矢量量化在钢水连铸下渣检测方面的应用

Application of vector quantization theory in steel slag detection of continuous casting

针对当前钢水连铸下渣检测系统(SWCC-SDS)成本高、难于安装等问题,提出了一种基于矢量量化(VQ)技术的钢水下渣检测方法.对钢水连铸下渣过程进行了研究,根据连铸过程中钢水、钢渣产生的振动差异,确定了把振动信号作为主要检测信号的实施方案.利用振动传感器,搭建了远离钢水的控制系统,解决了钢渣检测的传感器易耗问题.通过VQ技术对经过预处理的实时信号进行特征提取、码本训练,来判断钢水状态,实现连铸下渣的自动控制.实验结果表明,该方法成本低,对现在有设备改造小,下渣检出率在95%以上,符合钢水连铸的生产要求.

Abstract. To overcome problems of steel water continuous casting slag detection system （SWCC-SDS）, such as high cost and equipment inconvenience, a new SDS based on vector quantization （VQ） technology was put forward. The process detail of SWCC was investigated and analyzed, and the vibration signal was considered to be the main detecting signal according to the difference generated by steel water and slag during the course of SWCC. Automatic control system was established far away from steel water with vibration sensor, which could resolve the problem of the sensor＇s uncontrolled consumption. After pre-processing the real-time vibration signal, characteristic information was extracted and codebook was trained by VQ, then the status of steel water was identified to realize automatic control for SWCC. Experimental resuits proved that this system requires low cost and little re-building for current devices, and that the slag detection ratio can be more than 95%, and can meet the requirements of SWCC.