海洋变分同化背景场误差协方差矩阵构造技术

为改进背景场误差协方差矩阵构造技术,提高海洋资料变分同化效果,首先将背景场误差协方差矩阵分解为背景误差标准偏差矩阵、非平衡变量空间相关算子以及多变量间平衡算子3部分,然后利用广义扩散方程模拟空间相关算子,联合引入温盐约束关系、线性状态方程、海表高度诊断方程、静力平衡、地转平衡关系模拟多变量平衡,最后给出了一种适用于海洋资料变分同化的背景场误差协方差矩阵构造方案。单点海表温度观测模拟同化试验表明,该方案可实现观测信息在三维空间、多变量间的传播,充分体现了背景场误差协方差矩阵在同化中的作用。     

Preliminary results of a new global ocean reanalysis

Using a new global ocean reanalysis of the second generation Global Ocean Data Assimilation System of the Beijing Climate Center(BCC_GODAS2.0) spanning the period 1990-2009,we firstly quantify the accuracy of BCC_GODAS2.0 in representing the temperature and salinity by comparing with OISST and SODA data.The results show that the assimilation system may effectively improve the estimations of temperature and salinity by assimilating all kinds of observations,especially in the equatorial eastern Pacific.Moreover,the root mean square errors of monthly temperature and salinity are respectively reduced by 0.53℃ and 0.28 psu,compared with the model control simulation results.Then,the applicability of this ocean reanalysis for sea surface temperature(SST) anomaly variability in the tropical Pacific is evaluated with the observational HadISST data.The NINO3 index of the new reanalysis shows a good agreement with that of HadISST,with a correlation of 93.6%.Variations in SST from BCC_GODAS2.0 are similar to those obtained from HadISST data along the equator,showing the major large zonal-scale features such as the strong magnitude of seasonal cycle.The amplitude of SST anomaly standard deviation in the equatorial eastern Pacific is also closer to observations(HadISST) than NCEP GODAS does.Besides,the first two leading empirical orthogonal function(EOF) modes of the monthly SST anomalies over the tropical Pacific region are explored.The EOF1 pattern of BCC_GODAS2.0 captures a traditional El Ni o pattern,which improves magnitudes of the positive SST anomaly in the cold tongue of the eastern Pacific.The EOF2 pattern exhibits a El Ni o Modoki pattern.Comparatively,the EOF2 pattern of BCC_GODAS2.0 extends more strongly toward the subtropics.It also overcomes the problem that negative loadings are confined in the narrow equatorial eastern Pacific.Consequently,the magnitude and spatial distribution of the leading EOF patterns of BCC_GODAS2.0 are well consistent with those of HadISST.     

The time-saving numerical method for GPS/MET observation operator

The global positioning system (GPS) ray-shooting method is a self-sufficient observation operator inGPS/MET (meteorology) data variational assimilation linking up the GPS observation data and the atmosphere state vari-ables. But it cannot be applied to data assimilation and operational prediction so far because of huge computations. In order to reduce the amount of computation, a 2-order time-saving symplectic scheme is used to solve the equations of the GPS ray trajectory, due to its separable Hamiltonian nature, and good results are achieved. Not only does it save 75 % of CPU time taken by the old GPS ray-shooting model with 4th-order Runge-Kutta method , but also it improves the sim-ulation accuracy to some extent.     

耦合遗传算法的数据同化系统误差处理方法

针对数据同化系统中的误差估计与处理问题,介绍了集合滤波数据同化系统中各种误差来源及特征;侧重于在集合数据同化中为防止滤波发散的乘数放大法、附加放大法和松弛先验法等模型误差处理方案,利用经典的非线性模型-Lorenz模型开展了数值试验.在此基础上,提出了一种耦合遗传寻优算法的数据同化系统,来解决以往的误差调节因子由反复实...     

面向缩聚模型的结构损伤识别研究

为了减少结构健康监测系统中传感器的数量,创新性地结合缩聚原理和自适应二次乘方和误差法提出了一种面向缩聚模型的结构损伤识别方法.根据模型缩聚原理对结构有限元模型进行了缩聚,依据缩聚后有限元模型、利用自适应二次乘方和误差法对结构参数进行了识别和追踪,从而实现了结构损伤的识别.对某一大型平面桁架结构进行了仿真,期间考虑了白噪声和地震波的影响;对某一悬臂梁结构进行了实验,以研究白噪声、正弦、拟El-Centro 3种激励下的6种情况.结果表明:所提方法只需使用少量的传感器便可精确地识别出结构中损伤的发生时刻和损伤的程度及位置,仿真结果的最大误差小于3%,实验结果的最大误差小于5%.     

一维逆时热传导初值问题的反演方法

利用同化方法对一维逆时热传导问题进行了研究。首先构造一个正问题,其初值为给定时刻的观测值,目的是获得足够多的观测值;然后利用变分同化对初值进行反演;最后进行数值模拟。数值模拟结果显示该方法可行且具有一定的抗噪性。     

基于VMD-EWT-IASSP-EBILSTM的短期电力负荷预测

电力系统在国家工业基础设施中起着举足轻重的作用,维持系统负荷高精度预测是保障电力系统高效供应的关键。针对负荷数据的非平稳性、随机性与非线性,负荷预测误差较大的问题,结合变分模态分解(variational mode decomposition, VMD)、经验小波变换(empirical wavelet transform, EWT)、改进的空洞卷积金字塔模块(improved atros spatial pyramid pooling, IASSP)、集成双向长短时记忆模块(ensemble BiLSTM,EBiLSTM),提出了一种短期电力负荷预测模型。为解决负荷数据的非平稳性引起的模型预测波动问题,通过变分模态分解方法与经验小波变换的结合分解为若干子序列,显著降低了原始负荷序列的复杂性;为提高模型预测精度,将分解的负荷子序列利用过零率指标划分高低频序列,在低频序列中构建一种时序依赖捕获模块EBiLSTM提取长期负荷特征,高频序列中构建特征提取模块IASSP提取局部负荷特征,最后累加各子序列的预测结果,实现电力系统负荷的短期预测。选取行业通用客观评价指标：平均绝对误差、均方根误差,...     

基于神经网络内模控制的近红外光谱定量分析方法

以近红外光谱法为基础测定方法,结合内模控制,论述了采用自适应神经网络建立校正模型测定石油化工产品组成的可行性.基于dSPACE硬件平台,实验以直馏柴油、加氢精制柴油和催化裂化柴油为校正模型的训练样本,对自适应神经网络校正模型进行了检验,实验结果表明:该方法响应快、误差小、鲁棒性强,在近红外长波区内,校正样品和验证样品的均方误差小于10-6.     

动态最优插值方法及其同化应用研究

根据常规最优插值算法,在引入时间关联的基础上,详细推导了动态最优插值法的具体同化表达形式,利用这些公式,可以处理常见的在海洋观测数据同化中存在的时间错位问题.在陆架海模式HAM-SOM基础上,以7月份的渤海海表温度为例,验证了同化公式的可行性,并得到了较好的同化结果.验证结果表明:该方案能够兼顾观测数据与背景数据双方的特性,同化后的结果更加接近实际情况.     

A modeling study of relation between cloud amount and SST over western tropical pacific cloudy regions during TOGA COARE

The relationship between cloud amount and sea surface temperature (SST) over western tropical Pacific cloudy regions during TOGA COARE is investigated based on hourly grid simulation data from a two-dimensional coupled ocean-cloud resolving atmosphere model. The model is forced by the large-scale vertical velocity and zonal wind observed and derived from TOGA COARE for a 50-day period. The cloud amount becomes smaller when the ocean surface gets warmer, which is similar to previous relations obtained from observational analyses. As SST increases, the atmospheric temperature increases whereas the surface sensible heat flux decreases. The atmospheric water vapor is not sensitive to SST whereas the surface evaporation flux decreases as SST increases. These indicate that the oceanic effects do not play an important role in determining atmospheric heat and water vapor budgets. The cold atmosphere produces a larger amount of ice clouds that cover a larger area than the warm atmosphere does. The large amounts of ice clouds lead to cooling of the ocean surface through reflecting large amount of solar radiation back to the space. Thus, the negative correlation between the cloud amount and SST only accounts for the important atmospheric effects on the ocean.     

Evaluation of a 3dVAR system for the South China Sea

The authors evaluate a three-dimensional variational (3dVAR) system for the South China Sea (SCS) in this study. The assimilation method applied in the system takes into consideration error correlation along each ground track and uses recursive filter for optimization. Data from three R/V cruises during the spring and summer of 1998 and the summer of 2000 are used to evaluate the system. The root-mean-square error and bias are reduced significantly and when the altimeter data is assimilated, the distribution of the error is much closer to the Gaussian distribution. Precipitation and river discharge in the southwestern SCS are reproduced, and the variability of sea surface height is efficiently transferred to the subsurface. The 3dVAR system performs well for each of the three cruises, suggesting that it is steady for routine usage.     

基于自适应粒子群算法的重油热解模型参数估计

通过构造一个合适的目标函数,将化工模型参数估计问题转化为一个多维数值优化问题,然后提出一种参数自适应调整和维变异的改进粒子群优化算法来求解该问题。该算法首先利用佳点集方法初始化种群以保证粒子的多样性。惯性权重和学习因子随进化过程自适应调整,从而协调算法的全局和局部搜索能力。为了避免算法陷入局部最优,对收敛度最小的维进行变异。几个标准测试问题的实验结果表明该算法具有较强的全局寻优能力。最后将改进粒子群算法应用到重油热解模型参数估计中,并与基本遗传算法（SGA）和粒子群优化算法（SPS0）进行比较。研究结果表明：本文得到的平均相对误差为5．62％,比SGA和SPSO分别低1．08％和0．50％。     

Evaluation of a 3dVAR system for the South China Sea

The authors evaluate a three-dimensional variational （3dVAR） system for the South China Sea （SCS） in this study. The assimilation method applied in the system takes into consideration error correlation along each ground track and uses recursive filter for optimiza- tion. Data from three R/V cruises during the spring and summer of 1998 and the summer of 2000 are used to evaluate the system. The root-mean-square error and bias are reduced significantly and when the altimeter data are assimilated, the distribution of the error is much closer to the Gaussian distribution. Precipitation and river discharge in the southwestern SCS are reproduced, and the variability of sea surface height is efficiently transferred to the subsurface. The 3dVAR system performs well for each of the three cruises, suggesting that it is steady for routine usage.     

基于多种空间滤波方案的海温资料同化实验

为解决海洋资料同化中空间滤波方案选择的主观性问题,基于最优插值算法,借助现场资料分析系统(ISAS),分别试验高斯函数(Gauss)、二阶自回归函数(SOAR)以及贝塞尔函数(Bessel)作为拟合函数的空间滤波方案,同化全球实时海洋观测计划(Argo)温度剖面资料,并从多个角度比较分析场的质量,以期为寻找最优空间滤波方案提供借鉴。实验表明,Bessel函数方案同化结果的均方根误差最小,SOAR函数方案的均方根误差最大,但是在个别层次上,Bessel函数方案的误差大于其他2种方案;在多个典型层次上,Bessel函数方案的分析场能够分辨更多中小尺度的信息,这与Bessel函数本身可以揭示不同波谱能量误差分布的特性有关。结合海面动力地形数据和Argo剖面的分布状况分析,几个显著的中尺度冷暖中心分布是符合物理海洋学常识的。需要指出的是,在网格分辨率未提高的前提下,这些中小尺度的信号需要借助其他辅助资料和分析方法进行甄别。     

基于VMD改进算法的气体管道泄漏检测

为提取气体管道声波信号中的泄漏成分并进行重构,提出一种结合VMD( VMD: Variational Mode Decomposition) 和误差能量算法的特征提取方法。该方法首先利用油气管道泄漏检测系统模拟气体管道的微小泄漏,并采集泄漏声波信号; 然后利用VMD 算法将采集到的泄漏声波信号分解为一系列带宽受限的固有模态;随后,使用误差能量算法选择有效模态; 最后,利用有效模态进行信号重构。通过仿真分析发现,该改进算法可以提取有效模态,利用该方法处理气体管道微小泄漏声波信号,能有效滤除噪声并重构原始信号。     

大气边界层模式参数优化的数值试验

运用变分方法修正的正大气边界层模式中的若干参数，使模式解与观测拟合，利用ＨＥＩＦＦ实验观测资料所作的数值试验表明，参数修正后，不但在资料拟合阶段，而且在随后１８ｈ的预报阶段，模式模拟的边界层内感热，水汽和辐射通量均有显著改善。     

基于变指数分数阶全变差和整数阶全变差的图像恢复算法

结合变指数全变差(totalvariation, TV)和整数阶TV,提出一种变分图像恢复算法。该变分问题的能量泛函主要分为三个部分:变指数p(x)的分数阶TV正则化项、整数阶TV正则化项和数据保真项。该模型中的指数p(x)是与图像的梯度信息有关的函数。在理论上,由于分数阶导数和整数阶导数的结合,使得所提方法不仅能有效地去除图像噪音,保护图像的边界高频信息,还能更好地保留图像的纹理细节等中低频信息,同时可以极大地消除图像处理中产生的阶梯效应和散斑效应。在模型的求解上,利用变分法可以简单地将极小化泛函的优化问题转化为梯度下降流方程。最后,通过模拟数据和真实数据对本文所提方法进行了验证。试验结果表明,该方法可以去除噪声的同时,有效保持边界和纹理细节,并且对噪声是鲁棒的,具有一定的实际应用价值。     

遗传算法拟合变异函数的参数

 由于堆积体的空间变异特殊性,用加权多项式回归法进行球状模型的理论变异函数拟合误差很大,有时甚至无法拟合出变异函数.为解决这一问题,结合遗传算法原理,充分利用遗传算法在求解非线性问题时具有全局寻优的特点,拟合出球状模型的理论变异函数参数,并以小湾水电站左岸堆积体边坡的变形监测数据为例进行了计算.经过多方法对实例进行比较分析表明,该方法简便、通用,具有较高的拟合精度,是非线性、不连续可微模型参数估计的好方法.     

耦合集合卡尔曼滤波的非线性参数估计

准确的参数对于非线性模型或者函数有极为重要的意义。但是在实际应用中,输入的参数常常带有一定的偏差,因此在预测中非线性模型(函数)往往得不到满意的结果。把集合卡尔曼滤波方法引入到非线性模型(函数)的参数估计中,并采用基于联合状态向量的方法在同化时刻同时更新模型(函数)的状态和参数,该方法可以通过同化观测数据,动态地调整参数和结果,使得参数能够自适应地变化,同时也能较好地释放积累的模型误差。将该方法应用于随时间变化的二维非线性模型(函数)中,实验表明,该方法能够准确地估计出模型参数值,同时也表现出很好的鲁棒性和自适应性。