论文部分内容阅读
在分析混沌粒子群优化算法(CPSO)和最小二乘支持向量机(SVM)理论基础上,以某污水处理厂的氧化沟系统为对象,采用带有末位淘汰机制的混沌粒子群优化算法优化支持向量机的参数,建立了基于变异CPSO算法的LS-SVM的氧化沟出水水质COD软测量模型,并与PSO-LSSVM,LSSVM模型比较,研究表明,ICPSO-LSSVM模型预测准确,泛化性能好,且该模型预测结果中相对误差小于10%的样本达到90%,最大相对误差仅为12.5%,均方差MSE为0.0106,模型具有较高的精度,基本可以实现出水COD浓度的在线预估。
Based on the analysis of chaos particle swarm optimization (CPSO) and least square support vector machine (SVM) theory, aiming at the oxidation ditch system of a wastewater treatment plant, the chaotic particle swarm optimization algorithm with the last elimination mechanism is used to optimize SVM model of oxidation ditch effluent quality based on variant CPSO algorithm was established and compared with the PSO-LSSVM and LSSVM models. The results show that the ICPSO-LSSVM model has accurate prediction and good generalization performance , And the model predicts that the relative error of less than 10% of the samples to 90%, the maximum relative error of only 12.5%, the mean square error MSE of 0.0106, the model has a high accuracy, the basic output of COD concentration can be estimated online.