期刊文献+
共找到3篇文章
< 1 >
每页显示 20 50 100
面向频谱大数据处理的机器学习方法 被引量:21
1
作者 吴启晖 邱俊飞 丁国如 《数据采集与处理》 CSCD 北大核心 2015年第4期703-713,共11页
随着移动互联网与物联网的迅猛发展,个人无线设备的数量呈现指数级增长,随之产生的海量频谱数据与日俱增,频谱大数据的存在已成事实。同时,频谱赤字也日益严峻。为提高频谱利用率,有效的频谱大数据处理显得十分重要。本文从无线通信的角... 随着移动互联网与物联网的迅猛发展,个人无线设备的数量呈现指数级增长,随之产生的海量频谱数据与日俱增,频谱大数据的存在已成事实。同时,频谱赤字也日益严峻。为提高频谱利用率,有效的频谱大数据处理显得十分重要。本文从无线通信的角度,首先给出了频谱大数据的定义并分析了它的基本特征;然后总结了一些对于频谱大数据分析与利用颇具前景的机器学习方法,如分布式和并行式学习、极速学习、核学习、深度学习、强化学习、博弈学习和迁移学习;最后给出了几个开放性话题和研究趋势。 展开更多
关键词 大数据 频谱大数据 机器学习 数据挖掘 无线通信 物联网
下载PDF
机器学习在频谱大数据分析与处理上的应用 被引量:5
2
作者 史通 王洁 +2 位作者 罗畅 肖军 王世强 《火力与指挥控制》 CSCD 北大核心 2018年第6期47-51,共5页
作为通信的主要方式,无线通信产生了极具价值的频谱大数据。而机器学习作为新兴的智能算法,可以挖掘出频谱大数据中的有用信息,作出正确的频谱预测和决策,以提高频谱资源的利用率。对机器学习和频谱大数据进行了简要介绍;结合频谱大数... 作为通信的主要方式,无线通信产生了极具价值的频谱大数据。而机器学习作为新兴的智能算法,可以挖掘出频谱大数据中的有用信息,作出正确的频谱预测和决策,以提高频谱资源的利用率。对机器学习和频谱大数据进行了简要介绍;结合频谱大数据的四大特点,讨论了用于频谱大数据分析的4种机器学习方法——分布并行学习、极速学习机、核学习、深度学习;又以频谱大数据的智能预测和决策为目的,讨论了用于频谱大数据处理的两种机器学习方法——强化学习、博弈学习;最后对机器学习方法在频谱大数据上的应用前景进行了展望。 展开更多
关键词 无线通信 频谱大数据 机器学习 深度学习 博弈学习
下载PDF
Numerical simulation and preliminary analysis on ocean waves during Typhoon Nesat in South China Sea and adjacent areas 被引量:3
3
作者 王际朝 张杰 杨俊钢 《Chinese Journal of Oceanology and Limnology》 SCIE CAS CSCD 2014年第3期665-680,共16页
Using the wave model WAVEWATCH III(WW3), we simulated the generation and propagation of typhoon waves in the South China Sea and adjacent areas during the passage of typhoon Nesat(2011). In the domain 100°–145&#... Using the wave model WAVEWATCH III(WW3), we simulated the generation and propagation of typhoon waves in the South China Sea and adjacent areas during the passage of typhoon Nesat(2011). In the domain 100°–145°E and 0°–35°N, the model was forced by the cross-calibrated multi-platform(CCMP) wind fi elds of September 15 to October 5, 2011. We then validated the simulation results against wave radar data observed from an oil platform and altimeter data from the Jason-2 satellite. The simulated waves were characterized by fi ve points along track using the Spectrum Integration Method(SIM) and the Spectrum Partitioning Method(SPM), by which wind sea and swell components of the 1D and 2D wave spectra are separated. There was reasonable agreement between the model results and observations, although the WW3 wave model may underestimate swell wave height. Signifi cant wave heights are large along the typhoon track and are noticeably greater on the right of the track than on the left. Swells from the east are largely unable to enter the South China Sea because of the obstruction due to the Philippine Islands. During the initial stage and later period of the typhoon, swells at the fi ve points were generated by the propagation of waves that were created by typhoons Haitang and Nalgae. Of the two methods, the 2D SPM method is more accurate than the 1D SIM which overestimates the separation frequency under low winds, but the SIM method is more convenient because it does not require wind speed and wave direction. When the typhoon left the area, the wind sea fractions decreased rapidly. Under similar wind conditions, the points located in the South China Sea are affected less than those points situated in the open sea because of the infl uence of the complex internal topography of the South China Sea. The results reveal the characteristic wind sea and swell features of the South China Sea and adjacent areas in response to typhoon Nesat, and provide a reference for swell forecasting and offshore structural designs. 展开更多
关键词 TYPHOON WAVEWATCH III (WW3) cross-calibrated multi-platform (CCMP) South China Sea significant wave height SWELL
下载PDF
上一页 1 下一页 到第
使用帮助 返回顶部