基于海量用电数据的农业灌溉用户识别方法

Massive Power Consumption Data Based User Identification for Agricultural Irrigation

为了确定农业灌溉系统的使用者,优化水资源利用效率,提高农业生产效率,为以电折水系数计算提供支撑。利用农业灌溉的通用行为规律建立初步分类模型,进行数据标签与特征关联度构建,最终构建用户识别模型。首先针对种植不同作物的用户样本进行统计分析,结合不同作物普适的灌溉用电规律,建立基于降雨特征的农业灌溉用户识别规则模型。其次选取某个区县的用户,利用密度聚类算法,为不同簇群用户打上分类标签,构建基于随机森林的农业灌溉用户用电特征识别分类模型。最后进行模型融合,构建混合农业灌溉用户分类模型,实现农业灌溉用户、混合农业灌溉用户、非农业灌溉用户的精准识别。依据模型计算结果在现场核实,准确率达90%以上,减少了人力成本,为后续以电折水系数的精确计算提供了一定基础。

In order to identify the users of agricultural irrigation systems, optimize the efficiency of water resource utilization, improve agricultural production efficiency, and provide support for the calculation of the number of electricity-discounted water systems, this paper established a preliminary classification model by the generic behavioral laws of agricultural irrigation then constructed the associations of data labels and feature, and finally constructed a user identification model.Firstly, statistical analysis was carried out for the user samples planting different crops, combined with the universal irrigation power use law of different crops, to establish the recognition rule model of agricultural irrigation users based on rainfall characteristics.Secondly, the users in a certain district and county were selected, and the density clustering algorithm was utilized to label different clusters of users with classification labels, so as to construct a classification model based on Random Forest for identifying the power consumption characteristics of agricultural and irrigation users.Finally, model fusion was carried out to construct a hybrid agricultural irrigation user classification model to realize the accurate identification of agricultural irrigation users, hybrid agricultural irrigation users, and non-agricultural irrigation users.Based on the results of the model calculation in the field verification, the accuracy rate reached more than 90%,reducing labor costs, and providing a certain foundation for the subsequent accurate calculation of the number of water systems discounted by electricity.