基于随机森林特征优选的雪茄烟叶晾制过程含水率预测

针对雪茄烟叶晾制过程含水率人工判断主观性强、准确度低等不足,以及对影响雪茄烟叶晾制过程含水率预测的重要表观特征尚不明确等问题,该研究基于图像特征提取以及机器学习技术实现雪茄烟叶晾制过程含水率的预测。试验以雪茄烟品种“云雪2号”为试验材料,获取晾制过程的烟叶图像的颜色、轮廓、纹理以及部位四类特征并筛选出雪茄烟叶含水率预测的优选图像特征子集。在此基础上,构建了随机森林（random forest, RF）、支持向量机（support vector regression, SVR）与反向传播神经网络（back propagation neural network, BPNN）模型,并利用遗传算法（genetic algorithm, GA）对各模型超参数进行优化,将原始图像特征集与优选图像特征集输入3个机器学习模型,构建出6种模型-特征组合方案,依据晾制时期对原始数据集进行划分,并对测试集进行预测。最终结果显示：GA-SVR模型+优选图像特征子集的组合方案在测试集上表现最优,其决定系数（coefficient of determination,r²）与均方误差（mean square error,MSE）分别为0.980和0.001,且运行时间最短（运行时长=0.128 s）。研究结果可为雪茄烟叶晾制过程智能化控制提供理论依据。     

Characterization of mobile LiDAR data collected with multiple echoes per pulse from crowns during foliation

Abstract

As a state-of-the-art technology for fine-scale forest investigation, mobile terrestrial light detection and ranging (LiDAR), also referred to as mobile laser scanning (MLS), recently has been increasingly exploited to improve its performance on this task. One potential solution is to apply special MLS systems with the capability of collecting multiple echoes per pulse (multi-echoes, for short) from canopies. The methodologies of this rationale turned out to perform better but still insufficiently for canopy properties retrieval, owing to their common defective premise. That is, the knowledge of the characteristics of MLS scan data comprising multi-echoes, so far, is in shortage, especially when regarding the dynamic process of tree foliation. As a pioneering work for this challenge, this study attempted to comprehensively analyze the characteristics of MLS multi-echoes collected from tree crowns during foliation. Specifically, new stable multi-echoes-related features were deduced under the schematic frame of relative quantification, in both spatial and temporal sense. “Relative” here briefly means the division operation deployed on the attributes of multi-echoes, individually in terms of the number of echoes, echo width and crown volume integrity, between their different return orders. Then, the “relative” schematic was primarily validated for more stably representing crown properties during foliation, based on the real data that was collected by the Sensei MLS system with a maximum of three echoes per pulse. Further, a case of tree species classification was examined using a linear discriminant classifier, and it was testified that the resultant temporal statistical rules of multi-echoes as the reverse clues can enhance the performance of MLS in applications.     

螺旋榨油机关键设计的探讨

螺旋榨油机作为目前使用最广泛的榨油机械，其具有连续化、单机处理量大等特点，一般应用在高含油料的一次压榨以及预榨工艺中。为此，对螺旋榨油机的关键设计进行了探讨，从物料特性与机械结构相结合的角度讨论了压榨结构的设计过程，主要从物料的压榨特性，榨膛、榨排、榨螺的设计及理论压缩比的确定等方面展开阐述。     

不同腔长下调Q激光器输出脉冲特性研究

从Nd3+:YAG脉冲激光器调Q技术出发,研究了不同腔长条件下电光调Q激光器输出脉冲的宽度、能量和峰值功率特征,得到了不同腔长的调Q激光器输出脉冲特性曲线。研究结果表明,激光器腔长的改变能较好地调节脉冲的输出特性,对脉冲激光器的灵活应用具有一定的指导意义。     

基于Matlab的AHP分析程序设计

层次分析法是目前最为广泛应用的决策方法,但涉及大量的矩阵运算,影响了该方法的推广。Matlab软件精于矩阵运算,与C语言结构类似,移植性强,本研究基于Matlab设计出了便捷高效的AHP计算机程序。     

农民消费现状及主要特征分析

近年来,农村经济快速发展,农民收入增加,农民消费状况有所改善。农民衣着消费金额增加且比重下降,人均居住消费支出不断增加,教育支出和医疗保健支出受到政策影响有所减少,交通消费价格指数和文化娱乐消费价格指数有所上升,农民生活质量提高,消费结构趋于合理。     

塔克拉玛干沙漠公路沿线植被及特点

本文以塔克拉玛干浙江省公路为主线，调查由沙漠北缘至腹地自然概况，植物组成及分布特点，植被类型等，结果表明，塔克拉玛干沙漠公路沿线植物组成非常贫乏，仅１８种；生态学特性则以中生，耐盐的高位芽植物占主导地位；大多数植物以根蘖为主要繁殖形式；植被类型简单、隐域性盐化草甸植被占有一定的地位。     

典型草原种群特征对放牧制度的响应

采用放牧形式研究典型草原不同放牧制度群落动态变化规律.结果表明,牧草生长季前期干旱年份(2001年),划区轮牧区月现存量、生长量、生产力都显著高于自由放牧区(P＜0.05),而在降雨充沛的年份(2002年),两放牧处理无显著差异(P＞0.05);划区轮牧区的优势种的密度、高度、盖度以及重要值在两年均高于自由放牧区(P＜0.05).     

内蒙古东北部天然牧草生态气候适应性研究

应用积分回归、模糊数学中隶属函数和地理信息系统,对内蒙古东北部的天然牧草进行分区生态气候适应性综合评价。分析表明:牧草的自然生态适应性在半干旱、干旱环境中明显不如湿润、半湿润地区;对天然牧草的资源总量、生态系统进行了系统辩识;以定量、客观的方法分析和评价牧草气候资源区域分布特征,揭示出了资源的总体状况和各气候生态要素间的相互匹配情况。可为制定畜牧业发展规划提供决策依据。