Abstract: Aiming at the current problems of low irrigation efficiency, high labor intensity, and extensive management in blueberry orchards, a blueberry orchard precision irrigation system based on LoRa wireless longdistance communication and SVM-Markov combination model is designed. The system collects environmental parameters such as air temperature and humidity, soil humidity, illuminance, and wind speed in the blueberry garden through the LoRa wireless data acquisition system, and uploads the data packets to the cloud server through the LoRa gateway and the Internet of Things gateway, to realize irrigation volume prediction and irrigation decisionmaking, and feed back the decisionmaking results to the irrigation execution module. In order to improve the prediction accuracy, SVM-Markov algorithm was introduced. Taking blueberry Garden in Tianwang Town of Jurong City as the experimental object, the prediction results showed that the mean absolute error of SVM-Markov model was 0.188 7 mm/d, and the root mean square error was 0.239 4 mm/d. Compared with SVM model, the prediction accuracy of SVM-Markov model was higher and the data fitting effect was better. The system can realize realtime monitoring and precise irrigation of blueberry orchard environment, which provides a certain reference for the realization of precise irrigation in other orchards.

Key words: Keywords: 
precision irrigation, LoRa, SVM, Markov

摘要： 针对当前蓝莓园灌溉效率低、劳动强度大、管理粗放等问题，设计基于LoRa无线远距离通信和SVM-Markov组合模型的蓝莓园精准灌溉系统。该系统通过LoRa无线数据采集系统采集蓝莓园空气温湿度、土壤湿度、光照度、风速等环境参数，通过LoRa网关和物联网网关将数据包上传到云服务器，灌溉预测系统根据采集到的环境参数，实现灌溉量预测与灌溉决策，并将决策结果反馈到灌溉执行模块。为提高预测精度，引入SVM-Markov灌溉量预测算法。以句容市天王镇蓝莓园为试验对象，预测结果表明：SVM-Markov模型的平均绝对误差为0.188 7 mm/d，均方根误差为0.239 4 mm/d，相比于SVM模型，SVM-Markov的预测精度更高、数据拟合效果更好。该系统能够实现蓝莓园环境的实时监测与精准灌溉，为其它果园精准灌溉的实现提供一定的参考。

关键词: 精准灌溉, LoRa, 支持向量机(SVM), 马尔科夫(Markov)