分形维数在基于噪声的发动机故障诊断中的应用

Application of Fractal Dimension to Engine Fault Diagnosis Based on Noise

将分形理论引入发动机噪声信号分析中,介绍了基于信号重构的发动机噪声分形维数G-P算法,运用自相关函数法和伪相图法讨论了信号重构中延迟时间的选取,并通过分析比较分形维数随嵌入维数的收敛关系,确定了分形维数计算中合适的嵌入维数。从不同工况下柴油机噪声信号分形维数的变化关系中发现:基于发动机噪声的分形维数可以定性表征发动机故障,在一定程度上可作为内燃机工作状态监测和故障诊断的一个特征量。

By introducing the fractal theory to engine noise analysis, the G-P algorithm of noise fractal dimension based on the signal reconstruction was introduced, the delay time selection in the signal reconstruction was discussed with the autocorrelation function and false phase diagram method, and finally the appropriate embedding dimension in the fractal dimension calculation was determined by analyzing and comparing the convergence of fractal dimension with embedding dimension. According to the fractal dimension variation of engine noises under different conditions, it was found that the noise fracta[ dimension could characterize the engine faults and could be used as a feature to monitor state and diagnosing fault to some extent.