反刍家畜典型行为监测与生理状况识别方法研究综述

Review on Typical Behavior Monitoring and Physiological Condition Identification Methods for Ruminant Livestock

反刍家畜是人类获得肉、奶等食品的重要来源,随着人们对其产品产量与品质要求的提升,传统耗时耗力且高人工成本的人工监管模式已经难以满足规模化反刍家畜养殖的需要。反刍家畜行为中蕴含着许多身体状况信息,对反刍家畜行为的自动化监测有助于较早地识别其异常行为、评估其健康水平、预警其异常生理状态,辅助养殖人员及时调整养殖策略,实现低成本、高效率和高收益的生产过程。首先对反刍家畜基本运动(躺卧、行走、站立)、反刍、进食饮水、跛行等典型行为的监测方法进行总体阐述,然后详细分析了识别反刍家畜发情、分娩、疾病、疼痛状况的不同特征指标以及基于该特征指标的生理状况识别方法,最后探讨了反刍家畜行为监测方法目前存在的一些问题与难点,并指出未来的研究重点为:优化传感器功耗、融合多传感器数据、降低数据传输延时、减少大规模数据标注、轻量化深度学习模型以及深度解析和应用数据。

Ruminant livestock is an important source of meat,milk and other food for human beings.With the improvement of people's requirements for the output and quality of ruminant livestock products,the traditional manual supervision mode,which is time-consuming,labor-intensive and high labor cost,has been difficult to meet the needs of large-scale ruminant livestock breeding.Ruminant livestock behavior contains a lot of body condition information.The intelligent monitoring of ruminant livestock behavior is helpful to identify abnormal behavior of ruminant livestock earlier,evaluate the health level of ruminant livestock,early warning of abnormal physiological state of ruminant livestock,and assist farmers to adjust breeding strategies in a timely manner to achieve low cost-effective,efficient and profitable production process.Firstly,the monitoring methods for basic movements(lying,walking and standing),rumination,eating and drinking,lameness of ruminant livestock were overally described.Secondly,the different characteristic indicators to identify the condition of ruminant livestock in estrus,parturition,disease and pain were analyzed in detail and the physiological condition identification method was introduced based on the characteristic indicators.Thirdly,the problems and challenges of ruminant livestock behavior monitoring methods were summarized.Finally,the future development directions of relevant key technologies were prospected,including optimizing sensor power consumption,fusion of multi-sensor data,reducing data transmission delay,reducing large-scale data annotation,lightweight deep learning models and deep analysis and application data.