GA-BP神经网络在罗源湾口波浪模拟研究中的应用

由于BP神经网络存在收敛速度慢和易于陷入极小值等缺点,引入遗传算法(GA)对网络的权值和阈值加以优化,并采用不同组合的输入因子和样本数,对福建省罗源湾口的波浪进行模拟研究.对输入因子的分析结果表明,研究区域的波浪主要受台湾海峡波浪传播影响,与局地气象因子(风速、风向、海气温差)的相关性较弱.训练样本试验表明,30 d以上的波浪历史数据可使GA-BP神经网络充分学习研究区域的波浪特征,从而实现对波浪要素的高精度模拟.模拟结果显示,对春、夏季实测波浪数据的模拟效果均很好,其中相关性分别为0.967和0.938,均方根误差分别为0.112 m和0.107 m,表明GA-BP神经网络在近岸波浪模拟预报中有较广阔的应用前景.     

Onboard GRB trigger algorithms of SVOM-GRM

The Gamma-Ray Monitor （GRM） is a high energy detector onboard the future Chinese-French satellite named the Space-based multi-band astronomical Variable Object Monitor which is dedicated to studies of gamma-ray bursts （GRBs）. This paper presents an investigation of the algorithms that look for GRBs by searching for a significant increase in the photon count rate for the computer onboard GRM. The trigger threshold and trigger efficiency, which are based on a given sample of GRBs, are calculated with the algorithms. The trigger characteristics of onboard instruments GRM and ECLAIRs are also analyzed. In addition, the impact of solar flares on GRM is estimated, and a method to distinguish solar flares from GRBs is investigated.     

A New Strategy for Solving a Class of Constrained Nonlinear Optimization Problems Related to Weather and Climate Predictability

There are three common types of predictability problemsin weather and climate, which each involve different constrained nonlinearoptimization problems: the lower bound of maximum predictable time, theupper bound of maximum prediction error, and the lower bound of maximumallowable initial error and parameter error. Highly efficient algorithmshave been developed to solve the second optimization problem. And thisoptimization problem can be used in realistic models for weather and climateto study the upper bound of the maximum prediction error. Although afiltering strategy has been adopted to solve the other two problems, directsolutions are very time-consuming even for a very simple model, whichtherefore limits the applicability of these two predictability problems inrealistic models. In this paper, a new strategy is designed to solve theseproblems, involving the use of the existing highly efficient algorithms forthe second predictability problem in particular. Furthermore, a series ofcomparisons between the older filtering strategy and the new method areperformed. It is demonstrated that the new strategy not only outputs thesame results as the old one, but is also more computationally efficient.This would suggest that it is possible to study the predictability problemsassociated with these two nonlinear optimization problems in realisticforecast models of weather or climate.     

Testing an earthquake prediction algorithm

A test to evaluate earthquake prediction algorithms is being applied to a Russian algorithm known asM8 TheM8 algorithm makes intermediate term predictions for earthquakes to occur in a large circle, based on integral counts of transient seismicity in the circle. In a retroactive prediction for the period January 1, 1985 to July 1, 1991 the algorithm as configured for the forward test would have predicted eight of ten strong earthquakes in the test area. A null hypothesis, based on random assignment of predictions, predicts eight earthquakes in 2.87% of the trials. The forward test began July 1, 1991 and will run through December 31, 1997. As of July 1, 1995, the algorithm had forward predicted five out of nine earthquakes in the test area, which success ratio would have been achieved in 53% of random trials with the null hypothesis.     

Finding shortest paths on real road networks: the case for A*

The problem of identifying the shortest path along a road network is a fundamental problem in network analysis, ranging from route guidance in a navigation system to solving spatial allocation problems. Since this type of problem is solved so frequently, it is important to craft an approach that is as efficient as possible. Based upon past research, it is generally accepted that several efficient implementations of the Dijkstra algorithm are the fastest at optimally solving the ‘one‐to‐one’ shortest path problem (Cherkassky et al. 1996 Cherkassky, B. V., Goldberg, A. V. and Radzik, T. 1996. Shortest paths algorithms: theory and experimental evaluation.. Mathematical Programming: Series A and B, 73: 129–174. [Crossref], [Web of Science ®] , [Google Scholar]). We show that the most efficient state‐of‐the‐art implementations of Dijkstra can be improved by taking advantage of network properties associated with GIS‐sourced data. The results of this paper, derived from tests of different algorithmic approaches on real road networks, will be extremely valuable for application developers and researchers in the GIS community.     

基于稀疏AR模型的潮流信号建模与预报

潮流信号处理与预报在很多方面具有非常重要的意义和价值。本文引入信号稀疏表示理论,构建一种稀疏AR模型,寻找各潮流数据间的历史关联性,并进行预报分析。首先由实测潮流信号进行常规AR建模,获得一组过完备稀疏基;其次随机从该过完备稀疏基抽取部分建立欠定方程组,利用稀疏优化算法获得最稀疏的AR系数;多次重复上一步,获得稀疏AR系数的平均以增强稀疏AR模型的稳定性;最后利用这些稀疏AR系数来重构或预测潮流信号。文章针对实测潮流信号,特别是存在多峰值有回流现象的潮流信号,进行了稀疏AR建模与预测的多次实验。实验结果与传统的潮流信号调和预报方法相对比,发现基于稀疏AR模型的潮流预报对于潮流存在多变的现象时,具有明显优越性,从回报结果来看,稀疏AR模型的潮流预报均方差明显小于传统潮流调和分析预报方法。     

计算数据域特征值的两种算法

设计两种计算数据域特征值的算法。此特征值通常是用线性反馈移位寄存器（ＬＦ－ＳＲ）组成的特征分析器得到的．在数据已按字节存放和速度要求不高的场合下，本算法将是十分方便和有用的．     

卧虎山水库不同算法LS因子值适用性分析

在总结以往LS因子相关算法研究基础上,结合McCool的LS因子参照值,对LS算法的适用性进行评价分析。研究表明,除去复合算法和Remortel修正算法外,其他不同算法LS因子值都小于参照值;研究区LS因子值的最优算法为基于Remortel迭代运算的修正算法;其次为结合刘宝元的陡坡公式和Remortel改进L指数因子迭代运算的复合算法,以及Remortel第4版AML程序算法;再次为B?hner算法,而Moore算法和Desmet算法,由于其与参照值的相关性相对较差,而且其RMSE相对较大,不推荐在该区使用。     

Seismic hazard computations for regions with low earthquake activity — A case study for the Belgium,The Netherlands and NW Germany area

The algorithms to evaluate seismic hazard, used and/or developed by five teams participating in the TERESA project, applied to the low seismicity area Belgium, The Netherlands and NW Germany are compared. The main differences in the results can be explained by the majority criterion of Egozcue et al. (1989), the differences in the upper bound and zonification and, in some cases, by a higher attenuation.     

Quantifying flows along hydrological pathways by applying a new filtering algorithm in conjunction with master recession curve analysis

Quantifying the proportion of the river hydrograph derived from the different hydrological pathways is essential for understanding the behaviour of a catchment. This paper describes a new approach using the output from master recession curve analysis to inform a new algorithm based on the Lyne and Hollick ‘one‐parameter’ signal analysis filtering algorithm. This approach was applied to six catchments (including two subcatchments of these) in Ireland. The conceptual model for each catchment consists of four main flow pathways: overland flow, interflow, shallow groundwater and deep groundwater. The results were compared with those of the master recession curve analysis, a recharge coefficient approach developed in Ireland and the semi‐distributed, lumped and deterministic hydrological model Nedbør‐Afstrømings‐Model. The new algorithm removes the ‘free variable’ aspect that is typically associated with filtering algorithms and provides a means of estimating the contribution of each pathway that is consistent with the results of hydrograph separation in catchments that are dominated by quick response pathways. These types of catchments are underlain by poorly productive aquifers that are not capable of providing large baseflows in the river. Such aquifers underlie over 73% of Ireland, ensuring that this new algorithm is applicable in the majority of catchments in Ireland and potentially in those catchments internationally that are strongly influenced by the quick‐responding hydrological pathways. Copyright © 2013 John Wiley & Sons, Ltd.