基于K值的铁谱磨粒智能识别研究

Study on the Intelligent Identification Wear Particles Based on the K Value

摘要: 本文采用了K值法作为磨损颗粒分类的判据。通过对样本磨损颗粒图像的K值计算，得出了三种磨损颗粒图像K值的区间。虽然区间中有个别样本的K值存在重叠，但是能作为区分这三种磨损颗粒的依据，对这三种磨粒进行分类。将K值法作为区分上述三种磨损颗粒的判据，可以有效弥补用分形维数区分磨损颗粒时，识别不精确的情况，从而提高了识别的精确度。K值法为铁谱磨损颗粒智能识别提供了一种新的方法，具有一定理论意义和实用价值。

Abstract: A new method named K value method is adopted to classify the wear particles as the criterion of the wear particles. Through analyzing the K value obtained from wear particle image samples, the range of the three kinds of wear particle images’ K value are obtained. Although the K value ranges, the three different particles are still a little overlapped. It is better than the variable metric method, so it can be used as a good method to classify the three different kinds of wear particles. As a criterion to distinguish the above three kinds of wear particles, the K value method can effectively make up for ineffective results from fractal dimension. The K value method can improve the preci-sion of the identification. The K value method provides a new method for ferrographic wear par-ticles intelligent identification and has certain theoretical significance and practical value.