학술논문
基于改进DenseNet的固井质量评价新方法 / A New Method of Cementing Quality Evaluation Based on Improved DenseNet
Document Type
Academic Journal
Author
Source
计算机技术与发展 / Computer Technology and Development. 34(1):193-199
Subject
Language
Chinese
ISSN
1673-629X
Abstract
为解决固井质量评价效率低、准确率不高的问题,提出一种基于改进的DenseNet卷积神经网络的评价方法.该方法通过增加多尺度卷积层可以同时获取固井质量特征图的大尺度和小尺度特征,从而提高感受野的覆盖范围,增强模型对不同尺度的适应能力;通过嵌入CBAM机制使模型在空间和通道两个维度上充分提取对评价任务有用的信息,增强模型对特征的关注能力和感知能力,提升评价结果的准确度以及模型的鲁棒性;同时,通过缩减网络层数减少模型参数的数量,提升模型的计算效率以及泛化能力.实验结果表明,测试集中的 3 类评价样本的准确率为 95.86%,相比DenseNet-121 提升了4.9 百分点左右,且参数量显著减少;相比BP神经网络和支持向量机均提升了9 百分点左右.从而揭示出,采用改进DenseNet模型实施固井质量评价的研究方案不仅是可行的,而且优于同类机器学习方法.
In order to solve the problems of low efficiency and low accuracy of cementing quality evaluation,an evaluation method based on improved DenseNet convolutional neural network is proposed.In this method,the large-scale and small-scale features of the cementing quality feature map can be obtained simultaneously by adding a multi-scale convolutional layer,thereby improving the coverage of the receptive field and enhancing the adaptability of the model to different scales;by embedding the CBAM mechanism,the model fully extracts useful information for evaluation tasks in two dimensions of space and channel,enhances the model's ability to focus on features and perception capabilities,and improves the accuracy of evaluation results and its robustness;at the same time,by reducing the number of network layers,the number of model parameters is reduced,and the computational efficiency and generalization ability of the model are improved.The experimental results show that the accuracy rate of the three types of evaluation samples in the test set is 95.86%,which is about 4.9 percentage points higher than that of DenseNet-121,and the number of parameters is significantly reduced;compared with BP neural network and support vector machine,it is 9 points higher percent or so.Therefore,it is revealed that the research program of implementing the cementing quality evaluation using the improved DenseNet model is not only feasible,but also superior to other similar machine learning methods.
In order to solve the problems of low efficiency and low accuracy of cementing quality evaluation,an evaluation method based on improved DenseNet convolutional neural network is proposed.In this method,the large-scale and small-scale features of the cementing quality feature map can be obtained simultaneously by adding a multi-scale convolutional layer,thereby improving the coverage of the receptive field and enhancing the adaptability of the model to different scales;by embedding the CBAM mechanism,the model fully extracts useful information for evaluation tasks in two dimensions of space and channel,enhances the model's ability to focus on features and perception capabilities,and improves the accuracy of evaluation results and its robustness;at the same time,by reducing the number of network layers,the number of model parameters is reduced,and the computational efficiency and generalization ability of the model are improved.The experimental results show that the accuracy rate of the three types of evaluation samples in the test set is 95.86%,which is about 4.9 percentage points higher than that of DenseNet-121,and the number of parameters is significantly reduced;compared with BP neural network and support vector machine,it is 9 points higher percent or so.Therefore,it is revealed that the research program of implementing the cementing quality evaluation using the improved DenseNet model is not only feasible,but also superior to other similar machine learning methods.