基于聚类法的水泥分解炉工况识别研究

Study on Recognition of Working Conditions of Cement Decomposing Furnace By Clustering Method

针对分解炉系统非线性、大滞后、工况复杂且波动大、变量多且各变量间强耦合等特点,要对其进行有效控制以实现分解炉系统稳定运行且入窑物料分解率最大,则必须对其工况进行深入分析和研究。本文采用聚类法对分解炉系统进行工况识别,首先运用硬聚类方法对分解炉各参数的变化趋势和实时值进行识别,然后在此基础上改进为采用模糊聚类的方法对分解炉各参数的变化趋势进行识别,采用历史数据仿真得出的结果良好。在对分解炉各参数实现了参数趋势和实时值识别后,再根据分解炉系统的典型工况类可以准确判别分解炉系统此时工况,这对实现分解炉的有效控制,提高入窑生料分解率意义重大。基于该算法采用VB软件实现的分解炉工况识别器在生产现场的运用效果较佳。

For decomposing furnace system has characteristics like non-linearity,big lag,working condition of complex and volatile,variable,strong coupling between the variables,to control them effectively,we must analysis and research on its condition deeply.In this paper,the clustering method is used to identify the condition of decomposition furnace system.First of all,hard clustering method is used to identify the trends and real values of the parameters.But since the method exists inaccuracy itself,it can make the results in error.So on the basis of it,we have the improvement for the use of fuzzy clustering method to identify the trend of each parameter.using historical data,we got good simulation results.After identification of trends and real value of various parameters of calciner,according to the typical working conditions of decomposition furnace system,we can accurately distinguish decomposition furnace system's working condition at this moment,it is of great significance to achieve the effective control of decomposing furnace and improve the soil of decomposition rate.Recognizer of working conditions of decomposition furnace is realized by using vb software based on this algorithm.The use effect is better at the scene of the production.