| 改进人工蜂群算法优化的LSSVM在混合气体定量分析中的应用 |
| |
| 引用本文： | 李成兵,叶超,毛熙皓. 改进人工蜂群算法优化的LSSVM在混合气体定量分析中的应用[J]. 工程设计学报, 2020, 27(1): 94-102. DOI: 10.3785/j.issn.1006-754X.2020.00.015 |
| |
| 作者姓名： | 李成兵 叶超 毛熙皓 |
| |
| 作者单位： | 西南石油大学 机电工程学院，四川成都610500 |
| |
| 摘 要： | 针对在易燃易爆混合气体定量分析中因交叉敏感易产生测量误差以及最小二乘支持向量机（least squares support vector machine，LSSVM）参数难以确定的问题，提出一种改进人工蜂群（improved artificial bee colony，IABC）算法优化的最小二乘支持向量机。首先，在标准人工蜂群（artificial bee colony, ABC）算法中引入自适应递减因子以更新步长，并结合轮盘赌和反向轮盘赌改进待工蜂跟随概率公式，从而提高收敛精度；然后，利用改进后的人工蜂群算法对最小二乘支持向量机的惩罚参数C和核参数σ2进行优化；最后，利用优化后的参数重建最小二乘支持向量机定量分析模型，并与利用常用的混合气体定量分析方法——粒子群优化（particle swarm optimization，PSO）算法优化的最小二乘支持向量机定量分析模型进行对比。实验结果表明，在交叉敏感状态下，采用改进人工蜂群算法优化的最小二乘支持向量机时的建模总时间和各组分气体浓度测量的平均相对误差均低于采用粒子群算法优化的，有效提高了混合气体的浓度测量精度。研究表明，改进人工蜂群算法优化的最小二乘支持向量机可为混合气体定量分析提供理论支撑，具有一定的工程应用价值。
|
| 收稿时间： | 2020-02-28 |
Application of LSSVM optimized by improved artificial bee colony algorithm in quantitative analysis of mixture gas |
| |
| LI Cheng-bing,YE Chao,MAO Xi-hao. Application of LSSVM optimized by improved artificial bee colony algorithm in quantitative analysis of mixture gas[J]. Journal of Engineering Design, 2020, 27(1): 94-102. DOI: 10.3785/j.issn.1006-754X.2020.00.015 |
| |
| Authors: | LI Cheng-bing YE Chao MAO Xi-hao |
| |
| Abstract: | Due to the measurement errors caused by cross-sensitivity in the quantitative analysis of flammable and explosive mixture gas and the difficulty of parameters determination in least squares support vector machine (LSSVM), a LSSVM optimized by improved artificial bee colony (IABC) algorithm was proposed. Firstly, to improve the accuracy of convergence, the adaptive decreasing factor was introduced to update the step and the following probability formula of waiting worker bee was modified by combining the roulette and reverse roulette on the basis of the artificial bee colony (ABC) algorithm. Then, the penalty parameterC and the kernel parameter σ2 of the LSSVM were optimized by IABC algorithm. Finally, the quantitative analysis model of LSSVM was reconstructed by these optimized parameters, and compared with the quantitative analysis model of LSSVM optimized by particle swarm optimization (PSO) algorithm, which was a commonly used quantitative analysis method of mixture gas. The experimental results showed that under the cross-sensitive state, the modeling total time and the concentration measuring average relative error of each component gas with the LSSVM optimized by IABC algorithm were lower than those with the LSSVM optimized by PSO algorithm，which effectively improved the concentration measuring accuracy of the mixture gas. The research shows that the LSSVM optimized by IABC algorithm can provide theoretical support for quantitative analysis of mixture gas, and has certain engineering application value. |
| |
| Keywords: | |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《工程设计学报》浏览原始摘要信息 |
|
点击此处可从《工程设计学报》下载免费的PDF全文 |
|
