BP神经网络在焉耆盆地农田排水量

利用BP神经网络技术对焉耆盆地农田排水量进行预测。利用灰色关联度分析确定了排水量与各影响因素的关系，选取了对排水量影响最大的5个因素作为BP网络的输入，利用均匀设计方法，确定了最优的神经网络结构。估算结果表明利用BP神经网络可以准确的估算农田排水量，最大相对误差仅为-2.45％。     

基于径流分类的日径流量预测神经网络模型

利用聚类分析法将径流序列分解为若干个子径流序列 ,对这些子径流序列分别建立局部神经网络模型 ,而后把这些局部模型合并成一个混合模型。当新的信息进入该模型时 ,首先用分类器判别其类别 ,以确定用混合模型中的何种局部模型加以模拟。通过与不加分类的总体神经网络模型的模拟结果加以对比 ,结果表明这种基于径流分类的降雨 -径流模型表现出了更优良的性能 ,可以较大地提高径流模拟精度。     

神经网络模糊PID算法在温室温度控制中的仿真研究

为了更好地实现对温室温度的控制,提出了温室温度控制的一种新的控制方法;通过仿真比较证明了此方法的有效性,并对其动静态特性、鲁棒性和抗干扰能力进行了探讨。结果表明,采用神经网络模糊PID算法的温室温度控制方法具有较好的动静态特性、鲁棒性和抗干扰能力。     

基于双池化与多尺度核特征加权CNN的典型牧草识别

针对自然背景下牧草难识别的问题,提出一种基于双池化与多尺度核特征加权的卷积神经网络牧草识别方法。双池化特征加权结构通过将卷积层输出的特征图分别进行最大值池化和均值池化得到两组特征图,引入特征重标定策略,依照各通道特征图对当前任务的重要程度进行加权,以增强有用特征、抑制无用特征;多尺度核特征加权结构通过在卷积层中同时使用3×3和5×5两种卷积核,并将网络的前几层特征复用后进行加权,以提高重要特征的利用率。对10类牧草图像进行识别实验,结果表明,该方法识别率为94. 1%,比VGG-13网络提高了5. 7个百分点,双池化与多尺度特征加权有效提高了牧草识别精度。     

遗传神经网络在温室黄瓜霜霉病预测中的应用

在神经网络结构的优化、连续权重的训练中采用遗传算法的方法对日光温室黄瓜霜霉病进行预测，并对其发生程度进行了拟合，计算值与实际值接近，表明该模型具有良好的预测效果。     

一种改进深度卷积神经网络的海岛识别方法

受不规律潮汐的影响,现有的海岛地物类别自动识别方法存在精度低和时效性差等问题,通过改进深度卷积神经网络提出了一种基于遥感影像的海岛快速识别方法:(1)在深度卷积神经网络的卷积层中增设1×1的卷积核作为瓶颈单元,对多波段的遥感影像进行降维;(2)在池化层引入了重采样方法,基于灰度值对海量的遥感影像进行特征压缩。以300景Landsat-8遥感影像为源数据,分别采用CNN、RCNN和本文改进的深度卷积神经网络对遥感影像中的海岛进行识别,实验结果表明:(1)改进的深度卷积神经网络降低了海岛识别的计算耗时,其计算耗时仅为CNN的4.56%和RCNN的5.60%;(2)改进的深度卷积神经网络较CNN和RCNN提高了海岛识别的精度,识别精度分别为96.0%、93.3%和95.0%。结果说明,改进的深度卷积神经网络适用于面向遥感影像的海岛自动识别。     

Deep neural network algorithm for estimating maize biomass based on simulated Sentinel 2A vegetation indices and leaf area index

Accurate estimation of biomass is necessary for evaluating crop growth and predicting crop yield.Biomass is also a key trait in increasing grain yield by crop breeding.The aims of this study were(i)to identify the best vegetation indices for estimating maize biomass,(ii)to investigate the relationship between biomass and leaf area index(LAI)at several growth stages,and(iii)to evaluate a biomass model using measured vegetation indices or simulated vegetation indices of Sentinel 2A and LAI using a deep neural network(DNN)algorithm.The results showed that biomass was associated with all vegetation indices.The three-band water index(TBWI)was the best vegetation index for estimating biomass and the corresponding R2,RMSE,and RRMSE were 0.76,2.84 t ha−1,and 38.22%respectively.LAI was highly correlated with biomass(R2=0.89,RMSE=2.27 t ha−1,and RRMSE=30.55%).Estimated biomass based on 15 hyperspectral vegetation indices was in a high agreement with measured biomass using the DNN algorithm(R2=0.83,RMSE=1.96 t ha−1,and RRMSE=26.43%).Biomass estimation accuracy was further increased when LAI was combined with the 15 vegetation indices(R2=0.91,RMSE=1.49 t ha−1,and RRMSE=20.05%).Relationships between the hyperspectral vegetation indices and biomass differed from relationships between simulated Sentinel 2A vegetation indices and biomass.Biomass estimation from the hyperspectral vegetation indices was more accurate than that from the simulated Sentinel 2A vegetation indices(R2=0.87,RMSE=1.84 t ha−1,and RRMSE=24.76%).The DNN algorithm was effective in improving the estimation accuracy of biomass.It provides a guideline for estimating biomass of maize using remote sensing technology and the DNN algorithm in this region.     

基于粒子群算法优化BP神经网络的土壤含水量短期预测模型

为提高土壤含水量预测精度,基于物联网监测数据,提出了粒子群算法(PSO)优化BP神经网络的土壤含水量预测方法。首先应用主成分分析法筛选出影响土壤含水量的关键影响因子,然后构建8-5-1的BP神经网络拓扑结构,应用粒子群算法优化BP神经网络的初始权值和阈值。结果表明:与传统BP神经网络相比,新模型优化了网络结构,避免了陷入局部最优解,具有良好的预测效果;模型的评价指标平均绝对误差、平均绝对百分误差、误差均方根分别为0.259 2、0.010 5和0.135 6,与单一BP神经网络相比,预测精度更高,可满足实际的土壤含水量预测的需要。     

Performance and uncertainty analysis of a short-term climate reconstruction based on multi-source data in the Tianshan Mountains region,China

Short-term climate reconstruction, i.e., the reproduction of short-term(several decades) historical climatic time series based on the relationship between observed data and available longer-term reference data in a certain area, can extend the length of climatic time series and offset the shortage of observations. This can be used to assess regional climate change over a much longer time scale. Based on monthly grid climate data from a Coupled Model Inter-comparison Project phase 5(CMIP5) dataset for the period of 1850–2000, the Climatic Research Unit(CRU) dataset for the period of 1901–2000 and the observed data from 53 meteorological stations located in the Tianshan Mountains region(TMR) of China during the period of 1961–2011, we calibrated and validated monthly average temperature(MAT) and monthly accumulated precipitation(MAP) in the TMR using the delta, physical scaling(SP) and artificial neural network(ANN) methods. Performance and uncertainty during the calibration(1971–1999) and verification(1961–1970) periods were assessed and compared using traditional performance indices and a revised set pair analysis(RSPA) method. The calibration and verification processes were subjected to various sources of uncertainty due to the influence of different reconstructed variables, different data sources, and/or different methods used. According to traditional performance indices, both the CRU and CMIP5 datasets resulted in satisfactory calibrated and verified MAT time series at 53 meteorological stations and MAP time series at 20 meteorological stations using the delta and SP methods for the period of 1961–1999. However, the results differed from those obtained by the RSPA method. This showed that the CRU dataset produced a low degree of uncertainty(positive connection degree) during the calibration and verification of MAT using the delta and SP methods compared to the CMIP5 dataset. Overall, the calibrated and verified MAP had a high degree of uncertainty(negative connection degree) regardless of the dataset or reconstruction method used. Therefore, the reconstructed time series of MAT for the period of 1850(or 1901)–1960 based on the CRU and CMIP5 datasets using the delta and SP methods could be used for further study. The results of this study will be useful for short-term(several decades) regional climate reconstruction and longer-term(100 a or more) assessments of regional climate change.     

一种改进的双能量源电动汽车功率分配策略

以并联式双能量源(锂电池组&超级电容)纯电动汽车为研究对象,提出一种改进的功率分配策略。这种功率分配策略在模糊控制的基础上增加深度神经网络训练过程,以获得更精确的功率分配因子。通过整车模型构建、功率分配模型构建、功率分配策略制定、仿真验证结果分析后得出结论:与电量消耗功率分配策略相比,这种改进的功率分配策略能优化锂电池组和超级电容二者之间的功率分流,限制锂电池组充放电的峰值电流,延长复合储能系统的工作寿命,提升动力系统的工作效率。