Empirical mode decomposition based novel data compression algorithm for wireless data transmission in machine health monitoring
基於經驗模態分解 (EMD) 的用於設備健康監測中傳送無線數據的新型數據壓縮算法
Student thesis: Master's Thesis
|Award date||2 Oct 2009|
Monitoring the health of a machine requires a considerable amount of sensory data inputs. Nevertheless, inasmuch as monitored machines may operate in a hazardous environment or are moveable, wireless transmission becomes the only possible means for data acquisition. To ensure precision in data acquisition, especially in vibration-based machine fault diagnosis, high sampling rate and a large number of sampling points are required. However, they are impractical in a wireless data transmission context due to the usual low transmission rate and lengthy transmission time. One possible solution is to compress the numerous sensory data to an allowable number of samples before wireless transmission commences. In this research, a low-cost wireless data acquisition system suitable for monitoring the health condition of a machine has been developed. To facilitate fault diagnostic techniques which require two accelerometers to collect vibrations simultaneously, such as orbit analysis and Active Noise Cancellation (ANC), the wireless system is able to host at least two ICP type sensory inputs. In addition, a novel data compression algorithm which combines Empirical Mode Decomposition (EMD) with Differential Pulse Code Modulation (DPCM) has been developed. EMD is a new decomposition technique which can decompose and identify any instantaneous changes in non-linear and non-stationary signals caused by the anomalous operation of machines efficaciously. After the data have been decomposed and compressed by EMD, DPCM is applied to further compress the data through the embedded linear predictor and quantizer prior to the commencement of transmission. To enhance the effectiveness of the proposed data compression algorithm when it is applied to a noisy environment, the noise filtering ability of EMD is also investigated. The results prove that, after this new algorithm has been applied to real machines, it provides much faster wireless data transmission by significantly reducing the size of sensory vibration data. While the compressed data have been reconstructed back to their original form at the receiver end, the integrity of the sensory data can be maintained, with negligible reconstruction errors. With the help of this new algorithm, even the requirement of large data sampling – commonly required by vibration-based fault diagnosis – can be fulfilled in wireless data communication.
- Data compression (Telecommunication), Machinery, Wireless communication systems, Monitoring