Analysis On Power Plant Data Pre-processing Based On Steady State Detection

Steady-state detection is very important to the performance evaluation of the equipment in the thermal process,the modeling and optimization of the system process,the establishment of the fault detection mechanism and the process identification.In the thermal power plant mass historical data preprocessing process,only the steady state data can reflect the system objectively,and in the future system modeling and identification to obtain steady-state conditions,so the steady-state conditions is important to choose the variables from the massive database.SSD(Steady State Detection)algorithm has been widely used as an effective steady-state detection algorithm.Combining SSD with data filtering can greatly improve the accuracy of steady-state judgment of monitoring points.Combining EWMA filtering with SSD algorithm,the EWMA-SSD method is formed,which is the object of this paper.In this paper,we first analyze the mechanism of SSD and EWMA filtering for steady-state detection,and deduce the equivalence relation between window width n and EWMA filter factor ? in SSD algorithm.Considering the uncertainties in the distribution of monitoring data in the actual working conditions,based on the SSD algorithm,the improvement of the control limit and the monitoring index are made.Based on the nonparametric control limit algorithm of kernel function,the probability density function of monitoring index is fitted by kernel function.Then,the control limit of test water ? is obtained under the probability density function,and steady-state detection of the process data can be carried out after the monitoring index is improved by EWMA filter..In the end,the SSD steady-state detection method is applied to the actual process.In this paper,the steady-state detection of a 600 MW unit of a thermal power plant is carried out,and the improved SSD algorithm is more traditional.The validity of the method is verified.Finally,the steady-state data segments under similar loads after multivariable steady-state detection are screened out.After cluster analysis of the two variables involved,it is found that the steady-state conditions can be divided under different load conditions.The analysis of the state and economy of the operation and the optimization of the operation of the analysis showed that there were significant differences in the boiler efficiency of different clusters,which provided the basis for the follow-up research on this difference.