【作者】 李欣；

【导师】 刘万军；

【作者基本信息】 辽宁工程技术大学 ， 计算机科学与技术， 2006， 硕士

【摘要】 航空发动机是飞机的心脏,是机械专业中故障率最高、维护工作量最大的系统,其工作状态的好坏直接影响到飞机的安全可靠运行。目前,随着部队飞行训练任务的不断加重,飞机发动机的故障率明显呈上升趋势。因此,本文基于飞机发动机故障诊断的研究对改善和提高航空部队的保障能力具有重要意义。文中着重分析了人工神经网络的特点,指出了神经网络方法与故障诊断技术结合的可行性和必然性。并且从实际应用的角度研究了BP网络中网络层数、初始权值等参数的选取问题。针对BP算法的缺陷,本文提出了一种改进算法,并且分别应用两种算法通过故障诊断软件对待测涡扇-11型发动机进行了诊断,验证了改进算法的实用性以及在训练速度和误差方面较之原算法的优越性。同时程序的运行结果也进一步验证了本文研究工作的正确性和可行性。更多还原

【Abstract】 Aero-Engine is the heart of the airplane, is a system having the most ratio of trouble in the specialty of mechanism, and having the most work to do. The condition of the engine has the direct influence of the safety of the airplane. In the current, the Aero-Engine’faulting rate rising clearly with the improving task of the Aero-army. So, the research of the diagnosing of the trouble based on the plane’s engine has important significance to improve the Aero-army’s capability of the guarantee.In this article, it has placed the emphasis on the characteristic of the neutral net, and pointed the practicality and the necessity of the combination of the neutral network and the fault diagnosis. The article has researched the choices of the net’s layers, the primary’authority in the practical part.The article has produced a new improved method based on the primary Back-Prop algorithm’defects, and by the use of the two methods to the WS-11 Aero-engines with the fault diagnoses software, proved the practicality and the superiority to the old on the rate of study and the exact of fault. At the same time, the outcome of the running has proved the exactitude and the practicality of the research’s work.更多还原