查询结果:   夏祥礼,陈国彬,刘超.基于反学习飞蛾火焰算法优化的LSSVM模型及其软测量应用[J].计算机应用与软件,2019,36(5):322 - 326.
中文标题
基于反学习飞蛾火焰算法优化的LSSVM模型及其软测量应用
发表栏目
算法
摘要点击数
289
英文标题
OPTIMIZED LSSVM AND ITS SOFT SENSING APPLICATION BASED ON OPPOSITION-BASED LEARNING MOTH FLAME ALGORITHM
作 者
夏祥礼 陈国彬 刘超 Xia Xiangli Chen Guobin Liu Chao
作者单位
重庆工商大学融智学院 重庆 401320 贵州航天电器股份有限公司 贵州 贵阳 550009    
英文单位
Rongzhi College of Chongqing Technology and Business University, Chongqing 401320, China Guizhou Aerospace Electronics Co., Ltd., Guiyang 550009, Guizhou, China    
关键词
软测量 最小二乘支持向量机 飞蛾火焰算法 反学习 越界
Keywords
Soft sensing LSSVM Moth-flame optimization algorithm Opposition-based learning Out of bound
基金项目
重庆市物联网产业共性关键技术创新主题专项项目(cstc2015zdcy-ztzx40007)
作者资料
夏祥礼,讲师,主研领域:智能信息处理,机器学习,智能优化。陈国彬,副教授。刘超,高工。 。
文章摘要
针对最小二乘支持向量机LSSVM(least squares support vector machine)软测量模型参数难以估计问题,提出将参数估计转化为约束优化问题,基于反学习飞蛾火焰算法OMFO(opposition-based MFO)优化的LSSVM建模技术,并构建OMFO-LSSVM软测量模型。在MFO(moth-flame optimization)基础上增加新型反学习策略以提升算法性能。针对越界飞蛾,采用一种镜像越界策略保证飞蛾均在维度范围内,改善种群多样性。利用OMFO算法调整模型参数,并建立OMFO-LSSVM软测量模型。将OMFO-LSSVM模型用于机组热耗率预测,预测精度达到0.11%,验证了该模型的可行性与优越性。
Abstract
In order to solve the problem of parameter estimation in the LSSVM soft sensing model, we transformed the parameter estimation into a constrained optimization problem. We proposed an optimization LSSVM modeling technique based on opposition-based learning moth flame algorithm(OMFO) and constructed an OMFO-LSSVM soft sensing model. A new opposition-based learning strategy was added to improve the performance of the algorithm. For cross-border moths, a mirror cross-border strategy was adopted to ensure that moths were within the dimension range and improved population diversity. Then we the used OMFO algorithm to adjust the model parameters, and the OMFO-LSSVM soft sensing model was established. The OMFO-LSSVM model is used to predict the heat rate of the unit, and the prediction accuracy reaches 0.11%, which verifies the feasibility and superiority of the model.
下载PDF全文