基于深度学习的镜下矿石矿物的智能识别实验研究   总被引：1，自引：3，他引：1  

徐述腾  周永章 《岩石学报》2018,34(11):3244-3252

矿石矿物鉴定的智能化是智能地质学和智能矿床学的基础技术之一。计算机视觉技术和深度学习理论使矿石矿物鉴定的智能化成为可能。本研究基于深度学习系统Tensor Flow,以吉林夹皮沟金矿和河北石湖金矿的黄铁矿、黄铜矿、方铅矿、闪锌矿等硫化物矿物为例,设计有针对性的Unet卷积神经网络模型,有效自动提取矿相显微镜下矿石矿物的深层特征信息,实现镜下矿石矿物智能识别与分类。实验显示,模型在训练过程中,随着训练次数的增加,模型精度在不断增大,损失函数不断减小;经过3000个批处理之后,模型精度和损失函数基本趋于稳定。训练出的模型对测试集中的显微镜镜下矿石矿物照片的识别成功率均高于90%,说明实验所建立的模型,具有很好的图像特征提取能力,能完成镜下矿石矿物智能识别的任务。  相似文献   

基于客观分析的重庆雾的BP神经元网络预报模型研究   总被引：6，自引：4，他引：2  

刘德  李永华  喻桥  何险峰  周国兵 《气象科学》2005,25(3):293-298

本文利用数值预报产品、TTAA报文观测资料及重庆雾(能见度)的实况资料,采用客观分析方法将相关资料处理成二维网格资料,再运用车贝雪夫正交多项式实现二维网格图形的数学定量描述,最后建立重庆雾的BP神经网络方法预测模型,并对重庆雾进行预报试验。检验结果表明,这种方法预报效果较好,基于客观分析的BP神经网络模型进行雾的短期预报具有一定的应用前景。  相似文献   

基于RBF神经网络的多系统钟差预报算法          下载免费PDF全文

王瑞  柴洪洲  潘宗鹏 《海洋技术学报》2019,38(6):56-61

针对海上条件下,对于实时定位应用,实时数据流无法下载的情况,文中提出一种基于RBF神经网络的卫星钟差预报算法,给出基函数的中心、方差以及隐含层到输出层的权值的计算方法,采用滑动窗口的方法,用样本数据训练后的网络预测下一个历元的钟差值,依次往后训练网络直到预测完整个时间段,通过实验验证了算法的可用性。短期预报中,GPS预报精度在1 ns以下,BDS和GLONASS在2～3 ns左右;长期预报中,GPS预报精度在几十纳秒左右,而BDS和GLONASS在几百纳秒左右,文中给出了相应的结果分析。  相似文献   

利用端点延拓提高LS + NN模型的UT1-UTC预报精度     

赵丹宁  雷雨 《天文研究与技术》2019,(3):294-299

现有UT1-UTC预报模式在进行周期项与残差项拟合分离时,通常没有考虑最小二乘拟合序列的端部效应,预报精度难以取得较大提高。针对最小二乘拟合存在的端部效应,首先采用灰色模型在UT1-UTC序列的两端进行数据延拓,形成一个新序列,然后对新序列进行最小二乘拟合,最后再联合最小二乘和神经网络(LS+NN)模型对UT1-UTC原始序列进行外推。结果表明,对UT1-UTC序列进行端点数据延拓再进行最小二乘拟合,能够有效地改善最小二乘拟合序列的端部效应;相对于常规LS+NN模型,端部效应改善的LS+NN模型的UT1-UTC预报精度有一定提高,尤其对中长期预报精度提高更为明显。  相似文献   

基于GA-WNN的极化SAR海洋溢油检测方法研究     

陈伟民  丁亚雄  宋冬梅  王斌  刘善伟  甄宗晋  张婷  杨敏 《海洋科学》2018,42(1):70-81

海洋溢油对海洋生态和人类生活带来严重的影响。由于合成孔径雷达(Synthetic Aperture Radar,SAR)具有全天时全天候的工作能力,在海洋溢油检测中发挥重要作用。目前,极化SAR是SAR探测技术的先进手段。本文利用6个极化特征进行溢油检测,通过对比分析这些特征对不同溢油的检测能力,得出单一极化特征在溢油检测中存在不足。通过J-M特征优选方法,提取出溢油检测识别度较高的特征影像,并利用遗传算法优化的小波神经网络(Genetic Algorithm-Wavelet Neural Network,GA-WNN)进行溢油检测。利用2套Radarsat-2全极化数据进行了方法验证,结果表明,该方法优于其他检测方法,溢油检测精度分别达到90.31%和95.42%。  相似文献   

A neural network model for liquefaction-induced horizontal ground displacement   总被引：3，自引：0，他引：3  

J. Wang  M. S. Rahman   《Soil Dynamics and Earthquake Engineering》1999,18(8):5603-568

The horizontal ground displacement generated by seismically induced liquefaction is known to produce significant damage to engineered structures. A backpropagation neural network model is developed to predict the horizontal ground displacements. A large database containing the case histories of lateral spreads observed in eight major earthquakes is used. The results of this study indicate that the neural network model serves as a reliable and simple predictive tool for the amount of horizontal ground displacement. As more data become available, the model itself can be improved to make more accurate displacement prediction for a wider range of earthquake and site conditions.  相似文献   

天然地震识别与震相自动拾取技术进展     

孙印  潘素珍  刘明军 《中国地震》2018,34(4):606-620

震相拾取是地震数据处理过程中最基本的步骤之一。在传统的人工拾取技术不能满足庞大的地震数据处理需求的情况下,震相自动拾取技术从产生到发展至今经历了漫长的过程。本文回顾并总结了震相自动拾取技术的发展状况,重点介绍了长短时窗法、赤池准则法、模板匹配技术、基于自相关盲搜索的FAST法、S波偏振分析法、人工智能方法等,以及近年发展起来的多频率震相识别、全波形叠加、二次方自回归模型等方法,同时分析了每种方法的优势和局限性。  相似文献   

Acquisition design for direct reflectivity and velocity estimation from blended and irregularly sampled data     

Shotaro Nakayama  Gerrit Blacquière  Tomohide Ishiyama 《Geophysical Prospecting》2019,67(8):2127-2146

Blended acquisition along with efficient spatial sampling is capable of providing high-quality seismic data in a cost-effective and productive manner. While deblending and data reconstruction conventionally accompany this way of data acquisition, the recorded data can be processed directly to estimate subsurface properties. We establish a workflow to design survey parameters that account for the source blending as well as the spatial sampling of sources and detectors. The proposed method involves an iterative scheme to derive the survey design leading to optimum reflectivity and velocity estimation via joint migration inversion. In the workflow, we extend the standard implementation of joint migration inversion to cope with the data acquired in a blended fashion along with irregular detector and source geometries. This makes a direct estimation of reflectivity and velocity models feasible without the need of deblending or data reconstruction. During the iterations, the errors in reflectivity and velocity estimates are used to update the survey parameters by integrating a genetic algorithm and a convolutional neural network. Bio-inspired operators enable the simultaneous update of the blending and sampling operators. To relate the choice of survey parameters to the performance of joint migration inversion, we utilize a convolutional neural network. The applied network architecture discards suboptimal solutions among newly generated ones. Conversely, it carries optimal ones to the subsequent step, which improves the efficiency of the proposed approach. The resultant acquisition scenario yields a notable enhancement in both reflectivity and velocity estimation attributable to the choice of survey parameters.  相似文献   

A new fluid factor and its application using a deep learning approach     

Changcheng Liu  Deva Prasad Ghosh  Ahmed Mohamed Ahmed Salim  Weng Sum Chow 《Geophysical Prospecting》2019,67(1):140-149

Amplitude interpretation for hydrocarbon prediction is an important task in the oil and gas industry. Seismic amplitude is dominated by porosity, the volume of clay, pore-filled fluid type and lithology. A few seismic attributes are proposed to predict the existence of hydrocarbon. This paper proposes a new fluid factor by adding a correct item based on the J attribute. The algorithm is verified through stochastic Monte Carlo modelling that contains various rock physical properties of sand and shale. Both gas and oil responses are separated by the new fluid factor. Furthermore, an approach based on the neural network model is trained using the deep learning method to predict the new fluid factor. The confusion matrix shows that this model performs well. This model allows the application of the new fluid factor in the seismic data. In this study, the Marmousi II data set is used to examine the performance of the new fluid factor, and the result is good. Most hydrocarbon reservoirs are identified in the shale–sandstone sequences. The combination of deep learning and the new fluid factor provides a more accurate way for hydrocarbon prediction.  相似文献   

基于人工神经网络的台风浪高快速计算方法     

周水华  洪晓  梁昌霞  江丽芳 《热带海洋学报》2020,39(4):25-33

采用2010—2017年南海5个浮标波高观测资料和中国气象局热带气旋最佳路径集中的热带气旋参数, 基于前馈型误差反向传播(Forward Feedback Back Propagation, FFBP)神经网络(Artificial Neural Network, ANN)方法, 分别建立了各浮标站的台风浪高快速计算模型。研究显示, 基于热带气旋中心坐标、中心最低气压、近中心最大风速、热带气旋中心与浮标之间的距离和方位4个参数建立的神经网络模型经反复训练后, 模型输出结果可以很好地拟合观测数据, 各浮标有效波高计算值与观测值的均方根误差小于0.3m, 平均相对误差为5.78%~7.23%, 相关系数大于0.9, 属高度相关。独立测试结果显示, “山竹”( 国际编号: 1822)影响期间有效波高最大值的神经网络模型预报结果与观测值基本吻合, 相对误差为-31.06%~0.98%, 但计算的最大值出现时间和观测情况不完全一致。该计算方法可应用于热带气旋影响期间的有效波高最大值计算, 因而在海洋工程领域和海洋预报领域具有应用前景。  相似文献