论文部分内容阅读
采用基于Mie理论的激光散射法测量催化剂粉尘浓度时,催化剂粉尘浓度与监测参数——入射光强、散射光强、出射光强以及烟气流量之间存在着复杂的非线性关系,给粉尘浓度的准确测梁带来困难。利用支持向量机优良的非线性映射和强大的泛化能力,建立了一个基于最小二乘支持向量机的催化剂粉尘浓度软测量模型,给出了相应的系统结构和算法,并通过网格搜索和交叉验证的方法对支持向量机进行参数选择。采用遗忘因子法和数据滑动时间窗技术对工业软测量模型进行在线校正,克服了工况条件发生改变时的估计偏差,提高了估计精度。仿真和实际运行结果表明基于LS-SVM的软测量模型具有较高的估算精度与泛化能力,为催化剂粉尘浓度的在线测量提供了一种简单、可靠的新方法。
When using the laser scattering method based on the Mie theory to measure the catalyst dust concentration, there is a complicated nonlinear relationship between the catalyst dust concentration and the monitoring parameters - the incident light intensity, the scattered light intensity, the outgoing light intensity and the flue gas flow rate, Accurate beam measurement difficulties. Based on the excellent nonlinear mapping of support vector machine and powerful generalization ability, a soft dust concentration measurement model based on least square support vector machine was established, and the corresponding system structure and algorithm were given. Through grid search and Cross validation method to support vector machine parameter selection. The forgetting factor method and the data sliding time window technique are used to calibrate the industrial soft measurement model online to overcome the estimated deviation when the working conditions change and improve the estimation accuracy. The simulation and actual operation results show that the LS-SVM-based soft-sensing model has higher estimation accuracy and generalization ability, and provides a simple and reliable new method for on-line measurement of catalyst dust concentration.