基于Laplace先验和稀疏块相关性的旋转机械振动信号贝叶斯压缩重构  被引量:6

Bayesian Compression and Reconstruction for Rotating Mechanical Vibration Signal Based on Laplace Prior and Sparse Block Correlation

在线阅读下载全文

作  者:马云飞 白华军 温亮 郭驰名 贾希胜 MA Yunfei;BAI Huajun;WEN Liang;GUO Chiming;JIA Xisheng(Department of Armament,Noncomissioned Officer Academy of CAPF,Hangzhou 310023,Zhejiang,China;Equipment Command and Management Department, Shijiazhuang Campus, Army Engineering University,Shijiazhuang 050003,Hebei,China)

机构地区:[1]武警士官学校军械系,浙江杭州310023 [2]陆军工程大学石家庄校区装备指挥与管理系,河北石家庄050003

出  处:《兵工学报》2021年第12期2762-2770,共9页Acta Armamentarii

摘  要:为通过无线传输实时监测装备状态,针对机械振动信号采样频率较高导致压缩重构困难的问题,将Laplace先验模型和振动信号周期性稀疏块相结合,提出一种改进的贝叶斯压缩感知算法。建立基于Laplace分布的贝叶斯先验模型,相对于高斯先验具有更强的稀疏促进作用。根据机械设备转速和采样频率计算振动信号类周期,对信号进行周期性分块,并基于多稀疏块共享相同超参数的特点,采用快速相关向量机迭代估计出原始信号期望。选取两级平行轴齿轮箱作为研究对象,进行压缩重构仿真实验。结果表明,该方法在相同稀疏基下能有效改善机械振动信号的重构效果。For the difficult compression and reconstruction of mechanical vibration signal caused by high sampling frequency,an improved Bayesian compressive sensing algorithm is proposed by combining Laplace prior model with periodic sparse block of vibration signal for monitoring the equipment condition in real-time through wireless transmission.A Laplace distribution-based Bayesian priori model is proposed,which has stronger sparse promotion effect compared to Gaussian priori model.The vibration signal period is calculated according to the rotational speed and sampling frequency of mechanical equipment for dividing the signal periodically.The original signal expectation is estimated iteratively by fast correlation vector machine based on the feature of that multiple sparse blocks share the same hyperparameters.A two-stage parallel gearbox is selected as the research object.The compression and reconstruction simulations were carried out.It is found that the proposed method can effectively improve the reconstruction effect of mechanical vibration signals using the same sparse basis.

关 键 词:机械振动信号 Laplace先验 稀疏块 贝叶斯压缩 压缩感知 

分 类 号:TH113.1[机械工程—机械设计及理论]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象