航空发动机整机振动故障智能诊断技术研究

Intelligent Diagnosis Technology for Body Vibration Faults of Aero Engine

针对BP神经网络和支持向量机对整机振动故障诊断时的低效问题,提出了一种基于多类协同训练的航空发动机整机振动故障诊断方法。引入逻辑回归算法构建初始故障分类器,设计了一种新的属性划分算法来迭代优化故障分类器,基于优化后的故障分类器进行故障类别预测,并使用多数投票机制进行故障仿真识别。实验采用某航空发动机整机振动数据作为样本数据集,并从中选择80%的数据用于训练,20%的数据用于测试,同时验证了在噪声信号的干扰下该方法对故障数据的诊断准确性。结果表明,该方法可有效降低噪声信号对故障诊断结果的影响,且诊断准确性高,具有重要的工程实用价值。

To solve the inefficiency of BP neural network and support vector machine in body vibration fault diagnosis,a fault diagnosis method for body vibration based on multi-class co-training algorithm is proposed.The logistic regression algorithm is introduced to construct the initial fault classifier,a novel attribute partition algorithm is designed to iteratively optimize the fault classifier.Based on the optimized fault classifier,the fault classification is predicted,and the majority voting mechanism is used for fault simulation recognition.In the experiment,the body vibration data of aero engine is selected as the sample data set,and 80%of the data is selected for training and 20%for testing.Meanwhile,the diagnostic accuracy of this method for fault data under the interference of noise signals is verified.The results show that this method can effectively reduce the influence of noise signals on the fault diagnosis results and enhance the diagnosis accuracy,and has important engineering practical value.