【作者】 刘天豪； 周灿； 李勇刚； 朱红求； 王雨婕；

【Author】 LIU Tian-hao;ZHOU Can;LI Yong-gang;ZHU Hong-qiu;WANG Yu-jie;Central South University;

【机构】 中南大学自动化学院；

【摘要】 酸锌离子浓度是湿法炼锌电解工艺中的重要参数,直接影响着锌成品的质量。针对实际生产中电解液中酸锌离子浓度离线化验存在滞后的问题,提出了一种基于长短时记忆网络(long short-term memory,LSTM)的酸锌离子浓度预测方法。通过对氢-锌竞争模型与物料平衡模型的分析,提取关键特征序列,使用LSTM网络对多元特征序列与酸锌离子浓度之间的非线性关系进行动态时间建模并做出预测,采用中国湖南某冶炼厂的实测数据进行验证,分析结果表明所提预测方法的有效性。更多还原

【Abstract】 Zinc acid ion concentration is an important parameter in zinc electrolysis process, which directly affects the quality of zinc products.In order to solve the problem of lag in off-line test of zinc acid concentration in electrolytes in actual production, a method for predicting zinc acid concentration based on Long Short-Term Memory(LSTM) was proposed.Through the hydrogen-zinc competition model and the analysis of the material balance model, extract the key features of sequence, then use LSTM network characteristics of multiple sequences and the nonlinear relationship between the acid zinc ions concentration for dynamic modeling and forecasting time, validated using the measured data of a smelter of hunan province, analysis results show the effectiveness of the method.更多还原