基于计算智能和信息熵的故障组合预测研究

针对单项预测方法鲁棒性弱和稳定性差的不足,研究了一种基于计算智能和信息熵的故障组合预测方法.利用粗糙集、遗传算法、免疫算法改进基本神经网络的神经元结构、权重计算方法和隐含层激励函数,构造了融合各项算法特长的计算智能方法,实现了电子设备特征参数的退化趋势预测;基于信息熵理论,采用多准则评价对前述单项预测方法进行融合,实现了电子设备的故障组合预测.通过跟踪预测实际环境中的设备工作数据,验证了基于计算智能预测的有效性和基于信息熵组合预测的稳定性.

Combined Fault Prediction Based on Computing Intelligence and Information Entropy

A combined fault prediction method based on computing intelligence and information entropy was proposed to enhance the robust ability and stability of single prediction method.The rough set was used to ameliorate the structure of neurons of basic neural network,the genetic algorithm for the weight determination,and the immune algorithm for the activation function of hidden layer.Then the intelligent computing method with the advantages of all above methods was realized,and the characteristic parameters of electronic equipment were tracked and predicted with these methods.Based on information entropy theory,the combined fault prediction method was obtained which enhanced the reliability of prediction results by fusing the above prediction methods with multi-rules.The combined prediction method was applied to one-type electronic equipment for tracking and predicting the parameters.Results testified the validity of prediction based on intelligent computing and the stability of the combined prediction based on information entropy.