基于动态Markov的热风炉操作预测

针对马尔可夫预测传统的状态划分采用人为确定方法,由预测者的经验决定预测对象的初始状态,状态界限的划分定性分析因素极大的问题。提出根据预测对象数据自身的相似度,采用动态聚类方法对预测对象进行初始状态划分,确定初始状态概率和状态转移概率矩阵,进行马尔可夫预测的方法。该改进的方法在某厂的热风炉操作预测中进行了应用,预测结果表明该方法可以有效指导热风炉操作。

Application of Markov Method Based on Dynamic Clustering in hot-blast stove operation forecasting

In view the question of the traditional condition division of Markov is determined by person and the original state of fore-casting object is decided by people' s experience. The compartmentalizing of condition limit is related with the qualitative analysis. Proposes a method of Markov for compartmentalizing the original state of forecasting object based on method of dynamic clustering by resemblance degree of data own and solves the original state probability and the condition transition probability matrix. The forecast-ing result can effectively guide the hot-blast stove operation through the application of the improved method in hot-blast stove opera-tion forecast.