融合体尺信息的封闭式种猪性能测定站设计与试验

针对现有封闭式种猪性能测定站自动化程度不高、无法提供种猪体尺信息等问题,设计了一种集种猪自动识别、体质量自动称量、采食量自动统计、体尺自动测量于一体的封闭式种猪性能测定站。该系统机械部分采用前后端分离设计,通过设计采食门装置和门禁装置为种猪提供封闭测量环境,在此基础上,基于FIR滤波设计了种猪体质量动态称量算法,基于椭圆拟合设计了种猪理想姿态筛选算法,并进一步提出了基于包络分析的种猪体尺测量算法。分别进行了利用实际猪群模拟种猪生长性能验证试验和体尺测量试验,试验验证结果如下：生长性能试验猪群自由采食日均次数8.94次、日均采食时间92.93 min、群体料肉比2.66,Logistic拟合的生长曲线拐点日龄为126.18 d、拐点体质量72.70 kg,符合猪群的生长规律;体尺测量试验中猪群能够筛选出理想姿态帧,体长、体宽、臀宽、体高、臀高等体尺的平均相对检测误差分别为3.69%、2.53%、2.60%、2.59%、2.17%,满足体尺测量要求。试验结果表明,本文设计的封闭式种猪性能测定站可用于种猪的生产性能测定,能够同时提供种猪体质量、采食量和体尺等信息,提高育种效率。     

基于双目视觉技术的猪生长监测系统标定模

针对养猪生产中对猪体生长监测的需求,设计了基于双目视觉的猪生长监测系统的软、硬件,实现了基于非线性摄像机模型双目视觉系统的标定算法.根据摄像机成像原理建立了基于最小二乘法的空间点坐标检测算法.利用标定和检测算法,从标定板图像数目、标定板位置、旋转角度3个方面对系统的标定模式进行了研究.结果表明:利用19幅以上标定板图像能够得到稳定的标定结果;不同位置的标定板图像对检测精度影响较大,应当在全视场内采集标定板图像;标定板的旋转角度对检测精度影响不明显,但是旋转角度增大不利于标定点的完全提取.     

基于计算机视觉技术育肥猪体重分析研究

为了更好地解决育肥猪的体重预估问题,本研究通过获取育肥猪在不同生长阶段的图像和质量数据,利用计算机视觉技术将猪的侧视图像进行预处理、颜色特征处理、阈值分割及图像形态学处理,经过推导计算求出猪体的侧视面积,对一维体尺参数、侧视面积与体重进行数据拟合并建立数学模型。研究结果表明:在只考虑体尺单因素的影响时,拟合出的体重与体尺的相关性较小,其平均误差也较大。通过比较逐步回归法与MLP神经网络模型发现:MLP神经网络拟合模型相关性最好,相关性R2可达到0.993,平均相对误差为1.38%,可以很好地保证估测精度,为测量猪的体重提供新的方法。     

金华猪精准养殖与管理技术

在搜集、分析以家猪为代表的畜禽生长信息采集、畜禽图像处理的应用研究现状后,结合现代养猪工艺和金华猪生长习性,指出金华猪养殖产业迫切需要引入猪只体尺测量、估测猪体质量、智能饲喂设备（代谢笼）、数字化管理4个方面的精准养殖技术及其配套的智能化养殖设施,以期实现金华猪的现代化养殖与管理。      

无人机在大豆植保作业中的适用性评价体系研究

本论文通过研究和分析构建大豆植保无人机的适用性评价体系,提出了评价大豆植保无人机适用性的3个指标:匀速飞行速度波动、喷雾沉降量、喷雾均匀性.并提出了各性能指标的量化方法.通过运用德尔菲法和层次分析法相结合构成的综合权重分析法以及yaAHP权重分析软件对大豆植保无人机适用性评价的3个性能指标进行了权重分析,研究结果显示对...     

基于机器人的生猪健康养殖智能监控系统设计

为提高生猪养殖环境质量和行为识别率,克服规模化养殖过程中采用人工巡检生猪异常行为特征和监测养殖环境参数存在识别效率低、准确性差和劳动强度大等问题,设计一种以轨道式机器人为采集终端的生猪健康养殖智能监控系统。系统以FPGA控制器为硬件核心,结合猪只体表特征与环境感知传感器、智能控制、图像处理和无线通信技术,构建生猪异常行为与异常环境全景和局部两级联动监测平台,实现对生猪异常行为和环境参数的全方位观察、识别和数据采集功能,提高生猪养殖过程的智能化管理水平。试验结果表明,该系统能够按照预设指令自动、快速地依次对选取猪舍猪只个体的生长情况、行为特征和养殖环境信息进行自动采集,异常行为识别率可达93.5%,停车定位精度误差为±12 mm。该研究有利于技术人员快速、准确获取生猪生长环境和健康状况信息,为生猪异常环境及时调控、异常行为快速诊断、疫病防治和疫情预警提供科学依据。     

Energy and carbon inventory of Iowa swine production facilities

This study evaluates energy and carbon use by two types of facilities—conventional confinement and hoop barn-based—within farrow-to-finish pig production systems scaled to produce 5200 and 15,600 market pigs annually in Iowa. The United States is the world’s second largest producer of pork with pig production centered in the state of Iowa. Conventional confinement facilities are typical of pork industry practice in the United States and are characterized by individual gestation stalls and 1200 head grow-finish buildings with slatted concrete floors and liquid manure systems. The hoop barn-based alternative uses group pens in bedded hoop barns for gestation and finishing. Both systems use climate controlled farrowing facilities with individual farrowing crates as well as climate controlled nursery facilities. Feed is the single largest operating resource in pig production systems and feed fed to grow-finish pigs accounts for 63-65% of total energy use in raising pigs. The other stages of production are more reliant on non-renewable fuels and ignoring these stages of production misses 54-80% of the non-renewable fuel use associated with pig production. Taking into account demonstrated performance differences, hoop barn-based pig production requires 2.4% more feed and similar total energy as conventional pig production. Hoop barn-based pig production requires 63-64% less non-renewable fuel and results in 35% less emissions. There is little (<0.3%) energetic advantage to increase the scale of pig production from 5200 to 15,600 market pigs annually. Excluding the gross energy of feedstuffs fed to pigs, producing pigs in Iowa requires 7.2-8.2 MJ/kg live weight and results in emission of 1.0-1.6 kg CO₂ equivalents/kg live weight. This compares favorably with published energy assessments of pig production for European systems. Using hoop barns for grow-finish pigs and gestating sows is an effective strategy to reduce direct use of fossil fuels for pig production and may minimize global climate altering emissions.     

Calibration and sensitivity analysis of a model of the growing pig for weight gain and composition

The model concerned was a mechanistic model for pig growth and composition which was constructed using novel algorithms for the accumulation of protein and lipid. The model was parameterised using data collected from a serial slaughter trial of three different types of pigs and the sensitivity to parameters of biological importance was explored. The types (‘Landrace’ type, ‘Pietrain’ type and ‘Meishan’ type) were chosen to represent ‘lean’, ‘meaty’ and ‘fatty’ types.After optimisation, fitted parameters were found to lie close to the values that would be expected given the assumptions made during the construction of the model, with a predicted maximum protein retention rate of 0.20 kg d⁻¹ and an efficiency of recapture of amino acids during turnover of 0.94. Mean absolute percentage errors at slaughter point for modelled live weight were ca. 5%, compared with 8% for protein mass and 13% for lipid mass, suggesting that live weight gain is easier to predict than the partitioning of biomass into retained protein and lipid.     

陕西关中平原地区适栽羊肚菌菌株筛选试验

为明确陕西关中平原地区地区适栽羊肚菌菌株,本试验收集了10株羊肚菌菌株,从羊肚菌子实体农艺性状、产量及主要营养成分三个方面进行分析,以期筛选出适于本地区栽培优良菌株。结果表明：（1）10个不同菌株栽培后所得的羊肚菌子实体产量差异性显著,菌株 401和菌株 8号的鲜重和干重较其他品种具有明显优势;（2）子实体农艺性状方面有差异,从个头、重量和商品性三个方面分析,菌株8号、菌株 401 较其它菌株有显著优势;（3）羊肚菌子实体主要营养成分含量有差别,10个羊肚菌品种的粗纤维、多糖含量和总氨基酸较高,而脂肪含量较低。综上所述,菌株401和8号两个品种相对适宜在陕西关中平原地区进行栽培。     

基于AHP的热态除鳞系统往复式柱塞泵装置性能模糊评价方法

介绍了典型高压水热态除鳞系统的构成及工作原理,以除鳞系统核心装备的往复式柱塞泵装置在系统中的综合运行性能为研究对象,从装置的节能、可靠性和安全性3方面出发,以评价往复式柱塞泵装置综合运行表现力为目标,构建了完整的评价指标体系,以层次分析法(AHP)和模糊评价为核心,建立了评价指标体系评价要素权重分析方法,开展了评价指标体系评价要素实测物理参数模糊隶属度分析方法研究,在此基础上构造了基于AHP的高压水热态除鳞系统往复式柱塞泵装置综合运行性能模糊评价数学模型,形成了判定热态除鳞系统往复式柱塞泵装置综合运行力的评价方法,并通过实例分析了高压水热态除鳞系统往复式柱塞泵装置综合运行表现力评价指标体系评价要素的权重,以及评价要素实测物理参数模糊隶属度,对往复式柱塞泵装置综合运行表现力结果进行了分析,验证了该评价方法的可行性.     

Incorporation of environment and feed quality into a net energy model for beef cattle

The California net energy system has gained wide acceptance as a method for predicting rates of gain in growing and finishing beef animals. Researchers at the University of Kentucky, in co-operation with agricultural scientists in two regional research projects, NC-114 and S-156, have developed an interactive model that enhances the usefulness of the net energy system by including skeletal size, feed quality, temperature and relative humidity to determine feed intake and utilisation. Basically, the intake based on feed availability, nutrient composition, potential skeletal size and heat loss potential. After the feed is consumed, it is utilised according to the productive status of the animal, i.e. growth, lactation, foetal development, etc., using a modified form of the net energy system to predict weight change.The model allows a researcher to easily modify environment, feed, breed characteristics and productive status so as to evaluate their influence on the performance of the animal.     

Profit in a pig herd from performance testing and selection of breeding stock. The effect of variation in the accuracy and cost of testing

Computer modelling was used to compare a variety of performance testing and selection programmes whose objective was to increase the monetary value of pigs as determined by their rate and efficiency of growth and the leanness of their carcases.Factors varied over feasible ranges were the cost and accuracy of performance testing, the culling rates of breeding stock and the degree of subdivision of the herd into a nucleus supplying breeder replacements for the whole herd and a production unit producing slaughter stock.Net returns per sow, evaluated over 10 years of selection, increased, by 120 times, the standard deviation (sd) of the breeding objective for each improvement of 0·1 in the correlation between the selection criterion and the breeding objective (test accuracy).A unit of the cost of testing each pig equalled 0·4 sd of the breeding objective and net returns in the unsubdivided herd declined by 0·6 for each unit increase in cost. Herd subdivision increased net returns by reducing the number of pigs tested—and hence the cost of testing. The optimum ratios of production unit to breeding nucleus sizes which maximised net returns were found. When herds were subdivided in an optimum way, the decline in net returns with increasing testing costs was reduced sixfold.Returns were highest when boars were worked for only one breeding cycle (approximately 6 months) in both the nucleus and the production unit. Depending on the accuracy and cost of testing, maximum net returns occurred when nucleus sows were culled after one or two farrowings. Sows in the production unit were an optimum combination of culled ‘old’ sows and selected ‘young’ gilts from the nucleus. There were a number of other replacement schemes which were almost equally profitable.     

设施猪场生猪体温红外巡检系统设计与试验

针对规模化生猪养殖过程中难以及时获取猪只体温的问题,设计了基于红外技术的生猪体温巡检系统,以实现对设施猪场生猪体温的快速监测。硬件包括红外移动采集装置、系统控制装置和电源3部分。其中红外移动采集装置包括滑轨、移动小车、红外热像仪和保护壳;系统控制装置包括惠普MINI微型计算机、XP2-18R/RT型整体式控制器和DM542型数字式步进电机驱动器;电源包括电缆线和电缆滑车。控制系统集成红外图像的采集、存储控制,移动小车的运行控制及与远程服务器的数据通信等功能。在限位栏猪舍安装该装备,并进行了28 d的测试试验,试验表明:系统工作稳定,巡检过程中红外图像平均缺失率为1. 12%,能有效采集限位栏生猪红外图像。试验选择耳根区域作为生猪体温敏感区域,通过统计得到猪只耳根区域周期性巡检精度在90%以上,能有效监测猪只耳根区域温度情况。试验连续5 d对3～6号猪只耳根区域温度最大值进行监测,并统计其日均值,通过分析验证了温度数值的科学性,可为养殖人员提供有效的监测信息,对生猪的健康监测和疫情防治具有重要意义。     

基于模糊控制的畜禽舍环境温湿度监控系统

畜禽舍环境系统是一个多变量、非线性、时变和滞后的系统,各变量之间具有耦合关系,很难建立精确的数学模型.为此,利用模糊控制算法对畜禽舍环境温湿度状况进行实时监控,为商品猪提供最佳生长环境,缩短了生长周期,实现高效节能的工厂化生产,具有较好的实用价值和应用前景.     

基于熵权TOPSIS的农民合作社农超对接绩效评价

根据阳泉地区农民专业合作社走访调查的资料，参考平衡计分卡选取合作社农超对接绩效的评价指标，并用熵权法确定指标权重，结合TOPSIS模型对合作社参与农超对接的绩效进行评价，并用障碍度模型测算影响农超对接绩效水平的障碍因子。结果表明：阳泉市合作社参与农超对接绩效水平较低且提升空间巨大，销售培训、销售增长率以及技术培训市制约农超对接绩效的主要障碍因子。     

仔猪阶段对维生素的营养需要

仔猪阶段的营养是猪的所有成长阶段中,最复杂最难以掌握的一个阶段.若是维生素营养供给不合理,会直接造成仔猪生长缓慢,并出现腹泻,死亡率极高,最终导致中大猪阶段的生长慢,出栏时间大大延长.文章对仔猪阶段对维生素的营养需要做了探讨.     

基于双流跨模态特征融合模型的群养生猪体质量测定

针对生猪体质量准确测定问题,提出了一种跨模态特征融合模型（Cross modality feature fusion ResNet,CFF-ResNet）,充分利用可见光图像的纹理轮廓信息与深度图像的空间结构信息的互补性,实现了群养环境中无接触的生猪体质量智能测定。首先,采集并配准俯视猪圈的可见光与深度图像,并通过EdgeFlow算法对每一只目标生猪个体进行由粗到细的像素级分割。然后,基于ResNet50网络构建双流架构模型,通过内部插入门控形成双向连接,有效地结合可见光流和深度流的特征,实现跨模态特征融合。最后,双流分别回归出生猪体质量预估值,通过均值合并得到最终的体质量测定值。在试验中,以某种公猪场群养生猪为数据采集对象,构建了拥有9842对配准可见光和深度图像的数据集,包括6909对训练数据和2933对测试数据。本研究所提出模型在测试集上的平均绝对误差为3.019kg,平均准确率为96.132%。与基于可见光和基于深度的单模态基准模型相比,该模型体质量测定精度更高,其在平均绝对误差上分别减少18.095%和12.569%。同时,该模型体质量测定精度优于其他现有生猪体质量测定方法：常规图像处理模型、改进EfficientNetV2模型、改进DenseNet201模型和BotNet+DBRB+PFC模型,在平均绝对误差上分别减少46.272%、14.403%、8.847%和11.414%。试验结果表明,该测定模型能够有效学习跨模态的特征,满足了生猪体质量测定的高精度要求,为群养环境中生猪体质量测定提供了技术支撑。     

鄂尔多斯市杂交优良沙棘无性系的丰产性和抗旱性

观察了8个杂交优良沙棘无性系反映丰产性的单枝果实密度、百果重,反映抗旱性的落果率、果生长恢复率,反映丰产性和抗旱性综合表现的单株果产量,比较和评价了8个杂优沙棘丰产性和抗旱性的差异,给出了防落果,育大果,促高产措施的建议。     

Ranking and classification of irrigation system performance using fuzzy set theory: case studies in Australia and China

A methodology for ranking and classifying performance of irrigation systems through multidimensional performance indicators is developed using fuzzy set theory. The procedure uses the concepts of fuzzy resemblance and fuzzy dominance. Preference levels reflecting management priorities are incorporated into the analysis using appropriate weighting factors. The application of the procedure is demonstrated through two case studies: the Shi-Jin irrigation district in Hebei Province, China and the Goulburn irrigation region in Victoria, Australia. The classification of performance periods for the Australian system shows three clusters indicating the predominant effect of the increase in waterlogged area when higher priority is given to this indicator. No clear trend appeared when equal weight was assigned to all indicators. The effect of economic reforms implemented in 1977–78 in China is clearly reflected in the resulting ranking and clustering of the performance periods in the Shi-Jin irrigation district. Performance levels following reforms are shown to be consistently higher.Units mu unit of area equal to 1/15 ha (666 m²) - jin unit of weight equal to 0.5 kg.     

育肥猪精准下料控制系统设计与验证

为满足育肥猪获得正常生长所需采食量并达到精准饲喂控制等需求，以育肥猪为试验对象，设计一种育肥猪精准下料控制系统。研究通过使用双侧下料器，实现对称式双边下料，提高猪群的采食效率；增设触碰开关，猪只触动后执行预设料下料，避免饲料浪费；食槽内增设料位感应探针，猪只触碰感应开关时，若探针感应不到食槽内水料，系统自动补充水料到探针处，既保证猪只正常采食，又避免饲料浪费；增设防结拱装置，通过震动电机实现破拱，防止料粉受潮影响下料。试验结果表明，与理论采食量比较，智能饲喂方式下采食量低于理论采食量；饲喂采食量比例分别为10%，30%，30%，20%及10%时，获得较佳的采食量变化曲线。综上，所设计的双侧下料的育肥猪精准下料控制系统，采用拨轮下料的控制机构与嵌入式系统的协同工作，设备控制简单，下料稳定，计量准确，与进口设备比较，成本优势明显，适合在我国大、中、小型育肥猪场推广应用。