Research On Positioning Method Of Ladle Nozzle Based On Image Processing

With the continuous advancement of the national strategy of "made in China 2025",unmanned intelligent manufacturing of steel has become the only way for steel enterprises.In order to further enhance the intelligent degree of traditional iron and steel manufacturing enterprises,this paper takes a five meter wide heavy plate plant of an iron and Steel Group as an example to carry out unmanned transformation of two 120 ton BOF furnaces after converter.Due to the low degree of automation of the drainage sand feeding equipment behind the furnace,this problem has become one of the main factors restricting the unmanned transformation behind the furnace.In order to solve this problem,this paper proposes a new mechanical and electrical control scheme,and applies the machine vision positioning control technology to the drainage sand automatic delivery equipment.The research focus of this paper is to locate the position of ladle nozzle by image processing algorithm.Firstly,the hardware platform of drainage sand automatic delivery system is built.Drainage sand feeding system and drainage tube driving device are designed.The framework of drainage sand automatic feeding system is industrial control computer loaded with Win CC + MATLAB software.Siemens 1500 series PLC is used in the lower computer.MATLAB is used as visual algorithm processing software and Win CC as interface monitoring software.Then,the spatial domain algorithm of ladle nozzle image is studied,and the position coordinates of similar nozzle are obtained.The image of ladle nozzle is preprocessed by gray and binary,and the features of nozzle image are analyzed.The coordinates of similar nozzle are extracted by ?? circle detection algorithm.Next,the frequency domain algorithm of ladle nozzle image is studied to exclude non nozzle coordinates and determine nozzle coordinates.The window Fourier transform and the coordinates of the similar nozzle obtained by the spatial domain algorithm are used to obtain the spectrum of the similar nozzle window image.Through the analysis of the spectrum,the spectrum features of nozzle window and non nozzle window are determined by theoretical derivation and experimental demonstration,and the feature quantization classification algorithm and image edge feature extraction classification algorithm are designed to classify nozzle and non nozzle window.Finally,through three experiments,the feasibility of ladle nozzle image spatial domain algorithm and frequency domain algorithm for ladle nozzle location is verified.The experimental results of feature quantization classification algorithm and image edge feature extraction classification algorithm are compared to analyze which algorithm is more suitable for this system.