基于神经网络的特大型焦炉焦炭质量预测研究

Research on Prediction of Blended Coal for Coking Quality in Oversized Coke Furnace Based on Neural Network

预测焦炭质量是焦化企业面临的重要课题,对焦炭质量进行准确的预测,既保证焦炭质量符合要求,又合理利用煤炭资源。研究某焦化公司特大型焦炉炼焦配煤的工艺流程,采用具有非线性特性和自学习能力的BP神经网络对炼焦过程进行建模,确定焦炭质量预测模型的结构和参数,解决炼焦过程中变量之间复杂的非线性关系问题。使用MATLAB软件对模型进行训练和预测仿真,并对影响模型预测精度的因素进行详细的分析。最后采用MATLAB软件与VC++混合编程的方法来完成焦炭质量预测系统的开发。

Prediction of coke quality is an important topic which coking enterprise must face.The prediction of coke can not only assure the coke quality,but also utilize the coal resources effectively,which means coking enterprise can benefit from economy and the society.At the same time there is very important guiding significance in our coking industry.The technology process of coal blending for coke making in oversized furnace was studied.The model for prediction of coke quality based on the BP Neural Network,which had the ability of nonlinear and self-learning characters,was selected.And then the structure and the parameters were confirmed in the dissertation.The model resolved the problem from the complicated and nonlinear relation between all the variables in coke furnace.The causes bringing on error were analyzed after the training and testing simulation for the model for prediction of coke quality.The dissertation has constructed a system programmed by the joint programming of MATLAB and VC＋＋.The system is based on quality predictive model for coke.