混合神经网络在泵站机组故障诊断中的应用

Application of hybrid neural network to fault diagnosis for pumping stations

分析了专家系统ES和前向神经网络MLP对新事件的识别能力的不足 ,探讨了神经网络的数学模型和数学结构 ,建立了MLP和ART1相结合的混合型神经网络模型 ,以实现故障模式的有效识别 .在泵系统故障诊断与监测中的应用表明 ,混合型神经网络能可靠地进行水泵机组故障模式的识别 .

The shortage of expert system and multi layer back propagation neural network for the recognition of new faults is analyzed, and the mathematical model and structure of the neural network are studied. A hybrid neural network model utilizing both MLP and ATR1 is put forward so as to identify the fault modes effectively. Its application to fault diagnosis and state monitoring for a pump system shows that the hybrid neural network may recognize the fault modes in pump units reliably.