中文摘要：

提出一种基于时序趋势特征的回转窑喂煤支持向量机（SVM）分类方法，根据回转窑多个热工数据之间的动态变化规律预测喂煤变化趋势，指导人工喂煤调节的操作。首先对回转窑热工数据进行预处理，将喂煤时间序列分段线性表示，并提取相应趋势特征，形成训练样本；然后采用粒子群（PSO）算法优化SVM参数，构建SVM分类器，进而对测试样本进行分类，实现对喂煤趋势的分类预测。通过采用现场数据进行对比分析，证明本文提出的喂煤趋势预测方法具有较高的预测准确率，提高了窑前控制的鲁棒性，模型达到了现场应用的水平。

英文摘要：

A SVM classification for rotary kiln based on time series Coal feeding change in rotary kiln, so that data of rotary kiln is trend is predicted according to dynamic change trend characteristics is presented. rules of thermal engineering data regulation operation of manual feeding coal is guided. Firstly, thermal engineering preprocessed, time series of coal feeding is linearly represented segment, trend characteristics of rotary kiln time series are extracted, and training samples are obtained. Then, SVM parameters are optimized＇with particle swarm algorithm, and SVM classifier is constructed. Finally, test samples are classified and classification prediction of coal feeding trend is achieved. Prediction results are compared with field data and analyzed. It is proven that prediction method of coal feeding trend has higher prediction accuracy, robustness of rotary kiln control is enhanced, the model can be applied to production field.