【摘要】 针对柴油机缸盖表面振动信号的非平稳时变特点，提出了基于小波包分析的柴油机振动诊断方 法。给出了小波包变换的一种改进算法，通过移频处理，克服了原算法中的频率混叠现象，使分解序列 的排列顺序与频带的线性划分顺序—一对应。通过对完整工作循环内的缸盖振动信号进行小波包分 解，实现了整循环诊断特征向量的快速提取。试验结果表明，该方法在柴油机振动诊断中是有效可行 的，对其它复杂机械的振动诊断同样具有参考价值。更多还原

【Abstract】 An improved algorithm for wavelet packets (WP) using frequency-shift technique is proposed in this paper. The algorithm overcomes the frequency aliasing which exists in the original one,and makes the decomposed WP sequences correspond one by one to the linearly divided frequency bands. Such property provides WP as a powerful tool for analyzing the nonstationary and time-varying signals such as vibration signals measured on the cylinder head of a diesel engine. The feature vector for engine diagnosis can be constructed easily and quickly from the decomposed.WP coefficients of the vibration signal synchronously sampled during an entire working cycle. Then the Euclidean geometric distance function helps give fault information. The experimental results in real-world engines are presented to verify the theoretical developments. It is shown that this technique is simple to implement and may provide practical utility for vibration diagnosis in diesel engines as well as in other complex machines.更多还原