N-P准则下的多传感器信息融合算法设计  

Multi-Sensor Information Fusion Algorithm Design under N-P Criterion

作　　者：荀亚敏 张杰 翟芸翎 翟灵瑞 XUN Yamin;ZHANG Jie;ZHAI Yunling;ZHAI Lingrui(Weichai Power Co.,Ltd.,Weifang 261061,China)

机构地区：[1]潍柴动力股份有限公司,山东潍坊261061

出　　处：《自动化仪表》2024年第8期85-90,96,共7页Process Automation Instrumentation

摘　　要：传感器是环境感知中实现高度自动化的关键组件。为了实现多传感器系统高水平的抗干扰性和鲁棒性,针对局部传感器稳定性差及精度不足的问题,创新性地提出了一种基于统计判决的多传感器信息融合算法。首先,完成了多传感器融合的基本建模,并设计了单传感器的单值概率特性计算方法。然后,对传统单传感中卡尔曼滤波方法进行了优化设计,消除了累计误差的影响。最后,通过引入奈曼-皮尔逊(N-P)准则完成了多维度评估的多传感器信息融合算法设计。接收者操作特征(ROC)曲线评估试验结果证明了所提算法在理论上的正确性和合理性。相比其他算法,所提算法在抗干扰性和容错性等性能方面有显著优势。实际无人机测试证明了该算法具有一定的实用性。Sensors are the key components for realizing a high degree of automation in environmental sensing.To realize a high level of anti-interference and robustness of multi-sensor systems,a multi-sensor information fusion algorithm based on statistical judgment is innovatively proposed to address the problems of poor stability and insufficient accuracy of local sensors.Firstly,the basic modeling of multi-sensor fusion is completed,and the calculation method of single-value probability characteristics of single sensors is designed.Then,the Kalman filtering method in traditional single sensing is optimally designed and the effect of cumulative error is eliminated.Finally,the design of multi-sensor information fusion algorithm for multi-dimensional evaluation is accomplished by introducing the Neyman-Pearson(N-P)criterion.The receiver operating characteristic(ROC)curve evaluation test proves the theoretical correctness and rationality of the proposed algorithm.Compared with other algorithms,the proposed algorithm has significant advantages in the performance of anti-interference and fault tolerance.The actual unmanned aerial vehicle test proves that the algorithm has some practicality.

关 键 词：多传感器融合 信息融合 统计判决 卡尔曼滤波 奈曼-皮尔逊准则 接收者操作特征曲线 

分 类 号：TH741[机械工程—光学工程]