符号动力学信息熵参数优化方法研究

Study on the Method for Optimization of Parameters in Symbolic Dynamic Entropy

在机械振动信号特征提取中应用符号动力学信息熵具有很好的效果,然而其算法中的嵌入维数和延迟时间等参数严重依赖人为经验确定,成为了符号动力学信息熵工程应用中的瓶颈。根据符号动力学信息熵的算法原理,提出符号动力学信息熵延迟时间和嵌入维数独立确定和联合确定等两种参数确定方法,并使用试验信号对此两种方法进行了检验和对比。结果表明,独立确定延迟时间和嵌入维数方法相比联合确定方法具有更好的效果,为符号动力学信息熵的应用提供了新思路。

Symbolic dynamic entropy (SDE) has a good application effect in the feature extraction of mechanical vibration signals. However, determination of the parameters such as embedding dimension and delay time in the SDE algorithm strongly depends on human experience, which becomes a bottleneck for engineering application of the SDE. In this paper, according to the algorithm principle of SDE, the independent determination and joint determination methods about the parameters in SDE algorithm are proposed. The two methods are tested and compared using test signals. The result demonstrates that the independent determination method has better effect than the joint determination method in the determination of time-delay and embedded dimensions. This study provides a new idea for the application of SDE.