基于随机森林算法的阀冷系统膨胀罐液位预测  

Liquid Level Prediction Model of Expansion Tank in Valve Cooling System Based on Stochastic Forest Algorithm

作　　者：张浪文 邹培炯 ZHANG Langwen;ZOU Peijiong(Yueyang Goaland Energy Conservation Equipment Manufacturing Co.,Ltd.,Yueyang,Hunan Province,414000 China)

机构地区：[1]岳阳高澜节能装备制造有限公司,湖南岳阳414000

出　　处：《科技资讯》2022年第6期31-34,共4页Science & Technology Information

摘　　要：直流输电电流转换过程中产生大量的热量,阀冷系统的作用是对热量进行及时消散,以保证系统稳定运行,是直流输电工程的核心设备。膨胀罐是保证阀冷系统压力稳定的关键,提前预测膨胀罐排气、补气阀状态能提高系统运行效率。膨胀罐液位是反映其状态的重要指标,该文建立基于随机森林算法的阀冷系统膨胀罐液位预测方法。首先,采集阀冷系统重要参数数据,包括电极功率、环境温度、内冷进(出)阀温度等,分析关联数据对膨胀罐液位的影响关系;其次,开发随机森林回归算法,利用训练集数据对膨胀罐的液位进行预测建模,得到预测模型;最后,利用测试集数据对所建立的预测模型进行综合仿真,结果表明模型预测数据能够很好地与实测数据拟合,说明该文膨胀罐液位预测模型的有效性。A large amount of heat will be generated in the process of DC transmission current conversion. The function of valve cooling system is to dissipate the heat in time to ensure the stable operation of the system, which is the core equipment of DC transmission project. Valve cooling system is the core equipment of DC transmission project. The function is to effectively dissipate the heat in the process of current conversion. Expansion tank is the key to ensure the pressure stability of valve cooling system. Predicting the exhaust and air makeup valve status of expansion tank in advance can improve the operation efficiency of the system. The liquid level of expansion tank is an important index to reflect its state. The liquid level prediction method of expansion tank in valve cooling system based on random forest algorithm is established in this paper. Firstly, collect the important parameter data of valve cooling system, including electrode power, ambient temperature, internal cooling inlet(outlet) valve temperature,and analyze the influence of related data on the liquid level of expansion tank;secondly, the Stochastic Forest regression algorithm is developed, and the training set data is used to predict and model the liquid level of the expansion tank to obtain the prediction model;finally, the established prediction model is comprehensively simulated by using the test set data. The results show that the prediction data of the model can fit the measured data well, which shows the effectiveness of the expansion tank liquid level prediction model in this paper.

关 键 词：阀冷系统 膨胀罐 随机森林 液位 预测模型 

分 类 号：TM721.1[电气工程—电力系统及自动化]