摘要
火电汽轮机故障诊断系统中,存在故障信号非线性强,故障特征信号容易被干扰,很难建立稳定诊断模型。在传统的神经网络汽轮机故障诊断模型中,引入可溯源机制,利用贝叶斯网络的故障推理学习和边沿信号概率分析能力组成故障追溯模型,对汽轮机中的汽缸、隔板、喷嘴、静叶片、汽封和轴封模块高频故障信号进行追溯,再通过故障特征值和实际输出之差与设定汽轮机故障阈值的大小比较关系判断故障,实验结果表明,该方法有效地解决故障信号非线性带来的干扰,完成火电汽轮机高精度的故障检测。
Thermal power steam turbine fault diagnosis system, the fault signal nonlinear, strong fault feature signals are easy to be interference, it is difficult to establish stable diagnosis model. In the traditional steam turbine fault diagnosis model of neural network, the introduction of traceable mechanism, and the use of bayesian network inference ability to learn and edge signal probability analysis of the fault trace model, cylinder of steam turbine, baffle plate, nozzle, stator blade, seal and shaft seal module back on high frequency fault signal, again through the fault characteristic value and the difference between the actual output and setting the size of the steam turbine fault threshold comparison judgment fault, the experimental results show that this method is effective to solve the fault signal of nonlinear disturbance, complete thermal power turbine high accuracy of fault detection.
出处
《科技通报》
北大核心
2013年第11期153-156,共4页
Bulletin of Science and Technology
基金
山东建筑大学课题基金(120548)
关键词
故障诊断
贝叶斯网络
神经网络
fault diagnosis
bayesian network
neural networ