| 张海传,刘钟阳,许东卫,王宁会.基于RBF神经网络模型的臭氧浓度软测量研究[J].,2010,(6):1020-1023 |
| 基于RBF神经网络模型的臭氧浓度软测量研究 |
| Research on software sensor of ozone concentration based on RBF neural networks model |
| |
| DOI：10.7511/dllgxb201006032 |
| 中文关键词: 臭氧浓度 径向基函数(RBF) 神经网络 软测量 |
| 英文关键词: ozone concentration radial basis function (RBF) neural networks software sensor |
| 基金项目:辽宁省自然科学基金资助项目(20031079)；教育部科学技术研究资助项目(重点项目104062). |
|
| 摘要点击次数: 1552 |
| 全文下载次数: 863 |
| 中文摘要: |
| 目前的臭氧浓度在线分析测试仪器在臭氧发生器中的应用受到很大限制．通过监测影响臭氧产生浓度的6个参量，基于RBF神经网络模型实现了臭氧浓度软测量．该模型采用梯度下降法确定RBF基函数的中心及输出层权值，可离线和在线校正所建立的神经网络模型．实验证明，软测量模型输出结果与臭氧浓度分析仪测量结果的绝对误差小于5 g/m 3的达93%以上，绝对误差小于1 g/m 3的达33%以上，且响应时间小于1 s． |
| 英文摘要: |
| The ozone concentration on-line analyzer apparatus is limited in the application of ozone generator greatly. The software sensor of the ozone concentration based on RBF neural networks model is presented with measuring six parameters which influence the ozone concentration. A gradient descent algorithm is introduced to decide the center of the RBF function and the weights of output layer in the model. It can be calibrated off-line or on-line. The experimental results prove that the absolute error of more than 93% is less than 5 g/m 3 and more than 33% is less than 1 g/m 3 between the output of model and the result measuring with the ozone concentration on-line analyzer, and the response time is less than 1 s. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |