基于PCA-PSO-LSSVM的造纸废水厌氧处理系统出水COD软测量  

Soft-sensor modeling for paper mill effluent COD based on PCA-PSO-LSSVM

作　　者：孙军[1,2] 成忠 杨瑞芹[1,2] 单胜道 

机构地区：[1]浙江科技学院化工系,杭州310023 [2]浙江省农业生物资源生化制造协同创新中心,杭州310023 [3]浙江省废弃生物质循环利用与生态处理技术重点实验室,杭州310023

出　　处：《计算机与应用化学》2017年第9期706-710,共5页Computers and Applied Chemistry

基　　金：国家自然科学基金重点项目(U1609214);浙江省科技计划项目(2016C33105)

摘　　要：以某企业造纸废水厌氧处理系统为对象,基于装置运行及过程机理分析,结合专家经验知识,选择8个过程变量为输入变量,并借助现场传感器采集这些变量的工业运行数据,进而构建出水水质指标化学需氧量(COD)输出变量的PCA-PSO-LSSVM软测量模型:首先,选用主成分分析法(PCA)执行数据样本输入变量的预处理,以消除变量间的相关性,完成输入变量的降维和主成分提取;然后,实施主成分与出水COD间的最小二乘支持向量机(LSSVM)数据建模;考虑到LSSVM模型中核函数宽度和惩罚因子对模型性能有较大影响,再通过粒子群优化算法(PSO)完成上述两个参数的全局寻优;最后,将所建成的PCA-PSO-LSSVM软测量模型应用于未知样本数据的预测,得其均方根误差2.17%,极大误差4.19%。结果表明,本文所构建的软测量模型预测精度高,泛化性能与稳定性好,可为造纸废水厌氧出水COD在线预测及该处理系统的优化控制提供指导。With the rapid development of the paper industry, accompanied by pollutants severely influencing the environment, our country is paying more and more attention to the protection of water envirment, and strengthering scientific and technological research to solve the wastewater treatment. In this paper, a novel least squares support vector machines（LS-SVM） soft-sensor model is developed to infer the chemical oxygen demand（COD） of paper mill effluent from other process variables. According to the device operation and reaction mechanism, combined with expert experience, a total of 8 process variables have been chosen as the input variables to develop the COD prediction model. In order to reduce the dimensions of input variables, principal component analysis（PCA） was used to analyze the correlation of these variables. Meanwhile, particle swarm optimization（PSO） was used to obtain the optimal value of the important parameters（？ and？） in LSSVM model. Compared with PSO-LSSVM and LSSVM model, the experimental results showed that the PCA-PSO-LSSVM model was featured with some accuracy and better generalization ability. The maximum error and root mean square error（RMSE） were 4.19% and 2.17%, respectively. Moreover, the model has higher calculation speed. Therefore, the soft sensing model constructed in this paper can provide guidance for COD on-line prediction and optimal control of effluent from anaerobic digestion treatment of paper mill wastewater.

关 键 词：造纸废水 厌氧消化 化学需氧量 软测量模型 操作优化控制 

分 类 号：TE992.2[石油与天然气工程—石油机械设备] X703.1[环境科学与工程—环境工程]