APPLICATION OF ARTIFICIAL NEURAL NETWORK TO IDENTIFICATION OF DRILLED STRATA
-
摘要: 对用人工神经网络方法来解决钻探生产的实际问题, 在不取心的情况下识别所钻地层的岩性进行了研究.根据钻探生产的特点, 设计了人工神经网络的结构和输出方式, 开发了人工神经网络识别所钻地层的软件, 分析了影响人工神经网络应用效果的各因素, 在人工神经网络的优化设计方面作了较深入的研究.研究表明: 人工神经网络用于识别所钻地层有很好的效果; 人工神经网络的参数, 如学习率、隐含层层数、隐含层单元数和数据处理方式等对人工神经网络的应用效果有影响.Abstract: The artificial neural network (ANN) method is used to solve the puzzle that occur in the drilling: the identification of the lithology of the strata drilled. The drilling features are used to design the structure and output form of the ANN, to develop the ANN software for the identification of the strata drilled, to analyze various factors that affects the application of the ANN and to make a further research into the optimum design of the ANN. The research shows that the ANN perfectly matches the identification of the strata drilled. Various ANN parameters such as the training efficiency, the hidden layer number, the hidden layer unit number, and the data processing may affect the application of the ANN.
-
Key words:
- drill /
- drilled strata /
- identification /
- artificial neural network
-
表 1 人工神经网络结构参数、训练参数及训练结果
Table 1. Structure parameters, training parameters and training results of neural network
表 2 不同隐层单元数人工神经网络的应用效果
Table 2. Application effect of ANN to different unit numbers of the hide layer
表 3 不同学习率η人工神经网络的应用效果
Table 3. Application effect of neural network to different training efficiency (η)
-
[1] 史玉升. 基于人工神经网络的钻压优化模型的建模方法[J]. 地球科学——中国地质大学学报, 1999, 24(4): 432~435. https://www.cnki.com.cn/Article/CJFDTOTAL-DQKX199904023.htm [2] 焦李成. 神经网络系统理论[M]. 西安: 西安电子科技大学出版社, 1995. 34~36. [3] 姚姚. 人工神经网络和非线性反演[M]. 武汉: 中国地质大学出版社, 1993. 8~12. -
下载:
点击查看大图
图(3) / 表(3)
计量
- 文章访问数: 3428
- HTML全文浏览量: 109
- PDF下载量: 6
- 被引次数: 0
