Application Of Research On Parameters Self-learning Method Of VOD Production Guidance System

Based on the VOD refining technique in steel factories, this paper studies themethod of using neural network algorithm to realize self-learning and dynamicaladjustment of the processing parameters in endpoint carbon content prediction model ofVOD refining technique, in order that prediction accuracy can be improved.First, background information of the subject, the VOD refining technique anddevelopment status of self-learning algorithm are demonstrated. Technical principles ofVOD technique, basic functions of the production guidance software and therequirements for processing parameter self-learning and dynamical adjustment areintroduced. Then the technical scheme and implementation method for self-learning andadjusting of processing parameters with the endpoint carbon content model whichcombines BP neural network algorithm and RBF neural network algorithm with VODrefining technique is discussed in detail, simulation software is also developed. Basedon over twenty real steelmaking data, processing parameter self-learning and adjustingis realized with the software, the efficiency of the two neural network algorithms iscompared and analyzed.Finally, by calling matlab code in VB, the RBF neural network self-learningalgorithm, which has higher precision, is successfully embedded into the real VODproduction guidance software to realize the processing parameter self-learning anddynamical adjustment in the prediction model.