首页 | 本学科首页   官方微博 | 高级检索  
相似文献
 共查询到19条相似文献,搜索用时 250 毫秒
1.
宿州春季重旱发生年份的灰色神经网络预测模型   总被引:1,自引:1,他引:0  
宿州春季严重干旱序列数据偏少,可用传统GM(1,1)模型进行预测,但由于序列变化幅度较大,预测效果不理想。本文利用灰色与BP神经网络组合模型对宿州春季重旱发生年份进行预测,即首先弱化序列变化幅度,并改进GM(1,1)模型导数信息处理方式,构建可逼近精度目标的m—GM(1,1)预测模型,然后应用BP神经网络对m—GM(1,1)模型的残差进行拟合,对m—GM(1,1)预测模型进行修正。结果表明,灰色神经网络组合模型的精度(|Q|=0.0045)比单一的1.7-GM(1,1)模型(|Q|=4.18)和传统的单一GM(1,1)模型精度(|Q|=9.36)提高许多。预测2005年后的下一个宿州市春季严重干旱发生年份为2009年,可以作为预报当地春季干旱时的参考,并结合其他方法作进一步预测,为当地防灾减灾提供科学依据。  相似文献   

2.
基于分数阶灰色模型的农业用水量预测   总被引:1,自引:0,他引:1  
针对农业用水量序列的振荡特性以及传统灰色预测模型的过拟合问题,该文提出分数阶灰色预测模型。将农业用水量振荡序列转化为单调递减非负序列,并以转化序列为基础,根据"阶数最大(或最小)"、"历史数据拟合最好"2个目标函数构造优化模型,采用改进NSGA-Ⅱ(non-dominated sorting genetic algorithm Ⅱ,NSGA-Ⅱ)进行模型求解。根据验证集拟合结果优选出模型阶数,结合分数阶反向累加灰色模型(fractional order reverse accumulation grey model),以通辽市和宝鸡市为例,进行农业用水量的预测。为了检验模型性能,将该文模型分别与传统GM(1,1)模型、自回归模型、基于小波分析理论组合模型进行对比。结果表明,该文模型对于通辽市、宝鸡市与鄂尔多斯市的农业用水量预测的相对误差分别为2.33%、0.31%和1.77%。同时,该文模型预测误差最小(比自回归模型分别低1.11%(通辽)、6.18%(宝鸡);比传统GM(1,1)模型分别低3.32%(通辽)、0.97%(宝鸡)),具有一定实用性,研究结果可为区域农业用水量预测提供依据。  相似文献   

3.
GM(1,1)模型改进技术在咸阳市地下水动态预测中的应用   总被引:2,自引:0,他引:2  
郝健  刘俊民 《水土保持研究》2011,18(3):252-254,259
针对地下水埋深变化离散性程度较大的咸阳市,采用GM(1,1)模型改进技术对其地下水动态进行预测研究,为地下水埋深的准确预测提供支持。以灰色理论GM(1,1)模型为基础,运用滑动平均法对离散性程度较大的原始序列进行改造,使原始数据的变化变得缓慢,再利用改造后的序列建立GM(1,1)*模型,以咸阳市地下水埋深资料为研究对象,进行地下水动态预测,并与未改进的GM(1,1)模型的预测结果进行比较。咸阳市地下水动态的预测结果显示,该区地下水埋深有逐年减小的趋势,说明该区地下水资源得到了有效的保护与利用。利用2001-2007年的地下水埋深资料建立GM(1,1)*模型进行预测,相较于实测数据,GM(1,1)*模型的预测结果科学合理;相较于未改进的GM(1,1)模型的预测结果,改进后的GM(1,1)*模型具有更高的预测精度和实用性。GM(1,1)模型改进技术的应用,减小了原始序列的离散性程度,提高了预测精度,为地下水动态预测提供一种新思路。  相似文献   

4.
土地利用规划修编中粮食产量预测方法比较   总被引:2,自引:0,他引:2  
为探索提高土地利用总体规划修编中粮食产量预测精度的方法,该文应用1988-2005年晋城市粮食产量相关数据的分析,对线性回归模型、灰色GM(1,1)模型和灰色多元线性回归组合模型3种粮食产量预测方法进行了研究.首先,运用灰色关联分析对影响粮食产最的影响因素作出关联因子排序;其次,在灰色关联分析的基础上选取主要影响因子;再次,利用灰色GM(1,1)模型得到主要影响因素的预测值,同时,利用原始数据建立多元线性回归模型;最后,将灰色GM(1,1)模型的预测结果作为多元线性回归模型的输入值,得到灰色多元线性回归组合模型.通过比较这3种粮食产量预测方法的预测结果,得出灰色多元线性回归组合模型最适宜于晋城市粮食产量的预测.该研究可提高土地利用总体规划编制的科学性.  相似文献   

5.
基于新陈代谢GM(1,1)模型的玉米叶长动态预测   总被引:1,自引:1,他引:0  
为实现虚拟玉米生长动态预测,本文以玉米棒三叶为对象,采用新陈代谢 GM(1,1)模型分析了不同施氮水平(150、300和450 kg/hm2)下玉米叶片长度的生长动态,结果表明:各组数据序列所建模型的均方差比值均小于0.0811,平均相对误差均小于0.0471,模型精度优于二级,具有较高的准确性,表明新陈代谢GM(1,1)模型能有效模拟玉米叶长动态变化,具有一定的可行性,研究结果可为玉米生长过程中的动态变化模拟提供参考.  相似文献   

6.
传统灰色动态预测模型GM(1,1)未能考虑各个变量之间的权重,在收敛计算时易出现无收敛解的缺陷,因此引入改进的GM(1,1)模型。该模型引入权重系数对每个变量赋予权重值,从而解决传统模型存在无收敛解的缺陷,并将改进的GM(1,1)模型用于辽西地区地下水预测。结果表明改进的GM(1,1)模型可解决传统模型无收敛解的局限,且在地下水预测中具有较好的精度,可用于辽西地区地下水的预测。研究为辽西地区地下水预测提供参考。  相似文献   

7.
基于产量阶段的河南省夏玉米灰色——马尔柯夫预测模型   总被引:1,自引:3,他引:1  
针对较长产量序列纵向及横向可比性较差的问题 ,提出用划分产量阶段的方法进行分段处理。用 GM(1,1)灰色模型分段模拟预测河南省夏玉米单产 ,根据相对气象产量划分产量状态 ,再用马尔柯夫模型转移概率预测系统未来发展方向并修正前者的预报结果。通过模拟和预报 ,准确率都有所提高  相似文献   

8.
灰色GM(1,1)-小波变换-GARCH组合模型预测松花江流域水质   总被引:1,自引:0,他引:1  
为了研究松花江流域水质变化情况,预测未来水质变化趋势以及对松花江流域水质的保护提供理论依据和决策方案,通过对松花江抚远段2012年前15周的实测溶解氧(dissolved oxygen,DO)、高锰酸盐指数CODMn、氨氮NH3-N数据分析,以灰色GM(1,1)模型、小波分解与重构和广义自回归条件异方差(generalized auto-regressive conditional heteroskedasticity,GARCH)模型为基础,建立了灰色GM(1,1)和灰色GM(1,1)-小波变换-GARCH组合的混合预测模型,并以抚远段实测DO、CODMn、NH3-N数据为实例进行验证,预测结果极显著(P0.01),预测误差分别为3.39%、8.56%、7.83%,表明该预测模型精度较高,适用于对水质变化的预测研究。最后,利用该模型对松花江抚远、黑河、嘉荫和同江段2013年前8周的4个污染指标进行预测分析,预测结果与实测数据误差较小,基本符合水质未来变化趋势,为相关部门对松花江流域水质预测和保护提供参考。  相似文献   

9.
从山西省阳城县1957-2013年春季晚霜冻出现日期中找出最晚出现日期作为0点,求出各晚霜冻日期与0点的相对日期,将灾变点的相对日期确定为灾变阈值。绘制相对日期年际变化曲线和阈值水平线,把二者第二个交点及之后各交点分别到第一个交点的间距作为灰色GM(1,1)模型建模的原始数据序列,从而建立GM(1,1)传统模型,利用优化序列、优化背景值和灰数等维递补方法逐步改进模型,并对各改进型模型进行相对误差、方差比和小误差概率检验,将检验后的最优模型用于预测晚霜冻。结果表明,3种改进方法同时结合后所建模型的预测效果最理想,用来预测晚霜冻具有较高的可信度,根据模型预测,阳城县在2001年最后一次出现晚霜冻后,未来出现晚霜冻的年份为2030年和2044年。研究结果旨在为当地农业生产防御晚霜冻提供可参考的理论依据。  相似文献   

10.
对滑坡预测预报的非线性模型,存估计其参数时,传统的方法是将非线性模型在参数的近似值处展开成泰勒级数.并仅取至一次项,然后冉应用线性模型参数估计理论进行参数估计.因线性化时略去了,二阶及二阶以上的各高次项.所以必然会产生模型误差.介绍了高斯-牛顿法的基本原理,并以洒勒山新滑坡为例,在建立该滑坡厌色GM(1,1)模型和Verhulst模型的基础上,运用高斯牛顿法对两个非线性模型的参数进行优化.计算结果表明,参数优化后各模型的预测精度比优化前各模型的精度有显著提高.说明采用高斯-牛顿法优化非线性模型参数是提高滑坡预测预报精度的一种有效且切实可行的方法.  相似文献   

11.
孟祥健  李秀霞 《水土保持通报》2017,37(1):173-176,182
[目的]科学准确地预测城市建设用地,有利于把握城市发展的速度,了解城市化发展进程,为相关政府部门掌握土地利用情况,制定土地总体规划提供科学依据。[方法]把四平市作为研究对象,从"城市—农村"合力视角构建影响因子,利用因子分析探讨影响建设用地扩张的相关性,对指标进行筛选,在此基础上利用BP神经网络和灰色模型对四平市2012,2013和2014年建成区面积进行预测,最后对预测结果进行比较分析。[结果]通过预测与比较分析可知,BP神经网络结果的相对误差分别为0.8%,1.1%和2%,而灰色GM(1.1)模型预测结果相对误差分别为0.04%,0%和3.2%。可以看出,BP神经网络预测出的结果与实际相比较误差均在2%以内。[结论]BP神经网络预测的结果较精确,运用该方法可以有效提高预测的精度。  相似文献   

12.
黑龙江省农业机械化作业水平预测方法   总被引:6,自引:3,他引:3  
黑龙江省农机化作业水平的预测是一个复杂的非线性系统,其发展变化具有增长性和波动性,对于拟合的方法要求较高。该文对黑龙江省农机化作业水平预测方法进行研究,在传统预测模型灰色GM(1,1)模型、平滑预测模型和回归预测模型的基础上建立了基础预测模型,并与BP神经网络模型组合,建立了灰色神经网络、平滑回归神经网络等组合预测模型,并预测了黑龙江省2008~2015年的农机化耕、种、收、植保、灌溉作业水平。结果表明,新的预测方法拟合精度高、有效、可行,为农机化作业水平的预测提供了一条新的途径;黑龙江省机耕、机播、植保作业水平很高,但是机收作业水平不高,机械化灌溉是主要的瓶颈,需要进一步发展。  相似文献   

13.
Resuspension in the main basin of Lake Go?ci?? results approximately a doubling of the total sedimentation rate in the deepest region. The increase pollen flux in bottom traps during overturn was more four times higher than in the upper traps. The most intensive pollen sedimentation was recorded before and just after freeze-up. This could only have been the result of previous sediment resuspension. Mechanisms of sedimentation in the studied lake showed that resuspension and resedimentation can change "the environmental record" of events even in a lake with laminated sediments. Higher sedimentation of pollen during autumn circulation confirmed that in the sediment layers (varves or laminae) particles from previous seasons also could have been found. In the bay pollen sedimentation was lower and depended on plant flowering, there was no resuspension.  相似文献   

14.
Concentrations of mercury (Hg) were measured in six dated cores from four lakes in western Whatcom County, Washington, USA, that were at various bearings from a chlor-alkali plant, two municipal waste incinerators and a municipal sewage sludge incinerator. The importance of atmospheric emissions of Hg from these local municipal and industrial sources was evaluating by comparing the temporal trends in sedimentation of the lake cores with the emission history of each Hg species and by examining the geographical distribution of Hg sedimentation in relation to the region’s primary wind pattern. Local municipal and industrial sources of atmospheric Hg were not responsible for the majority of the Hg in the upper layer of sediments of Whatcom County lakes because of (1) the significant enrichment of Hg in lake sediments prior to emissions of local industrial and municipal sources in 1964, (2) smaller increases in Hg concentrations occurred after 1964, (3) the similarity of maximum enrichments found in Whatcom County lakes to those in rural lakes around the world, (4) the inconsistency of the temporal trends in Hg sedimentation with the local emission history, and (5) the inconsistency of the geographic trends in Hg sedimentation with estimated deposition. Maximum enrichment ratios of Hg in lake sediments between 2 and 3 that are similar to rural areas in Alaska, Minnesota, and New England suggest that global sources of Hg were primarily responsible for increases of Hg in Whatcom County lakes beginning about 1900.  相似文献   

15.
Nitrogen (N) removal within reservoirs can be substantial, but few studies have reported the relative importance of various N-retention pathways. Assessing N-removal processes in reservoirs is important for quantifying the impacts of reservoirs on downstream water quality. In this study, we used a time-series approach to quantify the relative importance of various N-removal processes in the Saylorville Reservoir in Iowa. Dynamic regression modeling of upstream–downstream changes in key water-quality surrogates (pH, hardness, alkalinity, and suspended solids) and their relation to N concentration changes were used to estimate the relative importances of denitrification, N assimilation by algal uptake, and sedimentation of N on N retention in the reservoir. Assuming that decreasing N concentrations in the reservoir are the sum of these three processes, we estimate that denitrification is the dominant N removal process (60.9 %) followed by algal assimilation (37.9 %) and sedimentation (1.2 %). Our approach represents a new method of establishing the relative importance of N-removal processes in reservoirs and quantifying the impacts of reservoirs on downstream water quality.  相似文献   

16.
基于灰色模型的黑龙江省水稻生育期热量指数分析及预测   总被引:1,自引:0,他引:1  
选择黑龙江省11个水稻农气观测站点为研究对象,利用1971-2016年逐旬气温资料和水稻发育期资料,将黑龙江省划分为东、西、南三个区域,使用微分方程动态建模方法建立3个区域5、6、7、8月热量指数的灰色预测模型,在此基础上滚动预报水稻生育期的总热量指数,以期开展黑龙江省水稻低温冷害的预测服务。结果表明热量指数能够很好地反映水稻生育期热量条件,且与低温冷害年有很好的对应关系。黑龙江省不同水稻产区热量指数灰色模型模拟结果与原序列关联度均达到0.88以上,通过了关联度检验和残差检验,1971-2010年拟合平均准确率为94.6%~97.6%,且7、8月的预报准确率普遍高于5、6月;2011-2016年的试报准确率均在97%以上,说明各模型的模拟效果很好。利用灰色模型预测黑龙江省水稻生长季热量指数是可行的,可以满足水稻生长发育过程中延迟性冷害的实时评估需求。  相似文献   

17.
Lake sediments are a potential source of mercury (Hg) for aquatic biota. Here, we investigated the predictive power of (a) key parameters for lake catchment morphometry and (b) organic matter composition of sediments in an effort to account for observed variations of total (THg) and methyl (MeHg) mercury concentrations in lake sediments. Using regression models we demonstrate that the morphometric parameters lake depth as well as inclination of catchment soils and lake bottoms can significantly predict variations of THg concentrations, but not MeHg, at profundal lake sediments. Although THg and MeHg concentrations in sediments could not be predicted by specific organic matter sources, as elucidated by atomic C/N ratios, our data suggest that wetland-derived total organic carbon (TOC) contained less THg concentrations than TOC derived from mostly forested watersheds. However, TOC concentrations could significantly predict MeHg concentrations and the proportion of methylated Hg at all sediment stations. Finally, from an ecotoxicological point of view, we propose that concentrations of TOC at surface lake sediments, rather than parameters of catchment morphometry, may predict dietary sources of MeHg for benthic consumers and consequently perhaps for organisms at higher trophic levels.  相似文献   

18.
Sediment cores collected from 12 lakes and eight marine sites along the Norwegian and Svalbard coast as part of a project investigating polycyclic aromatic hydrocarbons (PAH) in sediments were dated radiometrically using 210Pb, 137Cs and 241Am fallout radionuclides. In all lake cores, except on Svalbard, the 137Cs activity versus depth profile appears to have been significantly modified by post-depositional migration within the sediment column. The relatively low 137Cs inventories suggest that these sites were not significantly impacted by fallout from the 1986 Chernobyl accident. All the marine cores have 137Cs inventories that are substantially lower than in lake sediments almost certainly due to leaching of 137Cs from the marine sediments due to higher solubility in the seawater. In the marine surface sediments, the unsupported 210Pb concentrations are up to an order of magnitude lower than in the corresponding lake sediments reflecting the higher (dry mass) sedimentation rate at the marine sites. Five of the cores including marine sites and lakes have unusual high 210Pb flux most likely due to sediment focusing. Most of the irregularities in the 210Pb records seem to be due to slump events but some patterns are also due to possibly increased accumulation rates. Three of the marine cores show systematic increase in their sedimentation rate from c.1960 towards the present while only one lake shows the same systematic increase.  相似文献   

19.
本文应用GM(1,1)模型的前置零后的线性回归解法,就文献中的灰色消解模型进行了改进,实例表明,不仅计算工作量下降了,而且精度也有所提高。  相似文献   

设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号