【摘要】 在建立火电机组回热系统常见故障的故障集和征兆集基础上 ,利用模糊数学知识和相关理论 ,针对回热系统故障征兆参数的不同变化方向和程度 ,采用不同的变化等级和阈值 ,建立了回热系统典型故障样本模式知识库及实时故障模式集。同时利用基于MATLAB环境下的径向基函数网络 ,建立了回热系统故障诊断模型。并利用电站仿真机模拟典型故障进行了神经网络模型的验证。实践表明 ,这种方法可有效地进行回热加热器故障样本模式的模糊量化处理 ,极大地改善了神经网络训练的收敛性 ,有利于回热系统的故障诊断。更多还原

【Abstract】 A knowledge database of typical fault-sample modes and real-time fault modes for the regenerative heating system of a steam turbine has been set up based on the recognition of common faults and symptoms specific to the regenerative heating system of a thermal power plant. This was accomplished through the use of fuzzy mathematics knowledge and related theories with regard to the various change directions and degrees of the fault symptom parameters of the regenerative heating system, using different variation grades and thresholds. Meanwhile, by utilizing a radial base-function network based on a MATLAB environment established was a fault diagnostic model for the regenerative heating system. Moreover, a neural network model has been verified through the simulation of typical faults by a power plant simulator. Practice indicates that the method described above is highly effective in conducting the fuzzy quantitative treatment of fault sample modes of a regenerative heating device, dramatically improving the convergence of a neural network training and facilitating the fault diagnosis of the regenerative heating system.更多还原