Research On Key Technologies Of Continuous Casting Slab-end Face Information Recognition Based On The Convolution Neural Network

With the continuous development of intelligent manufacturing,all fields of life have the shadow of artificial intelligence,among them,the character information recognition technologies in the field of pattern recognition develops rapidly,which is mainly to recognize the characters on the measured objects and save the recognition results to the intelligent operation process of the computer.However,in industrial productions,the development of character information recognition technologies is slow,especially in iron and steel enterprises,continuous casting billet information recognition site environment is complex,it relies on manual identification of billet information,which leads to tedious work and low production efficiency,seriously affect the level of informatization and intelligence of steel production.In view of this phenomenon,this paper proposes to apply the information code recognition technologies based on convolution neural network to the continuous casting slab production line,so as to liberate the workers on the continuous casting billet identification station and improve the intelligent level of iron and steel enterprises.In this paper,the information location,information segmentation and information recognition of continuous casting slab end face are mainly studied,The main work and innovative research include: 1)According to the machine vision technologies and convolution neural network recognition algorithm,the model of continuous casting slab end face information recognition system is built,and the recognition principle of the system is studied in depth;2)Through the field investigation,the environment around the continuous casting billet recognition station is disordered,which interferes with the extraction of the effective region of characters.This paper proposes a method to solve this problem by combining the sliding window algorithm with the support vector machine algorithm.According to the obtained pixel coordinates,a character area is initially determined,and then an outlier detection method is designed to locate the effective region of characters accurately;3)After obtaining the image of the effective region of characters,this paper uses the sliding window algorithm to slide in the region,and calls the trained character segmentation classifier model to segment the single character accurately;4)In this paper,a recognition method of continuous casting slab end face information based on convolution neural network is proposed.Applied the Le Net-5 model in the convolutional neural network,and improve some links in the model,the accuracy of the improved model is increased by 4.76%.Finally,the applicability of the character recognition method based on convolution neural network is verified by the experiment;5)In this paper,the software of intelligent recognition system for continuous casting slab end face information is developed based on Py Qt5 platform,and the reliability of the research content in this paper is verified by field tests.The intelligent recognition system of continuous casting slab end face information designed in this paper can recognize the characters of continuous casting slab end face accurately,and provide technical support for intelligent upgrading of continuous casting slab line in iron and steel enterprises.The experiment shows that the information recognition system of continuous casting billet can free workers from the identification station,improve the intelligent level of production in steel enterprises.