基于于田气象站48a（1960--2007年）逐月平均气温和降水量数据，借助小波分析、人工神经网络等方法研究了塔里木盆地南缘于田绿洲近50a来在全球气候变暖的背景下气候变化的特征和总体趋势。结果表明，二维小波系数图可以清楚地显示出气...基于于田气象站48a（1960--2007年）逐月平均气温和降水量数据，借助小波分析、人工神经网络等方法研究了塔里木盆地南缘于田绿洲近50a来在全球气候变暖的背景下气候变化的特征和总体趋势。结果表明，二维小波系数图可以清楚地显示出气候变化在不同时间尺度上的演变特征，并能对未来的气候演变趋势做出定性估计。小波方差图显示于田绿洲逐月平均气温存在4a为主的短周期和以11～13a为主的长周期；逐月降水存在以7a为主的短周期和以准12a、27a左右为主的长周期。RBFN（Radial Basis Function Network）模型预测的结果显示，该区2008--2012年降水量的变化处于相对高值期，气温变化处于相对低值期。近50a于田绿洲气候总体趋于暖湿。展开更多
An improved RBFN algorithm is proposed.Its training algorithm includes K-mean cluster algorithm and improved recursive method with discounted measurements.The former is used in on-line leaming network centers,and the ...An improved RBFN algorithm is proposed.Its training algorithm includes K-mean cluster algorithm and improved recursive method with discounted measurements.The former is used in on-line leaming network centers,and the latter is used to adjust netword weights.Simulation results about estimating the catalyst activity in an industrial fixed-bed reactor show that the training algorithm provides simple structure,good track performance,fast learning speed.展开更多
文摘基于于田气象站48a（1960--2007年）逐月平均气温和降水量数据，借助小波分析、人工神经网络等方法研究了塔里木盆地南缘于田绿洲近50a来在全球气候变暖的背景下气候变化的特征和总体趋势。结果表明，二维小波系数图可以清楚地显示出气候变化在不同时间尺度上的演变特征，并能对未来的气候演变趋势做出定性估计。小波方差图显示于田绿洲逐月平均气温存在4a为主的短周期和以11～13a为主的长周期；逐月降水存在以7a为主的短周期和以准12a、27a左右为主的长周期。RBFN（Radial Basis Function Network）模型预测的结果显示，该区2008--2012年降水量的变化处于相对高值期，气温变化处于相对低值期。近50a于田绿洲气候总体趋于暖湿。
文摘An improved RBFN algorithm is proposed.Its training algorithm includes K-mean cluster algorithm and improved recursive method with discounted measurements.The former is used in on-line leaming network centers,and the latter is used to adjust netword weights.Simulation results about estimating the catalyst activity in an industrial fixed-bed reactor show that the training algorithm provides simple structure,good track performance,fast learning speed.
