摘要
依据地下探测的全空间特点,首次提出准全空间探测模型,推导准全空间MRS响应核函数表达式,有效剔除隧道已挖空区域在计算结果中的影响,提高MRS正演计算准确度。同时,反演方法采用序列二次规划法进行,完成了地下准全空间模型磁共振数据非线性反演。理论数据中,对比地面半空间、地下全空间的反演结果,准全空间模型能更精确地反演地下灾害水体位置及含量,且在含噪10%的情况下仍然得到较为理想的反演结果。在理论研究的基础上,结合金温铁路泽雅隧道的MRS现场测试,对掘进掌字面前方隐伏含水构造进行解释,并利用瞬变电磁数据对MRS探测结果进行了一致性的比对分析。该项研究表明准全空间探测模型结合SQP反演算法能够准确给出地下灾害水体的定位及富水程度,实现地下水害的超前准确预报,为隧道磁共振探测提供理论基础。
Based on of the whole-spatial characteristics of underground exploration, the quasi-whole space detection model was proposed for the first time, the quasi-whole space response kernel function of MRS was derived. The effect of the hollowed out area of the tunnel in the calculation result is eliminated effectively and MRS forward calculation accuracy is improved. Meanwhile, with the sequential quadratic programming （SQP） inversion method, the nonlinear inversion of magnetic resonance data of the underground quasi-whole space model is fulfilled. In the theoretical data, contracting the inversion results of ground half space with the underground space, quasi-whole space model can be more accurate in the inversion of underground disaster water＇s position and content, and a good inversion result even under the condition of containing 10% noise is gained. Based on the theoretical research and combining with MRS field testing of Zeya Tunnel of Jin Wen Railway, the water-bearing structure hidden in the tunneling tunnel face was explained, and the consistency of the detecting results comparatively was analyzed. The results show that the whole space exploration model combining SQP inversion algorithm can give the position of underground disaster water and water-bearing level accurately, and the prediction of groundwater harm accurately is achieved to lay a theoretical foundation for MRS detection of the tunnel.
出处
《中国有色金属学报》
EI
CAS
CSCD
北大核心
2013年第9期2413-2421,共9页
The Chinese Journal of Nonferrous Metals
基金
国家重大科学仪器设备开发和应用专项(2011YQ030133)
国家自然科学基金青年基金项目(21101071)
吉林省科技发展项目(20120326)
关键词
隧道
磁共振测深
准全空间
序列二次规划
tunnel
magnetic resonance detection
quasi-whole space
sequential quadratic programming
