基于多信号特征融合的煤料漏斗堵塞检测研究

A Coal Hopper Blockage Detection Based On Multi-Signal Feature Fusion

针对港口煤料堵塞检测的要求,提出一种基于多信号特征融合分析的煤料漏斗堵塞检测方案。首先分别对振动传感器,声音传感器单位窗内的短时能量,短时过零率,最大频谱组成的特征向量进行提取,然后利用相邻特征向量的欧氏距离结合卡尔曼滤波实现瞬间异常判断,最后通过上述异常特征综合判断煤料漏斗是否堵塞。利用Labview搭建的虚拟仪器平台开展了实验研究,结果表明,能快速实现漏斗堵塞检测,并且具有较高的准确率。

Aiming at the requirements of high real-time and high accuracy for a coal hopper blockage detection in port, the paper proposed a program based on multi-signal feature fusion analysis to detect the coal hopper blockage. First we respectively extracted short-time energy and short-time zero crossing ratio, maximum spectrum of feature vectors of unit box of the vibration sensors and acoustic sensors, then use the Euclidean of neighboring eigenvectors combined with Kalman filter to judge the transient abnormal. Finally, the coal hopper blockage was synthetically judged by anomaly features. The results of the experiments based on Labview virtual instrument platform showed that the scheme can quickly achieve funnel signals with high accuracy.