神经网络信息融合技术在故障诊断中的应用

在简述多传感器信息融合技术和BP诊断神经网络的基本概念后 ,详细讨论了基于神经网络多传感器信息融合的柴油机故障诊断技术 ,给出了测试系统框图和数据处理模式。在4 135柴油机上进行的 10种故障状态和 7种神经网络输入特征的故障监测和诊断实验表明 ,压力信息与振动信息的融合诊断效果比单一压力信息或单一振动信息要好 ,融合诊断的正确识别率比单一信息分别提高了 1 6%和 2 8 3% ,神经网络多源信息融合技术对复杂机械故障状态有较好的可诊断性和准确性。

Application of multisensor data fusion technique based on neural network in fault diagnosis

The basic conception of the multisensor data fusion technique and BP neural network are described briefly.Fault diagnosis technique for the diesel engine based on neural network multisensor data fusion technique is discussed in detail. The flow ohed of the testing system and data processing method are provided. The tests of the faults monitoring and diagnosing with 10 fault patterns and 7 characteristics input of the neural network have been carried out on model 4135 diesel engine. The test results show that the diagnosing effect combining the pressure data and vibrating data is better than that using single Pressure data or vibrating data, and the valididentifying ratio has been improved by l.6% and 28.3% respectively. The multisensor data fusion technique can Provide better idenhfong reliability and accuracy for the fault patterns of complex mechanism.