Computer Sciences Department and Wisconsin Institute for Discovery, University of Wisconsin-Madison, Madison 53706.
Computer Sciences Department and Wisconsin Institute for Discovery, University of Wisconsin-Madison, Madison 53706.
J Dairy Sci. 2020 Apr;103(4):3874-3881. doi: 10.3168/jds.2019-17199. Epub 2020 Feb 26.
Management decisions can be informed by near-real-time data streams to improve the economics of the farm and to positively benefit the overall health of a dairy herd or the larger environment. Decision support tools can use data management services and analytics to exploit data streams from farm and other economic, health, and agricultural sources. We will describe a decision support tool that couples data analytics tools to underlying cow, herd, and economic data with an application programming interface. This interface allows the user to interact with a collection of dairy applications without fully exposing the intricacies of the underlying system model and understand the effects of different decisions on outputs of interest. The collection of these applications will form the basis of the Dairy Brain decision support system, which will provide management suggestions to farmers at a single animal or farm level. Dairy operations data will be gathered, cleaned, organized, and disseminated through an agricultural data hub, exploiting newly developed ontologies for integration of multiple data sources. Models of feed efficiency, culling, or other dairy operations (such as large capital expenditures, outsourcing opportunities, and interactions with regulators) form the basis of analytical approaches, operationalized via tools that help secure information and control uncertainties. The applications will be independently generated to provide flexibility, and use tools and modeling approaches from the data science, simulation, machine learning, and optimization disciplines to provide specific recommendations to decision makers. The Dairy Brain is a decision support system that couples data analytics tools with a suite of applications that integrate cow, herd, and economic data to inform management, operational, and animal health improving practices. Research challenges that remain include dealing with increased variability as predictions go from herd or pen level down to individual cow level and choosing the appropriate tool or technique to deal with a specific problem.
管理决策可以通过近乎实时的数据流来提供信息，以提高农场的经济效益，并积极改善奶牛群或更大环境的整体健康状况。决策支持工具可以使用数据管理服务和分析来利用来自农场和其他经济、健康和农业来源的数据流。我们将描述一种决策支持工具，该工具将数据分析工具与底层牛、牛群和经济数据以及应用程序编程接口相结合。该接口允许用户与一组奶牛应用程序交互，而无需完全暴露底层系统模型的复杂性，并了解不同决策对感兴趣的输出的影响。这些应用程序的集合将构成 Dairy Brain 决策支持系统的基础，该系统将在单个动物或农场层面为农民提供管理建议。奶牛养殖数据将通过农业数据中心进行收集、清理、组织和传播，利用新开发的本体论来整合多个数据源。饲料效率、淘汰或其他奶牛养殖模型（如大笔资本支出、外包机会以及与监管机构的互动）构成了分析方法的基础，这些方法通过帮助确保信息安全和控制不确定性的工具来实现。这些应用程序将独立生成，以提供灵活性，并使用数据科学、模拟、机器学习和优化领域的工具和建模方法为决策者提供具体建议。Dairy Brain 是一种决策支持系统，它将数据分析工具与一系列应用程序相结合，这些应用程序集成了牛、牛群和经济数据，以提供信息、改善管理、运营和动物健康实践。仍然存在的研究挑战包括随着预测从牛群或畜栏层面下降到单个奶牛层面，处理增加的可变性，以及选择适当的工具或技术来处理特定问题。
