论文部分内容阅读
基于未确知测度理论,建立了砂土地震液化判别和液化势分级的未确知均值聚类分析模型和方法。针对砂土地震液化评价中的许多不确定性影响因素,选用地震震级,地面地震加速度幅值,标准贯入击数,比贯入阻力,砂土相对密实度,砂土平均粒径和场地地下水位等7个评价指标作为判别因子;选用17个砂土样本作为训练样本,建立各评价指标的未确知测度函数,以样本中的各评价指标数据的平均值表示其分类中心;利用相似权赋权方法确定评价指标的权重,依据未确知测度距离判别地震液化等级;根据建立的模型对训练样本回判,回判正确率为94.12%。将建立的模型对20个测试样本进行判别,将判别结果与地震液化的实际情况、BP神经网络和SOFM神经网络等方法的评价结果进行了对比。研究表明,该模型的评价结果与实测结果,以及BP神经网络、SOFM神经网络等方法的评判结果一致性较高。
Based on the unascertained measure theory, an unascertained mean clustering analysis model and method for liquefaction discrimination and liquefaction potential classification of sandstone are established. Aiming at many uncertain factors in the evaluation of sand liquefaction, the magnitude, magnitude of ground acceleration, penetration standard penetration rate, specific penetrating resistance, relative density of sand, sand average particle size and site groundwater And seven indexes were selected as discrimination factors. Seventeen sand samples were selected as training samples to establish the unascertained measure function of each evaluation index. The classification center was represented by the average of each evaluation index data in the sample. Weighting method is used to determine the weight of evaluation index, and the magnitude of earthquake liquefaction is judged according to the unascertained measuring distance. According to the established model, the correct rate of return to the training sample is 94.12%. The established model was used to discriminate 20 test samples, and the result of the discrimination was compared with the actual situation of seismic liquefaction, BP neural network and SOFM neural network. The research shows that the evaluation results of the model are in good agreement with the measured results, the BP neural network and the SOFM neural network.