刘煜辉  高艳强

（许昌职业技术学院机电与汽车工程学院，河南 许昌 461000）

摘要：针对工业生产线故障预警过程中，因不同传感器采集的数据维度不同，导致预警效率低的问题，提出一种多传感器融合的工业生产线故障预警技术。首先，通过投影的方式对不同传感器数据进行融合转换，将传感器坐标系下工业生产线设备的位置信息转换成平面坐标系下的数据形式；然后，利用光束平差算法进行坐标系融合，得到工业生产线设备的位姿；最后，利用深度置信网络计算工业生产线设备的位姿与目标状态之间的关系，确定工业生产线设备是否处于故障状态，并及时做出相应的预警。经测试，该技术对不同类型、不同程度的工业生产线故障均实现了高效预警，时间开销稳定在1.0 s以内。

关键词：多传感器融合；工业生产线；故障预警；投影；光束平差算法；坐标系融合转换

中图分类号：TP277            文献标志码：A          文章编号：1674-2605(2024)02-0007-04

DOI：10.3969/j.issn.1674-2605.2024.02.007

Industrial Production Line Fault Warning Technology Based on       Multi-sensor Fusion

LIU Yuhui  GAO Yanqiang

(School of Mechanical-electronic and Automotive Engineering, Xuchang Vocational Technical College,  Xuchang 461000, China)

Abstract: In order to solve the problem of low early warning efficiency caused by different data dimensions collected by different sensors in the process of industrial production line fault early warning, a multi-sensor fusion technology for industrial production line fault early warning is proposed. Firstly, different sensor data are fused and converted by projection, and the position information of industrial production line equipment in the sensor coordinate system is converted into the data form in the plane coordinate system; Then, using the beam adjustment algorithm for coordinate system fusion, the pose of industrial production line equipment is obtained; Finally, using deep confidence networks to calculate the relationship between the pose of industrial production line equipment and the target state, determine whether the industrial production line equipment is in a faulty state, and make corresponding warnings in a timely manner. After testing, this technology has achieved efficient early warning for industrial production line faults of different types and degrees, with a stable time cost of less than 1.0 second.

Keywords: multi-sensor fusion; industrial production line; fault warning; projection; beam adjustment algorithm; coordinate system fusion conversion