基于QPSO组合优化的发酵过程LS-SVM建模

LS-SVM modeling for fermentation process based on QPSO combinatorial optimization

利用最小二乘支持向量机(LS-SVM)对发酵过程进行建模,辅助变量和模型参数的选择对建模效果有很大影响。因此提出了一种基于量子粒子群算法(QPSO)的组合优化建模方法,构造基于赤池信息量准则(AIC)的适应度函数,利用QPSO同步选择最优的辅助变量组合和参数对,对模型进行自动优选。将该方法用于诺西肽发酵过程的建模,仿真结果表明,通过QPSO组合优化能获得更好的建模效果。

The selection of instrumental variables and parameters has an important impact on least square support vector machine（LSSVM） model performance in fermentation process.A new combinatorial optimization method based on quantum-behaved particle swarm optimization（QPSO） is proposed to solve this problem.A fitness function based on Akaike information criterion（AIC） is constructed, and thenQPSO is applied to select the optimal combination of input variables and couple of parameters simultaneously.The result ofmodeling simulation of Nosiheptide fermentation process shows that this combinatorial optimization method can get better model performance.