Estimation of peanut biomass based on feature selection and particle swarm optimization
-
摘要:
为解决花生植株生物量估算精度低、破坏性大等问题，该研究提出一种无人机低空遥感技术结合高光谱特征筛选的花生生物量估算方法。通过无人机搭载高光谱成像仪，获取田块尺度多个花生品种的高光谱影像数据，首先对获取的影像进行拼接、辐射定标、大气校正等预处理，提取出地面采样点位置的光谱反射率，计算光谱反射率的一阶微分和植被指数，使用变量投影重要性（variable importance in projection，VIP）方法对光谱反射率、一阶微分和植被指数等三种数据进行特征筛选，利用筛选后的特征和地面实测数据构建支持向量机回归（support vector regression，SVR）、反向传播神经网络回归（back propagation neural network，BPNN）和随机森林回归（random forest regression，RFR）模型，并使用粒子群优化算法（particle swarm optimization，PSO）进行模型优化。结果表明：相比原始光谱反射率和植被指数，一阶微分光谱反射率与花生生物量具有较好的相关性；使用一阶微分光谱反射率与植被指数组合的RF回归模型精度最高（决定系数R2为0.754，均方根误差RMSE为0.085 kg/m2），使用粒子群优化后的PSO-RF模型可进一步提高模型精度（R2为0.80 ，RMSE为0.076 kg/m2）。该研究为花生生物量精准估算提供了有效的方法，为智慧乡村建设中的精细化农田管理提供技术支持。
Abstract:Peanut is one of the most widely cultivated oil crops globally, with China leading in both production and consumption. As the demand for oil crops increases, ensuring stable peanut production and oil supply security has become a key agricultural goal. Peanut biomass, as a crucial parameter reflecting crop growth status, is essential for precision agriculture management and efficient resource utilization. The aboveground parts of peanut plants can be used not only as animal feed but also as a resource for bioenergy production. Therefore, comprehensive and accurate biomass estimation provides valuable references for yield prediction and resource management. Traditional biomass measurement methods are often labor-intensive and time-consuming, with spatial and temporal limitations. Recently, with the development of UAV remote sensing, especially the widespread application of hyperspectral imaging technology, crop biomass estimation has become more efficient. Hyperspectral imaging, known for its high resolution and rich spectral information, has been used for growth monitoring and yield estimation of crops such as soybean, rice, and wheat, demonstrating superior performance in predicting parameters like yield, chlorophyll, and nitrogen content, as well as in disease diagnosis. However, research on peanut remains limited, particularly regarding the spectral characteristics of different peanut varieties and their impact on biomass estimation accuracy. This study, using UAV hyperspectral imaging, investigated sensitive spectral bands and feature combinations for efficient and accurate field-scale peanut biomass estimation. An experimental field with 11 peanut varieties in Xingyang, Henan, was used as the study area. First, UAV hyperspectral images of the test field were collected and preprocessed with radiometric calibration and atmospheric correction to ensure data accuracy. Spectral reflectance data from ground sampling points were then extracted, and the first derivative of spectral reflectance and multiple vegetation indices were calculated to enhance the feature dimensions related to biomass. The variable importance in projection (VIP) method was used to select sensitive spectral bands and feature combinations closely related to biomass, effectively eliminating data redundancy and isolating highly relevant features. Using the selected features and ground-truth data, Support Vector Regression (SVR), back propagation neural network (BPNN), and random forest regression (RFR) models were constructed, and the estimation accuracy of different machine learning models was compared. Additionally, selected sensitive features were combined in multiple ways and input into the models to further improve estimation accuracy. The particle swarm optimization (PSO) algorithm was employed to optimize model hyperparameters, achieving the best model performance. Results showed that the sensitive features derived from the first derivative of spectral reflectance were highly correlated with peanut biomass, yielding better model performance than those derived from raw spectral reflectance and individual vegetation indices. The RF model combining the first derivative of spectral reflectance and vegetation indices achieved the highest estimation accuracy (R2 = 0.75, RMSE = 0.08). Further improvement was achieved with the PSO-optimized RF model (PSO-RF), which resulted in an accuracy of R2 = 0.80 and RMSE = 0.07. This study demonstrates the potential of combining UAV hyperspectral imaging with machine learning models for non-destructive peanut biomass estimation, providing essential theoretical and technical support for large-scale agricultural biomass monitoring. This study provides an effective method for accurate peanut biomass estimation and offers technical support for precision farmland management in the construction of smart villages.
-
Keywords:
- peanut /
- biomass /
- smart village /
- feature selection /
- machine learning /
- particle swarm optimization
-
0. 引 言
气候变化和人口增长的综合影响使水和能源的需求及成本呈指数级增长[1]，农业节水已成为缓解中国水资源短缺的必然选择与根本出路[2]。喷灌是农业节水中一种有效的灌溉方式，喷头被视为喷灌系统中确保节水的关键部件[3-5]。为了达到节能节水的目的，降低喷头工作压力的同时保证均匀性是当前喷头研究的重点。
在固定式旋转喷灌系统中，当喷头的工作水压力较低时，存在喷头附近和射程中段的水量较少，而射流末端水量集中的问题，为了改善这种情况，已有研究者采用多种方法来分散水射流。由美国Nelson公司生产的R2000、R33等系列喷头，采用环形散水盘外圈分布散水齿的结构，射流冲击散水齿达到分散射流的目的[6]，李红等[7]采用正交试验方法研究了散水齿的结构参数对旋转式喷头 R33 水力性能的影响，确定了散水齿的最佳结构参数，并建立了散水齿影响区域最短射程的计算方法。范兴科等[8]在摇臂式全园喷洒喷头的基础上进行了结构改造，增加了喷洒雾化度调节机构，该机构在喷嘴下方装置了有利于射流水雾化的散水撞针，以增强射流的破碎效果，从而提高喷头的喷洒雾化度。JIANG等[9]设计了不同类型的喷嘴，并在低压条件下使用散水针进行了试验，通过对初始射流破碎长度、分散角、喷嘴尺寸和工作压力等参数的曲线拟合，建立了具有分散装置的喷头的射程公式。PAN等[10-11]通过试验与数值模拟相结合的方法，分别研究了散水针与散水齿对射流破碎和水力性能的影响，结果表明两种射流分散方法增强了射流破碎的同时造成了较高的能量及喷头射程损失。
射流撞击作为一种常见的流体力学现象，经常出现在气体和液体的注入、混合和分离等过程中。射流撞击现象十分复杂，受到多种因素的影响，如射流初速度、射流直径、射流撞击角度、射流撞击距离等[12-14]。射流撞击常用于导弹与火箭推进剂雾化、水煤浆气化、药物吸入治疗、固体颗粒制备等很多领域，近年也将射流撞击应用于植保喷头[15-16]，设计出的喷头具有雾滴粒径均匀性较好、雾滴谱较窄、调压范围宽和抗飘性较强的特点。而现有灌溉喷头射流分散方法（如散水齿、散水针等）多以液-固撞击为主，射流撞击固体造成了部分射流动量减小甚至反向撞击前部射流，最终导致较高的射程损失，而液-液射流撞击不会造成动量损失，当两股射流的撞击角小于90°时，合成动量会比任一射流的动量更大，这种射流撞击方式不仅有助于射流的分散，还可能对射程产生促进作用，因此考虑将液-液射流撞击应用在喷灌领域。
鉴于此，本文基于射流撞击方法研制一种通过主副射流相撞击的旋转喷头，采用正交试验确定喷头副喷嘴主体结构的最佳结构参数，并通过与射流不撞击喷头对比得到射流的撞击对于水力性能的影响。
1. 射流撞击式喷头设计
1.1 射流撞击原理
传统对称射流撞击的雾化机理是两个射流速度和射流直径相等的圆柱形液体相互碰撞。这种撞击产生了一个垂直于两股射流动量矢量的扁平液膜。液膜上的不稳定性会促进破碎。液膜分解成圆柱形的液带，然后液带分解成液滴。破碎主要是由空气动力和惯性力引起的，并受到表面张力的影响[17-18]。在灌溉喷头设计中，除追求更高的喷洒均匀性外，射程也是重要的性能参数，对称式同径的射流撞击模式虽然雾化效果出众，但会造成射程显著降低，因此本研究设计的喷头采用非对称异径射流撞击模式，喷头有主副两个喷嘴，其中主喷嘴孔径大于副喷嘴孔径，通过副射流撞击主射流，以改变主射流的运动及破碎模式，达到优化射流分散效果进而提高水力性能的目的。根据射流撞击原理设计的射流撞击喷头工作原理如图1所示。
![]()
图 1 射流撞击喷头工作原理示意图1.主射流 2.副射流 3.射流撞击点 4液膜 5液滴Figure 1. Schematic diagram of working principle of jet impingement sprinkler1. Primary jet 2. Secondary jet 3. Jet impingement point 4. Liquid sheet 5.Droplets1.2 喷头结构设计
射流撞击喷头的结构尺寸及主副喷嘴相对位置及角度变化会对射流撞击距离、射流撞击角度等产生影响，进而影响喷头水力性能。为了确保射流撞击喷头满足规定的射程要求，喷头主喷管结构与传统15PY2保持一致，为实现主副射流撞击，适当增加主副喷管之间的距离。本研究将副喷嘴结构作为研究重点，副喷嘴主体结构参数包括副喷管内径、副喷嘴仰角、副喷管长度、副喷嘴进口锥角，喷头结构如图2所示。试验加工喷头采用铝合金材质以保证喷头强度要求，为了便于试验更换及互配，采用分段的结构设计，将下喷体分为上下两部分，将副喷嘴分为副喷管与副喷嘴两部分，并将本研究所有结构加工成零件，由于螺纹连接的喷嘴在装配过程中射流方向会随安装时有一定旋转，可能导致主副射流无法形成撞击，为保证主副射流方向保持在同一平面，喷嘴部分未采用传统的螺纹连接结构，改用插接内部安装止水垫片，并在两侧设限位孔安装螺丝限位，通过面于面对齐保证喷嘴安装时不会改变射流轴线方向，本次试验样机结构主体参数为进水口公称直径15 mm，主喷嘴仰角26°，根据规范GB/T 22999−2008《旋转式喷头》可知，15 mm进水口公称直径双喷嘴喷头选取主喷嘴4～6 mm，副喷嘴取2.5～3.0 mm，本研究选取主喷嘴孔径4 mm，副喷嘴孔径3 mm进行试验。将主喷体反装，使副喷嘴射流与主喷嘴射流在相反的方向上得到射流不撞击喷头，用于和射流撞击喷头进行对比试验，本次试验所用喷头结构如图3所示。
![]()
图 2 喷头主体结构1.摇臂 2.主喷嘴 3.主喷管 4.副喷嘴5.下喷体 6.下轴承Figure 2. Main structure of sprinkler1. Rocker arm 2. Primary nozzle 3. Primary spray tube 4. Secondary nozzle 5. Lower spray body 6.Lower bearing 注： α 为副喷嘴仰角，(°)； θ 为副喷嘴进口锥角，(°)； D 为副喷管内径，mm； L 为副喷管长度，mmNote: α is secondary nozzle elevation angle, (°); θ is secondary nozzle inlet cone angle, (°); D is inner diameter of secondary spray tube, mm; L is secondary spray tube length, mm1.3 试验设计与方法
1.3.1 试验方案
选取副喷嘴仰角α、副喷管内径D、副喷管长度L、副喷嘴进口锥角θ为正交试验因素。副喷嘴仰角选择区间27°～33°，可以实现射流撞击且不会对摇臂运动产生影响；副喷管内径根据传统15PY2副喷嘴入口直径为6 mm，选择4～6 mm；传统15PY2副喷管长度较小，仅为5～6 mm，根据DOMBROWSKI 等[19]的研究，射流撞击的工作液体需要经过较长的管道以获得充分发展的速度剖面，因此本研究的副喷管选择了10～30 mm的较长范围区间；副喷嘴进口锥角选择较为常用的35°～55°进行研究[20]。选择因素水平如表1所示。选用L9（34）正交表，正交试验方案如表2所示，并将最终优化结果与传统15PY2喷头对比。
表 1 因素水平表Table 1. Factor and level水平
Level因素Factors 副喷嘴仰角
Secondary nozzle elevation angle
α/(°)副喷管内径
Inner diameter of secondary spray tube
D/mm副喷管长度
Secondary spray tube length
L/mm副喷嘴进口锥角
Secondary nozzle inlet cone angle
θ/(°)1 27 4 10 35 2 30 6 20 45 3 33 8 30 55 表 2 试验方案Table 2. Test scheme试验号
Test No.α/(°) D/mm L/mm θ/(°) 1 27 4 10 35 2 27 6 20 45 3 27 8 30 55 4 30 4 20 55 5 30 6 30 35 6 30 8 10 45 7 33 4 30 45 8 33 6 10 55 9 33 8 20 35 1.3.2 试验装置及设计
样机试验在直径为44 m的无风喷灌实验大厅进行。喷头安装高度为1.4 m，在工作压力分别为150、200、250和300 kPa的工况下做全圆喷洒。由于喷灌大厅内无风，因此可以用两条射线上的数据平均代替圆周内的各条射线。如图4所示，在这条射线上每隔1m放置一个雨量桶，并在12 m后每0.5 m放置一个雨量桶用以更加精确的获得射程数据，在稳定运转10 min后开始试验，测试时间为1 h，测定各点上的喷灌强度。
通过本次试验可以得出副喷嘴几何参数对水量分布、射程、组合均匀系数的影响规律，根据射程与组合均匀度判定各结构喷头性能优劣。通过与射流不发生撞击对比，得到射流撞击对水力性能的影响，通过与传统15PY2喷头对比，得到射流撞击优化提升程度。
1.3.3 指标计算方法
根据 GB /T 27612. 3−2011，喷头的射程指的是正常旋转情况下，喷头中心线距测出的喷灌强度为某一数值的最远点的距离。对流量大于75 L/h的喷头，该点的灌水强度为0.26 mm/h。若出现A 点雨量筒测得的点喷灌强度大于0.26 mm/h，但距离喷头更远的 B 点雨量筒测得的点喷灌强度小于 0.26 mm/h时，喷头的射程位于 A 点和B 点之间的 C 点采用插值算法计算喷头射程[21-22]。
本研究使用Matlab软件，通过插值数学模型对间距系数为喷头的射程R时进行方形布置的4个喷头进行数学叠加计算，计算得到组合均匀性。间距系数描述了不同布局形式下相邻两个喷头的重叠距离。均匀系数反映了水量分布的均匀程度，是衡量喷灌质量的重要指标。本研究采用克里斯琴森均匀系数作为评价指标[23-24]。计算式为
Cu=[1−∑|xi−xm|∑xi]×100% (1) 式中Cu为克里斯琴森均匀系数，%；xi为i位置测点的喷灌强度，mm/h；xm所有测点中测得的平均喷灌强度，mm/h。
利用2DVD视频雨滴谱仪对射流撞击喷头和传统15PY2喷头开展水滴粒径测量试验，选择水滴测量位置分别为在低压150 kPa和中压250 kPa下前中段6 m位置，低压150 kPa下后段12 m及中压250 kPa下后段14 m位置进行测量。
2. 结果与分析
2.1 水力性能对比
对射流不撞击喷头进行水力性能试验可以得到射流不发生撞击，即主副射流不会相互影响的水量叠加情况下水力性能，对正交试验中的9种结构组合进行了射流撞击与射流不撞击喷头水力性能的对比，由于各压力下水量分布趋势相似，因此选择150和250 kPa作为低压与中压下的水量分布代表，其水量分布曲线如图5所示，可以看出射流撞击喷头产生的效果并非主副射流水量的叠加，水量分布近处主要为摇臂打击导致的水分散，各试验方案相差不大。射流撞击对射程中部的水量稍有提升，但提升效果较小，这避免了刘俊萍等[25]提到的副喷嘴为圆形时径向水量分布中部较高不利于组合的问题，而对于水量分布曲线中10 m到射程末端产生较大影响，射流撞击使得末端水量分布的高点在降低的同时后移，且相比于射流不发生撞击，射流撞击的射程更远，射程平均提升4.39%。分析其原因：副射流与主射流汇合，导致主射流流量变大，进而导致射程提高；副射流从下部撞击主射流，一定程度抬升了主射流角度，使其射流末端向后延申；副射流与主射流发生撞击时的动量合成后的方向对射程提升产生帮助。射程的延长导致后端的水量分布得到拉伸，解释了水量分布产生的变化。较高的局部喷灌强度容易引起土壤压实与地表径流[26]，射流撞击使得射流末端水量集中的问题得到改善，相较于不发生撞击喷头，在实际应用中更有优势。
![]()
图 5 不同压力下射流撞击喷头与射流不撞击喷头水量分布曲线Figure 5. Water distribution curve of jet impingement sprinkler and jet non-impingement sprinkler under different pressures2.2 结构参数优化
不同压力下的流量如表3所示，各压力下流量变异系数分别为1.94%，1.47%，1.65%，1.18%，均小于2%，考虑到加工精度及试验误差，可以得出本次研究中的4个结构参数的改变不会对流量产生影响。将9组流量平均后，得到本次试验样机流量与压力成正比关系，且各试验压力下的流量均大于0.08 m3/h，即喷头的射程为喷头中心线距测出的灌水强度为0. 26 mm/h的最远点的距离。通过插值法计算的射程与通过Matlab软件计算得到的组合均匀系数结果如表4所示。
表 3 各压力下流量Table 3. Flow at various pressures(m3·h-1) 试验号
Test No.150 kPa 200 kPa 250 kPa 300 kPa 1 1.10 1.27 1.42 1.55 2 1.12 1.29 1.44 1.57 3 1.11 1.27 1.42 1.55 4 1.07 1.24 1.37 1.53 5 1.15 1.31 1.45 1.58 6 1.12 1.27 1.44 1.56 7 1.10 1.27 1.42 1.53 8 1.11 1.27 1.44 1.53 9 1.10 1.27 1.42 1.54 平均值Avg. 1.11 1.27 1.42 1.55 表 4 各压力下射程与组合均匀系数Table 4. Wetted radius and combination uniformity coefficients under various pressures试验号
Test No.150 kPa 200 kPa 250 kPa 300 kPa R/m Cu/% R/m Cu/% R/m Cu/% R/m Cu/% 1 14.25 59.22 15.40 65.44 16.00 65.61 16.00 67.34 2 14.98 63.72 15.48 65.87 17.50 70.24 17.38 71.68 3 14.38 61.06 15.71 67.12 16.33 65.79 17.43 71.00 4 14.94 63.37 15.50 65.96 17.21 67.97 17.21 70.63 5 15.45 65.42 15.93 65.03 16.33 65.13 17.16 65.28 6 14.97 68.67 16.31 47.47 16.00 68.57 16.50 69.71 7 14.00 62.67 15.08 65.91 15.65 68.79 15.75 70.66 8 13.13 67.20 14.33 68.69 14.39 70.47 14.73 71.54 9 14.00 62.45 15.58 66.97 16.10 69.61 16.19 69.47 对正交试验结果进行极差分析，将各压力下极差进行平均，得到150～300kPa全压力下因素优水平和优组合及因素对各指标影响的主次顺序，结果见表5。可以看出对射程的影响由大到小依次为α，L，θ，D，最优组合α为30°，L为20 mm，θ为45°，D为8 mm，对组合均匀系数的影响由大到小依次为α，θ，D，L，最优组合α为33°，θ为55°，D为6 mm，L为20 mm，副喷嘴仰角α对射程与组合均匀系数影响最大。
表 5 射程和组合均匀系数极差分析Table 5. Wetted radius and combination uniformity coefficient range analysis指标Index α D L θ 射程
Wetted radius
RK1 47.71 46.75 45.50 47.10 K2 48.38 46.70 48.02 47.40 K3 44.73 47.38 47.30 46.32 k1 15.90 15.58 15.17 15.70 k2 16.13 15.57 16.01 15.80 k3 14.91 15.79 15.77 15.44 r 1.22 0.23 0.84 0.36 组合均匀
系数
Combination uniformity
coefficient CuK1 198.52 198.39 197.48 196.74 K2 195.80 202.57 201.99 198.49 K3 203.61 196.97 198.47 202.70 k1 66.17 66.13 65.83 65.58 k2 65.27 67.52 67.33 66.16 k3 67.87 65.66 66.16 67.57 r 2.60 1.86 1.50 1.99 注：K 为每个因素各水平的数值之和； k 为每个因素水平的平均值； r 为极差。
Note: K is the sum of the values of each factor at each level; k is the average value of each factor level; r is extreme range.此试验是两指标的试验设计，极差分析对射程和组合均匀系数得到不同的最优组合，因此通过CRITIC 权值法[27]计算权重，得到各压力组下射程与组合均匀系数的权重：150 kPa下0.49，0.51；200 kPa下0.48，0.52；250 kPa下0.43，0.57；300 kPa下0.5，0.5。发现组合均匀系数的权重略高于射程，符合传统喷头设计思路。通过式（2）计算得到各组试验综合评分yi
yi=w1×R+w2×Cu (2) 式中w1，w2分别为射程与组合均匀系数权重系数。
为得到适用于150～300 kPa全压力下最优结构组合，将各压力下综合评分平均，得到每组试验全压力下综合评分并对综合评分进行极差分析，综合评分计算结果如表6所示，综合评分均值极差分析如表7所示。从综合评分结果发现，试验号2、4、8、9组评分较高，其中的2、4、9组均包含L=20 mm，与极差分析中得到副喷管长度对综合评分影响最大的结果一致，因素对于综合评分的影响从大到小依次为L、θ、D、α，对比射程和组合均匀系数极差分析的结果可以发现虽然副喷嘴仰角对射程和组合均匀系数的影响均为最大，但在综合评分中影响最小，可能是由于副喷嘴仰角对射程与组合均匀系数影响趋势不同，所得到的优结构不同导致的。最终可以得到150～300 kPa全压力下对射程和组合均匀系数权衡后的最优副喷嘴几何参数分别为副喷嘴仰角α为33°、副喷管内径D为6 mm、副喷管长度L为20 mm、副喷嘴进口锥角θ为55°。
表 6 各压力下的综合评分及其均值Table 6. Comprehensive scores and its average value under various pressures试验号
Test No.工作压力
Pressure/kPaˉy 150 200 250 300 1 37.03 41.38 44.34 41.45 41.05 2 39.67 41.64 47.62 44.30 43.31 3 38.03 42.40 44.58 43.99 42.25 4 39.47 41.70 46.20 43.69 42.77 5 40.76 41.42 44.20 41.02 41.85 6 42.17 32.49 46.03 42.88 40.89 7 38.65 41.47 46.00 42.97 42.27 8 40.52 42.55 46.42 42.89 43.10 9 38.54 42.26 46.66 42.60 42.52 表 7 综合评分均值极差分析Table 7. Range analysis of average comprehensive scores指标Index 因素Factors α D L θ K1 126.61 126.09 125.04 125.42 K2 125.51 128.26 128.59 126.47 K3 127.89 125.66 126.37 128.11 k1 42.20 42.03 41.68 41.81 k2 41.84 42.75 42.86 42.16 k3 42.63 41.89 42.12 42.70 r 0.79 0.87 1.19 0.90 2.3 优化结构性能对比
将射流撞击式喷头与改进前的传统15PY2型摇臂喷头进行水力性能对比，传统15PY2主副喷嘴孔径与射流撞击式喷头主副喷嘴孔径均为4 mm和3 mm，保证了两种喷头在相同流量下进行对比，各压力下射程、组合均匀系数及综合评分如表8所示。
表 8 射流撞击式喷头与传统15PY2喷头水力性能对比Table 8. Comparison of hydraulic performance between the jet impingement sprinkler and the traditional 15PY2 sprinkler.喷头类型
Sprinkler type150 kPa 200 kPa 250 kPa 300 kPa R/m Cu/% yi R/m Cu/% yi R/m Cu/% yi R/m Cu/% yi 射流撞击喷头
Jet impingement sprinkler13.47 66.54 40.35 14.65 68.96 42.85 15.47 70.03 46.63 15.70 71.61 43.42 传统15PY2
Traditional 15PY213.95 64.43 39.52 14.68 67.71 42.22 15.21 65.40 43.88 15.43 66.83 40.91 注：R为射程，Cu为组合均匀系数，yi为综合评分。 Note: R is wetted radius, Cu is combination uniformity coefficients, yi is comprehensive score. 通过对比射程可以发现，在150和200 kPa下，射流撞击式喷头比传统15PY2喷头射程有所降低，平均降低1.82%，但250和300 kPa下，射流撞击式喷头比传统15PY2喷头射程有所提高，平均提高1.73%，总的来说降低与提高幅度均不大。对比组合均匀系数可以看出，射流撞击式喷头组合均匀系数均高于传统15PY2喷头，在150和200 kPa下平均提升2.56%，在250 kPa和300 kPa下平均提升7.12%，即在低压下均匀度有所提升，在中压下提升较大。射流撞击喷头在低压下射程有所降低而均匀度提高，其原因有两个：一是在低压力条件下，喷头所形成的射流具有较低的能量。当射流发生撞击时，造成主射流部分能量散失，导致射流的射程受限，无法达到远距离；二是由于低压力下射流的能量较低，射流的弥散效应也较小（弥散指的是射流在空气中逐渐扩散和散开的现象），射流的撞击提高了弥散效应即射流不稳定性，使得射流提前破碎，主射流无法到达原本距离，射程降低，同时导致均匀性提高。
对比综合评分，可以看到射流撞击式喷头均高于传统15PY2喷头，在150和200 kPa下平均提升1.8%，在250和300 kPa下平均提升6.2%，即在低压下综合评分有所提升，在中压下提升较大。
试验结果表明射流撞击喷头在150～300 kPa的全压力下组合均匀系数均高于传统15PY2型喷头，平均提升4.84%。射流撞击喷头和传统15PY2喷头的综合评分均值分别为43.31和41.63，也可看出射流撞击喷头综合评分平均提升4%，有较为明显的优势。
对8组水滴粒径数据进行测量，得到的水滴粒径区间为0～7.5 mm，将此区间分为19个等级，数据在喷头旋转状态下测量，为了降低试验误差，在同一位置同一压力下测量3次进行平均，结果如图6所示。
![]()
图 6 射流撞击喷头与传统15PY2喷头雨滴谱对比Figure 6. Comparison of droplet size distribution between the jet impingement sprinkler and the traditional 15PY2 sprinkler水量累积频率为 10%、50%以及 90%对应的粒径比较如表9所示，试验结果表明压力升高，水量累积频率为 10%、50%以及 90%对应的粒径均减小，这与JIANG等[28]的研究结果是一致的，即喷头雾化程度随着压力的增加而增加，前中段水滴粒径较小，后段水滴粒径较大，这与向清江等[3]的试验结果相似。从图6中可以看到，在射程前中段，射流撞击喷头与传统15PY2喷头水滴粒径分布较为相似，且没有出现较大粒径的水滴，射流撞击喷头的水滴粒径普遍大于传统15PY2喷头，传统15PY2喷头中段的小水滴是由射流主体剥落形成的，而稍大的水滴是由副喷嘴雾化形成，射流撞击喷头中段的水滴均属于射流主体剥落形成，而其水滴粒径较大的原因在于副射流撞击导致的主射流发展不稳定，部分边缘射流因撞击波影响而提前破碎，避免了过小的水滴粒径导致的较多的蒸发飘移损失[29-30]。在射程后段，水滴的粒径区间较大，150 kPa下射流撞击喷头与传统15PY2喷头相比整体水滴粒径更大，但在250 kPa下射流撞击喷头整体水滴粒径更小，其原因在于，传统15PY2喷头只有主射流能够到达射程后段，而射流撞击喷头到达射程后段的是主副射流撞击后合成的射流，其总体流量更大，结果造成了同时有更好的雾化和更大的水滴粒径，而压力增大，撞击波造成的影响增大，克服了流量大带来的影响，导致了较小的水滴粒径区间分布。
表 9 射流撞击喷头与传统15PY2喷头的平均粒径对比Table 9. Comparison of average drop size between jet impingement sprinkler and traditional 15PY2 sprinkler喷头类型
Sprinkler type喷头工作压力
Working pressure/kPa雨滴测量位置
Measurig position/m累计频率10%粒径
Cumulative percentage 10% drop size d10/mm累计频率50%粒径
Cumulative percentage 50% drop size d50/mm累计频率90%粒径
Cumulative percentage 90% drop size d90/mm射流撞击喷头
Jet impingement sprinkler150 6 0.68 1.11 1.79 12 2.67 3.94 6.02 250 6 0.45 0.85 1.41 14 1.63 2.76 3.78 传统15PY2喷头
Traditional 15PY2 sprinkler150 6 0.49 0.99 1.25 12 2.12 3.91 5.45 250 6 0.45 0.81 1.18 14 2.06 3.13 4.15 3. 结 论
1）正交试验得到影响喷头综合工作性能的结构参数依次为副喷管长度，副喷嘴进口锥角，副喷管内径，副喷嘴仰角。且射流撞击喷头最优结构尺寸为：副喷管长度20 mm，副喷嘴进口锥角55°，副喷管内径6 mm，副喷嘴仰角33°。
2）在150～300 kPa的全压力下，射流撞击式喷头的组合均匀系数及综合评分均值均高于传统15PY2喷头，其中组合均匀系数平均提升4.84%，综合评分均值平均提升4%。
3）对比了射流撞击喷头和传统15PY2喷头前中段和后段的水滴分布，发现射流撞击导致主射流发展不稳定，部分边缘射流因撞击波影响而提前破碎，使射流撞击喷头前中段水滴粒径普遍大于传统15PY2喷头，避免了过小的水滴粒径导致的较多的蒸发飘移损失，然而射流撞击对水滴分布的影响机理尚不明确，后续将通过射流撞击喷头与射流不撞击喷头对比进一步研究液滴分布变化规律。
-
![]()
图 1 研究区位置与样地点分布图
Figure 1. Location of the study area and distribution of sampling sites
![]()
图 2 不同品种采样点植株生物量实测值分布
Figure 2. Distribution of measured plant biomass at sampling points of different varieties
![]()
图 3 采样点光谱反射率和一阶微分值
Figure 3. Spectral reflectance and first order differential values of sampling points
![]()
图 6 一阶微分数据与植被指数数据组合基于PSO-RF、PSO-SVM、PSO-BPNN模型对花生生物量预测模型
注：a、b、c中左图为回归拟合图，右图是PSO优化模型的寻优过程
Figure 6. Peanut biomass prediction models based on the combination of first-order derivative data and vegetation indices using PSO-RF, PSO-SVM, and PSO-BPNN models
Note: In Fig. a, b, and c, the left panels show the regression fitting plots, while the right panels depict the optimization process of the PSO model.
表 1 高光谱成像仪的详细参数
Table 1 Detailed parameters of the hyperspectral imager
参数名称
Parameter name参数值
Parameter value光谱范围
Spectral range/nm400～ 1000 光谱分辨率
Spectral resolution/nm3.5 数值孔径
Numerical aperture/nm像方F/2.8 光谱采样率
Spectral sampling rate/nm0.5 全幅像素
Full frame pixels/像素1936× 1456 像素间距
Pixel pitch/µm4.54 横向视角(FOV)
Field of view/(°)30.25 单幅图像分辨率
Single image resolution/像素1920× 2080 空间分辨率
Spatial resolution/m0.085 
下载: 导出CSV
表 2 部分植被指数及计算公式
Table 2 Partial vegetation indices and calculation formulas
植被指数
Vegetation index (VI)计算公式
Calculation formula修正地面叶绿素指数MMTCI
Modified medium resolution transformed chlorophyll index((R750-R680+0.03)(R750-R710))/(R710-R680) 双峰冠层氮素指数DCNI
Double peak canopy nitrogen index((R750-R670+0.09)(R750-R700))/(R700-R670) 增强植被指数EVI
Enhanced vegetation index2.5(R800-R670)/(1+R800+6R670-7R479) 红边叶绿素指数RECI
Red edge chlorophyll index(R750/R710)-1 中分辨率陆地叶绿素成像指数 MTCI
Medium resolution imaging spectrometer chlorophyll index(R750-R710)/(R710-R680) 绿色归一化植被指数GNDVI
Green normalized difference vegetation index(R801-R550)/(R800+R550) 新植被指数NVI
New vegetation index(R777-R747)/R673 复归一化差值植被指数RDVI
Renormalized difference vegetation index(R800-R670)/(√R800+R670) 植被衰老反射率指数PLSRI
Plant leaf senescence reflectance index(R680-R500)/R750 新型植被指数LICI
Light-use efficiency chlorophyll indexR735/R720-(R573-R680)/（R573+R680) 注：Rλ表示波长为λ nm的光谱反射率。 Note:Rλis spectral reflectance at a wavelength of λ nm.
下载: 导出CSV
表 3 基于三种机器学习方法构建花生生物量预测模型精度对比
Table 3 Comparison of peanut biomass prediction models constructed using three machine learning methods
数据
Data建模方法
Modeling method变量个数
Number of variables训练集R2
Training set R2训练集RMSE
Training set RMSE/
(kg·m-2)测试集R12
Testing set R12测试集RMSE1
Testing set RMSE1/
(kg·m−2)原始波段反射率
Raw band dataSVM 2 0.197 0.167 0.147 0.158 BPNN 2 0.278 0.159 0.104 0.162 RF 2 0.541 0.076 0.126 0.157 植被指数
Vegetation index dataSVM 2 0.590 0.119 0.364 0.137 BPNN 2 0.622 0.115 0.342 0.139 RF 2 0.851 0.072 0.564 0.113 一阶微分
First derivative dataSVM 2 0.482 0.134 0.302 0.143 BPNN 2 0.487 0.134 0.330 0.140 RF 2 0.846 0.073 0.686 0.096
下载: 导出CSV
表 4 基于特征组合的三种机器学习方法模型精度对比
Table 4 Comparison of peanut biomass prediction models constructed using three machine learning methods
数据源
Data建模方法
Modeling method变量个数
Number of
variables训练集R2
Training set R2训练集RMSE
Training set RMSE/
(kg·m−2)测试集R12
Testing set R12测试集RMSE1
Testing set RMSE1/
(kg·m−2)一阶微分+植被指数
First derivative data + Vegetation index dataSVM 4 0.656 0.109 0.618 0.106 BPNN 4 0.570 0.122 0.480 0.124 RF 4 0.734 0.096 0.754 0.085 一阶微分+原始波段
First derivative data + Raw band dataSVM 4 0.611 0.116 0.564 0.113 BPNN 4 0.364 0.149 0.355 0.138 RF 4 0.751 0.093 0.674 0.098 植被指数+原始波段
Vegetation index data + Raw band dataSVM 4 0.791 0.081 0.501 0.121 BPNN 4 0.261 0.161 0.216 0.152 RF 4 0.554 0.125 0.469 0.125 一阶微分+植被指数+原始波段
First derivative data + Vegetation index data+ Raw band datSVM 6 0.660 0.107 0.530 0.117 BPNN 6 0.479 0.135 0.344 0.139 RF 6 0.651 0.110 0.640 0.103
下载: 导出CSV
表 5 基于粒子群优化机器学习方法的花生生物量估测模型精度对比
Table 5 Comparison of peanut biomass prediction models using particle swarm optimization-based machine learning methods
数据源
Data建模方法
Modeling method变量个数
Number of variables训练集R2
Training set R2训练集RMSE
Training set RMSE/
(kg·m-2)测试集R12
Testing set R12测试集RMSE1
Testing set RMSE1/
(kg·m-2)一阶微分+植被指数
First derivative data + Vegetation index dataPSO-SVM 4 0.637 0.112 0.636 0.103 PSO-BPNN 4 0.590 0.119 0.536 0.117 PSO-RF 4 0.810 0.081 0.800 0.076 一阶微分+原始波段
First derivative data + Raw band dataPSO-SVM 4 0.621 0.115 0.601 0.108 PSO-BPNN 4 0.540 0.127 0.414 0.131 PSO-RF 4 0.815 0.080 0.709 0.092
植被指数+原始波段
Vegetation index data + Raw band dataPSO-SVM 4 0.792 0.085 0.568 0.113 PSO-BPNN 4 0.285 0.158 0.270 0.146 PSO-RF 4 0.818 0.079 0.499 0.121 一阶微分+植被指数+原始波段
First derivative data + Vegetation index data+
Raw band datPSO-SVM 6 0.662 0.108 0.564 0.113 PSO-BPNN 6 0.556 0.124 0.504 0.121 PSO-RF 6 0.830 0.077 0.742 0.087
下载: 导出CSV
-
[1] 廖伯寿. 我国花生生产发展现状与潜力分析[J]. 中国油料作物学报,2020,42(2):161. LIAO Boshu. A review on progress and prospects of peanut industry in China[J]. Chinese Journal of Oil Crop Sciences, 2020, 42(2): 161-166. (in Chinese with English abstract)
[2] 闫琰,王秀东,王济民,等. “双循环” 背景下国家粮食安全战略研究[J]. 中国工程科学,2023,25(4). YAN Yan, WANG Xiudong, WANG Jimin, et al. National food security strategy against the backdrop of domestic and international dual circulation[J]. Engineering Sciences in China, 2023, 25(4). (in Chinese with English abstract)
[3] QI H, WU Z, ZHANG L, et al. Monitoring of peanut leaves chlorophyll content based on drone-based multispectral image feature extraction[J]. Computers and Electronics in Agriculture, 2021, 187: 106292. doi: 10.1016/j.compag.2021.106292
[4] 曹中盛,李艳大,黄俊宝,等. 监测花生叶面积指数和地上部生物量的最优植被指数及适宜波段带宽[J]. 中国油料作物学报,2022,44(6):1320-1328. CAO Zhongsheng, LI Yanda, HUANG Junbao, et al. Sensitive vegetation indices and optimal bandwidths for monitoring peanut LAI and AGB[J]. Journal of the Chinese Oil Crops, 2022, 44(6): 1320-1328. (in Chinese with English abstract)
[5] SHU M, LI Q, GHAFOOR A, et al. Using the plant height and canopy coverage to estimation maize aboveground biomass with UAV digital images[J]. European Journal of Agronomy, 2023, 151: 126957. doi: 10.1016/j.eja.2023.126957
[6] 曾斌,唐敏,唐伟,等. 花生粕和花生秸秆在动物饲用中的研究进展[J]. 中国饲料,2023(5):152-162. ZENG Bin, TANG Min, TANG Wei, et al. Research progress of peanut meal and peanut straw in animal feeding[J]. China Feed, 2023(5): 152-162. (in Chinese with English abstract)
[7] ZHAO F T, WANG F, ZHANG F, et al. Catalytic hydropyrolysis of crop straws with different biochemical composition[J]. International Journal of Hydrogen Energy, 2023, 48(19): 6927-6936. doi: 10.1016/j.ijhydene.2022.03.047
[8] 冯惠芬,李映雪,吴芳,等. 机器学习结合高光谱植被指数与SPAD值估算冬小麦氮含量[J]. 农业工程学报,2024,40(1):227-237. FENG Huifen, LI Yingxue, WU Fang, et al. Estimating winter wheat nitrogen content using SPAD and hyperspectral vegetation indices with machine learning[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2024, 40(1): 227-237. (in Chinese with English abstract)
[9] 刘一博,裴杰,方华军,等. 利用无人机影像反演水稻SPAD值的最优空间窗口确定[J]. 农业工程学报,2023,39(19):165-174. doi: 10.11975/j.issn.1002-6819.202304205 LIU Yibo, PEI Jie, FANG Huajun, et al. Optimizing spatial window selection for rice SPAD value retrieval using multispectral UAV images[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2023, 39(19): 165-174. (in Chinese with English abstract). doi: 10.11975/j.issn.1002-6819.202304205
[10] 刘修宇,张锦水,吴俊旭,等. 时-空-谱融合自动纯化生成样本的农作物遥感分类[J]. 农业工程学报,2024,40(10):157-167. doi: 10.11975/j.issn.1002-6819.202311098 LIU Xiuyu, ZHANG Jinshui, WU Junxu, et al. Automatic purifying and generating graded samples for crop remote sensing classification using temporal-spatial-spectral fusion[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2024, 40(10): 157-167. (in Chinese with English abstract) doi: 10.11975/j.issn.1002-6819.202311098
[11] 兰仕浩,李映雪,吴芳,等. 基于卫星光谱尺度反射率的冬小麦生物量估算[J]. 农业工程学报,2022,38(24):118-128. LAN Shihao, LI Yingxue, WU Fang, et al. Winter wheat biomass estimation based on satellite spectral-scale reflectance[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2022, 38(24): 118-128. (in Chinese with English abstract)
[12] 赵越,徐大伟,范凯凯,等. Landsat 8和机器学习估算蒙古高原草地地上生物量[J]. 农业工程学报,2022,38(24):138-144. doi: 10.11975/j.issn.1002-6819.2022.24.015 ZHAO Yue, XU Dawei, FAN Kaikai, et al. Estimating above-ground biomass in grassland using Landsat 8 and machine learning in Mongolian plateau[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2022, 38(24): 138-144. (in Chinese with English abstract) doi: 10.11975/j.issn.1002-6819.2022.24.015
[13] 刘涛,张寰,王志业,等. 利用无人机多光谱估算小麦叶面积指数和叶绿素含量[J]. 农业工程学报,2021,37(19):65-72. LIU Tao, ZHANG Huan, WANG Zhiye, et al. Estimation of the leaf area index and chlorophyll content of wheat using UAV multi-spectrum images[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2021, 37(19): 65-72. (in Chinese with English abstract)
[14] 王志业,刘涛,张寰,等. 基于多光谱影像特征的小麦锈病反演模型构建[J]. 河南农业大学学报,2023,57(5):826-834. WANG Zhiye, LIU Tao, ZHANG Huan, et al. Construction of wheat rust disease inversion model based on multi-spectral image features[J]. Journal of Henan Agricultural University, 2023, 57(5): 826-834. (in Chinese with English abstract)
[15] 苏宝峰,刘砥柱,陈启帆,等. 基于时间序列植被指数的小麦条锈病抗性等级鉴定方法[J]. 农业工程学报,2024,40(4):155-165. SU Baofeng, LIU Dizhu, CHEN Qifan, et al. Method for the identification of wheat stripe rust resistance grade using time series vegetation index[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2024, 40(4): 155-165. (in Chinese with English abstract).
[16] 许鑫,张力,岳继博,等. 农田环境下无人机图像并行拼接识别算法[J]. 农业工程学报,2024,40(9):154-163. XU Xin, ZHANG Li, YUE Jibo, et al. Parallel mosaic recognition algorithm for UAV images in farmland environment[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2024, 40(9): 154-163. (in Chinese with English abstract)
[17] ZHANG Z, ZHU L. A review on unmanned aerial vehicle remote sensing: Platforms, sensors, data processing methods, and applications[J]. Drones, 2023, 7(6): 398. doi: 10.3390/drones7060398
[18] LI W, WEISS M, JAY S, et al. Daily monitoring of effective green area index and vegetation Chlorophyll content from continuous acquisitions of a multi-band spectrometer over winter wheat[J]. Remote Sensing of Environment, 2024, 300: 113883. doi: 10.1016/j.rse.2023.113883
[19] 帅爽,张志,张天,等. 特征优化结合随机森林算法的干旱区植被高光谱遥感分类方法[J]. 农业工程学报,2023,39(9):287-293. SHUAI Shuang, ZHANG Zhi, ZHANG Tian, et al. Hyperspectral image classification method for dryland vegetation by combining feature optimization and random forest algorithm[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2023, 39(9): 287-293. (in Chinese with English abstract)
[20] ABURAED N, ALKHATIB M Q, MARSHALL S, et al. A review of spatial enhancement of hyperspectral remote sensing imaging techniques[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2023, 16: 2275-2300. doi: 10.1109/JSTARS.2023.3242048
[21] 唐子竣,张威,黄向阳,等. 基于地面高光谱遥感的大豆产量估算模型研究[J]. 农业机械学报,2024,55(1):145-153,240. TANG Zijun, ZHANG Wei, HUANG Xiangyang, et al. Soybean Seed Yield Estimation Model Based on Ground Hyperspectral Remote Sensing Technology[J]. Transactions of the Chinese Society for Agricultural Machinery, 2024, 55(1): 145-153,240. (in Chinese with English abstract)
[22] 齐浩,吕亮杰,孙海芳,等. 基于无人机高光谱与机器学习的小麦品系产量估测研究[J/OL]. 农业机械学报,[2024-06-03]. http://kns.cnki.net/kcms/detail/11.1964.S.20240520.1707.002.html. QI Hao, LV Liangjie, SUN Haifang, et al.Yield estimation of wheat lines based on UAV hyperspectral and machine learning[J/OL]. Transactions of the Chinese Society for Agricultural Machinery,[2024-06-03]. http://kns.cnki.net/kcms/detail/11.1964.S.20240520.1707.002.html. (in Chinese with English abstract)
[23] FENG H, TAO H, FAN Y, et al. Comparison of winter wheat yield estimation based on near-surface hyperspectral and UAV hyperspectral remote sensing data[J]. Remote Sensing, 2022, 14(17): 4158. doi: 10.3390/rs14174158
[24] CHEN X, LI F, SHI B, et al. Estimation of winter wheat plant nitrogen concentration from UAV hyperspectral remote sensing combined with machine learning methods[J]. Remote Sensing, 2023, 15(11): 2831. doi: 10.3390/rs15112831
[25] XU T, WANG F, SHI Z, et al. Dynamic estimation of rice aboveground biomass based on spectral and spatial information extracted from hyperspectral remote sensing images at different combinations of growth stages[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2023, 202: 169-183. doi: 10.1016/j.isprsjprs.2023.05.021
[26] SAHOO R N, REJITH R G, GAKHAR S, et al. Drone remote sensing of wheat N using hyperspectral sensor and machine learning[J]. Precision Agriculture, 2024, 25(2): 704-728. doi: 10.1007/s11119-023-10089-7
[27] 江凯伦,安吉庆,赵雨薇,等. 采用RNCA-PSO-ELM的水稻叶绿素光谱特征分析与反演[J]. 农业工程学报,2022,38(8):178-186. JIANG Kailun, AN Jiqing, ZHAO Yuwei, et al. Analysis and inversion of rice chlorophyll spectral characteristics using RNCA-PSO-ELM[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2022, 38(8): 178-186. (in Chinese with English abstract)
[28] 梁文广,吴勇锋,石一凡,等. 基于无人机高光谱的太浦河水质反演[J]. 测绘通报,2024(4):29-34. LIANG Wenguang, WU Yongfeng, SHI Yifan, et al. Inversion of Taipu River water quality parameters by UAV hyperspectral imaging technology[J]. Bulletin of Surveying and Mapping, 2024(4): 29-34. (in Chinese with English abstract)
[29] 彭俊杰,汪泓,王宇,等. 基于GA-PLS-SPA的辣椒叶片叶绿素含量高光谱估测[J]. 江苏农业科学,2024,52(7):184-192. PENG Junjie, WANG Hong, WANG Yu, et al. Hyperspectral estimation of chlorophyll content in pepper leaves based on GA-PLS-SPA[J]. Jiangsu Agricultural Sciences, 2024, 52(7): 184-192. (in Chinese with English abstract)
[30] MAHIEU B, QANNARI E M, JAILLAIS B. Extension and significance testing of Variable Importance in Projection (VIP) indices in Partial Least Squares regression and Principal Components Analysis[J]. Chemometrics and Intelligent Laboratory Systems, 2023, 242: 104986. doi: 10.1016/j.chemolab.2023.104986
[31] BELGIU M, DRĂGUŢ L. Random forest in remote sensing: A review of applications and future directions[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2016, 114: 24-31. doi: 10.1016/j.isprsjprs.2016.01.011
[32] WANG L, ZENG Y, CHEN T. Back propagation neural network with adaptive differential evolution algorithm for time series forecasting[J]. Expert Systems with Applications, 2015, 42(2): 855-863. doi: 10.1016/j.eswa.2014.08.018
[33] MAVROFORAKIS M E, THEODORIDIS S. A geometric approach to support vector machine (SVM) classification[J]. IEEE Transactions on Neural Networks, 2006, 17(3): 671-682. doi: 10.1109/TNN.2006.873281
[34] WANG D, TAN D, LIU L. Particle swarm optimization algorithm: an overview[J]. Soft Computing, 2018, 22: 387-408. doi: 10.1007/s00500-016-2474-6
[35] 师翊,王应宽,赵龙,等. 基于CART重要度排序和混合ELM模型的蒸散预测[J]. 农业工程学报,2023,39(14):89-96. doi: 10.11975/j.issn.1002-6819.202301095 SHI Yi, WANG Yingkuan, ZHAO Long, et al. Evapotranspiration prediction using CART importance ranking and hybrid ELM models[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2023, 39(14): 89-96. (in Chinese with English abstract) doi: 10.11975/j.issn.1002-6819.202301095
[36] 严格齐,赵婉莹,于镇伟,等. 基于超参数优化算法的随机森林模型预测奶牛呼吸频率[J]. 农业工程学报,2024,40(11):195-203. YAN Geqi, ZHAO Wanying, YU Zhenwei, et al. Predicting respiratory rate of dairy cows using hyperparameter-optimized random forest models[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2024, 40(11): 195-203. (in Chinese with English abstract)
[37] CHEN P, HABOUDANE D, TREMBLAY N, et al. New spectral indicator assessing the efficiency of crop nitrogen treatment in corn and wheat[J]. Remote Sensing of Environment, 2010, 114(9): 1987-1997. doi: 10.1016/j.rse.2010.04.006
[38] 董恒,孟庆野,王金梁,等. 一种改进的叶绿素提取植被指数[J]. 红外与毫米波学报,2012,31(4):336-341. doi: 10.3724/SP.J.1010.2012.00336 DONG Heng, MENG Qingye, WANG Jinliang, et al. A modified vegetation index for crop canopy chlorophyll content retrieval[J]. Journal of Infrared and Millimeter Waves, 2012, 31(4): 336-341. (in Chinese with English abstract) doi: 10.3724/SP.J.1010.2012.00336
[39] 张凡凡,贾子玉,庄前友,等. 基于高光谱遥感的针茅群落生物量估算[J/OL]. 草业科学, [2024-07-30]. http://kns.cnki.net/kcms/detail/62.1069.s.20240723.1508.002.html. ZHANG Fanfan, JIA Ziyu, ZHUANG Qianyou, et al. Estimation of Stipa biomass based on hyperspectral remote sensing[J]. Pratacultural Science, [2024-07-30]. http://kns.cnki.net/kcms/detail/62.1069.s.20240723.1508.002.html. (in Chinese with English abstract)
[40] 赵玉清,杨慧丽,张悦,等. 基于特征组合与SVM的小粒种咖啡缺陷生豆检测[J]. 农业工程学报,2022,38(14):295-302. doi: 10.11975/j.issn.1002-6819.2022.14.033 ZHAO Yuqing, YANG Huili, ZHANG Yue, et al. Detection of defective arabica green coffee beans based on feature combination and SVM[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2022, 38(14): 295-302. (in Chinese with English abstract) doi: 10.11975/j.issn.1002-6819.2022.14.033
[41] 王晗,向友珍,李汪洋,等. 基于无人机多光谱遥感的冬油菜地上部生物量估算[J]. 农业机械学报,2023,54(8):218-229. WANG Han , XIANG Youzhen , LI Wangyang , et al. Estimation of winter rapeseed above-ground biomass based on UAV multi-spectral remote sensing[J]. Transactions of the Chinese Society for Agricultural Machinery, 2023, 54(8): 218-229. (in Chinese with English abstract)
[42] 周敏姑,闫云才,高文,等. 基于多光谱遥感和CNN的玉米地上生物量估算模型[J]. 农业机械学报,2024,55(9):238-248. ZHOU Mingu, YAN Yuncai, GAO Wen, et al. Estimating aboveground biomass of maize based on multispectral remote sensing and CNN[J]. Transactions of the Chinese Society for Agricultural Machinery, 2024, 55(9): 238-248. (in Chinese with English abstract).
计量
- 文章访问数: 157
- HTML全文浏览量: 11
- PDF下载量: 46
