Detection of cows with insemination problems using selected classification models

In the present study, the detection of cows with artificial insemination (AI) difficulties using selected statistical and machine learning methods is presented. Cows were divided into two classes: those that conceived after one or two services (“GOOD”) and those that required more than two services per conception (“POOR”). The best performance was exhibited by one of the artificial neural networks (ANN) and the multivariate adaptive regression spline (MARS) method (AIC, BIC, RMS and accuracy); whereas logistic regression (LR) and classification functions (CF) were of somewhat lower quality. The detection of cows with AI difficulties, performed on the basis of the test set comprising new instances, showed that the ANN and MARS were more precise in comparison with the statistical methods. Sensitivity and specificity were over 85% for the perceptron with two hidden layers (MLP2) and MARS and approximately 80% or lower for LR and CF. From among variables determining the AI category, the average calving interval and cow body condition index were the most important. Other significant variables were lactation number, pregnancy length, sex of calf from previous calving and cow age. The prognoses obtained using ANN and MARS can be used for the appropriate preparation of cows for AI.     

机器学习算法在森林生长收获预估中的应用

森林生长收获预估是森林经理学的一个重要方向,采用模型技术进行森林生长收获估计是森林经营决策的重要前提。传统的统计模型如线性及非线性回归模型、混合效应模型、分位数回归、度量误差模型等统计方法已被广泛应用于研究林木生长,但这些统计方法在应用时常常需满足一定的统计假设前提,诸如数据独立、正态分布和等方差等。由于森林生长数据的连续观测和层次性,上述假设通常难以满足。近年来随着人工智能技术的发展,机器学习算法为森林生长收获预估提供了一种新的手段,它具有对输入数据的分布形式没有假设前提、能够揭示数据中的隐含结构、预测结果好等优点,但在森林生长收获预估中的应用仍十分有限。文章对分类和回归树、多元自适应样条、bagging回归、增强回归树、随机森林、人工神经网络、支持向量机、K最近邻等方法在森林生长收获预估中的应用、软件及调参等进行了综述,讨论了机器学习方法的优势和挑战,认为机器学习方法在森林生长收获预估方面有很大的潜力,必将得到广泛应用,并和传统统计模型相结合成为生长收获模型发展的一种趋势。      

云冷杉天然林林分年龄预测——以金沟岭林场为例

应用BP神经网络模型、PPR神经网络模型以及多元逐步回归模型,依据林分因子预测了金沟岭林场云冷杉天然林林分年龄。对比分析了人工神经网络计算模型算法与多元逐步回归分析模型预测结果的精度以及稳定性。结果表明:3种模型均可用于天然林林分年龄的预测,BP神经网络模型的预测平均相对误差为0.04,模型稳定性差;PPR神经网络模型的预测相对误差为0.06,模型稳定性好;多元逐步回归模型的预测相对误差为0.08,模型稳定性好。     

基于核极限学习机奶牛日粮能量消化预测方法

为准确评价饲料营养价值和提升饲喂管理水平,精准预测奶牛日粮能量消化指标具有重要意义。传统上主要基于线性回归(LR)方法预测奶牛日粮能量消化指标,但受参数模型假设限制,预测结果精度低,甚至偏离实际。文章首次将核极限学习机(KELM)方法应用于奶牛日粮消化能(DE)和能量消化率(ED)预测,KELM作为一种典型非参数机器学习模型,无需提前对预测模型作任何假设,仅通过学习训练样本数据,便可拟合出最接近实际的函数,特别适用于奶牛日粮能量消化等复杂系统预测问题。与传统LR方法和其他非参数模型RBF-ANN、SVM及标准ELM对比验证与讨论,结果表明,基于KELM预测方法在MAE、MAPE、RMSE及RT等多数指标上优于其他方法,特别是与传统LR方法相比,KELM方法预测精度更高,可作为对奶牛日粮DE和ED预测新型参考方法,为人工智能与机器学习在预测和评价动物饲粮营养价值应用研究提供借鉴。     

Artificial neural network models of the rumen fermentation pattern in dairy cattle

The objectives of this study were: (1) to predict the rumen fermentation pattern from milk fatty acids using a machine learning technique, i.e. artificial neural networks (ANN) combined with feature selection and (2) to compare the prediction accuracy of the resulting model to that of a statistical multi-linear regression model, based on odd and branched chain milk fatty acids. Data were collected from 10 experiments with rumen fistulated dairy cows, resulting in a dataset of 138 observations. Feature selection was based on correlation and principal component analysis, and background physiological knowledge. Different ANN architectures and training algorithms were assessed. The evaluation of the model performance, based on the test dataset, showed a root mean square prediction error, expressed relative to the observed mean, of 2.65%, 7.67% and 7.61% of the observed mean for acetate, propionate and butyrate, respectively. Compared to a multi-linear regression model, the ANN revealed not to perform significantly better. However, the results confirm that milk fatty acids have great potential to predict molar proportions of individual volatile fatty acids in the rumen.     

基于机器视觉技术的小粒中药材种子净度快速检测

为探究机器视觉技术用于小粒中药材种子净度快速检测的可行性,以黄芩、桔梗、黄芪、紫苏和柴胡5种常见小粒中药材种子为材料,使用扫描仪获取净种子、其他植物种子和所含杂质的图像,采用种子自动化分析系统(PhenoSeed)批量提取种子、其他植物种子及所含杂质的颜色、尺寸及纹理信息,通过相关性分析和主成分分析进行特征变量的筛选,采用多层感知器(MLP)和二元逻辑回归(BLR)建立上述5种中药材种子净度快速检测模型。结果表明,净种子、其他植物种子及所含杂质在物理指标方面存在显著差异,针对不同种子,采用不同指标建立的MLP净度模型的训练集和测试集准确率均在96.0%以上,该模型在不同中药材种子上的稳定性均优于BLR模型;以特征指标建立的模型稳定性优于全部指标的建模效果,运用特征变量建立的MLP模型对不同净度梯度(75.0%~100.0%)的混合样本进行预测,回归曲线的决定系数均达到0.99以上。采用机器视觉技术获取种子、其他植物种子及所含杂质颜色、尺寸和纹理等信息,以特征指标建立MLP模型可用于小粒中药材种子的净度快速检测。     

Discrimination between organically and conventionally grown winter wheat farm pair samples using the copper chloride crystallisation method in combination with computerised image analysis

Organic and conventional winter wheat farm pair grain samples were tested with the copper chloride crystallisation method and submitted to computerised image analyses followed by pattern recognition and classification with multivariate statistical tools.Appropriate discriminant analyses (DA) models were established. Depending on the analysed region of interest up to 100% of “unknown” samples could be correctly predicted using the DA models.     

超数排卵对奶牛产奶量及繁殖能力的影响

用FSH+PGF_(2)对16头黑白花奶牛进行超数排卵处理。超排前4d和超排1～10d日均产奶量差异不显著(P>0.05)。超排采胚后1头因年老淘汰,1头冲胚时发现患有慢性卡他性脓性子宫内膜炎进行治疗未配,其余14头冲胚后发情正常,人工授精后均妊娠,和正常牛发情配种妊娠率无差异。超数排卵对奶牛产奶量及繁殖能力无影响。     

约束标准化线性回归法估计合成品种动物基因组品种构成

【背景】合成品种是由至少两种纯种（祖先）培育的新品种,旨在兼顾祖先品种的有利遗传特征,并且可以长期保持后代的杂种优势而不需要每个世代都杂交。合成品种的遗传稳定,不同于杂交群体,因而可以像纯种一样繁育。实践中,估计合成品种的祖先品种对每个动物个体基因组的遗传贡献比例,即基因组品种构成（genomic breeding composition, GBC）,在畜禽品种登记、品种培育历史和品种构成分析、品种保护和杂交优势预测等方面有着非常重要的意义。利用基因组SNP基因型数据,采用合适的数学模型和统计方法,可以鉴定现有纯种品种的动物个体或纯种品种在杂交个体基因组的遗传贡献比例,而估计合成品种GBC的方法和研究都较少。【目的】线性回归是估计GBC的常用方法之一,但也存在诸多的问题。本研究旨在提出和评估一种约束的标准化线性回归方法（restricted standardized linear regression, RSLR）,作为传统线性回归方法的改进方法,应用于估计合成品种动物个体的GBC。【方法】采用肉牛王牛（Beefmaster）及其3个祖先品种（婆罗门牛、海福特牛和短角牛）的GGP 50K SNP芯片所测定的基因型数据,通过计算其基因频率和欧氏距离,利用层次聚类分析方法解析了4个动物群体的遗传关系,然后提出了RSLR方法,估计合成品种动物个体GBC的原理和方法。为了检验该方法的估计效果,从基因型数据中选择了均匀分布的分别包含1 000、5 000、10 000、20 000、30 000、40 000个SNP以及3个祖先品种共有的47 900个SNP的7个子集,分别采用RSLR和传统线性回归（linear regression, LR）两种方法估计了4 323头肉牛王牛的GBC,并比较了两种方法的计算结果。【结果】聚类分析的结果与4个品种间的遗传关系相吻合,表明肉牛王牛与婆罗门牛的遗传关系最近,遗传距离小于其与海福特牛和短角牛的遗传距离。LR方法估计的GBC会低估婆罗门牛（0.459—0.462）和短角牛（0.208—0.212）对于肉牛王牛的基因组贡献,同时高估海福特牛（0.326—0.333）的基因组贡献。但RSLR方法估计的肉牛王牛GBC的平均值与3个祖先品种预期的基因组贡献比例比较吻合：婆罗门牛为0.497—0.503,海福特牛为0.262—0.274,短角牛为0.229—0.231。此外,LR方法估计GBC的标准差和变异系数明显大于用RSLR估计的结果。当SNP子集数量在20 000以上时,LR方法估计牛肉王牛的3个祖先品种婆罗门牛、海福特牛和短角牛基因组贡献的标准差分别为0.048、0.032和0.051—0.052,变异系数分别为10.46%—10.50%、9.61%—9.76%和23.94%—25.00%,而RSLR方法估计的标准差,3个祖先品种对应为0.021、0.021—0.022和0.024—0.025,变异系数分别为4.18%—4.20%、7.89%—8.33%以及10.26%—10.68%。【结论】用RSLR方法估计的合成品种肉牛王牛动物个体的GBC,比LR方法的估计结果更加准确,估计的结果比LR方法估计的结果更稳定,且估计的一致性也更好,可以作为线性回归方法的改进,应用于估计合成品种动物个体GBC。     

中药复方制剂对奶牛免疫机能的影响

选取30头健康奶牛和30头隐性乳房炎奶牛进行中药复方制剂饲喂试验,对奶牛的B淋巴细胞EAC花环形成率、中性粒细胞吞噬率、中性粒细胞吞噬指数和T淋巴细胞转化率进行测定,观察中药复方制剂对奶牛免疫机能的影响.结果显示,健康奶牛的EAC花环形成率、中性粒细胞吞噬率、中性粒细胞吞噬指数、T淋巴细胞转化率均显著高于隐性乳房炎奶牛(P＜0.01或P＜0.05);中药复方制剂可以显著提高健康奶牛和隐性乳房炎奶牛的EAC花环形成率、中性粒细胞吞噬率、中性粒细胞吞噬指数、T淋巴细胞转化率(P＜0.01或P＜0.05).试验表明,中药复方制剂可以增强奶牛机体的体液免疫和细胞免疫功能,促使隐性乳房炎奶牛免疫机能的恢复.     

基于SOFM网络的山东省水资源承载力评价

水资源承载能力与评价指标组成了一个复杂的非线性系统,评价的难点在于确定各评价指标的权值。结合地理信息系统（GIS）和人工神经网络（ANN）技术,提出对水资源承载力进行评价的一种新方法;根据构建的水资源承载力评价指标体系,以山东省为例,利用自组织神经网络模型（SOFM）对水资源承载力进行了评价。结果表明,山东省17地市水资源承载力可划分为5类,模拟结果比较理想。     

阿尔泰山森林土壤温度动态变化及其预测

阿尔泰山作为干旱区典型的山地系统,其土壤温度的日、月、季节和年际动态及其影响因素研究,是深入理解干旱区山地森林生态系统能量循环过程的关键所在。基于阿尔泰山森林生态站2014年11月-2019年7月的气象因子和土壤温度数据,应用相关分析、回归分析和BP人工神经网络分析了阿尔泰山5、10、20 cm和30 cm深度土壤温度的动态变化及其对气象因子的响应,同时,应用多元线性回归和BP人工神经网络对土壤的温度进行了模拟。结果表明:1）近5 a各层土壤温度月均值年际变化一致,最低最高温度和日较差最大值均出现在20 cm,仅30 cm土壤温度的月变化出现自表层至深层滞后现象,年内月较差最大值出现在30 cm深度;各土壤层温度在春夏秋季变化较大,冬季变化较小;2）空气温度、气压和太阳辐射等与土壤温度的相关性达到极显著水平,其中与空气温度的相关性最强;3）回归模型和BP人工神经网络对20 cm土壤层的模拟结果最好,且BP人工神经网络模型的性能总体上优于回归模型。     

间隔覆盖条件下坡面产流产沙状况的BP神经网络模拟

本文以塿土和黄绵土作为实验材料,尝试使用BP神经网络方法(Back-Propagation neural network)模拟人工降雨条件下,间隔覆盖坡面的产流产沙状况。通过设置不同坡度、降雨强度、面积比,获得各种因素不同水平组合下的实测数据;以实际降雨强度、坡度、面积比、径流起始时间和初始含水率5个因子为输入变量、坡面产流量和产沙量为输出变量,利用BP神经网络模型与多元线性回归模型对数据进行模拟分析,并检验其模拟效果。研究结果表明:训练样本集平均相对误差为18.23%,预测样本集平均相对误差为5.21%;与多元线性回归模型相比,BP神经网络模型拟合精度较高,拟合效果更理想,表现出更强的预测能力。另外,比较不同土质坡面产流量与产沙量模拟效果,塿土优于黄绵土。从本研究的结果看,BP神经网络模型应用于坡面产流产沙模拟预测,省时省力,方便快捷,具有一定的应用潜力,但其实际的模拟预测能力尚需进一步探索。     

奶牛乳铁蛋白基因启动子区PCR-RFLP分析与乳房炎的相关性

采用CMT方法检测奶牛的乳房炎发病情况.筛选 90头分别设为对照组(健康奶牛)、隐性乳房炎组 (试验组Ⅰ )和临床乳房炎组(试验组Ⅱ),每组 30头.检测每头奶牛的NAGase活性,并用限制性片段长度多态性 (RFLP)分析技术,检测乳铁蛋白(Lf)基因启动子区域的RFLP多态性.结果表明:不同组别的奶牛之间NAGase活性差异显著 (P<0. 01),且Lf基因启动子区域存在RFLP多态性,说明该多态性与乳房炎存在一定关系,可能是奶牛乳房炎的一个分子标记.     

Automatic real-time monitoring of locomotion and posture behaviour of pregnant cows prior to calving using online image analysis

Monitoring the locomotion and posture behaviour of pregnant cows close to calving is essential in determining if there is a need for human intervention to assist parturition. In this study an automatic real-time monitoring technique is described in detail which allows identifying the locomotion and posturing behaviour of pregnant cows prior to calving. For this purpose video surveillance images of eight cows for the last 24 h prior to their calving were analysed. Data on seven different variables with time were obtained for each cow using an automatic real-time monitor. These were namely: x–y coordinates of the geometrical top view centre point of the cow; walking trajectory; distance walked; orientation of the main axis; body width/length ratio; hip length and back area. These variables were then used to classify specific behaviours such as standing or lying (including incidences of motion during lying), and eating or drinking. On average 85% of the standing and lying and 87% of the eating or drinking behaviour of the eight cows during the last 24 h before calving could be correctly classified. However, the developed technique needs to be further validated with additional tests in the field.     

Real-time automatic lameness detection based on back posture extraction in dairy cattle: Shape analysis of cow with image processing techniques

In this paper results on utilizing image analysis techniques towards early lameness detection in dairy cattle are presented. Data from two different dairy farms in Belgium were gathered. Preprocessing on raw data is required because of non-predictable behaviours of cows such as stopping for a while in front of the camera or non-uniform walking behaviour during experiments. Prelocalization of cow in each frame has been done based on two steps separation: (1) A coarse estimation of moving objects was obtained through background subtraction, (2) second statistical analysis of intensities in gray-scale image along with binarization was utilized to detect moving object in video. A common problem in on-farm collected videos is the similarity of the background and the cow's body colour since the use of classic algorithms for segmentation purposes does not work. Here a hierarchy background/foreground exaggeration is proposed to segment the cow in each frame and track it in video. The combination of logarithm and exponential, background subtraction as well as statistical filtering are used to find the accurate shape of the cow. Furthermore, the back posture of each cow during standing and walking was extracted automatically. It was done by detecting the arc of back posture and fitting a circle through selected points on the spine line. The average inverse radius of four frames displaying the hind hoofs in contact with the ground (two frames for each hoof in a row) was assigned to the cow. Based on this curvature value, a score representing the status of lameness in the individual cow was given automatically. Experimental results from two different databases show promising results in automatic lameness detection based on back posture information.     

Determination of body measurements on the Holstein cows using digital image analysis and estimation of live weight with regression analysis

In this study, the body measurements (BMs) of Holstein cows were determined using digital image analysis (IA) and these were used to estimate the live weight (LW) of each cow. For this purpose, an image capture arrangement was established in a dairy cattle farm. BMs including wither height (WH), hip height (HH), body length (BL), hip width (HW), plus the LWs of cows were first determined manually, by direct measurement. Then the digital photos of cows were taken from different directions synchronously and analyzed by IA software to calculate WH, HH, BL and HW of each cow. After comparing the BMs obtained by IA with the manual measurements, the accuracy was determined as 97.72% for WH, 98.00% for HH, 97.89% for BL and 95.25% for HW. The LW estimation using BMs was then performed by the aid of the regression equations, and the correlation coefficient between the estimated and real (manual) LW values obtained by weighing was calculated as 0.9787, which indicates the IA method is appropriate for LW estimation of Holstein cows.     

Optimization of an artificial neural network topology using coupled response surface methodology and genetic algorithm for fluidized bed drying

In this study, an integrated response surface methodology (RSM) and genetic algorithm (GA) are recommended for developing artificial neural networks (ANNs) with great chances to be an optimal one. A multi-layer feed forward (MLFF) ANN was applied to correlate the outputs (energy and exergy) to the four exogenous inputs (drying time, drying air temperature, carrot cubes size, and bed depth). The RSM was used to build the relationship between the input parameters and output responses, and used as the fitness function to measure the fitness value of the GA approach. In the relationship building, five variables were used (number of neurons, momentum coefficient and step size in the hidden layer, number of epochs and number of training times). A polynomial model was developed from training results to mean square error (MSE) of 50 developed ANNs to generate 3D response surfaces and contour plots. Finally, GA was applied to find the optimal topology of ANN. The ANN topology had minimum MSE when the number of neurons in the hidden layer, momentum coefficient, step size, number of training epochs and training times were 28, 0.66, 0.35, 2877 and 3, respectively. The energy and exergy of carrot cubes during fluidized bed drying were predicted with R² values of greater than 0.97 using optimal ANN topology.     

奶牛头胎产犊日龄与305d泌乳量的相关分析

[目的]探讨奶牛头胎产犊日龄对其产奶量的影响。[方法]以宁夏平吉堡奶牛场一分场系谱资料比较完整的50头奶牛头胎产犊记录为资料,对奶牛头胎产犊日龄与305 d泌乳量进行相关分析。[结果]50头奶牛头胎产犊日龄与305 d泌乳量的相关系数为0.4482,达显著水平(P<0.05),即奶牛头胎产犊日龄与305 d泌乳量呈中等正相关。305 d泌乳量(y)与产犊日龄(x)之间的回归方程为y=8.173x+11.460。在744~887 d之间产犊的奶牛,其305 d泌乳量随头胎产犊日龄的增加而增大,可以预测该时期内奶牛的产奶量。[结论]奶牛头胎产犊日龄对其305 d泌乳量有较大影响。     

Neural network modelling of flat-plate solar collectors

In this study, different approaches to the modelling of flat-plate solar collectors are introduced and analysed. Among the physically based models, the heat network model and Hottel–Vhillier (H–V) models are discussed. The parameters of the latter model are identified for three different types of these solar collectors. The identification exhibited good agreement with the measured values. Finally, modelling simulations with an artificial neural network (ANN) technique were carried out. A sensitivity study was performed on the parameters of the neural network. The possible ANN structures, the size of training data set, the number of hidden neurons, and the type of training algorithm were analysed in order to identify the most appropriate model. The same ANN structures were trained and validated for the three solar collectors, using data generated from the H–V model and long-term (17 days) measurements.