摘要
针对新发展的管式连续结晶器和生产工艺,提出一种基于高斯过程回归(Gaussian Process Regression,GPR)的管式结晶器温度分布区间建模方法。通过测量DN15管式结晶器4个管段区间的温度,构建基于高斯过程回归的各区间温度分布模型。而且,建立针对不同夹套流速操作条件的连续结晶管段温度分布预测模型,采用北方苍鹰优化算法确定模型中超参数,以提高模型预测准确性。通过L-谷氨酸连续结晶过程在不同夹套流速下的各管段温度分布测试实验,验证了本研究提出方法的有效性与优点。
For the newly developed tubular continuous crystallizer and production process,this paper proposes a modeling method for the temperature profile of the tubular crystallizer based on Gaussian Process Regression(GPR).By measuring the temperatures in four sections of the DN15 tubular crystallizer,temperature profile models for each section are constructed based on GPR.Moreover,prediction models are established for the temperature distribution of continuous crystallization sections under different jacket flow rate operating conditions.Meanwhile,the northern goshawk optimization algorithm is adopted to determine the hyperparameters of these models,aiming to improve the prediction accuracy.Through experimental tests on the temperature distribution of each section during the L-glutamic acid continuous crystallization process under different jacket flow rates,the effectiveness and advantages of the proposed method are verified.
作者
李元鋆
赵洺延
宋博
刘涛
LI Yuanjun;ZHAO Mingyan;SONG Bo;LIU Tao(School of Control Science and Engineering,Dalian University of Technology,Liaoning,Dalian 116024,China)
出处
《化学工业与工程》
CSCD
北大核心
2024年第6期169-175,共7页
Chemical Industry and Engineering
基金
国家自然科学基金面上项目(62173058)
国家重大科研仪器研制项目(62327807)
关键词
管式连续结晶器
温度分布
预测模型
高斯过程回归
智能优化算法
continuous tubular crystallizer
temperature distribution
prediction model
gaussian process regression
intelligent optimization algorithm
