基于遗传神经算法与证据理论的柴油机燃油系统故障诊断

Diesel engine fault diagnosis based on AGNNA and evidence theory

利用遗传算法优化神经网络的初始权值和阈值,进而提高神经网络诊断故障时的准确性和快速性,解决了网络本身固有的收敛速度慢和容易陷入局部极小值问题.结合证据理论,对多信息源的柴油机燃油系统故障数据进行融合,从而更加准确、全面地反映故障状态,提高诊断的可靠性.

Initial weights and thresholds of neural network are optimized with AGNNA, and both the problems of network it- self inherent slow convergence speed and easy to get into the lo- cal minimum value are solved, which enhanced the accuracy and rapidity of neural network in fault diagnosis. In addition, com- bining with evidence theory, more information sources on the diesel engine fuel system fault data are fused, and the fault states are reflected more accurately and comprehensively so as to achieve the reliability of the diagnosis.