作物生长模拟的研究进展

作物生长模拟的研究进展戚昌瀚,殷新佑(江西农业大学南昌330045)作物生长模拟是作物科学中引进系统分析方法和应用计算机后兴起的个研究领域。它是通过对作物生育和产量的实验数据加以理论概括和数据抽象,找出作物生育动态及其与环境间关系的动态模型,然后在计……     

定量的感参数与作物肥水模型链接初探

针对当前作物肥水模型和定量遥感参数在农业上应用的各自局限性，提出通过完善定量遥感的输出接口和作物模型的输入接口，将定量遥感参数与作物肥水模型进行链接，通过建立不同生育时段订正的绿度指数和叶片纱分段函数以及订正的湿度指数与耕层土壤、植株水分及产量品质函数，使得生产上大范围、动态的肥水监测成为可能，最后对应用链接模型通过不同肥水管理措施对作物产量和品质进行调控的前景进行了展望。     

定量遥感参数与作物肥水模型链接初探

针对当前作物肥水模型和定量遥感参数在农业上应用的各自局限性 ,提出通过完善定量遥感的输出接口和作物模型的输入接口 ,将定量遥感参数与作物肥水模型进行链接 ,通过建立不同生育时段订正的绿度指数和叶片氮素分段函数以及订正的湿度指数与耕层土壤、植株水分及产量品质函数 ,使得生产上大范围、动态的肥水监测成为可能 ,最后对应用链接模型通过不同肥水管理措施对作物产量和品质进行调控的前景进行了展望     

神经网络自变量输入模式的视电阻率求解算法

根据瞬变电磁场理论公式中的响应和自变量之间的关系特点,提出用非线性方程模式的BP神经网络求解电阻率。通过构造单输入单输出网络结构,建立以不同时间点上的电流归一化的感应电压值为输入、视电阻率值为输出的神经网络,来拟合瞬变电磁场的二次涡流曲线。利用数值方法计算出的数据验证该训练网络的精确性,比较了不同算法对训练精度和收敛速度产生的影响。以重庆大学某处的防空洞探测实验为例验证了该算法的有效性,该算法避开具体的复杂电磁场计算或数值反问题计算,从而实现电阻率快速计算,为快速成像准备必要条件。     

基于LM-BP神经网络的耕地土壤养分等级划分模型──以皖南山区为例

研究旨在通过BP神经网络方法,构建起LM-BP网络结构(5-M-1)模型,达到对土壤养分等级划分的目的,为合理的土壤养分管理提供可靠依据。采用Levenberg-Marquardt (LM)训练算法,构建3层网络模型：一个输入层、一个隐含层、一个输出层,利用3层网络作为耕地土壤养分等级划分模型。利用土壤养分各级评价标准作为模型的训练样本和测试样本,以此来对BP神经网络进行训练和测试,并对歙县土壤养分进行综合评价。结果表明：LM-BP网络结构对测试样本输出的预测值和实际参考值是一致的。最终通过灰色关联模型和主成分分析方法对歙县土壤养分的综合评价结果与BP神经网络的模拟结果相对比,发现也是基本一致的。LM-BP网络结构应用于土壤养分等级划分中,得到了很好的预测效果,为智能算法应用于农业领域奠定了良好的基础。     

利用神经网络提取棉花叶片数字图像氮素含量的初步研究

选取6种输入向量组合，利用线性网络、BP网络以及径向基网络等3种神经网络模型进行比较研究，筛选最适宜网络模型和最佳输入组合，建立叶片数字图像彩色信息和叶片氮含量的关系模型，探索利用神经网络技术获取叶片数字图像信息的方法。结果表明，径向基网络在利用数字图像(B，H，G-R，G/R)指标作为网络输入向量时，能够实现获取棉花叶片数字图像氮含量的目标。径向基网络训练的180组样本的训练精度均达到极显著水平（r = 0.9022**），30组测试样本的预测值与实测值也达到极显著相关（r = 0.8674**），径向基网络和(B，H，G-R，G/R)向量是一种适合本研究的数学模型。对利用神经网络提取棉花叶片数字图像氮含量技术的初步探索，拓展了神经网络和数字图像技术在农业生产中的应用。     

作物高产群体LAI动态模拟模型的建立与检验

针对目前已有群体叶面积指数（LAI）模拟模型形式多样、参数较多以及应用性不强等问题,对春玉米、水稻和冬小麦的LAI及出苗至成熟天数进行归一化处理,分别将最大LAI和出苗至成熟天数定为1,以相对LAI (0~1)和相对时间(0~1)为参数进行LAI动态模拟,筛选、建立了一个适用于这3种作物的相对化LAI动态模拟模型y= (a+bx) / (1+cx+dx²)。其中,春玉米y= (0.0134+0.3234x) / (1－2.774x+2.4178x²),r=0.9859**;水稻y= (0.0777+0.0205x) / (1－2.73744x+2.0484x²),r=0.9865**;冬小麦y= (0.0131+0.0035x) / (1－2.4515x+1.5273x²),r=0.9719**。利用该模型,自拔节期起就能够较准确地进行LAI的动态预测,其在春玉米、水稻和冬小麦上的准确度（以k表示）分别为1.050、1.0357和1.1168,精确度（以R²表示）分别为0.8728、0.9270和0.9254。3种作物整个生育期内模型的模拟值与测量值的精确度均在0.98以上,准确度达0.86以上,表明相对化LAI动态模型能够准确地反映作物群体动态变化。不仅可以计算出作物生育期间的平均LAI、总光合势,还能计算任一时刻的LAI以及任一时段的光合势。结合田间调查还可得到作物生长期间的平均净同化率和平均作物生长率等产量相关的重要生理参数。根据作物群体中各光合生理参数与产量的关系,提出了3种作物进一步增产的可能途径。     

基于BP神经网络的玉米膳食纤维挤压改性预测模型

以玉米为原料进行挤压试验,开发基于双螺杆挤压的玉米膳食纤维改性的BP神经网络预测模型。此网络模型以螺杆转速、喂料速度、含水量和机筒温度为输入单元,以糊化度、吸水性(WAI)和水溶性(WSI)为输出单元,拥有一个8单元的隐含层。网络输出和目标输出之间的相关系数为0.984 46,预测误差小于10%,具有较好网络性能,能够实现对玉米糊化度、吸水性和水溶性等挤压性能的预测。进一步开发人机交互图形化用户界面设计(GUI),方便预测模型的应用。     

基于无迹卡尔曼滤波神经网络的光伏发电预测

针对光伏发电系统在不同天气状况下发电功率预测精度不高的问题,在分析传统方法的基础上,提出一种无迹卡尔曼滤波神经网络光伏发电预测方法。该方法利用无迹卡尔曼滤波实时更新神经网络模型的权重,以直流电压和电流作为系统的输入,以有功功率和无功功率作为系统的输出,分别建立两个独立的双输入单输出功率预测模型。实验结果表明：所提出的方法对有功功率和无功功率的预测精度分别为97.3%和94.2%,并且对天气具有良好的鲁棒性。     

紧凑型玉米叶龄模式栽培促控指标初探

本文对紧凑型玉米掖单１３号，用叶龄作指标采用以肥水为中心的促控措施，即“促，攻，补”肥水运筹技术，符合掖单１３号生育规律。研究结果表明，无论产量，群体结构，产量性状以及与产量密切相关的主要生育和生理指标－叶面积系数变化动态，干物质积累与分配，光势，净同化率的变化，光能利用率，能量和累动态等方面都明显优于常规栽培，因此叶龄模式促控栽培技术在玉米生产上值得应用和推广。     

analysis and application of demand forecasting model with multi element variable parameters

A forecasting model for logistics demand was presented to overcome the limitations of single goal forecasts of logistics demand and forecast data complexity. Based on the forecasting evaluation index and pretreatment of rough set theory, a multi input and multi output wavelet network (MMWNN) model for forecasting multi element regional logistics demand was studied. The network configuration was confirmed using the stepwise checkout and iterative gradient descent methods. After rough set reduction, the evaluation index was used to forecast the multi element regional logistics demand. The results of the numerical example indicate the feasibility and effectiveness of the model.     

Energy Use and Output Assessment of Food-Forage Production Systems

An assessment of the energy requirements of different intensive forage production systems was carried out at the Indian Grassland Fodder, and Agroforestry Research Institute, Jhansi. This included assessment of energy use and output for five intensive crop production systems: (1) sorghum (single‐cut)–berseem, (2) cowpea–sorghum (single‐cut)–berseem + mustard–maize + cowpea, (3) sorghum (multi‐cut)–berseem + oats, (4) guar–oats–maize and (5) sorghum (single‐cut)–wheat–fallow. In all the systems, the Napier bajra hybrid (IGFRI‐3) was transplanted in regular plots of 50 m × 13 m. Results revealed that the total annual energy use was highest for sorghum (multi‐cut)–berseem + oats (36 606 MJ ha^?1), followed by sorghum (single‐cut)–berseem + mustard, sorghum (single‐cut)–berseem–cowpea, guar–oats–maize and sorghum (single‐cut)–wheat–fallow (for which values were 31 086, 30 449, 29 867 and 25 956 MJ energy ha^?1, respectively). The high value found for sorghum (multi‐cut)–berseem + oats might be attributable to the multi‐cuts in this system. Energy use by fertilizers represented the major part of the total energy use, amounting to 28–38 % in all treatments, followed by energy used in electricity, machinery, seeds, human labour and farmyard manure (FYM), in case of all with slight increase in input. In sorghum–wheat, energy use by seeds occupied the second position, followed by energy used in electricity, human labour, FYM and machinery/diesel. Pesticides contributed the least energy utilization in all the treatments. Herbicides were used for the control of weeds. Among the five forage production systems, sorghum (single‐cut)–berseem + mustard–maize + cowpea was found to be the most energy efficient, followed by sorghum (multi‐cut)–berseem + oats–sorghum (multi‐cut), sorghum (single‐cut)–berseem–cowpea, guar–oats–maize and sorghum (single‐cut)–wheat–fallow. Sorghum (single‐cut)–berseem + mustard–maize + cowpea increased the fertility of the soil, resulting in a higher percentage of organic carbon, higher availability of nitrogen and optimal balancing of the C:N ratio in the upper layers of the soil. These intensive crop production systems also maintain the optimum microbial population in the crop root zone. The benefit–cost ratio (B:C ratio) for the most energy‐efficient forage production system was 1.37 : 1. However, the highest B:C ratio was found in the sorghum–wheat rotation (1: 1.57).     

基于人工神经网络理论的土壤水分预测研究

土壤水分含量是影响作物生长的重要因素,精确的预测技术对水资源的合理利用与管理具有重要的指导意义。利用人工神经网络理论,建立了以降水量、蒸发量、相对湿度和地下水埋深为输入因子,土壤水分含量为输出因子的预测模型,并对其预测精度进行了评价。结果表明,BP神经网络模型预测土壤含水率的最大误差为8.66%,平均误差为4.27%,预测精度达到0.989。模型具有较高的预测精度,其结果可为制定合理的水资源调配方案和调度计划提供科学依据。     

Design and experimental validation of a novel double ring plate gear reducer

A novel double ring plate gear reducer (DRPGR) is designed and its experimental validation is conducted. By using the closed graph method, the modification coefficients of internal gearing in DRPGR are determined. Based on this, the overall structure of the reducer is designed and its kinematics simulation is achieved. With APDL programming language of ANSYS, a 3D finite element (FE) model of DRPGR is established, and its modal analysis and loaded tooth contact analysis (LTCA) are conducted respectively. The modal analysis reveals that the lower natural frequencies of DRPGR are much higher than the input frequency and the lower modes can be classified as vertical and horizontal bending of input and output shafts. The LTCA indicates that three pairs of gear teeth contact simultaneously at rated output torque. Due to the multi teeth contacts, the maximum stress is much lower than the permissible stress of internal gearings. The experimental tests of the prototype validate that DRPGR is adapted to both single driven and dual driven work condition. When single droved, DRPGR can eliminate the kinematic uncertainty by non 180 degree phase difference design. DRPGR claims a better dynamic performance in dual driven work condition. The efficiency of DRPGR increases with the increment of output torque and the reducer has an excellent short term overloaded capacity. The experimental results validate that the proposed method can be used to guide the design of that kind of reducer.     

An empirical analysis of risk in conventional and organic arable farming in The Netherlands

This paper assesses and compares risk in conventional and organic arable farming in The Netherlands with respect to family farm income and underlying price and production variables. To investigate the risk factors the farm accountancy data network was used containing unbalanced panel data from 196 conventional and 29 organic representative Dutch arable farms (for the period 2002 up to and including 2011). Variables with regard to price and production risk were identified using a family farm income analysis scheme. Price risk variables are input and output prices, while yield volatility of different crops is the main production risk variable. To assess risk, an error components implicit detrending method was applied and the resulting detrended standard deviations were compared between conventional and organic farms. Results indicate that the risk at the level of family farm income is higher in organic farming. The underlying variables show higher risk for organic farms in crop yields, crop prices and variable input costs per crop.     

Biomass yield and energy balance of giant reed (Arundo donax L.) cropped in central Italy as related to different management practices

In order to evaluate the possibility of reducing energy input in giant reed (Arundo donax L.) as a perennial biomass crop, a field experiment was carried out from 1996 to 2001 in central Italy. Crop yield response to fertilisation (200–80–200 kg ha⁻¹ N–P–K), harvest time (autumn and winter) and plant density (20,000 and 40,000 plants per ha) was evaluated. The energy balance was assessed considering the energy costs of production inputs and the energy output obtained by the transformation of the final product. The crop yield increased by +50% from the establishment period to the 2nd year of growth when it achieved the highest dry matter yield. The mature crop displayed on average annual production rates of 3 kg dry matter m⁻², with maximum values obtained in fertilised plot and during winter harvest time.

Fertilisation mainly enhanced dry matter yield in the initial period (+0.7 kg dry matter m⁻² as years 1–6 mean value). The biomass water content was affected by harvest time, decreasing by about 10% from autumn to winter. With regard to plant density, higher dry matter yields were achieved with 20,000 plants per ha (+0.3 kg dry matter m⁻² as years 1–6 mean value).

The total energy input decreased from fertilised (18 GJ ha⁻¹) to not fertilised crops (4 GJ ha⁻¹). The higher energetic input was represented by fertilisation which involved 14 GJ ha⁻¹ (fertilisers plus their distribution) of total energy costs. This value represents 78% of total energy inputs for fertilised crops.

Giant reed biomass calorific mean value (i.e., the calorific value obtained from combustion of biomass sample in an adiabatic system) was about 17 MJ kg⁻¹ dry matter and it was not affected by fertilisation, or by plant density or harvest time. Fertilisation enhanced crop biomass yield from 23 to 27 dry tonnes per ha (years 1–6 mean value). This 15% increase was possible with an energy consumption of 70% of the overall energy cost. Maximum energy yield output was 496 GJ ha⁻¹, obtained with 20,000 plants per ha and fertilisation. From the establishment period to 2nd–6th year of growth the energy production efficiency (as ratio between energy output and energy input per ha) and the net energy yield (as difference between energy output and energy input per ha) increased due to the low crop dry biomass yield and the high energy costs for crop planting. The energy production efficiency and net energy yield were also affected by fertilisation and plant density. In the mature crop the energy efficiency was highest without fertilisation both with 20,000 (131 GJ ha⁻¹) and 40,000 plants per ha (119 GJ ha⁻¹).     

大连市种植业主要作物经济效益分析

本研究旨在及时掌握大连市种植业生产现状、布局特点,指导农民进行高效、高产、高质的种植业生产。论文依托大连农业经济运行动态监测信息系统覆盖大连市6个区县的动态监测网络,采集2008年大连市种植业760个农户全作物种植的投入产出构成信息等方面的定向动态数据作为市种植业经济运行的样本。结果表明：苹果和樱桃具有一定的规模优势,蔬菜并不具备规模优势,玉米具有一定的规模效益,水稻的规模效应不明显。     

Piezoelectric Control of Lifting Surface Flutter

Piezoelectric material was used to establish a single input/single output (SISO) model of a lifting surface with distributed piezoelectric actuators and piezoelectric sensors. The flutter equations and state space equations of piezoelectric laminated structures were also derived. MATLAB, NASTRAN and other software programs compiled by the authors also were used to analyze the flutter and dynamic characteristics. The stability of the closed loop was analyzed for the SISO. The result indicates that the control law can be employed for flutter suppression.     

Hierarchical control method of low voltage looped microgrid

The low voltage (LV) looped microgrid has high reliability of power supply and good accessibility of the DG. The medium voltage distribution network with single loop network and multi sectioned & multi linked is studied to analyze the wiring mode of the LV looped microgrid. A research on the two hierarchical control scheme of the looped microgrid is conducted, and the sub microgrid, which meets island operation requirements, is divided according to their segments. Then the incidence matrix is formed based on the connection relation of different segments and components, and the plan for switching the microgrid operation status is concluded from the reform of incidence matrix on the basis of the judgment of island and reconnection segments. A resynchronization strategy using microgrid central controller is proposed. The MATLAB/SIMULINK simulation software is used to analyze the dynamic characteristics of different operation status switching and synchronizing, and the results verify the effectiveness and correctness of the proposed control method.     

Using BP Neural Network to Determine the Membership Function of the Input of Fuzzy Control System

While design the fuzzy controller, it is very important to determine the membership function of fuzzy variables.The data can be broadly classified as fuzzy sets by using the classification property of the BP neural network. The author selects a BP neural network with one hide layer and uses S function to the input and hide layer, and linear function to the output layer.Advanced BP algorithm isused to train the BP neural network in the environment of MATLAB . The nearer to the target values is the better the last output is.With the trained BP network , the membership values of the inputs can be got ten. This method has high rate and low error.