论文部分内容阅读
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。
In order to detect the online SO2 concentration in thermal power plants, the detection accuracy is affected by many factors such as temperature, pressure (atmospheric pressure and flue gas pressure), coal quality, moisture content, electronic device noise, optical lens aging and gas absorption peak cross interference, It is difficult to improve this issue in a single way. Based on the actual production data of a medium-sized thermal power plant in China, the pretreatment device was designed to complete the gas pre-treatment to achieve the analyzer’s analysis requirements (temperature, flow, water content, etc.) as much as possible to reduce the foreseeable interference. The Adaboost algorithm The integrated BP neural network is optimized to reduce the interference of other factors on the test results. The simulation test analysis proves the effectiveness of the method.