一种用于气候室相对湿度预测的MSPOA-LSSVM模型研究  

Study on MSPOA-LSSVM model for predicting relative humidity in climate chamber

作　　者：王一诺 郑焕祺 杨胜坤 周玉成 WANG Yinuo;ZHENG Huanqi;YANG Shengkun;ZHOU Yucheng(School of Information and Electrical Engineering,Shandong Jianzhu University,Jinan 250101,China;School of Architecture and Urban Planning,Shandong Jianzhu University,Jinan 250101,China;National Center of Quality Inspection and Test for Decoration Materials,Jinan 250102,China)

机构地区：[1]山东建筑大学信息与电气工程学院,济南250101 [2]山东建筑大学建筑城规学院,济南250101 [3]国家装饰装修材料质量检验检测中心,济南250102

出　　处：《重庆理工大学学报(自然科学)》2025年第2期97-105,共9页Journal of Chongqing University of Technology:Natural Science

基　　金：泰山学者优势特色学科人才团队项目(2015162);山东建筑大学博士基金项目(X21110Z);山东省产品质量检验研究院院储备项目(2024ZJKY0008)。

摘　　要：针对通风条件下,气候室相对湿度控制精度对甲醛检测准确性的影响,提出一种相对湿度预测模型。模型选取控温水箱、控制露点水箱和气候室相对湿度等7个数据采集点的数据作为输入和输出。基于多策略改进鹈鹕优化算法和最小二乘支持向量机构建MSPOA-LSSVM相对湿度预测模型。针对鹈鹕优化算法寻优能力不足的问题,使用随机对立学习初始化种群,引入融合鲸鱼优化的正余弦策略和动态权重因子策略,提高算法性能。将MSPOA-LSSVM模型与4种机器学习模型进行对比实验,结果表明,MSPOA-LSSVM模型决定系数、均方根误差分别为0.964和0.07389,均低于其他模型,可为解决相对湿度控制精度不足问题提供参考。Considering the significant impact of the relative humidity control accuracy of the full-scale chamber under ventilation conditions on the accuracy of formaldehyde detection,we propose a relative humidity prediction model to provide the necessary support for the development of the relative humidity control strategy.The model selects data from seven data collection points,such as temperature control water tank,control dew point water tank and full-scale chamber relative humidity,as input and output.The MSPOA-LSSVM relative humidity prediction model is built based on the multi-strategy improved pelican optimization algorithm and the least squares support vector mechanism.To address the problem that the pelican optimization algorithm lacks the ability to find the optimum,random opposition learning is used to initialize the population,and the sine-cosine strategy incorporating whale optimization and dynamic weight factor strategy are introduced to improve the algorithm performance.Comparison experiments of the MSPOA-LSSVM model and five machine learning models are conducted.Our results reveal the coefficient of determination and root mean square error of the MSPOA-LSSVM model are 0.964 and 0.07389 respectively,which are lower than those of the other models.Our model provides a viable solution to address the low control accuracy in relative humidity.

关 键 词：气候室 相对湿度预测 鹈鹕优化算法 最小二乘支持向量机 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]