SBIR Phase I: Data Analytics on Honeybee Hives Using IoT Sensor Data

这个小企业创新研究（SBIR）第一阶段项目的更广泛的影响/商业潜力是更健康的蜜蜂种群，以增加粮食安全。根据联合国粮食及农业组织（Food and Agriculture Organization）的数据，到2050年，每年的粮食产量必须比2007年增加60%，才能为估计的91亿人生产足够的粮食，如果没有蜜蜂授粉，这是不可能的（Food and Agriculture Organization, 2012）。在过去的60年里，蜜蜂的数量急剧下降。总共损失了300万个蜂箱（国家农业统计局，2007年）。每年蜂房损失使美国经济损失20亿美元（美国国家农业统计局，2016年；白宫，2014年）。该项目旨在通过为养蜂人寻找蜂窝丢失的实际解决方案，推进物联网在养蜂业中的应用，以及为养蜂人提供自动化数据分析，增强对问题的科学理解。研究人员和专家怀疑蜂箱数量的下降是由多种因素造成的，但由于缺乏大规模追踪蜂箱的技术，研究受到了限制。研究人员无法对大型、多样化的数据集进行分析，这阻碍了他们得出蜂巢丧失原因的结论，并推动传统养蜂方法的创新。该项目将通过算法构建，为养蜂人自动化数据分析，推进物联网在养蜂业中的应用。一些研究建立了模型来描述蜂巢内部的活动。然而，这些都是建立在有限的数据或理论基础上的。Henry et al.， 2016提到需要将预测算法扩展到更大的数据集，以减少蜂群损失。数据驱动的养蜂行业尚处于起步阶段，但基于农业科技领域的成功，数据监测将是解决蜂群健康问题的必要步骤。该项目将建立一个包含4000多个蜂箱的数据库，以改进实际监测的研究模型。该数据库将与来自项目蜂巢的传感器和监测数据配对。通过该项目，该公司将确认传感器的商业化质量。该公司将创建一个健康蜂巢的基线模型来检测异常情况。这些异常情况将被纳入一个预测模型，以检测蜂巢中与病虫害有关的问题。该公司将寻找对受威胁蜂巢的广泛检测，然后深入研究具体问题（如瓦螨）。这些发现将包括在提供给养蜂人的监测产品中。

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase I project is a healthier honeybee population to increase food security. According to the Food and Agriculture Organization, annual production must increase by 60% from 2007 to produce enough food for an estimated 9.1 billion people by 2050, which will be impossible without honeybee pollination (Food and Agriculture Organization, 2012). Honeybee populations have declined precipitously for the past 60 years?a total loss of 3 million hives (National Agricultural Statistics Service, 2007). Annual hive losses cost the U.S. economy $2 billion (National Agricultural Statistics Service, 2016; The White House, 2014). This project aims to enhance scientific understanding of the problem by finding practical solutions to hive loss for beekeepers, advancing IoT applications in beekeeping, and automating data analysis for beekeepers. Researchers and experts suspect the declines are due to multiple factors, but research is limited by a lack of technology tracking the hives on a large scale. Researchers' inability to perform analysis on a large, diverse dataset has obstructed their ability to draw conclusions on causes of hive loss and drive innovation on traditional beekeeping methods. This proposed project will advance Internet of Things applications in beekeeping by automating data analysis for beekeepers through algorithm building. Several studies built models to describe actions inside the hive. However, these are built on limited data or are theoretical. Henry et al., 2016 mentioned the need for predictive algorithms to be expanded to larger data sets to reduce hive loss. Data driven beekeeping is in its infancy as an industry, but based on the success in the AgTech space, data monitoring will be a necessary step in solving colony health problems. The project will take a database of over 4,000 hives to improve research models for practical monitoring. This database will be paired with sensor and monitoring data from project hives. Through the project hives the company will confirm the sensor quality for commercialization. The company will create a baseline model of a healthy hive to detect anomalies. These anomalies will be worked into a predictive model to detect problems related to pests and diseases in the hive. The company will look for broad detection of a threatened hive, then drill down into specific problems (like Varroa mites). These finds will be included in monitoring products offered to beekeepers.