Design And Research Of Automatic Observation System For Wire Icing

Wire icing has a profound impact on the normal operation of power network in China.Knowing the ice condition of wire icing timely attaches great importance for power departments to make a right decision of ice melting and avoid the ice disaster.This project designed a set of automatic observation system of wire icing based on the meteorological standard.On the one hand,it can meet the requirements of automatic observation of meteorological department and improve the observation efficiency.On the other hand,it can also provide important reference data for the power department to perform routine task of wire icing observation and location selection of wire grid layout.The system determines the weight of wire icing by measuring the weight of electric wires in both east-west direction and south-north direction automatically.It observes temperature,atmospheric pressure,wind speed and wind direction at the same time.The experiment sites are located at Niba mountain,Erlang mountain and Longwangmiao et.al in western Sichuan.In this paper,the data observed from different observation stations were processed and analyzed.The filtering processing of data was completed in this paper and the correlation between wire icing and the observed meteorological elements was discussed.Besides,the singularity analysis of wire icing weighing data was analyzed.Finally,the regression effect between wire icing and meteorological factors was discussed.Wavelet transform method,EKF and UKF were used in the filtering of the wire icing weighing data and the filtering effect of single filtering method was compared finally.It shows that the single filtering method can suppress the noise interference in the original signal well but the filtering effect of the wavelet filtering method and the UKF method are better than the EKF method.Then the wavelet filtering method was used to filter the data after the process of the UKF method again and the result shows that it has a better filtering effect than any one of the single filtering method in the 4 filtering methods according to indexes of RMSE,SNR and r.In the aspect of data analysis,it analyzed the relevance between the wire icing weighing data after filtering processing and the temperature,atmospheric pressure,wind speed with grey comprehensive correlation degree analysis theory.The results show that three kinds of meteorological factors have strong relationships with the wire icing.The average grey comprehensive correlation degree between temperature,atmospheric pressure and wind speed and wire icing weighing data are 0.632,0.594 and 0.555 at the Niba mountain site.At the Erlang mountain site,the average grey comprehensive correlation degree are 0.565,0.530 and 0.525.The average grey comprehensive correlation degree are 0.605,0.575 and 0.516 at Longwangmiao site.Temperature has the strongest effect on wire icing followed by atmospheric pressure and wind speed.It analyzed separately in both east-west and south-north directions under static and non-static wind conditions when discussed the effect of wind direction on wire icing.The result shows that the influence of wind direction on wire icing is not obvious under static wind conditions.Under the condition of non-static wind with suitable wind speed,wind direction may help the growth of wire icing where the wire is facing the wind direction and it has less influence on the wire which is following the wind.The singularity analysis of wire icing weighing data shows that meteorological factors may be the main reasons which lead to the mutation of wire icing weighing data.The natural shedding of wire icing system and external environment interference will also cause the singularity of wire icing weighing data but the effect is smaller than that of the meteorological reasons.Finally,the regression prediction effect of ice accretion and meteorological elements is analyzed based on SVM.In the direction of east-west,the mean square error of the original data and the regression prediction data is 0.0075 and the correlation coefficient is 0.859.In the direction of south-north,the mean square error of the original data and the regression prediction data is 0.0055 and the correlation coefficient is 0.923.According to the correlation between prediction curve and actual data curve and relative error,it indicates that wire icing-meteorological factors regression curve based on SVM can reflect the development trend of wire icing well and it will have a better prediction effect when the development of wire icing is relatively stable.