胡银磊, 张裕明. 人工神经元网络方法对首都圈地区潜在震源区的划分[J]. 华北地震科学, 1996, 14(3): 54-59.
引用本文: 胡银磊, 张裕明. 人工神经元网络方法对首都圈地区潜在震源区的划分[J]. 华北地震科学, 1996, 14(3): 54-59.
Hu Yinlei, Zhang Yuming. Identification of Potential Seismic Sources in Beijing Area Using ANN[J]. North China Earthquake Sciences, 1996, 14(3): 54-59.
Citation: Hu Yinlei, Zhang Yuming. Identification of Potential Seismic Sources in Beijing Area Using ANN[J]. North China Earthquake Sciences, 1996, 14(3): 54-59.

人工神经元网络方法对首都圈地区潜在震源区的划分

Identification of Potential Seismic Sources in Beijing Area Using ANN

  • 摘要: 指标选取是人工神经元网络方法应用中最基础、最重要的一环,是判别工作成败的关键。但目前尚没有比较成熟的指标选取或指标显著性评价方法。因此,指标问题往往成了方法应用效果进一步提高的"瓶颈"问题。该问题在潜在震源区划分应用中也同样存在。在分析了潜在震源区划分中的原则、方法和依据的基础上,提取了判定震级上限的指标体系,即:不同时代活动断层的分段长度、活断层运动速率、活动盆地参数、不同震级档地震数、布格重力异常、航磁△T异常、地形变和地壳应力场等。在此基础上,利用人工神经元网络方法对首都圈地区的潜在震源区作了定量划分。结果表明,由于指标体系较全面地考虑了震级上限确定中的主要因素,判别准确性和精细程度有较大提高。

     

    Abstract: Index selection is essential in the application of Artificial Neural Network(ANN).In this paper, on the basis of analyzing the factors involved in the determination of upper-bound limit magnitude of potential seismic source, 14 indexes are selected to form the index-set, and then the potential seismic sources in the Beijing Area are identified using those indexes and ANN. The result indicates that, because the index-set well incorporates main factors in the determination of the upper-bound limit magnitude, the accuracy is elevated, and the result is more detailed.

     

/

返回文章
返回