基于模糊SOM神经网络的汽轮机通流部分故障诊断技术  

Fault Diagnosis Technology of Turbine Flow Based on Fuzzy SOM Neural Network

作　　者：黄瑜 Huang Yu(Guoneng Shenfu(Longyan)Power Generation Co.,Ltd.,Longyan 364000,China)

机构地区：[1]国能神福(龙岩)发电有限公司,福建龙岩364000

出　　处：《科学技术创新》2022年第36期1-4,共4页Scientific and Technological Innovation

摘　　要：汽轮机通流部分长时间受到高温、高压、高流量等因素的影响,容易发生结垢、腐蚀、磨损、叶片断裂、主汽门卡涩等故障。由于故障类型多样、故障成因复杂,给故障的诊断和排查增加了难度。提出了一种基于模糊SOM神经网络的汽轮机通流部分故障诊断技术,该技术融合了模糊系统与神经网络,以汽轮机通流部分的故障样本作为训练对象,将经过训练后的模糊SOM神经网络进行模糊聚类分析,以便于快速、准确完成故障诊断。随后以某发电厂的300 MW机组作为研究对象,利用模糊SOM神经网络进行故障诊断,诊断结果为第5类故障(高压缸级组结构),SOM神经网络诊断结果与实际拆解诊断结果一致。The flow through part of steam turbine is affected by high temperature, high pressure, high flow rate and other factors for a long time, which is prone to scale, corrosion, wear, blade fracture, main valve stuck and other faults. Due to the variety of fault types and complex fault causes, the fault diagnosis and troubleshooting are more difficult. Proposes a flow part of steam turbine based on fuzzy SOM neural network fault diagnosis technology, the technology combines the fuzzy system and neural network, with flow of steam turbine fault samples as the training objects, after training, the fuzzy SOM neural network to fuzzy cluster analysis, so that the rapid and accurate fault diagnosis. Then, a 300 MW unit of a power plant is taken as the research object, and the fuzzy SOM neural network is used for fault diagnosis. The diagnosis result is the fifth type of fault(high pressure cylinder level group structure), and the diagnosis result of SOM neural network is consistent with the actual dismantling diagnosis result.

关 键 词：SOM神经网络 汽轮机 模糊聚类 故障诊断 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TK268[自动化与计算机技术—控制科学与工程]