| With the strict requirements for controlling the emission of mercury pollutants in flue gas from coal-fired power plants,it is particularly important to monitor the different forms of mercury concentration in flue gas accurately and in real time.Based on the online monitoring method of cold vapor atomic fluorescence spectrometry(CVAFS),it will gradually take the place of off-line test methods for its simple operation,high measurement accuracy and fully automatic control.The mercury online monitoring instruments,with high measurement accuracy,strong adaptability and low operation cost,have become the development trend in this field.This paper starts with the signal processing of the photomultiplier output in mercury online monitor and the design of the control system of the whole instruments,and then carries out related research work.During the detection of mercury concentration in the mercury analyzer,the noise of its own and the corresponding signal conversion circuit will lead to the final concentration of mercury having a big deviation.In order to solve this problem,a mathematical model of voltage peak based on piecewise polynomial fitting(PPF)is proposed based on the voltage peak signal from PMT in mercury analyzer and validated its rationality.According to the established PPF model and the noise characteristics of the signal,the corresponding state-space equation is obtained and the signal is filtered using the extended Kalman filter algorithm.In order to verify the applicability of the model,the established PPF model is used to deal with other noisy voltage peaks.The simulation results show that the voltage peak fitted by the PPF model is the closest to the true values,its RMSE and MAE are respectively 51.79%and 53.13%of the global polynomial fit(GPF)model,which are respectively 9.00%and 11.45%of that of the Gaussian model.The extended Kalman filter based on the PPF model has a significant filtering effect on a certain peak voltage of a noisy voltage.This model is used to process other voltage peaks and the RMSE and MAE of the filtered voltage peaks are around 1,which shows the extended Kalman filter algorithm based on this model has strong adaptability in dealing with various voltage peaks output by PMT.In the process of detecting mercury concentration,the output voltage peak signal noise statistical characteristics will change with the changes in working conditions,it leads to the conventional extended Kalman filter algorithm being poor to filter the signal,so a fuzzy extended Kalman filter(FEKF)algorithm is proposed.The algorithm adjusts the model variance matrix of system noise and observed noise through the established fuzzy controller according to the mean and covariance of the observed new information,so that the Kalman filter can adaptively filter the noisy voltage peak signal.The simulation results show that the RMSE and MAE of the filtered voltage peak by the FEKF algorithm are reduced by 4.16%and 9.73%respectively compared with the conventional extended Kalman filter when the actual noise variance is less than the model noise variance.When the actual noise variance is greater than the model noise variance,the RMSE and MAE of the filtered voltage peak by the FEKF algorithm are reduced by 11.50%and 11.45%respectively compared with the conventional extended Kalman filter algorithm,which shows that the FEKF algorithm has strong applicability in dealing with model noise deviating from the actual working conditions.On the basis of the detailed analysis of the composition and working principle of the mercury online monitoring instrument,the lower machine is designed by Siemens STEP7 programming software,and the upper machine is designed by WinCC configuration software,at the same time,the extended Kalman filter and its improved algorithm were applied to the system.By the online simulation debugging between the upper machine and the lower machine,the designed control system meets the basic technical requirements.The control system is an important exploration in the development of mercury online monitoring instrument,providing a reference for further research and development in the future. |
PDF Full Text Request