基于粗糙集的电机机械状态特征的约简研究

RESEARCH ON FEATURES REDUCTION IN MECHANICAL CONDITION RECOGNITION ABOUT MOTORS BASED ON ROUGH SET

针对集装箱起重机起升电机振动特征参数较多的特点,在利用小波包分解算法进行特征提取的过程中,引入基于粗糙集理论的属性约简方法,在保证识别精度的前提下约简特征参数的维数,以便更高效地进行机械状态识别。同时,引入Wallace测度对约简后的特征属性集与约简前进行识别精度比较。实验结果显示,约简前后的识别结果和精度基本相同,而特征属性维数大大减少,从而大大降低了聚类识别过程的复杂程度和计算量。

In view of too many attributes in the condition recognition of driving system of Container Crane, an attributes reduction method based on rough set is introduced in feature selection using wavelet packets in order to achieve the recognition more rapidly. Then, the Wallace measure is introduced for comparison between the clustering results based on the original data set and the reduced one. the experiments show that both sets give almost the same results, while the dimension of later set is much smaller, which simplifies greatly the complexity and computation of clustering.