基于置信规则库的飞控系统故障诊断

Fault diagnosis of flight control system based on belief rule base reasoning

摘要:

针对传统飞控系统故障诊断中存在的因引入专家知识引起的主观偏差问题和使用数据驱动方法因数据量不足导致的过拟合问题，提出了基于置信规则库推理的飞控系统故障诊断。根据已有故障知识构建飞控系统故障诊断置信规则库，利用测试过程中获得的故障数据，以数值样本优化学习模型对置信规则库参数进行训练。实例表明，经少量样本训练后的置信规则库可以很好地解决初始置信规则库参数存在主观偏差的问题，经实验证明该方法能够实现高效可靠的飞控系统故障诊断。

Abstract:

Aiming at the subjective deviation caused by the introduction of expert knowledge and the over-fitting problem caused by the insufficient data in the data-driven method in the traditional flight control system fault diagnosis, proposes the fault diagnosis of the flight control system based on belief rule base reasoning. Based on the existing fault knowledge, the belief rule base for fault diagnosis of flight control system was constructed, and the parameters of the belief rule base were trained by numerical sample optimization learning model using the fault data obtained in the test process. The example shows that the belief rule base trained with a small number of samples can solve the problem that the parameters of the initial belief rule base have subjective deviation, the experimental results proves the method can realize efficient and reliable fault diagnosis of the flight control system.