To address challenges of high energy consumption in data transmission and significant delays in fault prediction during full-lifecycle monitoring of rotating machinery systems, a novel condition monitoring method is proposed by integrating dynamic sparse optimization with edge collaborative computing. A three-tier architecture—comprising the terminal sensing layer, edge computing layer, and cloud analysis layer—is constructed. At the terminal layer, a condition-adaptive compressed sensing mechanism is designed to dynamically adjust the sampling rate. At the edge layer, a spatiotemporal two-dimensional clustering algorithm is introduced to perform correlation analysis between vibration signals and temperature data. At the cloud layer, an Unscented Kalman Filter (UKF) is employed to enable trend prediction and real-time early warning over the entire lifecycle. Experimental results demonstrate that, compared to traditional methods, the proposed approach reduces data transmission volume by 87.6% while maintaining the same level of monitoring and prediction accuracy.

wireless sensor networks,edge computing,life-cycle monitoring,cloud-edge collaborating,Unscented Kalman Filter