切缸运行工况下低压缸次末级温度增益分类识别

Classification and Recognition of L-1 Temperature Gain of Low Pressure Cylinder under Cylinder Cutting Operating Condition

供热机组低压缸切缸运行下,低压缸次末级温度对低压缸进汽流量实际对象存在强非线性问题,其增益随进汽流量大小及正负变化。根据低压缸次末级温度对进汽流量增益进行分类,把研究可解决对象强非线性问题的方法转换为已经研究成熟的多模型控制方法。提出一种低压缸次末级温度对进汽流量增益进行分类识别的方法:将次末级温度及进汽流量作为特征变量,使用模糊C均值聚类算法将实际对象增益分为4类,通过D-S证据理论对待判别数据的两个特征变量进行数据融合,最大隶属度对应类别即为判别结果。以某电厂实际对象为例,利用D-S证据理论能将各待判别的数据依据聚类中心进行判别,判别结果相较于普通欧氏距离判别更加准确可靠。

Under the cutting operation of low-pressure cylinder of heating unit,the L-1 temperature of low-pressure cylinder has a strong nonlinear problem to the actual object of low-pressure cylinder inlet flow,and its gain varies with the size and positive and negative of inlet flow.The intake flow gain was classified according to the L-1 temperature of the low-pressure cylinder,and the method which can solve the object strong nonlinear problem was transformed into the mature multi-model control method.We proposed a method to identify the gain of inlet steam flow of low pressure cylinder penultimate order temperature:the L-1 temperature and steam flow as characteristic variables,we used the fuzzy c-means clustering algorithm to divide the actual object gain into four types.Then,the two characteristic variables of the data to be discriminated were fused by D-S evidence theory,and the category corresponding to the maximum affiliation was the discriminant result.Taking the actual object of a power plant as an example,the D-S evidence theory can be used to discriminate the data to be discriminated based on the cluster center,and the discriminant result is more accurate and reliable than the ordinary Euclidean distance discriminant.