基于遗传算法优化BP神经网络的玉米遥感估产方法     

于海洋  陈圣波  杨北萍  安秦 《世界地质》2020,39(1):208-214

以内蒙古自治区开鲁县玉米作物为研究对象,将生育期内玉米遥感影像所提取的多种植被指数和实地采样点的测产数据作为训练值,利用BP(back propagation)神经网络和遗传算法优化BP(GA-BP)神经网络估产模型,得出网络预测的玉米产量数值。通过决定系数R 2和均方根误差RMSE,比较实测产量与预测产量之间的精度,BP神经网络模型R^2为0.8452,RMSE(%)为28.37;遗传算法优化BP神经网络模型R^2为0.9850,RMSE(%)为6.70,表明遗传算法优化BP神经网络估产模型具有一定可行性和可信度。  相似文献   

基于共轭梯度搜索的病态问题处理方法     

刘杰  张娟娟 《大地测量与地球动力学》2019,39(8):863-868

在最小二乘平差准则基础上，把病态平差问题转化为无约束的二次规划问题，并利用优化理论分析病态对平差解的影响。通过共轭梯度搜索算法在可行域中寻找最优步长因子，自动寻找最速下降方向，并给出迭代初值的设置方法。分析近似计算中病态问题与局部最优解的关系，讨论局部最优解的快速迭代方法，并通过实例验证算法的有效性，计算迭代的速度。由于整个过程没有对法方程系数矩阵进行求逆计算，该算法可用于处理大规模系数矩阵高病态的平差问题。  相似文献   

花岗岩风化料弃土快速堆填过程中不排水抗剪强度评估     

詹良通  孙倩倩  郭晓刚  陈锐  陈云敏 《岩土力学》2021,(1):50-58

沿海沿江城市地下开挖产生的工程渣土含泥量大、含水率高且较松散,主要运往渣土场进行堆填处置。由于产量巨大而处置场地有限,许多渣土场在运营过程中存在堆填速度快、缺乏排水设施、超高超库容堆填等问题,容易引发堆填体失稳事故。目前对于非饱和工程渣土堆体在快速堆填过程中的失稳机制认识尚不清晰,尤其是对这一过程中的高饱和度工程渣土强度增长规律缺乏足够的认识。以深圳红坳渣土场填料——花岗岩风化料(CDG)填土为研究对象,对不同初始饱和度土样进行三轴不排水不排气等向压缩和剪切试验,结果表明:非饱和CDG填土不排水抗剪强度随围压增大呈非线性增长,增长速率与试样初始饱和度密切相关;当土样压缩后的饱和度超过0.7,不排水强度随围压的增长速率显著降低。基于有效固结应力法的原理,结合Hilf孔压公式和修正剑桥模型,提出了一种工程渣土不排水抗剪强度估算方法,并通过与试验结果对比验证了该方法对初始饱和度高于0.6的CDG填土的适用性。利用该方法确定的不排水强度cu与正应力σn的关系可应用于高饱和度工程渣土快速堆填中的稳定性分析。  相似文献   

CINRAD-SA双偏振雷达资料在降水估测中的应用初探   总被引：2，自引：0，他引：2       下载免费PDF全文

陈超  胡志群  胡胜  张扬  李珊珊 《气象》2019,45(1):113-125

对基于水平反射率ZH和差分传播相移率K_(DP)的降水估测综合法R(C)进行了改进,并对广州S波段双偏振雷达2016年2次飑线和2次台风降水过程的Φ_(DP)使用小波分析进行滤波处理,在此基础上使用变距最小二乘法拟合得到K_(DP)的值。分别使用R(C)和R(Z_H)法对2次飑线和2次台风降水过程进行降水估算,将估算结果和雨量计小时雨量进行了对比,并将两种方法的评估结果进行了对比。结果表明:(1)对于飑线类型降水,R(C)法对5 mm·h~(-1)以上的降水估测精度要好于R(Z_H)法,且降水率越大,R(C)法优势越明显,当降水率≥20 mm·h~(-1)时,两次过程R(C)法比R(Z_H)法的平均相对误差(RE)降低了17. 2%,平均绝对误差(AE)减少了1.89 mm,平均均方根误差(RMSE)减少了1.66 mm;(2)对于台风类型降水,R(C)法对5 mm·h~(-1)以上的降水估测精度也好于R(Z_H)法,当降水率≥20 mm·h~(-1)时,两次过程R(C)法比R(Z_H)法的平均RE降低了33. 19%,平均AE减少了3. 95 mm,平均RMSE减少了4.05 mm;(3)对于飑线和台风两种类型降水R(C)法都明显改善了降水率较大时的R(Z_H)法低估问题,但R(C)法在降水率10 mm·h~(-1)时也存在低估,可能是由雨滴谱资料观测误差导致拟合的系数偏小或雷达硬件造成的观测偏差等造成的。  相似文献   

Linking pesticides and human health: A geographic information system (GIS) and Landsat remote sensing method to estimate agricultural pesticide exposure     

《Applied geography (Sevenoaks, England)》2015

Accurate pesticide exposure estimation is integral to epidemiologic studies elucidating the role of pesticides in human health. Humans can be exposed to pesticides via residential proximity to agricultural pesticide applications (drift). We present an improved geographic information system (GIS) and remote sensing method, the Landsat method, to estimate agricultural pesticide exposure through matching pesticide applications to crops classified from temporally concurrent Landsat satellite remote sensing images in California. The image classification method utilizes Normalized Difference Vegetation Index (NDVI) values in a combined maximum likelihood classification and per-field (using segments) approach. Pesticide exposure is estimated according to pesticide-treated crop fields intersecting 500 m buffers around geocoded locations (e.g., residences) in a GIS. Study results demonstrate that the Landsat method can improve GIS-based pesticide exposure estimation by matching more pesticide applications to crops (especially temporary crops) classified using temporally concurrent Landsat images compared to the standard method that relies on infrequently updated land use survey (LUS) crop data. The Landsat method can be used in epidemiologic studies to reconstruct past individual-level exposure to specific pesticides according to where individuals are located.  相似文献   

Using simulated data to investigate the spatial patterns of obesity prevalence at the census tract level in metropolitan Detroit     

《Applied geography (Sevenoaks, England)》2015

Obesity is a serious public health problem in the United States. It is important to estimate obesity prevalence at the local level to target programmatic and policy interventions. It is challenging, however, to obtain local estimates of obesity prevalence because national health surveys such as the Centers for Disease Control and Prevention (CDC) Behavioral Risk Factor Surveillance System (BRFSS) are not designed to produce direct estimates at the local levels (e.g. census tracts) due to small population samples and the need to preserve individual confidentiality. In this study we address the problem of estimating local obesity prevalence rates by implementing a spatial microsimulation modeling technique to proportionally replicate the demographic characteristics of BRFSS respondents to census tract populations in metropolitan Detroit. Obesity prevalence rates are examined for high and low spatial clusters and studied in relation to the U.S. Department of Agriculture's (USDA) measures of low-income neighborhoods and local food deserts and CDC's measure of healthy and less healthy food environments currently used to target obesity reduction initiatives. This study found that obesity prevalence was largely clustered in the City of Detroit extending north into contiguous suburbs. The spatial patterns of highest obesity prevalence tracts were most similarly aligned with USDA-defined low-income tracts and CDC's less healthy food tracts. The locations of USDA's food desert tracts rarely overlapped with the highest obesity prevalence tracts. This study demonstrated a new methodology by which to assess local areas in need of future obesity interventions.  相似文献   

方程误差系统的多新息辨识方法     

丁锋 《南京气象学院学报》2015,7(5):385-407

多新息方法可以用于线性系统和非线性系统的自适应滤波、参数估计、自校正控制、自适应故障检测与诊断等.线性系统包括两种基本类型:方程误差类系统和输出误差类系统.本文将多新息辨识应用到方程误差滑动平均(EEMA)系统(即CARMA系统),研究多新息增广随机梯度算法和多新息增广最小二乘算法,应用到方程误差自回归滑动平均(EEARMA)系统(即CARARMA系统),提出基于分解的多新息广义增广随机梯度算法和基于分解的多新息广义增广最小二乘算法,以及基于滤波的多新息广义增广随机梯度算法和基于滤波的多新息广义增广最小二乘算法.  相似文献   

雷达雨量计联合估算降水在城市内涝模型中的应用   总被引：1，自引：0，他引：1  

陈靖  解以扬  东高红  李培彦  邱晓滨  李英华 《气象科技》2015,43(5):866-873

为了满足城市暴雨内涝模型对面雨量精细化的需求,在天津城市暴雨内涝模型的基础上,将雷达估算降水产品应用到模型面雨量计算中,针对2012年7月25—26日天津的大暴雨过程,考察4种雷达估算降水产品和两种插值方法计算的内涝模型面雨量。经过对比发现,利用变分方法计算的雷达估算降水产品VAR用曲面插值方法计算内涝模型的面雨量整体效果最好。  相似文献   

A Comparison of De-noising Methods for Differential Phase Shift and Associated Rainfall Estimation          下载免费PDF全文

HU Zhiqun  LIU Liping  WU Linlin  WEI Qing 《Acta Meteorologica Sinica》2015,29(2):315-327

Measured differential phase shift ΦDP is known to be a noisy unstable polarimetric radar variable, such that the quality of ΦDP data has direct impact on specific differential phase shift KDP estimation, and subsequently, the KDP-based rainfall estimation. Over the past decades, many ΦDP de-noising methods have been developed; however, the de-noising effects in these methods and their impact on KDP-based rainfall estimation lack comprehensive comparative analysis. In this study, simulated noisy ΦDP data were generated and de-noised by using several methods such as finite-impulse response(FIR), Kalman, wavelet,traditional mean, and median filters. The biases were compared between KDP from simulated and observedΦDP radial profiles after de-noising by these methods. The results suggest that the complicated FIR, Kalman,and wavelet methods have a better de-noising effect than the traditional methods. After ΦDP was de-noised,the accuracy of the KDP-based rainfall estimation increased significantly based on the analysis of three actual rainfall events. The improvement in estimation was more obvious when KDP was estimated with ΦDP de-noised by Kalman, FIR, and wavelet methods when the average rainfall was heavier than 5 mm h-1.However, the improved estimation was not significant when the precipitation intensity further increased to a rainfall rate beyond 10 mm h-1. The performance of wavelet analysis was found to be the most stable of these filters.  相似文献   

雷达-雨量计联合估测区域降水量方法检验与评估   总被引：3，自引：3，他引：0  

李建通  李柏  杨洪平  刘晓阳  张玲  郭林 《气象》2015,41(2):200-211

文章在"黄河淮河流域暴雨洪水预报系统"项目试验整理的高密度地面雨量站网数据和雷达连续采集体积扫描数据基础上,首次提出了对稠密雨量站网数据科学分组的方法,并采用分组数据对雷达-雨量计联合校准的10种方法进行了多参数的客观检验和评估。研究表明:(1)雨量站分组方法科学合理,可以满足评估的要求;(2)集成法估测区域降水量好于其他非集成的8种方法;(3)雷达对不同降水强度的估测精度具有三段式分布特征;(4)50~100 km为雷达估测降水的最佳区间,150~200 km区间的估测精度最差;(5)通过时间累积,可以提高各种估测方法区域降水量估测的精度和稳定度。  相似文献