相似规则建模方法在火电机组设备状态监测中的应用研究

Application of Similarity Modeling based on Clustering in Equipment Condition Monitoring of Power Plant

以火电机组的主要设备为研究对象,研究按相似性规则建模的方法在设备状态监测中的应用。针对火电机组存储数据的高维、体量大等特征,引入K均值聚类方法提取设备运行的典型工况,结合设备的具体参数模型构建出其相似性规则模型中的动态矩阵,进而可以对设备参数值进行实时估计。利用国内某600 MW机组的实际运行数据,选取给水泵和高压加热器两种典型的主要设备对相似性规则的建模过程进行阐述,实际运行数据的计算结果表明,方法能够对设备参数值进行实时的准确估计,仿真结果还进一步表明,在测点故障和异常工况下,该方法能够及时发现并给出预警。

With the main equipments of a thermal power plant as studying object,the application of similarity modelling method in equipment condition monitoring was studied. According to the data characteristics of power plant in terms of high dimension,large volume and etc.,the K-means clustering method was introduced to extract typical working conditions for the equipments. By combining the specific parameter models of the equipments the dynamic matrix was established which can real-time estimate the parameters. In this study,the feed water pump and high pressure heater were selected as the primary equipments and the similarity principle modelling process was detailed. The calculation results with the field data showed that the presented method has an excellent real time approximation capability. Moreover,the simulation test indicated the system abnormal can be detected and alarm will be issued whenever the sensor or system fault occurs.