【摘要】：針對(duì)甲烷氣體定量分析過程中,傳統(tǒng)SVM模型預(yù)測(cè)精度低、收斂速度慢等問題,提出了一種基于改進(jìn)PSO算法的SVM回歸模型。該模型在傳統(tǒng)PSO算法尋優(yōu)的基礎(chǔ)上,引入動(dòng)量項(xiàng)的同時(shí)增加隨機(jī)粒子個(gè)體極值的追隨因子,使粒子不僅追隨全局最優(yōu)解和局部最優(yōu)解,還跟隨種群中任一粒子的個(gè)體極值,使得尋優(yōu)算法后期收斂速度較快,不易陷入局部最小值。實(shí)驗(yàn)中,對(duì)0~5.05%濃度的25組標(biāo)準(zhǔn)甲烷樣氣進(jìn)行建模分析,并與傳統(tǒng)PSO算法尋優(yōu)模型和Grid搜索法尋優(yōu)模型進(jìn)行對(duì)比。結(jié)果表明,采用改進(jìn)PSO算法建立的SVM回歸模型均方根誤差小,收斂速度快。
[Abstract]:In order to solve the problems of low prediction accuracy and slow convergence rate in the process of methane gas quantitative analysis, a new SVM regression model based on improved PSO algorithm is proposed. On the basis of the traditional PSO algorithm, the momentum term is introduced and the following factor of individual extremum of random particle is added, so that the particle not only follows the global optimal solution and the local optimal solution, but also follows the individual extremum of any particle in the population. The convergence speed of the optimization algorithm is fast and it is not easy to fall into the local minimum. In the experiment, 25 groups of standard methane sample gas with 0 5. 05% concentration were modeled and analyzed, and compared with the traditional PSO algorithm and Grid search optimization model. The results show that the root-mean-square error of the SVM regression model based on the improved PSO algorithm is small and the convergence rate is fast.
【作者單位】： 中國(guó)計(jì)量大學(xué)機(jī)電工程學(xué)院;
【基金】：浙江省大學(xué)生科技創(chuàng)新活動(dòng)計(jì)劃暨新苗人才計(jì)劃項(xiàng)目(省級(jí))(2016R409)
【分類號(hào)】：O212.1;TD712.5
，

本文編號(hào)：2318344