Research On Ice Thickness Extraction Based On Image Recognition And Early Warning Model For Icing Accretion On Transmission Lines

National economic has a high degree of dependence on electricity,while economic growth will also promote the increase in demand for electricity consumption.With the large-scale construction and development of power grid,the transmission lines of long distance and large capacity are inevitable to be through the micro meteorological areas which are caused by the cold,high altitude,high humidity and strong precipitation area,or basin,plateau,watershed and other special micro terrain.That is very easy to cause the icing disasters on transmission lines,and icing disasters on the transmission lines have been aggravated because of global warming.So people pay more attention to the icing on transmission lines.It is particularly important to study the image recognition technology of icing thickness on transmission line,and establish the icing warning model.And then managers can know the icing field condition and master icing development trend in real time.In the icing monitoring of transmission line,the current research in China mainly concentrates on the analog wire method and the weighing method.But there are few researches on image recognition technology,which has the advantages of real-time,intuitive and so on,as the research trend of icing monitoring of transmission line.On the other hand,the researches on the icing warning model are mainly related to icing prediction,yet further involving ice warning level,early warning mechanism and so on.And then the established icing prediction models have some problems,such as the difficulty in measuring the parameters of the model,the low accuracy of the model.At the same time,there are some problems such as the warning value is not standardized;the model is not perfect and so on.In view of the above problems,Canny operator was used to get the image edge,in addition,Hough transform and least squares are combined to solve the problems of traditional Hough transform in the parameter space which are easy to be disturbed by image background and noises,eventually,the edges of icing transmission line and un-icing transmission line were accurately detected in images which have low contrast,complex grayscale,and many noises.Furthermore,based on the imaging principle of camera,a new geometric model for icing thickness was established by using the radius of transmission line as a reference,and automatic calculation of icing thickness was achieved.A large number of icing and micro meteorological monitoring data were analyzed,and then the icing period was segmented and the key influencing factors of icing were identified with the growth law of icing and the micro meteorological characteristics.At last,based on quantitative analysis of icing micro meteorological experiments and qualitative analysis of icing and the key influencing factors,a multivariate nonlinear regression model of icing rates and micro meteorological factors was established,at the same time,a comprehensive prediction model of icing was established according to the segmentation characteristics of ice coating period.Based on the four level of the icing ratio,the double warning models of real time and forecasting are established.