网格搜索法改进SVM的出水COD软测量建模
|
摘要
针对污水处理过程出水水质参数难以在线检测这一问题,提出一种基于网格搜索法优化SVM的软测量模型.根据污水处理过程的输入输出数据,选取进水COD、进水NH3-N、溶解氧(DO)作为模型输入,出水COD作为模型输出,采用SVM方法建立出水COD的软测量模型,同时采用网格搜索法对SVM的惩罚因子和核函数参数进行寻优操作,提高软测量模型的预测精度.仿真结果表明,基于网格搜索法优化的SVM软测量模型具有良好的预测精度和泛化能力,应能满足处理过程的需求.
In view of the difficulty of on-line detection of effluent quality parameters in sewage treatment process, a soft sensor model based on grid search method was proposed to optimize SVM. According to the input and output data of the sewage treatment process, the input of the in-fluent COD, the in-fluent NH_3-N and the dissolved oxygen(DO) was used as the model input, and the effluent COD was the model output. The soft measurement model of the effluent COD was established by the SVM method, and the grid search method was used to optimize the SVM penalty factor and the kernel function parameters to improve the prediction accuracy of the soft sensing model. The simulation results show that the SVM soft sensor model based on grid search method has good prediction accuracy and generalization ability, and can meet the actual needs of the sewage treatment process.
In view of the difficulty of on-line detection of effluent quality parameters in sewage treatment process, a soft sensor model based on grid search method was proposed to optimize SVM. According to the input and output data of the sewage treatment process, the input of the in-fluent COD, the in-fluent NH_3-N and the dissolved oxygen(DO) was used as the model input, and the effluent COD was the model output. The soft measurement model of the effluent COD was established by the SVM method, and the grid search method was used to optimize the SVM penalty factor and the kernel function parameters to improve the prediction accuracy of the soft sensing model. The simulation results show that the SVM soft sensor model based on grid search method has good prediction accuracy and generalization ability, and can meet the actual needs of the sewage treatment process.
引文
[1]宋剑杰.污水处理过程COD指标集成软测量模型[J].计算机工程与应用, 2012, 48(17):243-248.
[2]徐方舟,潘丰.基于PSO-LSSVM污水处理系统出水数据的软测量[J].江南大学学报(自然科学版), 2010, 9(3):253-256.
[3]郭美丽.基于GA-SVR和改进的网格搜索SVR的话务量预测研究[D].乌鲁木齐:新疆大学, 2013.
[4]王兴玲,李占斌.基于网格搜索的支持向量机核函数参数的确定[J].中国海洋大学学报(自然科学版), 2005(5):859-862.
[5]胡长鹏,洪晓斌,刘桂雄,等.基于软测量技术的液相乙醇浓度在线检测系统[J].科学技术与工程, 2008(5):1183-1187.
[6]王永智,付忠广,刘炳含,等.基于GA-SVM的烟气含氧量软测量[J].化工自动化及仪表, 2017, 44(10):937-939.
[7]]李荣雨,戚桂洪.基于CSSE-OSELM算法的软测量建模及其工业应用[J].计量学报, 2017, 38(5):650-655.
[8]李明河,王健,孟凡许.基于支持向量机α阶逆系统方法的污水处理内模控制[J].控制工程, 2016, 23(2):185-189.
[9]宋夫华,李平.基于支持向量机α阶逆系统方法的非线性内模控制[J].自动化学报, 2007, 33(7):778-781.
[10]王定成.支持向量机建模预测与控制[M].北京:气象出版社,2009.
[11]郭超,宋卫华,魏威.基于网格搜索-支持向量机的采场顶板稳定性预测[J].中国安全科学学报, 2014, 24(8):31-36.
[12]乔志刚.基于网格搜索和支持向量机的矿用风机故障诊断研究[J].中州煤炭, 2015(2):81-83.
[13]亢生彩.网格搜索法SVM参数优化在主扇风机故障诊断中的应用[J].煤炭技术, 2015, 34(1):295-297.
[14]刘佳,施龙青,韩进,等.基于Grid-Search_PSO优化SVM回归预测矿井涌水量[J].煤炭技术, 2015, 34(8):184-186.
[15]张公让,万飞.基于网格搜索的SVM在入侵检测中的应用[J].计算机技术与发展, 2016, 26(1):97-100.
[2]徐方舟,潘丰.基于PSO-LSSVM污水处理系统出水数据的软测量[J].江南大学学报(自然科学版), 2010, 9(3):253-256.
[3]郭美丽.基于GA-SVR和改进的网格搜索SVR的话务量预测研究[D].乌鲁木齐:新疆大学, 2013.
[4]王兴玲,李占斌.基于网格搜索的支持向量机核函数参数的确定[J].中国海洋大学学报(自然科学版), 2005(5):859-862.
[5]胡长鹏,洪晓斌,刘桂雄,等.基于软测量技术的液相乙醇浓度在线检测系统[J].科学技术与工程, 2008(5):1183-1187.
[6]王永智,付忠广,刘炳含,等.基于GA-SVM的烟气含氧量软测量[J].化工自动化及仪表, 2017, 44(10):937-939.
[7]]李荣雨,戚桂洪.基于CSSE-OSELM算法的软测量建模及其工业应用[J].计量学报, 2017, 38(5):650-655.
[8]李明河,王健,孟凡许.基于支持向量机α阶逆系统方法的污水处理内模控制[J].控制工程, 2016, 23(2):185-189.
[9]宋夫华,李平.基于支持向量机α阶逆系统方法的非线性内模控制[J].自动化学报, 2007, 33(7):778-781.
[10]王定成.支持向量机建模预测与控制[M].北京:气象出版社,2009.
[11]郭超,宋卫华,魏威.基于网格搜索-支持向量机的采场顶板稳定性预测[J].中国安全科学学报, 2014, 24(8):31-36.
[12]乔志刚.基于网格搜索和支持向量机的矿用风机故障诊断研究[J].中州煤炭, 2015(2):81-83.
[13]亢生彩.网格搜索法SVM参数优化在主扇风机故障诊断中的应用[J].煤炭技术, 2015, 34(1):295-297.
[14]刘佳,施龙青,韩进,等.基于Grid-Search_PSO优化SVM回归预测矿井涌水量[J].煤炭技术, 2015, 34(8):184-186.
[15]张公让,万飞.基于网格搜索的SVM在入侵检测中的应用[J].计算机技术与发展, 2016, 26(1):97-100.