CAJ | 학술논문

故障预测是PHM体系的重要特征,提出基于过程神经网络(PNN)对机载设备进行故障预测;首先建立了基于PNN的多步预测模型,推导了模型在函数正交基展开形式下的计算公式;基于正交基函数展开和矢量矩免疫算法训练PNN,推导出了适用于进化算法的优化问题模型;以设备转换部件Out2信号的全寿命故障规律为例进行了故障预测,基于MATLAB进行了仿真,结果表明,基于PNN进行故障预测精度较高,是一种有效的方法.
고장예측시PHM체계적중요특정,제출기우과정신경망락(PNN)대궤재설비진행고장예측;수선건립료기우PNN적다보예측모형,추도료모형재함수정교기전개형식하적계산공식;기우정교기함수전개화시량구면역산법훈련PNN,추도출료괄용우진화산법적우화문제모형;이설비전환부건Out2신호적전수명고장규률위례진행료고장예측,기우MATLAB진행료방진,결과표명,기우PNN진행고장예측정도교고,시일충유효적방법.
Prognostic is an important characteristic of PHM system. Thus Prognostic of an Aero Equipment Based on Procedure Neural Network (PNN) is brought out. Firstly, the multi- steps model based on PNN is established, then the formula of the model based on orthogonal trigonometric function basis is deduced. Using both orthogonal function basis expansion and Immune Algorithm (IA) to train PNN. Finally, take the trend of the signal 'Out2' that is of an converter in the device in its coherent life for example, simulation is carried out based on MATLAB, results show that using PNN to carry out prognostics can obtain more accurate value, and effectiveness of the way is also proved.