Study On The On-line Diagnosis And Data Reconcilation Technology Of Heating System Perception Information

With the process of information age speeding up,the operation and management of heating system rely more and more on heating data.At present,the application of data reconciliation is widely used in the chemical industry while little in the heating field.In this paper,some researches of the on-line diagnosis and data reconciliation of sensor upload data in heating system are carried out as follows.First of all,the upload data of actual heating system was collected.After analyzing and summarizing the characteristics of the sensor upload data,the abnormal phenomena were listed while observing the upload measurement data from sensors,as well as the relationship among temperature,pressure and flow these three variables.Secondly,to detect the gross error,three kinds of methods were put forward,which are wavelet filtering,median filtering and BP neural network.The results of these three methods detecting the gross error were compared by the 2F measure evaluation index.Finally,the correction process of upload sensor measurement data was put forward in heating system.After the meters' system error influence on the upload data was eliminated,median filtering method was used to detect the gross error of measurement data.Among the detected gross error,the abnormal value caused by the sensor reason was filtered out and corrected.In the course of the research,the main conclusions are as follows:Among the measurement variables upload by sensors of heating system,a corresponding relationship was found up between the variables,such as the flow in heat source and primary network,the instantaneous flow in secondary network and the water temperature in primary network,the instantaneous flow and the water temperature in secondary network,the instantaneous flow and the water pressure in secondary network,and the flow and the pressure difference.When detecting the gross error in the data measured by sensors,the median filter worked better than the wavelet filter and the BP neural network.The system error of instrument can be detected successfully by moving-window method,and the influence of instrument system error on the measurement data can be eliminatedIn dealing with the dramatic high and low flow data detected,the "abnormal data" caused by heating system operation were screened and filtered according to the relationship between the pressure difference and temperature difference of supply and return water in the primary net.Using the data replacement method,the medi an filtered data is corrected as the correction value.