(School of Nuclear Resources and Nuclear Fuel Engineering, University of South China, Hengyang 421001, China)
摘 要: 针对现有位移反分析方法所存在的反分析模型复杂、计算量大、分析过程人为干预多、求解难度高等问题，提出采用模式搜索算法优化支持向量机，建立一种位移反分析的模式搜索支持向量机方法(PSA-SVM)。以湖南某矿山一个地下矿房为例，应用所建方法对其围岩力学特性参数进行反演，并对该矿房围岩稳定性进行数值模拟计算，分析矿房回采后对地面沉降的影响；并将模拟计算结果与长期监测结果进行比较分析。结果表明：这种位移反分析的模式搜索支持向量机方法的反演结果与实际值吻合良好，完全能够满足工程应用的要求；此外，该方法还具有模型建立简捷、参数调节明确且结果为全局最优等特点，是一种优异的位移反分析方法。
关键字: 模式搜索算法；支持向量机；位移反分析；围岩；数值模拟
（School of Nuclear Resources and Nuclear Fuel Engineering, University of South China, Hengyang 421001, China）
Abstract:Aiming at the problems of current approaches for back analysis of displacements on complexity of back analysis model, large amount of calculation, human intervention and difficulty in solving orthogonal test design method, a PSA-SVM approach for back analysis of displacements is proposed by support vector machine (SVM) optimized by pattern search algorithm (PSA). A case study was performed in a stope underground in a mine, Hunan Province. This new approach for back analysis of displacements was used to inverse the mechanical parameters of the stope. And the reverse results have been used to carry out numerical calculation on the stope, and the effect of the stope on ground subsidence was analyzed. Furthermore, a comparison between the numerical calculation results and monitoring results was given and the both results were in good agreement with each other. The results show that the new approach can satisfy the requirements of engineering application. And it also show that this approach has some features as simple modelling, definitive parameters adjusting and accurate solution within a relatively large solution space, and is an excellent approach for back analysis of displacements.
Key words: pattern search algorithm; support vector machine; displacements back analysis; surrounding rock; numerical simulation
