本文關(guān)鍵詞： 蛋雞 物聯(lián)網(wǎng) 分布式流式計(jì)算 蛋雞行為 決策分析　出處：《中國(guó)農(nóng)業(yè)大學(xué)》2017年博士論文　論文類型：學(xué)位論文

【摘要】：本研究面向現(xiàn)代蛋雞行業(yè)健康養(yǎng)殖的需求,探索如何將蛋雞養(yǎng)殖從傳統(tǒng)粗放管理轉(zhuǎn)化成實(shí)際生產(chǎn)過程精細(xì)養(yǎng)殖,并對(duì)感知的蛋雞生理、環(huán)境及生產(chǎn)過程數(shù)據(jù)進(jìn)行實(shí)時(shí)處理,對(duì)照廠家提供的相關(guān)育種標(biāo)準(zhǔn)數(shù)據(jù),對(duì)蛋雞在禽舍的生存狀態(tài)及生產(chǎn)經(jīng)濟(jì)效益予以實(shí)時(shí)監(jiān)測(cè),提高了蛋雞養(yǎng)殖企業(yè)安全預(yù)警的水平及管理者的決策能力。針對(duì)上述目標(biāo),本研究研發(fā)了如下關(guān)鍵技術(shù):基于物聯(lián)網(wǎng)的蛋雞養(yǎng)殖生產(chǎn)過程數(shù)據(jù)的實(shí)時(shí)獲取與傳輸技術(shù),基于分布式流式計(jì)算的蛋雞養(yǎng)殖生產(chǎn)數(shù)據(jù)實(shí)時(shí)處理分析技術(shù),基于機(jī)器學(xué)習(xí)的自動(dòng)追蹤蛋雞行為算法,通過有效整合,研發(fā)了蛋雞設(shè)施養(yǎng)殖數(shù)字化智能監(jiān)測(cè)與遠(yuǎn)程管理系統(tǒng),并在實(shí)際雞場(chǎng)得到應(yīng)用。該系統(tǒng)實(shí)現(xiàn)了對(duì)多個(gè)異地蛋雞場(chǎng)的環(huán)境、生理及生產(chǎn)過程數(shù)據(jù)的實(shí)時(shí)監(jiān)測(cè)、獲取、處理和預(yù)警,提高了養(yǎng)殖場(chǎng)對(duì)蛋雞養(yǎng)殖生產(chǎn)實(shí)時(shí)數(shù)據(jù)的存儲(chǔ)、分析和管理的能力;實(shí)現(xiàn)了對(duì)蛋雞行為視頻的自動(dòng)分析。該系統(tǒng)在蛋雞場(chǎng)的應(yīng)用實(shí)現(xiàn)了對(duì)養(yǎng)殖情況的精細(xì)感知,提升了我國(guó)蛋雞養(yǎng)殖業(yè)的管理水平,降低人工勞動(dòng)強(qiáng)度,對(duì)支撐我國(guó)蛋雞行業(yè)的福利養(yǎng)殖、實(shí)時(shí)決策都有很大幫助。主要結(jié)論如下:(1)通過文獻(xiàn)閱讀和現(xiàn)場(chǎng)調(diào)研,本研究總結(jié)了蛋雞養(yǎng)殖中的三類數(shù)據(jù):環(huán)境、生理及生產(chǎn)過程數(shù)據(jù)。針對(duì)這三類數(shù)據(jù)的特點(diǎn),使用物聯(lián)網(wǎng)技術(shù)自動(dòng)獲取可精細(xì)感知的數(shù)據(jù);對(duì)其它數(shù)據(jù),改變?cè)屑堎|(zhì)報(bào)表留存的方式,開發(fā)了基于WEB和APP的人工填報(bào)方式,實(shí)現(xiàn)了數(shù)字化,方便了現(xiàn)場(chǎng)工作人員的工作。應(yīng)對(duì)蛋雞養(yǎng)殖場(chǎng)一般地處偏僻,信息通訊難度大的特點(diǎn),并基于節(jié)省流量和帶寬的目標(biāo),本研究提出了異步傳輸機(jī)制,編寫了監(jiān)測(cè)數(shù)據(jù)更新的程序,最大程度的避免了傳輸過程中的數(shù)據(jù)丟失,并可顯著降低通信運(yùn)營(yíng)成本。(2)為實(shí)現(xiàn)對(duì)蛋雞養(yǎng)殖生產(chǎn)大數(shù)據(jù)的實(shí)時(shí)處理、計(jì)算和預(yù)警,自主研發(fā)了應(yīng)用于蛋雞設(shè)施養(yǎng)殖中的分布式流式計(jì)算框架Data-Canal。Data-Canal使用控制流集中、數(shù)據(jù)流分散的模型,以分布式文件系統(tǒng)為中間結(jié)果的存儲(chǔ),具有很強(qiáng)的可擴(kuò)展性和可靠性。運(yùn)行結(jié)果說明,使用Data-Canal計(jì)算框架可以解決蛋雞養(yǎng)殖生產(chǎn)數(shù)據(jù)實(shí)時(shí)計(jì)算的問題,在部署8臺(tái)機(jī)器的情況下,Data-Canal集群的處理能力峰值達(dá)到160 MB/s,延遲在分鐘級(jí)別。(3)視頻數(shù)據(jù)已經(jīng)成為目前蛋雞養(yǎng)殖生產(chǎn)大數(shù)據(jù)的主體�，F(xiàn)有系統(tǒng)通過傳感器設(shè)備收集了大量的現(xiàn)場(chǎng)視頻圖像數(shù)據(jù),但往往缺乏應(yīng)用分析手段,進(jìn)而無法從中得到有價(jià)值的信息�；诖吮狙芯刻岢隽嘶跈C(jī)器學(xué)習(xí)的自動(dòng)追蹤蛋雞行為算法,實(shí)現(xiàn)了對(duì)蛋雞行為視頻數(shù)據(jù)的自動(dòng)分析。該算法使用HOG特征描述方法提取雞只樣本特征,訓(xùn)練混合SVM模型計(jì)算雞只最優(yōu)位置,可自動(dòng)識(shí)別小群體蛋雞中任意雞只的行為軌跡,計(jì)算其運(yùn)動(dòng)距離、速度等,量化雞只的運(yùn)動(dòng)行為。結(jié)合設(shè)施內(nèi)區(qū)域分布,自動(dòng)統(tǒng)計(jì)雞只對(duì)不同區(qū)域的使用情況,間接對(duì)采食/飲水行為進(jìn)行量化。(4)本研究以規(guī)模化養(yǎng)殖蛋雞為研究對(duì)象,在分布式架構(gòu)下,使用物聯(lián)網(wǎng)與流式計(jì)算技術(shù),實(shí)現(xiàn)了對(duì)蛋雞養(yǎng)殖生產(chǎn)過程實(shí)時(shí)數(shù)據(jù)的自動(dòng)獲取、異步傳輸和實(shí)時(shí)處理,并結(jié)合基于機(jī)器學(xué)習(xí)的自動(dòng)追蹤蛋雞行為的算法,實(shí)現(xiàn)了對(duì)非結(jié)構(gòu)化數(shù)據(jù)(視頻圖像)的自動(dòng)處理,最后以數(shù)據(jù)分析服務(wù)的方式在實(shí)際蛋雞場(chǎng)應(yīng)用。數(shù)據(jù)分析成果主要有養(yǎng)殖經(jīng)濟(jì)效益分析、生產(chǎn)過程實(shí)時(shí)預(yù)警服務(wù)、環(huán)境數(shù)據(jù)分析、多維數(shù)據(jù)分析、生產(chǎn)信息管理服務(wù)等。
[Abstract]:This research is focused on the healthy culture of modern egg industry needs to explore how the laying hens are transformed from traditional extensive management to fine breeding process, and laying hens on physiological perception, environment and production process real-time data processing, control manufacturers to provide relevant breeding standard data to real-time monitoring of laying hens in poultry house living state the production and economic benefits, improve the level of enterprise management and laying hens safety warning decision makers'ability. According to the above objectives, this study developed the following key technologies: laying hens production process real-time data networking and transmission technology based on the analysis of real time processing technology of laying hens production data stream based on distributed computing. Automatic tracking behavior of layers algorithm based on machine learning, through the effective integration and development of laying hens breeding facilities and digital intelligent monitoring The remote management system, and has been applied in actual farms. The system implements the multiple remote egg farm environment, real-time monitoring, physiology and production process data, processing and early warning, improve the farm production of laying hens in real-time data storage, analysis and management ability; to realize the automatic analysis of behavior of layers video. The system realizes the fine perception of the situation in the application of breeding egg chicken, enhance the management level of China's poultry industry, reduce labor intensity, to support China's egg industry of animal welfare, real-time decision-making are of great help. The main conclusions are as follows: (1) through the literature reading and the field investigation, this paper summarizes three types of data cultivationoflayers: environment, physiology and production process data. According to the characteristics of these three kinds of data, can automatically obtain the fine sensing data to use networking technology; Other data, change the original paper statements retained, developed WEB and APP artificial reporting methods based on digitization, convenient on-site staff. To deal with the poultry farm general remoteness, difficulty of information communications, and based on the save traffic and bandwidth of the target, this study proposes an asynchronous the transmission mechanism, preparation of monitoring data to update the program, the maximum to avoid the loss of data in the transmission process, and can significantly reduce the operating costs of communication. (2) in order to realize the real-time processing of laying hens production data, calculation and early warning, Data-Canal.Data-Canal computing framework using control flow concentrated flow of independent research and development of distributed application in the laying hens breeding facilities, distributed data flow model, intermediate results to a distributed file system for storage, has strong scalability and reliability. The operation. The results illustrate that the use of Data-Canal computing framework can solve the real-time calculation of laying hens production data, for the deployment of the 8 machine case, the peak processing ability of Data-Canal cluster is up to 160 MB/s, the delay in the minute level. (3) the video data has become the subject of Laying Hens Production data. The existing collection of live video system a large number of image data through the sensor devices, but often lack the application of analysis method, and then to get valuable information from it. This research is put forward based on the behavior of layers automatic tracking algorithm based on machine learning, to realize the automatic analysis of the behavior of layers of video data. The algorithm uses HOG feature extraction method only samples the characteristics of chicken training, mixed SVM model to calculate the optimal position of chicken, chicken can track any automatic recognition of small groups of laying hens, calculate the moving distance, speed The degree of movement, quantitative chicken behavior. Combined with the regional distribution facilities, automatic statistics on the use of chickens in different areas, indirectly on foraging / drinking behavior was quantified. (4) based on the large-scale breeding hens as the research object, in the distributed architecture, the use of networking and stream computing technology to realize the automatic acquisition of real-time data in laying hens production process, asynchronous transmission and real-time processing, and the combination of automatic tracking behavior of layers of machine learning algorithm based on the realization of non structured data (video) automatic processing, and finally to the data analysis service in practical application. The data analysis results of hen the main analysis of the economic benefits of aquaculture, real-time early warning service production process, data analysis, multidimensional data analysis, production information management services.

【學(xué)位授予單位】：中國(guó)農(nóng)業(yè)大學(xué)
【學(xué)位級(jí)別】：博士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP391.44;TN929.5;S831

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 Cheng Wang;Hongqian Chen;Xuebin Zhang;Chaoying Meng;;Evaluation of a laying-hen tracking algorithm based on a hybrid support vector machine[J];Journal of Animal Science and Biotechnology;2017年01期

2 朱寧;秦富;;蛋雞產(chǎn)業(yè)發(fā)展的國(guó)際趨勢(shì)及中國(guó)展望[J];中國(guó)家禽;2016年20期

3 李保明;王朝元;楊柳;;從歐洲精準(zhǔn)畜牧業(yè)研討會(huì)看蛋雞精準(zhǔn)養(yǎng)殖技術(shù)的研究進(jìn)展[J];中國(guó)家禽;2016年03期

4 楊柳;李保明;;蛋雞福利化養(yǎng)殖模式及技術(shù)裝備研究進(jìn)展[J];農(nóng)業(yè)工程學(xué)報(bào);2015年23期

5 陳紅茜;滕光輝;邱小彬;孟超英;曹晏飛;汪成;;基于分布式流式計(jì)算的蛋雞養(yǎng)殖實(shí)時(shí)監(jiān)測(cè)與預(yù)警系統(tǒng)[J];農(nóng)業(yè)機(jī)械學(xué)報(bào);2016年01期

6 臧素敏;;蛋雞標(biāo)準(zhǔn)化養(yǎng)殖技術(shù)[J];北方牧業(yè);2015年12期

7 劉翔;鄒婷婷;李仁旺;王亞沛;;基于HMM模型的老年人出行異常檢測(cè)[J];浙江理工大學(xué)學(xué)報(bào);2015年05期

8 袁曉慶;孔箐鋅;李奇峰;李琳;李道亮;;水產(chǎn)養(yǎng)殖物聯(lián)網(wǎng)的應(yīng)用評(píng)價(jià)方法[J];農(nóng)業(yè)工程學(xué)報(bào);2015年04期

9 熊本海;楊振剛;楊亮;潘曉花;;中國(guó)畜牧業(yè)物聯(lián)網(wǎng)技術(shù)應(yīng)用研究進(jìn)展[J];農(nóng)業(yè)工程學(xué)報(bào);2015年S1期

10 曹晏飛;余禮根;滕光輝;趙淑梅;劉旭明;;蛋雞發(fā)聲與機(jī)械噪聲特征提取及分類識(shí)別[J];農(nóng)業(yè)工程學(xué)報(bào);2014年18期

相關(guān)博士學(xué)位論文 前10條

1 漆海霞;畜禽舍關(guān)鍵環(huán)境因子檢測(cè)系統(tǒng)的研究[D];華南農(nóng)業(yè)大學(xué);2016年

2 李卓;基于立體視覺技術(shù)的生豬體重估測(cè)研究[D];中國(guó)農(nóng)業(yè)大學(xué);2016年

3 林林;水利云平臺(tái)若干技術(shù)問題研究[D];中國(guó)水利水電科學(xué)研究院;2016年

4 郭雷風(fēng);面向農(nóng)業(yè)領(lǐng)域的大數(shù)據(jù)關(guān)鍵技術(shù)研究[D];中國(guó)農(nóng)業(yè)科學(xué)院;2016年

5 陳東成;基于機(jī)器學(xué)習(xí)的目標(biāo)跟蹤技術(shù)研究[D];中國(guó)科學(xué)院研究生院(長(zhǎng)春光學(xué)精密機(jī)械與物理研究所);2015年

6 董正浩;基于支持向量機(jī)的移動(dòng)互聯(lián)網(wǎng)用戶行為偏好研究[D];北京郵電大學(xué);2014年

7 鄭博;基于云計(jì)算的農(nóng)業(yè)圖像處理系統(tǒng)設(shè)計(jì)與算法研究[D];江蘇大學(xué);2012年

8 王新政;禽舍環(huán)境智能控制關(guān)鍵技術(shù)研究[D];東北林業(yè)大學(xué);2012年

9 朱軍;種豬數(shù)字化養(yǎng)殖平臺(tái)的系統(tǒng)集成與應(yīng)用研究[D];內(nèi)蒙古農(nóng)業(yè)大學(xué);2010年

10 李海生;支持向量機(jī)回歸算法與應(yīng)用研究[D];華南理工大學(xué);2005年

相關(guān)碩士學(xué)位論文 前3條

1 張小龍;基于超聲波技術(shù)的豬只定位與其異常行為分析[D];太原理工大學(xué);2015年

2 馬蓓蓓;基于HOG特征的車輛檢測(cè)技術(shù)研究[D];華南理工大學(xué);2015年

3 孫月強(qiáng);基于Labview的蔬菜大棚環(huán)境監(jiān)測(cè)系統(tǒng)的研究[D];中國(guó)海洋大學(xué);2010年

，

本文編號(hào)：1551281