机车车辆车轮多边形检测算法研究
Algorithm Research on Locomotive Wheel Polygon Detecting

车轮不圆度作为机车车辆车轮运行安全检测的最重要的参数之一，其是否能被准确的检出直接决定着列车的行车安全。文章介绍了一种机车车辆领域的不圆度检测系统，针对周期不圆的特殊情况–多边形做了详细的研究。该系统通过测量车轮的轮缘高度变化量，进而得到反应车轮不圆度状况的波形，针对波形中电噪声及钢轨磨耗及安装偏差等起伏干扰，提出基于传统谱减算法的改进方法，对波形进行去噪预处理，去噪效果明显，降低了多边形漏报风险，提高了多边形检测准确率。
The out of roundness of wheels is one of the most important parameters for the safe operation of locomotive and vehicle wheels, and whether it can be accurately detected directly determines the safety of train operation. This article introduces a non circularity detection system in the field of locomotives and vehicles, and conducts detailed research on the special case of periodic non circularity—polygons. The system measures the height variation of the wheel flange to obtain a waveform that reflects the out of roundness of the wheel. In response to fluctuations such as electrical noise, rail wear, and installation deviation in the waveform, an improved method based on traditional spectral subtraction algorithm is proposed to preprocess the waveform for denoising. The denoising effect is significant, reducing the risk of polygon miss reporting, and improving the accuracy of polygon detection.