| 基于BP神经网络的稳定湖相和风成沉积物粒度数据表征方法 |
| |
| 引用本文： | 刘懿馨,郭雪莲,刘懿伟.基于BP神经网络的稳定湖相和风成沉积物粒度数据表征方法[J].甘肃地质学报,2016,25(1):84-89. |
| |
| 作者姓名： | 刘懿馨 郭雪莲 刘懿伟 |
| |
| 作者单位： | 兰州大学地质科学与矿产资源学院，甘肃省西部矿产资源重点实验室，甘肃 兰州 730000,兰州大学地质科学与矿产资源学院，甘肃省西部矿产资源重点实验室，甘肃 兰州 730000,甘肃省地矿局第二地勘院，甘肃 兰州 730020 |
| |
| 基金项目： | 国家自然科学基金项目（批准号：41202129）资助 |
| |
| 摘 要： | 沉积物的形成受到多种地质因素的综合控制。通过粒度分析可判别沉积物的成因类型,推断其形成的沉积环境,解释环境演变;而沉积物的粒度组分除了受到原岩的控制外,还受到机械沉积作用的影响难以准确预测。运用人工神经网络对稳定湖相沉积物和风沉积物的粒度参数进行研究,将沉积物的4个粒度参数作为网络模型的输入变量,在对168个浙闽沿海迎风岸风成老红砂样品和282个苏贝淖湖滨湖泊沉积物样品所对应的粒度参数进行数据样本训练之后,获得了基于BP神经网络的稳定湖相和风沉积物预测模型。然后利用448个大树摆鱼湖相沉积物粒度参数样本和100个兰州榆中黄土风沉积物粒度参数样本作为测试样本对该模型进行了测试和验证,结果显示模型的可靠性较好,能够对沉积物的形成环境做出正确的判断。
|
| 关 键 词： | 粒度分析 沉积环境 BP神经网络 预测模型 |
Stable Lacustrine and Eolian Sediments Grain Size Representation Method Based on BP Neural Network |
| |
| LIU Yi-xing,GUO Xue-lian and LIU Yi-wei.Stable Lacustrine and Eolian Sediments Grain Size Representation Method Based on BP Neural Network[J].Acta Geologica Gansu,2016,25(1):84-89. |
| |
| Authors: | LIU Yi-xing GUO Xue-lian and LIU Yi-wei |
| |
| Abstract: | The formation of sediments is synthetically controlled by a variety of geologic factors. The grain size composition of sediments is controlled by the original rock; in addition, it is mainly affected by the mechanical sedimentation. Through analyzing characteristic of grain size, types of sediments can be distinguished, the characteristic of the sedimentary environment can be concluded, and the environmental evolution can be interpreted. In this paper, we used the artificial neural network to study the grain-size parameters of stable lacustrine sediments and typical eolian sediments. The four grain-size parameters considered as a the neural network input vector, through 168 grain-size parameters samples of crust red earth and 282 grain-size parameters samples of lacustrine sediments of Subeinao Lake training, we got a predict model to judge the stable lacustrine sediments and typical eolian sediments based on BP neural network. After that, we used 448 lacustrine sediments grain-size parameters samples of Baiyu village Dashu town of Hanyuan and 100 eolian sediments grain-size parameters of Yuzhong of Lanzhou to test this model. The classification result presented the model have the ability to make a right judgment for the sediment environment and it also with the feature of dependability. |
| |
| Keywords: | grain-size analysis sediment environment BP neural network predict model |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《甘肃地质学报》浏览原始摘要信息 |
| 点击此处可从《甘肃地质学报》下载免费的PDF全文 |
|
