汽轮发电机组故障智能诊断方法研究

INTELLIGENT FAULT DIAGNOSIS METHOD FOR TURBO-GENERATOR UNIT

该文针对汽轮发电机组故障特点,研究了用于故障智能诊断的神经网络结构,对传统的BP神经网络进行了改进,提出了神经网络的逐层优化法,优化后的神经网络更加适用于故障诊断。通过对汽轮发电机组常见故障的分析,提出了适用于汽轮机故障诊断的原因-征兆表,在此基础上对某热电厂的汽轮发电机组故障进行了诊断,诊断结果表明,该方法能够正确地诊断出存在的故障,具有实用价值。

This article presents an approach to fault diagnosis of turbo-generator unit by using intelligent neural networks. The conventional back-propagation network was optimized by Using a method of layer by layer, the action mech-anism of signal processing in neural networks was probed, the structure of this system was designed and the performance was discussed based on this. An intelligent method which can adaptively determine the fault of the turbo-generator unit is proposed. A relation table for causes and manifestations to turbine fault was proposed, using this table, three-layer percep-trons was improved, a fault in turbine was identified. This method is used in an example of the fault diagnosis, the results show that the fault can be accurately diagnosed.