Quantitative Detection Model of Pernicious Gases in Pig House Based on BP Neural Network

To find a neural network model suitable to identify the concentration of mixed pernicious gases in pig house, the quantitative detection model of pernicious gases in pig house was set up based on BP （ Back propagation） neural network. The BP neural network was trained separately by the three functions, trainbr, traingdm and trainlm, in order to identify the concentration of mixed pernicious gases composed of ammonia gas and hepatic gas. The neural network toolbox in MATLAB software was used to simulate the detection. The results showed that the neural network trained by trainbr function has high average identification accuracy and faster detection speed, and it is also insensitive to noise; therefore, it is suitable to identify the concentration of pemidous gases in pig house. These data provide a reference for intelligent monitoring of pemicious gases in pigsty.

Numerical Simulation of Natural Ventilation in Northern Civil Pig Houses

With northern pig house as study objective,the 3-D numerical simulation of natural ventilation airflow field of five pig houses with different window area was done by using Fluent software,and the comparative analysis was carried out on simulation result and measured result.The results showed that the effect of ventilation was decided by the reasonable area ratio of window and earth and the ratio of length and width of window.According to this real pig house in experiment,there are the best uniformity of ventilation and the least area of having no ventilation by 1：10 of the area ratio of the window and earth in the pig houses,the smaller air retention zone by the smaller ratio of length and width of window,the better effect of ventilation;the results of simulation agree well with the results of measured,the max relative error and the average error was 11.1% and 7.7%,so the software of Fluent was proved effective in the numerical simulation of natural ventilation in the pig houses and the optimal design of pig houses＇ structure.

Design of Environmental Monitoring and Control System for Large-scale Pig House with Fermentation Bed

The design and assembly of environmental monitoring and control system for large-scale pig house with fermentation bed helped to solve the problem of environmental automatic control in piggery.The sensors would monitor the temperature,humidity,light,wind direction,wind speed,CO2,NH3and other parameters.On-line real-time data collection was achieved.The expert system was constructed to control the temperature in piggery below 30℃,to control the air and mattress humidities higher than 65%.Under the conditions of different season or different wind speed,even in day and night,the control actuators were different.The actuators included fanning wet curtain,lighting,micro spraying,spraying,propeller fan,electric aluminum alloy shutter and spraying systems on the roof.The actuators were integrated,and they control the piggery environment simultaneously.The system also designed the remote video monitor interface,parameter-monitoring curved interface and operation interface,which provided a good man-machine interface.

Characteristics of Microbial Aerosol Pollution in Pig Houses

[ Objective] To explore the characteristics of microbial aerosol pollution in different pig houses and provide theoretical foundation for risk assessment of microbial aerosols on human and animal health. [Method] Gestation house, nursery house and fattening house in suburb of Beijing were chosen as research objects, and the concentration, size distribution and composition characteristics of the airborne microorganisms were determined. [ Result] The concentrations of airborne microorganisms were completely different in different houses under the influences of ventilation, methods of cleaning manure and pig populations. The highest concentration of airborne microrganisms was the nursery house. The particles with diameter size of 0.65 -2.10 μm carried 26% -27% heterotrophic bacteria and 39% -43% fungi. They could reach alveolus of humans and animals and thus made infectious threat to health of humans and animals. The dominant heterotrophic bacteria genera were composed of Bacil/us and Pseudomonas; and the dominant fungi were Penicillium and Muco. [ Conclusion] Concentrations of airborne microorganisms are completely different in different houses. However. there is no sianificant difference between the microbial composition and tyee of oia houses.

Study on Multi-variables Decoupled Fuzzy Controller for Confined Pig House in Northern China

In winter, the confined pig house of northern China is severe. The environment variables are nonlinear, time-varying and coupled, which seriously affect the health of pigs and the qualities of the meat. In order to solve the problem multi-variables coupling, a multi-variables decoupled fuzzy logic control method was proposed. Two fuzzy logic controllers were designed based on fuzzy logic theory. The fans, heaters and humidifiers were used to control temperature, humidity and ammonia. The reductions of temperature and humidity caused by ventilating were compensated by heaters and humidifiers respectively which realized the multivariables decoupling. The proposed methods were validated through theoretical, experimental and simulation analysis. The results suggested that the methods were able to regulate the confined pig house environment effectively. In addition, comparing to the manual regulation, the proposed methods could reduce 19% power consumption as well.

基于PigsTrack跟踪器的群养生猪多目标跟踪

基于视频的生猪行为跟踪和识别对于实现精细化养殖具有重要价值。为了应对群养生猪多目标跟踪任务中由猪只外观相似、遮挡交互等因素带来的挑战,研究提出了基于PigsTrack跟踪器的群养生猪多目标跟踪方法。PigsTrack跟踪器利用高性能YOLOX网络降低目标误检与漏检率,采用Transformer模型获取具有良好区分特性的目标外观特征;基于OC-SORT(observation-centric sort)的思想,通过集成特征匹配、IoU匹配和遮挡恢复匹配策略实现群养生猪的准确跟踪。基于PBVD(pigs behaviours video dataset)数据集的试验结果表明,PigsTrack跟踪器的HOTA(higher order tracking accuracy),MOTA(multiple object tracking accuracy)和IDF1得分(identification F1 score)分别为85.66%、98.59%和99.57%,相较于现有算法的最高精度,分别提高了3.71、0.03和2.05个百分点,证明了PigsTrack跟踪器在解决外观相似和遮挡交互引起的跟踪过程中身份跳变问题方面的有效性。随后,利用Slowfast网络对PigsTrack跟踪器的跟踪结果进行了典型行为统计,结果显示PigsTrack在群养生猪个体行为统计方面更准确。此外,通过在ABVD(aggressive-behavior video)数据集上的试验,PigsTrack跟踪器的HOTA、MOTA和IDF1得分分别为69.14%、94.82%和90.11%,相对于现有算法的最高精度,提高了5.33、0.57和8.60个百分点,验证了PigsTrack跟踪器在群养生猪跟踪任务中的有效性。总而言之,PigsTrack跟踪器能够有效应对外观相似和遮挡交互等挑战,实现了准确的生猪多目标跟踪,并在行为统计方面展现出更高的准确性,为生猪养殖领域的研究和实际应用提供了有价值的指导。

The Influence of Displacement to the Success of Sustainable Multi-Storey Housing Development for Low-Income Society in Urban Area： Case Study- Multi-Storey Housing in Jakarta

Multi-storey housing for low income society is a concept of redevelopment from many alternatives of housing development for slum in Jakarta. However, redevelopment directly or indirectly leads to displacement of low-income residents. The new residents would be of a higher socio-economic status than the previous occupants. For controlling this phenomenon, a study about the influence of displacement to the successful of sustainable multi-storey housing development for low-income society has been done with descriptive method, observation and literature study. The result of this study give knowledge that displacement always happened when the quality of building and environment is increase, the location of multi-storey housing is near city center and commercial area. Low income society voluntary displaced their units to middle or high income society, because they are not affordable for paying operational and maintenance costs. The government has to make regulations for controlling displacement on multi-storey housing for low-income society and for the successful of sustainable multi-storey housing development in urban area.

严寒地区混合垂直通风猪舍环境质量评价

为探究垂直通风系统(屋顶与粪坑混合排风)在寒区猪舍中的适应性,该研究在秋冬季节对内蒙古某规模化猪舍的热湿与空气质量环境进行了实时连续监测,并根据测试结果对猪舍环境质量进行评价。结果表明:1)测试期间,舍内平均温度、相对湿度以及氨气(NH_(3))、二氧化碳(CO_(2))、总悬浮颗粒物(total suspended particulate, TSP)浓度分别为22.7℃、59.6%、6.5 mg/m^(3)、2 200 mg/m^(3)、2 915μg/m^(3),舍内通风量为0.34~0.39 m^(3)/(h·kg),各环境指标基本满足生产环境阈值要求;保育期间舍内有效环境温度范围为27.1~30.7℃,存在轻度热应激情况,占总监测时间的12.9%;2)舍内温度、相对湿度与NH_(3)、CO_(2)、TSP浓度空间分布不均匀系数分别为0.01、0.15、0.45、0.03、0.15,各环境参数空间分布均较为均匀;3)试验单元(U1、U3)猪的平均日增质量、料重比与死淘率分别842和818 g/d、2.68和2.79、1.68%和1.31%,生产性能良好。综上,不设吊顶的侧墙进风,屋顶与粪坑混合排风的垂直通风系统,总体上能够满足严寒地区猪舍秋冬季热湿环境与空气质量调节需求,保证良好的生产性能。

漏缝地板下粪水贮存周期内猪舍空气环境的监测

[目的]研究漏缝地板下粪水贮存周期内猪舍空气环境,为猪健康生长和生产提供参考。[方法]选取1栋分娩猪舍,测定粪水贮存周期(21 d)内猪舍的空气环境指标。沿猪舍对角线方向均匀选5个测定点,粪水贮存周期内每3 d测1次,每次分别在6:00、11:00、13:00、18:00测定。[结果]猪舍粪水贮存周期内平均温度(26.08±1.45)℃,相对湿度(93.44±3.96)%,氨气质量浓度为(3.58±1.62)mg/m^(3),二氧化碳质量浓度为(1760.63±284.98)mg/m^(3)。第18天的温度、相对湿度、二氧化碳浓度最高,各指标基本与其他测定时间差异显著;第21天氨气浓度最高,与第3天、第6天差异显著;[结论]漏缝地板下粪水贮存周期内,温度略高,相对湿度过高,氨气浓度持续上升但未超过标准,二氧化碳浓度较高,应增加通风设备,改善猪舍空气环境。

生猪生长环境舒适度实时监测系统设计与实现

为了提高猪舍环境舒适度评价的准确性,克服猪舍多环境因子相互耦合,避免采用单一环境因子评价猪舍内环境质量状况不准确问题。本研究根据猪舍环境养殖标准,构建了猪舍舒适度评价指标体系和权重,建立了温度、湿度、风速、CO_(2)、NH_(3)和H_(2)S各环境因子的隶属度函数,提出了基于层次分析-模糊理论的猪舍多环境因子综合评价方法,搭建了生猪生长环境舒适度实时监测系统。以育肥猪舍环境监测的24组数据为例对本文提出的猪舍舒适度模糊综合评价方法进行验证。结果表明,该系统建立的猪舍多环境因子舒适度模糊综合评价方法能准确反映猪舍空气质量状况,可为猪舍环境精准调控提供科学决策依据,具有一定的应用价值。

基于CFD与TRNSYS的圆形猪舍内环境控制研究

为了给寒冷地区密闭圆形猪舍提供内环境控制策略,试验以地处寒冷地区的张家口市某在建密闭圆形猪舍为研究对象,通过ANSYS FLUENT软件搭建计算流体力学(computational fluid dynamics,CFD)仿真模型,利用CFD模型计算猪舍在最低通风量标准条件下舍内污染气体质量浓度与风速分布,研究通风量对舍内污染气体质量浓度的影响;利用TRNSYS 17软件搭建猪舍温控仿真模型,以月舍温、月舒适时间占比和月暖通不保证率为评价指标,研究非供热策略、供热策略和调整养殖密度策略下猪舍全年通风量对舍内环境温度与舒适性的影响,筛选较为合适的猪舍全年温控策略,并对最适温控策略进行优化研究。结果表明:在圆形猪舍内每60°布置一组送、排风侧立管的“上送侧排”通风方案中,污染气体浓度随通风量增大而降低;在全年最低通风量[0.35 m~3/(h·kg)]标准条件下,养殖区域猪只背部高度平面的平均风速为0.3 m/s,猪只呼吸高度平面NH_(3)、H_(2)S、CO_(2)质量浓度最大值分别为1.02,0.021,525 mg/m~3,上述参数均满足猪舍养殖环境要求,即通风量、污染物浓度、猪只吹风感之间得到较好平衡。调整养殖密度策略比供热策略的全年平均月舒适时间占比高1.5百分点,全年平均月暖通不保证率低0.2百分点;调整养殖密度策略在夏季的平均月舒适时间占比比非供热策略高4.3百分点,月平均舍温低0.3℃,因此育肥猪舍采用调整养殖密度策略调节猪舍温度更为合适。猪舍采用调整养殖密度策略调节猪舍温度时,将5,9月份的通风量从0.50 m~3/(h·kg)提升至0.68 m~3/(h·kg),使这两个月份的月平均舍温均下降2.3℃,月暖通不保证率分别从1.6%和7.2%下降至0.9%和3.6%,月舒适时间占比分别从35.9%和26.1%提升至70.4%和59.4%;夏季采用遮光板辅助降温措施可使午间猪舍温度降低0.3~1.2℃,使猪舍温度更加舒适。说明应用新风全热交换器和湿帘-风机系统并适当调整通风量或养殖密度可以达到密闭圆形猪舍的环境控制要求。

屋顶光伏—土壤源热泵系统在不同气候区分娩猪舍的应用

可再生能源逐步替代常规能源,是规模化猪场未来的趋势。为了探索土壤源热泵系统(ground-coupled heat pump systems,GCHPs)和附加光伏屋顶(building attached photovoltaic, BAPV)系统在规模化猪场的应用,该研究提出一种屋顶光伏—土壤源热泵系统(building attached photovoltaic-ground-coupled heat pump systems,BAPV-GCHPs)保障某分娩猪舍的冷/热需求,并选择绥化、青岛和重庆分别作为严寒、寒冷和夏热冬冷地区3类典型气候区代表城市,对比了该系统在3类气候区的运行特性,分析了BAPV系统对GCHPs运行性能及综合效益的影响。研究结果表明,3个地区BAPV发电量由大到小依次为:青岛、绥化、重庆;GCHPs耗电量由大到小依次为:绥化、青岛、重庆;绥化、青岛、重庆3个地区的年太阳能分数分别为0.62、0.71、0.53;BAPV系统的加入明显提高了GCHPs的性能系数、一次能源利用率和CO_(2)减排量;3个地区BAPV-GCHPs相对于GCHPs年性能系数分别提高了64.2%、97.6%和39.6%;一次能源利用率分别提升了1.6倍、2.4倍和1.1倍。与GCHPs相比,BAPV-GCHPs可分别减少5.82、6.45和2.17 t CO_(2)排放量。研究结果为该系统在中国不同气候区分娩猪舍的推广应用提供了一定参考。

猪舍环境的多参数无模型自适应控制算法设计

【目的】针对猪舍环境控制中场景复杂、多参数控制难以及系统呈现耦合特性的问题,基于无模型自适应控制(Model-free adaptive control,MFAC)算法,设计了适用于猪舍的多参数无模型自适应控制(Multi-parameter model-free adaptive control,MMFAC)算法。【方法】通过将MFAC算法与用于尺度伸缩变换的权值矩阵相融合,设计MMFAC算法。该算法利用紧格式动态线性化(Compact form dynamic linearization,CFDL)技术和最优化数学方法,进行伪雅可比矩阵(Pseudo-jacobian matrix,PJM)的参数辨识和调控装置的控制量计算。【结果】仿真试验结果表明,MMFAC算法具备对多参数进行自适应控制的能力,并可通过调整权重矩阵来降低关键参数的误差。真实猪舍环境控制试验结果表明,该算法可根据实时环境参数测量值计算风机控制量,并根据该控制量驱动风机,以保持环境参数在舒适区间范围内的稳定控制。【结论】MMFAC算法展现出良好的多参数自适应控制能力,能够适应复杂的猪舍环境,有效解决多参数控制难以及系统耦合的问题,在猪舍环境控制中具有较好的应用前景。

基于物联网的猪舍环境监测系统

为了实时掌握猪只生长环境的情况,满足猪舍环境的监测需求,同时降低系统的成本,设计了一款基于物联网的猪舍环境实时监测系统。系统选用STM32F103C8T6单片机作为主控器,应用空气温度、空气湿度、光照强度、CO_(2)浓度、NH_(3)浓度等传感器对猪舍环境进行实时检测,并根据检测结果对异常情况进行报警或控制。系统选用ESP8266 Wi-Fi模块通过MQTT协议连接OneNET物联网平台,实现数据的远程传输,便于数据的可视化显示和管理。研究中根据物联网架构模式搭建了系统,对系统所需硬件进行了分析,经核算总计成本在150元左右,并对网络、功耗和数据准确性进行了测试,结果表明系统总体稳定可靠。该系统集多传感器于一体,可以同时监测多种环境因子,且成本低、简捷易用,能够为猪只提供较好的生长环境,能够帮助养殖人员进行科学化管理,从而提高产能和收益。

基于注意力机制与自适应特征融合的群养猪身份识别

针对群养猪攻击过程中身体形变和重叠影响猪身份识别精度的问题,开发一种基于注意力机制和自适应特征融合的深度学习方法以提高猪身份识别精度。录制猪栏中8只猪每天8小时共3天的视频,并筛选含攻击的16830帧作为数据集。首先,采用ResNet50提取猪的卷积神经网络(CNN)特征;然后,采用特征金字塔网络(FPN)在ResNet50中选择3层不同尺度的特征,以优化这些特征的定位和语义信息;接着,采用自注意力机制提高这些特征的区分度,并采用自适应空间特征融合(ASFF)以融合不同尺度的特征;最后,利用卷积和Sigmoid函数相结合的检测器对猪身份进行识别。使用该方法后,猪身份识别的均值平均精度(mAP)达到95.59%,速度达到37.6 f/s。结果表明,该方法能够在攻击场景下有效识别猪身份,有助于将攻击识别从群体级细化为个体级。

移动式智能猪舍的研发及应用

生猪养殖业由传统的小规模分散养殖模式向规模化养殖模式转变,带来更大的环保压力和占用耕地等问题,也无法满足居民的多样性消费需求。针对目前生猪养殖行业存在的诸多问题,浙江省农业机械研究院依托金华市农业重点项目,联合相关企业,研发了一套移动式智能猪舍系统并投入应用。该移动式智能猪舍由可移动箱体、智能控制系统、中控系统、智能环保系统组成,可通过自动控制模块实现猪舍微环境的实时监测,并根据猪群最佳生长环境实时调整环境参数和饲养参数,给生猪提供适宜其生长发育的舒适环境,避免由于高温高湿引起生猪的热应激性行为。既能提高生产性能、改善猪肉品质,又可减少饲料浪费和猪场环境污染。该科研成果已经在衢州牧子家庭农场有限公司和金华大堰河农牧场推广使用。

猪舍氨气与二氧化碳浓度变化时序预测模型优化

NH3质量浓度和CO_(2)质量浓度是猪舍环境精准控制的重要指标。由于畜禽舍气体浓度具有时变性、非线性耦合等特点,目前有害气体浓度预测模型存在预测精度低的问题。提出了基于门控制循环单元(Gated recurrent unit,GRU)、改进麻雀搜索算法(Improved sparrow search algorithm,ISSA)并融合差分整合移动平均自回归模型(Autoregressive integrated moving average model,ARIMA)的有害气体浓度时序数据预测模型ISSA-GRU-ARIMA。首先构建了GRU气体浓度时序预测模型,然后通过引入Tent混沌序列、混沌扰动和高斯变异增强ISSA算法的局部寻优能力,实现GRU模型超参数优化;然后利用统计学习ARIMA方法提取优化后的ISSA-GRU模型预测残差的线性特征,最终达到提升模型预测精度的目的。以采集的52 d猪舍环境的1248组数据对模型进行训练和测试。结果表明,ISSA-GRU-ARIMA模型NH3质量浓度预测的均方根误差(RMSE)、平均绝对百分比误差(MAPE)和决定系数R2分别为0.263 mg/m^(3)、8.171%和0.928,CO_(2)质量浓度预测的分别为55.361 mg/m^(3)、4.633%和0.985。本文构建的ISSA-GRU-ARIMA模型具有较高的预测精度,可为猪舍有害气体浓度精准控制提供科学依据。

基于改进Cascade Mask R-CNN与协同注意力机制的群猪姿态识别

猪体姿态识别有助于实现猪只健康状况预警、预防猪病爆发,是当前研究热点。针对复杂场景下群猪容易相互遮挡、粘连,姿态识别困难的问题,该研究提出一种实例分割与协同注意力机制相结合的两阶段群猪姿态识别方法。首先,以Cascade Mask R-CNN作为基准网络,结合HrNetV2和FPN模块构建猪体检测与分割模型,解决猪体相互遮挡、粘连等问题,实现复杂环境下群猪图像的高精度检测与分割;在上述提取单只猪基础上,构建了基于协同注意力机制(coordinate attention,CA)的轻量级猪体姿态识别模型(CA-MobileNetV3),实现猪体姿态的精准快速识别。最后,在自标注数据集上的试验结果表明,在猪体分割与检测环节,该研究所提模型与MaskR-CNN、MSR-CNN模型相比,在AP_(0.50)、AP_(0.75)、AP_(0.50:0.95)和AP_(0.5:0.95-large)指标上最多提升了1.3、1.5、6.9和8.8个百分点,表现出最优的分割与检测性能。而在猪体姿态识别环节,所提CA-MobileNetV3模型在跪立、站立、躺卧、坐立4种姿态类上的准确率分别为96.9%、99.1%、99.5%和98.6%,其性能优于主流的MobileNetV3、ResNet50、DenseNet121和VGG16模型,由此可知,该研究模型在复杂环境下群猪姿态识别具有良好的准确性和有效性,为实现猪体姿态的精准快速识别提供方法支撑。

复合微生物制剂对生长猪生长性能、血清生化指标和猪舍有害气体的影响

试验旨在研究复合微生物制剂对生长猪生长性能、血清生化指标和猪舍有害气体的影响。选择60头健康的“杜×长×大”生长猪,随机均分为对照组和试验组,每组设3个重复,每个重复10头猪。对照组猪饲喂基础日粮,试验组在基础日粮中添加0.75%复合微生物制剂。预试期7 d,试验期30 d。结果显示:与对照组相比,试验组猪平均日增重提高10.44%(P<0.05),料重比降低8.07%(P<0.05)。与对照组相比,试验组生长猪血清中碱性磷酸酶活性升高18.91%(P<0.05),尿素氮含量降低12.87%(P<0.05)。试验前两组猪舍氨气(NH_(3))和硫化氢(H_(2)S)浓度基本接近,差异不显著(P>0.05)。试验后,对照组猪舍NH_(3)和H_(2)S浓度比试验前略微升高,试验组猪舍NH_(3)和H_(2)S浓度比试验前明显降低。与对照组相比,试验组猪舍NH_(3)浓度降低31.13%(P<0.05),H_(2)S浓度降低26.04%(P<0.05)。研究表明,在基础日粮中添加复合微生物制剂可以提高生长猪生长性能,改善血液生化指标,降低猪舍有害气体含量。

生物型除臭剂对猪舍氨气浓度的影响

为降低猪舍内氨气浓度,该研究使用生物型除臭剂在山东潍坊市一典型聚落化猪场进行粪沟除臭剂喷洒试验,并连续监测猪舍氨气浓度变化,为猪舍臭气治理问题提供参考。结果表明,生物除臭剂A对猪舍粪沟4个位置的氨气平均去除率分别为71.12%、64.78%、71.44%和64.30%,风机口处氨气平均浓度降低65.59%;效果稳定时间为5~8 d;生物除臭剂B对猪舍粪沟4个位置的氨气平均去除率分别为75.60%、76.90%、74.42%、84.87%,风机口处氨气平均浓度降低77.53%,效果稳定时间为8~11 d。试验表明,粪沟喷洒除臭剂工艺可有效降低粪沟氨气浓度64%以上,减少风机端排出氨气浓度65%以上。