燃煤电厂二氧化硫排放质量浓度的软测量技术  

Soft-sensing Technology of SO_2 Emission for Coal-fired Power Plants

作　　者：郑海明[1] 杨志[2] 

机构地区：[1]华北电力大学能源动力与机械工程学院,保定071003 [2]邢台技师学院,邢台054001

出　　处：《动力工程学报》2013年第2期130-134,共5页Journal of Chinese Society of Power Engineering

基　　金：国家自然科学基金资助项目(11174078);河北省自然科学基金资助项目(E2012502046);中央高校基本科研业务费专项资金资助项目(12MS102)

摘　　要：针对电厂烟气污染物SO2监测中出现的安装费用高、维护困难等问题,提出了一种烟气SO2排放量预测方法.首先应用BP神经网络建立模型,然后应用遗传算法对BP神经网络的连接权值和阈值进行优化配置,建立新的软测量模型,并对改进后的软测量模型的预测结果与实际运行数据进行了比较,结果表明:所建立的新模型具有较高的准确性和稳定性,可以较好地预测电厂烟气SO2排放质量浓度.To solve the problem of high installation and maintenance cost of continuous emission monitoring system （CEMS） for flue gas pollutants from coal-fired power plants, a soft sensing technology is proposed for prediction of the SO2 emission. The specific way is to firstly build up a model using BP neural network, then configure and optimize the link weights and threshold value of BP network using genetic algorithm, and finally establish a new soft-sensing model. Predicted results based on the new soft sensing model were compared with actual measurements. Results show that the new model has a high prediction accuracy and stability, which may be used to predict SO2 emission from coal-fired power plants.

关 键 词：SO2 排放监测 软测量技术 BP神经网络 遗传算法 

分 类 号：TK314[动力工程及工程热物理—热能工程] X831[环境科学与工程—环境工程]