INGOA结合LSTM的自来水混凝投药预测模型  

INGOA combined with LSTM for prediction model of tap water coagulation dosage

作　　者：张庭源 张长胜[1] 张健忠 田海勇 毛辉 丁鑫 ZHANG Ting-yuan;ZHANG Chang-sheng;ZHANG Jian-zhong;TIAN Hai-yong;MAO Hui;DING Xin(Faculty of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650500,China;Yunnan Shuye Technology Co.,Ltd.,Kunming 650032,China;Kunming Branch of North China Municipal Engineering Design and Research Institute Co.,Ltd.,Kunming 650051,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,昆明650500 [2]云南树业科技有限公司,昆明650032 [3]中国市政工程华北设计研究总院有限公司昆明分公司,昆明650051

出　　处：《兰州大学学报(自然科学版)》2025年第2期222-230,共9页Journal of Lanzhou University(Natural Sciences)

基　　金：国家自然科学基金项目(61963022,62173169)。

摘　　要：为解决自来水厂混凝投药量的控制问题,提出一种基于改进的北方苍鹰优化算法(INGOA)结合长短期记忆神经网络(LSTM)的混凝投药预测模型.为均衡NGOA的全局勘探和局部开发能力,根据种内竞争机制,加强算法的局部持续开发能力,引入柯西-高斯变异增加算法在局部最优值及其邻域附近的逃逸能力,提升算法的全局优化能力;利用基于Bernoulli混沌的自适应权重加快算法收敛速率.用Pearson相关系数法降低模型的输入维度,通过INGOA优化LSTM网络,得到最优参数组合策略,建立INGOA-LSTM非线性混凝投药量预测模型.对某自来水厂的数据进行训练和测试,结果表明,与LSTM及传统的预测模型相比,INGOA-LSTM模型效果最佳,在各项指标中均取得较优的性能,均方根误差为82.2μg/L,平均绝对误差为13.9μg/L,平均绝对百分比误差为0.29%.To address the challenge of controlling coagulant dosage in water treatment plants,we presented a coagulant dosage prediction model that combines an improved northern goshawk optimization algorithm(INGOA)with long-short term memory neural network(LSTM).The NGOA enhanced local exploitation based on intraspecific competition,introduced Cauchy-Gaussian perturbation to improve escapement from local optima,and strengthened global optimization.Bernoulli chaos-based adaptive weights were employed to accelerate convergence;the Pearson correlation coefficient used to reduce the input dimensionality of the model,and the INGOA optimized the LSTM to determine the optimal parameter combination,leading to the development of the INGOA-LSTM nonlinear coagulation dosing prediction model.Upon training and testing with data from a water plant,the results indicated that the INGOALSTM model outperformed all comparison models optimized by LSTM and traditional prediction models,achieving a root mean square error of 82.2μg/L,a mean absolute error of 13.9μg/L,and a mean absolute percentage error of 0.29%.

关 键 词：混凝投药预测 长短期记忆神经网络 改进的北方苍鹰优化算法 种内竞争机制 Bernoulli混沌权重 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]