基于神经网络的区域地壳稳定性评价

Evaluation of regional crust based on ANN

在重大工程的选址中,由于多种影响因素的控制和复杂的非线性相关关系,如何建立统一的数学模型,充分利用已有地质资料,对区域地壳稳定性做出早期的准确评价一直是一个难题。基于人工神经网络方法,建立了区域地壳稳定性评价的数学模型,并介绍了计算方法。以沂沭断裂带区域地壳稳定性定量评价为例,介绍了模型结构和关联系数矩阵的确定方法,在模型中,考虑地形地貌、含水层岩性、地下水活动、地质构造、地震活动等多种因素的影响,对区域地壳稳定性进行了定量评价。同时,选取多种运算参数,分析了模型参数对运算的影响。依据计算结果,对研究区的铁路选线和今后的研究工作提出了几点认识和建议。

Because the stability of regional crust was controlled by many factors, and the nonlinear correlation among influencing factors was complicated, it was difficult to construct a mathematical model by traditional methods to evaluate the stability precisely during geological surveying, and many geological data could not be applied adequately. Based on artificial neural network （ANN）, the evaluating model of regional crust stability was established, and the calculating method was investigated. The structure of evaluating model was introduced by taking example for Yishu faulted zone, and the correlative coefficient matrices were explained. In this model, topography, lithology of aquifer, underground water, geological structure and seismic activity were considered. So the regional crust stability in Yishu faulted zone was calculated as a basis of this model. Furthermore, model preferences were discussed and their influence on the calculated result was analyzed, According to the calculated results, some advice for the construction of railway in this area and further researches were proposed.