Missing-Measurements-Tolerant Compressed Sensing in Wireless Sensor Networks for Mechanical Vibration Monitoring

Compressed sensing (CS) can significantly improve the transmission efficiency of large amounts of vibration data in wireless sensor networks (WSNs) for mechanical vibration monitoring. To address the issue of irrecoverable measurements loss due to unstable communication links in WSN, this article proposes a missing-measurements-tolerant CS (MMTCS) in WSN for mechanical vibration monitoring. First, the embedded compressed sampling (ECS) is designed to compressed sampling the original signals in the acquisition nodes, thereby enhancing transmission efficiency. Moreover, the article analyzes the missing measurements perturbation error caused by compressed sampling and measurements loss in wireless transmission. An objective optimization function is derived for missing measurements. Combined residual adaptive sparse reconstruction (CRASR) is proposed for accurate data reconstruction. The experimental results demonstrate that the proposed method achieves a better trade-off between reconstruction accuracy and reconstruction time in comparison with other popular methods. More importantly, the proposed method can achieve satisfactory fault detection accuracy for rotating machinery under some degree of compressed sampling and missing measurements. This is of great value to practical engineering applications and provides a practical and effective solution.