基于人工神经网络的转炉冶炼终点控制模型的优化-钢铁冶金专业论文.docxVIP

河北理I学院硕十学何论文摘 河北理I学院硕十学何论文 摘 要 用计算机控制转炉冶炼过程，是科学的组织炼钢生产的重要手段。它可以规范 转炉冶炼操作，克服盲目性，从而提高钢水质量，降低炼钢成本，提高终点命中 率。本研究对以神经网络为基础的转炉冶炼终点控制模型进行优化。模型以转炉炼 钢基础理论为理论依据，采用误差反向传播网络——BP网络为原形，对其算法进行 改进，经过300个样本的学习训练，确定了网络的权值矩阵，并建立了可保存100 炉数据的学习样本，用束对网络进行实时训练。使模型适应生产条件的变化。优化 后使模型的学习时间缩短，预报精度提高。并使预报模型与加料系统连接，实现了 计算机按预报模型确定的原料加入量自动控制加料过程。 模型在邢钢30t转炉上进 行现场试用．效果良好。 在转炉炼钢过程中。枪位变化直接关系到造渣、脱碳、升温及冶炼过程的平稳 进行，因此合理控制枪位是转炉冶炼的一个非常重要的操作。基于这一点，本研究 通过采集现场操作数据，利用Visual Basic6．0为编程语言开发实时枪位变化曲线模 型作为对枪位控制的参考曲线。氧枪进出水温差数据可间接反映炉内的过程温度， 绘制氧枪进出水温差曲线，有利于控制过程温度平稳上升，最终准确到达终点目标 温度。 本研究在软件制作过程中，注重界面设计，利用VB语言的强大多媒体功能， 玎发出一套集声音、动画、图片、文字为一体的交互式软件系统。 关键词：转炉冶炼，神经网络， 终点预报，枪位曲线，终点命中率 河北理【学院硕十学仲论文Abstract 河北理【学院硕十学仲论文 Abstract It is an imporlant method for science production organization of steelmaking，which control converter blowing process by computer．It can standard the operation of converter blowing，against the blindness，thereby improve the quality of molten steel，reduce cost of steelmaking，increase the end—point hit rate．The research improved a model of static control for converter smelting operation based on neural network．The model by the basic theory of converter process，choose the Back Propagation(BP)network as primary form， and improved the BP algorithm．Matrix of power value of network is fixed after learning 300samples．And establish a data file which capacity is 1 00 samples for the real—time learning of network，in order to its adapt of the environmental impact．it shorten training time and increased prediction accuracy after the improved．But also the predict model connect with charging system，which make it became reality that computer autocontrol charge process according to the material quantity determined by the model．And the model get a good result after it was putted into practice in converter mill ofXG． In converter steel making process，the lance position directly relates to slag forming， decarbonization of molten iron，raise of temperature and stability of process，it is very important to control the lance posi