摘要
针对火电厂在线SO2浓度检测中,检测精度受到温度、压力(大气压及烟气压力)、燃煤质量、水分含量、电子器件噪声、光学镜片老化、气体吸收峰值交叉干扰等多种因素的干扰,很难以单一方法进行改进这一问题。以国内某中型火电厂2015年实际生产数据为依据,设计预处理装置完成气体的前期处理,以尽可能达到分析仪分析要求(温度、流量、含水量等),减少可预见干扰,采用Adaboost算法集成BP神经网络进行优化,降低其他因素对检测结果的干扰,仿真测试分析,证明了该方法的有效性。
Aiming at problem of online S02 concentration detection in power plant, detection precision may be interfered by all kinds of factors ,such as temperature ,pressure (which includes atmosphere and smoke pressure), coal mass, moisture content, noise of electric devices, optical lens aging, interference at spectral of gas absorption, it is difficult to improve by a single way. According to real production data of a large thermal power plant in 2015, a kind of gas pretreatment device is developed in order to meet the requirements of analyzer and reduce predictable interference. Adopt Adaboost algorithm to optimize integrated BP neural network in order to avoid unpredictable interference. By simulation, test and analysis, it is verified that this method is effective.
出处
《传感器与微系统》
CSCD
2016年第9期148-151,共4页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(61502063)
重庆市教委科研项目(KJ1500639)