针对井架钢结构冲击载荷振动信号非线性、非平稳性对损伤识别的干扰问题,提出了一种基于希尔伯特振动分解(Hilbert Vibration Decomposition,HVD)与遗传算法优化的神经网络(Genetic BP Neural Networks,GA-BP)相结合的智能故障诊断方法.首先,利用HVD分解的方法处理冲击载荷作用下的加速度非平稳振动信号;其次,由斯皮尔曼相关系数选取HVD分解后的最优(Intrinsic Mode Function,IMF)分量,以最优IMF分量能量变化率构造特征向量;最后,通过特征向量建立数据集进行神经网络训练,完成信号的特征学习和故障分类.利用ZJ70型井架钢结构模型进行冲击载荷作用下的单处损伤和多处损伤的不同工况实验验证,结果表明:对于单处损伤位置识别率达到90%,多处损伤位置识别率高达96%,利用HVD分解与GA-BP神经网络相结合的方法具有较好的稳定性,能够准确判断出井架钢结构损伤位置,具有一定的实际应用价值.
Damage Identification of Oil Derrick Steel Structure Based on HVD Decomposition and GA-BP Neural Network
Aiming at the interference of non-linear and non-stationary vibration signal of the impact load of an oil derrick steel structure on damage identification,an intelligent fault diagnosis method based on Hilbert vibration decomposition(HVD)and genetic algorithm optimization(GA-BP)neural network is proposed.Firstly,the HVD decomposition method is used to deal with the non-stationary vibration acceleration signal under the action of impact load.Then,the optimal IMF component after HVD decomposition is selected by the Spearman correlation coefficient,and the eigenvector is constructed by the optimal energy change rate of the IMF component.Finally,a data set for neural network training is established through the feature vector,and the signal feature learning and fault classification are completed.The ZJ70 oil derrick steel structure model is used to verify the single damage and multiple damages under different working conditions and impact load.The results show that the recognition rate of single damage location is up to 90%,and the recognition rate of multiple damage locations is as high as 96%.Therefore,the method of combining HVD with GA-BP neural network has good stability and can accurately determine the damage locations of the derrick steel structure,which has a certain practical application value.
国家科技期刊平台
NSTL
万方数据
