摘要
针对船舶舱室温湿度保持困难、数据难以预测的问题,提出了基于克隆选择算法的支持向量机集成方法。首先,利用克隆选择算法优化个体支持向量机,根据个体预测误差进行自适应集成;然后,对舱室温湿度时间序列数据样本化,采用支持向量机集成进行训练、测试;最后通过统计测试结果以及与BP神经网络、单支持向量机、GM(2,1)模型的预测误差对比发现,支持向量机集成模型可有效预测空调故障条件下船舶舱室温湿度的变化规律,为装备的使用和维护提供技术支持。
For challenges in maintaining cabin temperature and humidity,as well as data predicting,ensemble of support vector machine(ESVM)based on clonal selection algorithm(CSA)was proposed.Firstly,individual SVMs were optimized by CSA,and then the cabin temperature and humidity time data series were sampled.Lastly,ESVM was used for training and testing.finally,Statistical testing results and comparison of prediction errors with BP neural network,individual SVM and GM(2,1)models show that the ESVM model can effectively predict the changes in humidity and temperature in submarine cabins under air conditioning fault conditions.The method provides a technical support for the use and maintenance of equipment.
作者
刘丙杰
侯慕馨
冀海燕
LIU Bingjie;HOU Muxin;JI Haiyan(Navy Submarine Academy,Qingdao 266199,China)
出处
《海军工程大学学报》
CAS
北大核心
2024年第3期21-25,32,共6页
Journal of Naval University of Engineering
基金
国家社会科学基金资助项目(2021-SKJJ-B-011)。
关键词
支持向量机集成
船舶舱室
温湿度预测
ensemble of SVM
ship cabin
prediction of temperature and humidity
