论文部分内容阅读
针对飞机发电机振动特征参数多、故障特征参数难以准确识别飞机发电机健康状况的现状,设计了发电机振动信号实时采样装置对飞机发电机转动时的多种频域参数及幅域参数进行采样,并引入小波分析计算各频带能量值,构建神经网络进行故障判定,选用不同的振动特征参数组合对检验样本进行验证以期获得指向性较好的飞机发电机故障特征参数。诊断结果表明,利用RBF网络对发电机故障诊断,采用基于幅值域的特征参数峭度指标、峰值因子、脉冲指标、裕度指标、歪度和基于频域的重心频率、均方根频率、频率标准差,再考虑进小波包分频带能量值作为神经网络的输入参数指标,可取得良好的诊断准确率。
Aiming at the status of aircraft generator vibration characteristic parameters and fault characteristic parameters difficult to accurately identify the status of aircraft generator health, the generator vibration signal real-time sampling device is designed to sample various frequency domain parameters and domain parameters of aircraft generator when it rotates , And introduce the wavelet analysis to calculate the energy value of each band, construct the neural network to judge the fault, choose different vibration characteristic parameters to validate the test samples to obtain the directional characteristic fault characteristics of the aircraft generator. The diagnostic results show that RBF neural network is used to diagnose generator faults by using kurtosis index, crest factor, pulse index, margin index, skewness and center-of-gravity frequency based on amplitude domain, root mean square frequency, Frequency standard deviation, and then consider the sub-band wavelet packet energy value as an input parameter index of neural network can get a good diagnostic accuracy.