锚杆受荷条件下声学特征及承载力智能预测方法研究

The anchoring technology of bolt is an important supporting means in underground engineering and side slope management. Anchoring quality is one of the key factors influencing engineering quality and safety. Nowadays, it is the difficult in the world for the high precision intelligent detection of bearing capacity. Therefore, our research has important theoretical and practical significance for the safety of underground engineering. .The research is about bolt bearing mechanism with the tools of pattern recognition, modern information processing technology and intelligent control. Theoretical model of acoustic characteristics related to bolt's mechanical properties was established, distribution characteristics of bolt' bearing capacity under different loading and working condition were calculated by using numerical simulation method. Bearing capacity of bolt could be predicted with the non-interval GM (1, 1) grey model by ant colony algorithm identification when the parameters of bolt-grout system were suitable selected. Frequency response function between the test signal and input excited signal are pretreated using spectrum analysis, characteristic values are got, which is used to be the input values in a network. A prediction method would be proposed based on multi-source information fusion, so the theoretical basis are founded for bearing capacity intelligent prediction.

锚杆锚固技术是地下工程及边坡治理的重要支护手段，锚固质量的好坏关系着工程质量和安全，目前锚杆承载力的高精度智能检测是该领域面临的世界性难题，因此，本研究对保障地下工程安全等具有重要的理论意义和实用价值。.项目以模式识别、现代信息处理技术、智能控制理论为工具，建立与锚杆力学特性相关的声学特性理论模型，运用数值模拟方法计算锚杆处于不同受荷状态和工作状态下锚杆承载力分布特性和声波传播特性，研究锚杆不同承载力与声波波场的响应关系；研究锚杆承载条件下波场机理，选择表征锚杆-围岩结构系统的参数，利用蚁群算法辨识灰色模型参数，非间隔GM(1,1)灰色模型预测锚杆承载力；利用谱分析对锚杆受力的声波信号与激励信号进行处理，得到频率响应函数，求其特征值，以此作为神经网络的输入，提出基于多源信息融合的锚杆承载力预测方法，为建立锚杆承载力智能检测系统奠定理论基础。

锚杆锚固技术是地下工程及边坡治理的重要支护手段，锚固质量的好坏关系着工程质量和安全，目前锚杆承载力的高精度智能检测是该领域面临的世界性难题，因此，本研究对保障地下工程安全等具有重要的理论意义和实用价值。.项目以模式识别、现代信息处理技术、智能控制理论为工具，建立与锚杆力学特性相关的声学特性理论模型，运用数值模拟方法计算锚杆处于不同受荷状态和工作状态下锚杆承载力分布特性和声波传播特性，研究了锚杆不同承载力与声波波场的响应关系；构建频率响应函数矩阵，利用基于主元分析的频响函数主成分提取，最后应用基于PSO-RBF神经网络完成锚杆锚固质量检测；提出了基于改进的三次多项式函数的锚杆承载力预测方法；对传统的灰色GM（1,1）模型进行了改进，建立了非等间距传统灰色GM(1，1)模型，完成了锚杆承载力预测；针对改进基于改进的三次多项式函数，非等间距灰色GM(1，1)模型、改进指数、改进指数-幂函数等得到的预测结果，使用D-S证据理论融合算法来实现多模型组合预测锚杆极限承载力。.