System Design For Cohesiveness Steel Leakage Of Crystallizer In Continuous Casting Process

With the development of Continuous Casting production process, develop moreefficient Continuous Casting technology is the main research direction in this field. Due tothe casting speed increasing, the risk of breakout also will increase. Breakout in theproduction of Continuous Casting is the most common and easy to cause significant loss ofaccident, in the form of many breakout sticking breakout proportion is the largest. Thethermocouple temperature measurement is the most common method of sticking breakout,it can effectively detect the surface temperature of the Crystallizer internal,and thetemperature can be intuitive, rapid reflect the internal condition of Crystallizer,so it playsthe key role of breakout prediction. This article is mainly analyzed from the thermocouplecollected on the surface of Crystallizer temperature. analyzing and researching of a largenumber of the temperature data, and eventually realize the breakout prediction systemdesign in PC LabVIEW.In this paper, starting from the formation mechanism of breakout, Study of theformation process of breakout, Mainly analyzes the factors of sticking breakout, andpreventive measures. First of all, denoise the temperature of thermocouple collected data,Respectively using five sliding method, polynomial fitting method, Fourier transformmethod and wavelet analysis method for noise reduction processing to these data, andcompare four methods of noise reduction effect; Secondly, according to the embeddedthermocouple installation of the Crystallizer, the breakout prediction mathematical modelis established, and the BP neural network is introduced into the breakout prediction system.Aimed at the disadvantages of BP network, using LM algorithm to raise the networkconvergence speed, PSO algorithm to improve into the local optimal solution; Finally,using the MATLAB Script node method, wavelet analysis and the optimized BP neuralnetwork is seamless connected with LabVIEW, and realize the breakout prediction systemdesign in PC. Through the existing breakout sample data to verify this system, prove that this system can achieve the purpose of breakout prediction, but in the actual application offorecast effect remains to be further testing.