Abstract: 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.