论文部分内容阅读
公开号:CN103884670A公开日:2014.06.25申请人:西安交通大学摘要一种基于近红外光谱的烟气成分定量分析方法,采集火力发电厂烟气信号构成现场历史数据库,数据库包括火电厂烟气在近红外光的各个波长的吸收率和各成分浓度,采用异常点检测将异常样本从数据库中剔除,得到新数据库,利用猫群聚类方法将新数据库进行聚类划分,建立烟气成分定量分析的预测模型并对模型参数进行训练,最终通过所得到
Publication number: CN103884670A Publication date: 2014.06.25 Applicant: Xi’an Jiaotong University Abstract A method for quantitative analysis of flue gas composition based on near infrared spectroscopy is proposed. The flue gas signal from the thermal power plant is collected to form the on-site historical database. The database includes flue gas from thermal power plant Absorbance of each wavelength of near-infrared light and the concentration of each component, Abnormal point detection will be used to remove abnormal samples from the database to obtain a new database, using cat clustering method to cluster the new database to establish a quantitative analysis of flue gas composition The prediction model and model parameters training, and finally obtained by