Dissertation > Excellent graduate degree dissertation topics show

Research on Breakout Prediction System for Continuous Casting Beased on RBF Neural Network and Fuzzy Theory

Author: JinRong
Tutor: RenTingZhi
School: Yanshan University
Course: Mechanical Design and Theory
Keywords: Continuous casting Breakout Prediction RBF Neural Network Nearest neighbor clustering algorithm T - S fuzzy neural network MATLAB
CLC: TF777
Type: Master's thesis
Year: 2009
Downloads: 115
Quote: 0
Read: Download Dissertation

Abstract

Breakout catastrophic accident in the continuous casting production is the result of deviation from the steady state role by a variety of process factors , leading to a large number of molten steel outflow , not only endanger the safety of persons and equipment , but also seriously affect the quality of slab , causing huge economic losses. Therefore , the formation mechanism of breakout accident and further develop the key technologies that accurately breakout prediction , an important theoretical research value and significant social and economic significance to avoid or reduce the occurrence of this type of serious accident . This paper first introduces the continuous casting process , analysis of the formation mechanism of the kind of breakout accident and bonding breakout and recommending preventive measures . On this basis , the breakout prediction using RBF neural network and the TS fuzzy neural network consisting of multi-level network model . Then , RBF neural network model to identify a single thermocouple temperature waveform . Improved dynamic nearest neighbor clustering algorithm , RBF network has better approximation ability , and can get better network performance . Thermocouple buried position different combination of space , a horizontal network - based regional discrimination law , and is used to identify the extension of the the adhesive solidified shell rupture line in space . In view of the thermocouple temperature waveform vulnerable to the impact of the casting process parameters , as an important factor in the the breakout diagnosis forecast , TS fuzzy neural network modeling . Simulation results show that the breakout prediction system established in this paper can further reduce the false alarm rate of the system , to improve forecast accuracy . Finally , in the WINDOWS platform , using MATLAB language breakout prediction system off-line simulation , timely alarm is given for a dangerous situation . Development of the visual interface of the system at the same time , intuitively reflect the thermocouple temperature changes and display the location of dangerous thermocouple which , for good human-computer interaction .

Related Dissertations

CLC: > Industrial Technology > Metallurgical Industry > Steelmaking > Ingot > Continuous cast steel,near net shape casting
© 2012 www.DissertationTopic.Net Mobile