acquisition system were designed.In the experiment,the moisture content fuzzy recognition was consistent with the real moisture content obtained by weighting method.The electroencephalogram guided by different colors of light was acquired.The influence of human concentration was determined through the analysis of Beta wave components.The color sequence of alarm light which suitable for human concentration was determined.The acoustic emission parameters were mined to assess the damage.Uncertainty derivation of the logical correlation between environmental factors and damage results was conducted through the reliability theory,information entropy theory and causal reasoning theory.A time-domain wave and damage information entropy vector were proposed,which contain the correlation information of parameters and the confidence of damage occurrence,respectively.A fatigue index was proposed based on the rock damage mechanics and probability statistics knowledge.The sensitivity of this index was verified in moisture damage assessment experiments.A cause strength index was proposed based on the Bayesian theory.The probability location of the damage area was conducted using the arrival time parameter.