摘要
考虑到空间运动特性差异对机械性能退化的体现更为直接,且振动信号蕴含丰富的机械状态信息,提出一种利用振动信号对空间运动特性进行表征的断路器健康状态识别方法。首先,利用位移信号获取能够反映关键机构机械状态的运动特性参数;其次,采用AAF-AAKR模型构建运动特性健康指标;然后,基于关键动作阶段三维振动信号特点提取多域特征参数,选取相关性较高的特征进行层次聚类并计算与运动特性的互信息,得到对运动特性表征能力强的关键退化特征矢量;最后,将退化特征矢量作为输入,运动特性健康指标作为输出,构建1D-CNN性能退化回归模型,以实现储能机构健康状态识别。实例验证表明,三维振动信号相对于一维振动信号对运动健康指标的拟合效果更好,回归分析RMSE为0.0186,MAE为0.0112,可精准地识别断路器的健康状态。
Considering that the difference in spatial motion characteristics is more direct to the mechanical property degradation,and the vibration signal contains rich mechanical state information,a circuit breaker health state identification method is proposed using vibration signals to characterize spatial motion properties.First,the displacement signal is used to obtain the motion characteristic parameters that can reflect the mechanical state of the key mechanism.Secondly,the AFF-AAKR is utilized to construct motion-characteristic health indicators offline.Then,multi-domain feature parameters are extracted based on the characteristics of the three-dimensional vibration signal in the key action phase.The features with higher correlation are selected for hierarchical clustering and the mutual information with the motion characteristics is calculated to achieve the key degradation feature vectors with strong characterization ability of the motion characteristics.Finally,the degradation feature vectors are used as the input and the health indicator of motion characteristics is used as the output to construct a 1D-CNN performance degradation regression model.In this way,the health state identification of the energy storage mechanism is realized.The example validation shows that the three-dimensional vibration signal fits the motion health indicator better than the one-dimensional vibration signal,and the regression analysis RMSE is 0.0186 and MAE is 0.0112,which can accurately identify the health status of the circuit breaker.Considering that the difference of spatial motion characteristics is more direct to the mechanical property degradation,and the vibration signal contains rich mechanical state information,we propose a circuit breaker health state identification method using vibration signals to characterize spatial motion properties.First,the displacement signal is used to obtain the motion characteristic parameters that can reflect the mechanical state of the key mechanism.Secondly,the AFF-AAKR is used to construct motion characteristic health indicator offline.Then,multi-domain feature parameters are extracted based on the characteristics of three-dimensional vibration signal in the key action phase,and the features with higher correlation are selected for hierarchical clustering and the mutual information with the motion characteristics is calculated to obtain the key degradation feature vectors with strong characterization ability of the motion characteristics.Finally,the degradation feature vectors are used as the input and the health indicator of motion characteristics is used as the output to construct a 1D-CNN performance degradation regression model,in order to realize the health state identification of the energy storage mechanism.The example validation shows that the three-dimensional vibration signal fits the motion health indicator better than the one-dimensional vibration signal,and the regression analysis RMSE is 0.0186 and MAE is 0.0112,which can accurately identify the health status of the circuit breaker.
作者
孙曙光
王浩宇
王景芹
李奎
郝永耀
Sun Shuguang;Wang Haoyu;Wang Jingqin;Li Kui;Hao Yongyao(School of Artificial Intelligence,Hebei University of Technology,Tianjin 300130,China;Provincial and Ministerial Co-construction Collaborative Innovation Center on Reliability Technology of Electrical Products,Hebei University of Technology,Tianjin 300130,China;State Key Laboratory of Reliability and Intelligence of Electrical Equipment,Hebei University of Technology,Tianjin 300130,China;Jiangsu Sifang Clean Energy Equipment Manufacturing Co.,Ltd,Xuzhou 221000,China)
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2024年第10期50-62,共13页
Chinese Journal of Scientific Instrument
基金
河北省中央引导地方科技发展资金(246Z2101G)
河北省自然科学基金创新群体(E2024202298)
河北省自然科学基金(E2021202136)项目资助。
关键词
万能式断路器
三维振动信号
空间运动特性
健康状态识别
conventional circuit breaker
three-dimensional vibration signal
spatial motion characteristics
health state identification