基于D-S理论的智能故障诊断关键技术研究与实现

以智能检测和故障诊断技术作为研究对象,分析了在智能故障诊断系统中用到的特征提取、模式识别和故障预测等关键技术,探讨了用于实现上述关键技术中的小波变换和人工神经网络算法。鉴于人工神经网络算法容易出现系统故障定位不准确性的不足,引入了D-S(Dempster Shafer)证据理论。首先用BP神经网络对故障信息进行局部诊断,然后利用D-S证据理论对局部诊断结果进行融合,进行全局诊断,得到最终的故障结果。D-S证据理论的引入,增大了诊断结果的可靠性和准确性,提高了诊断算法的适应性。在LabWindows/CVI的平台下,实现了智能故障诊断算法。

Design and Realization of the Intelligent Fault Diagnosis Key Technologies Based on D-S Evidence Theory

Taking the intelligent examination and the faults-diagnostic technology as the object of study,the feature extraction,the pattern recognition and the faults forecasting technology and so on the key technologies are analyzed,and the wavelet transformation algorithm and the artificial neural networks algorithm are discussed.In view of the fact that the artificial neural networks algorithm easy to appear the system failure localization not accurate insufficiency,the D-S(Dempster Shafer) evidence theory is introduced.First,the BP neural network to carry on the partial diagnosis to the fault information is used,then the fusion using the D-S evidence theory to the partial diagnosis result is carried on,and the final fault result is obtained.The introduction of D-S evidence theory increases the diagnosis result reliability and the accuracy,enhances the diagnosis algorithm compatibility.Under the LabWindows/CVI platform,the intelligent faults-diagnostic algorithm has been realized.