针对生化过程软测量建模过程中样本数据可能包含的测量误差对模型性能的影响,提出一种自适应加权最小二乘支持向量机(Adaptive weighted least squares support vector machine,AWLS-SVM)回归的软测量建模方法。该方法基于最小二乘支持向量机模型,根据样本拟合误差,并结合改进的正态分布赋权规则,自适应地为每个建模样本分配不同的权值,以降低随机误差对模型性能的影响;同时采用混沌差分进化—模拟退火(Chaos differential evolution simulated annealing,CDE-SA)算法对模型参数进行优化选择,以提高模型的泛化能力。仿真实验表明,AWLS-SVM模型的预测精度及鲁棒性能优于最小二乘支持向量机(Least squares support vector machine,LS-SVM)和加权最小二乘支持向量机(Weighted least squares support vector machine,WLS-SVM)。利用Pensim仿真平台的数据,将AWLS-SVM方法用于青霉素发酵过程软测量建模,获得了较好的效果。
【英文摘要】
The presence of outliers in sample data can corrupt the model's performance,giving undesirable results.A novel adaptive weighted least squares support vector machine(AWLS-SVM) regres-sion method is presented for modeling of penicillin fermentation process.In AWLS-SVM,least square support vector machine regression is employed for the sample data to develop model and obtain the sample datum fitting error.According to the fitting error,the adaptive sample weights are obtained via the proposed improved normal d...