基于概率密度函数的软测量方法及其在烟气含氧量中的应用

Probability Density Function Based Soft Sensor Technology and Its Application for Flue Gas Oxygen Content

现场采集的数据多为动态的生产数据,常规的数据预处理方法不能完全剔除数据中存在的各种有色噪声。针对该问题,提出了一种基于概率密度函数的烟气含氧量软测量方法,建立基于神经模糊系统的烟气含氧量软测量模型,将概率密度函数控制的思想引入到模型参数的辨识算法中,以克服有色噪声的影响。仿真结果表明,文中提出的新型软测量方法具有较高的建模精度,特别是在测量数据存在噪声的情况下,能够有效地解决烟气含氧量的预测问题。

The effective way to measure flue gas oxygen contents in thermal power plant is to adopt the soft sensor technology and depends on the quality of the measured data.However,the collected data are normally dynamic,and traditional method cannot completely remove the noise existed in the data.This paper proposes a probability density function based soft sensor technology for flue gas oxygen contents,builds a neural fuzzy based soft sensor model of the flue gas oxygen contents and employs the concept of the probability density function into the model parameter identification algorithm,which can overcome the affects of colored noises.The simulation results show that the proposed method has higher modeling accuracy,especially when the data is corrupted by noises.