摘要
用人工种经网络技术，对波音747—200型飞机的JT9D发动机的故障诊断进行了研究，并构成了诊断装置。该研究使用北京飞机维修工程有限公司提供的发动机性能监控数据，在脱机后根据性能排队情况用经验推定法，对发动机的一些常见故障和突发性故障进行了诊断。在诊断过程中，首先搜集发动机的故障状态数据，并对这些数据进行归纳选择，制成了诊断用的教师信号“故障模型”，通过神经网络系统对教师信号的学习，在一定范围内，对没有经过学习的实故障数据进行了诊断。其结果表明了诊断装置的有效性。
There is a strong demand for the off-line type fault diagnosis device for aircraft jet engines because of the decreasing veteran mechanics.The device is more useful if it can diagnose only by using the recorded engine data which are gathered by the conventionally- equipped sensors during the flight. A single engine datum, however, is not enough to specify the fault spot even against the air flowing system itself. Therefore, four engine data of the revolution of the low pressure compressor, the exhaust gas temperature and the fuel flow rate are used for this diagnosis as a data pattern. In order to specify the relevant fault, neuro- network is adoptal. Relatively good results are obtained by using the fault diagnosis device.
出处
《中国民航学院学报》
1999年第2期7-12,共6页
Journal of Civil Aviation University of China
基金
中国民航总局科技项目
