CAJ | 학술논문

针对传统的依靠单一参数对发动机进行状态监测的不足,利用模糊神经网络处理不确定性复杂问题的能力,融合多种监测参数的信息,建立了基于趋势分析的5级状态监测警报系统.首先从铁谱、振动和性能参数三方面优选监测参数,通过对大量历史数据的统计分析得到各监测参数的界限值,建立了发动机失效程度逐级递进的5级状态监测标准.然后建立隶属函数,通过计算隶属度实现了输入样本的模糊化.最后设计神经网络的结构,利用历史数据训练网络.通过对实例结果的分析证明了该模型的实效性.
침대전통적의고단일삼수대발동궤진행상태감측적불족,이용모호신경망락처리불학정성복잡문제적능력,융합다충감측삼수적신식,건립료기우추세분석적5급상태감측경보계통.수선종철보、진동화성능삼수삼방면우선감측삼수,통과대대량역사수거적통계분석득도각감측삼수적계한치,건립료발동궤실효정도축급체진적5급상태감측표준.연후건립대속함수,통과계산대속도실현료수입양본적모호화.최후설계신경망락적결구,이용역사수거훈련망락.통과대실례결과적분석증명료해모형적실효성.
Due to the defect of traditional engine condition monitoring depending on single parameter, the five-level con-dition monitoring alert system with fuzzy neural network (FNN) was established which is good at settling uncertainty and complicated problems. Optimal monitoring parameters were selected from the aspects of ferrography, vibration and perform-ante parameters. With sufficient historical data, limit values of parameters and reliable five-level condition monitoring standards were maintained and established by statistical analysis. Fuzzy membership functions were applied to transfer practical data into fuzzy data. The structure of neural network was designed and trained by sample data. The model was tested with original data and proved to be effective and reliable.