【作者】 孙亮；

【导师】 周建新；

【作者基本信息】 华中科技大学 ， 材料加工工程， 2013， 硕士

【摘要】 根据调查统计，铸铁件产量占我国铸件总产量的70%以上，铸铁生产仍是我国铸造生产最重要的方面，因此，提高铸铁件质量、降低其成本对促进我国国民经济的发展有着重要意义。而铸铁熔炼过程是影响铸铁件质量的关键因素，因此，熔炼水平的提高对实现低能耗绿色铸造以实现国家降低单位产值能耗的目标具有重要的现实意义。冲天炉作为铸铁最主要的熔炼方式之一，对其熔炼过程的研究至关重要。本文主要对冲天炉熔炼成分体系成分和温度预测模型的建立和系统实现两方面开展研究工作，并最终自主开发冲天炉成分与温度预测系统。首先，将冲天炉熔炼成分体系成分变化过程转换成数学问题，并比较平衡常数法和最小自由能法这两种方法，最终选用最小自由能法来建立冲天炉熔炼体系成分预测的热力学模型，并选用遗传算法作为上述模型的求解算法，然后利用Matlab平台实现了该模型，并且设计了算例来说明成分与温度预测系统的实际应用效果。其次，由于冲天炉熔炼是一个高度复杂的非线性模型，从机理上建立温度预测模型十分困难，本文利用各种神经网络模型计算并把结果进行比较，最后证明用遗传算法优化过的BP神经网络可以迅速的达到要求。本文通过建立冲天炉熔炼过程成分和温度预测系统数学模型、选择合适的算法求解模型、编程实现算例，结果表明，预测值与实际值误差在工业允许误差以内，证明了冲天炉熔炼过程成分和温度预测系统的可行性，最后对全文进行了总结，并对后续工作提出了展望。更多还原

【Abstract】 According to surveys，the production of iron casting which is still the most importantcasting metal in China occupies more than70%of the total output of castings. Therefore,it is of great significance on development of national economy to improve the quality andreduce the cost. As a key factor influencing its quality and cost, the level of the meltingprocess has an important practical meaning for realizing the goal of green casting withlower energy consumption, and reducing energy consumption per unit of output value.This article researched one of the main methods of iron melting, the cupola meltingprocess, and focused on modeling and programming of the composition and Temperatureprediction system.Firstly, turning the cupola melting process of composition variation system intomathematics problems. Then, comparing the advantages and disadvantages of the twomathematic methods, namely, the equilibrium constant and the Gibbs free energyminimization method. Lastly, chose the latter one to create the thermodynamics models ofcomposition prediction system in cupola. Besides, this article chose genetic optimizationas the solution algorithm to above-mentioned models and did some improvement ofgenetic optimization to raise solving efficiency, based on the solving specialties ofcomposition forecasting model. Then The Composition Forecasting System was realizedby the Matlab platform，and application cases were analyzed at the end of this article，through which the application effects of the System were introduced. It is difficult tocreate temperature forecasting system theoretically because cupola melting process is sucha complicated and nonlinear model. So, this article try to use various neural networkmodels, compare the results, and finally found out that BP neural network after geneticoptimization.The results of this article testified the feasibility of the composition and temperatureprediction in cupola melting process，and deeply stated the factors influencing the system,which pointed the way of future research. The research work in this paper is a beneficialexploration to simulation of the cupola melting process，and is a new direction of theapplication of computer technology in the foundry processes.更多还原