基于噪声容错UWB定位的高程越界检测算法研究

Research on Noise-tolerant UWB Localization for Elevation Cross-boundary Detection

针对变电站改扩建施工过程中高程越界违章检测的实际需求,探讨采用一种基于超宽带(UWB)定位技术的虚拟电子围栏方案替代传统的物理围栏。UWB本质上是一种基于测距的无线定位技术,其定位精度依赖于UWB节点间距离信息的完备和准确程度。然而,在变电站改扩建施工的复杂场景下,UWB节点所获取到的距离信息往往是缺失或含噪的。为此,提出一种基于混合指数幂分布的噪声容错超宽带定位算法,该算法一方面引入混合指数幂分布拟合不可预知的复杂噪声,另一方面采用低秩矩阵分解技术补全缺失的距离信息。与已有的其他方法相比,所提出的噪声容错UWB定位算法能很好地重构UWB节点间的距离信息,从而确保了复杂环境下高程越界违章检测方案的可行性。

Aiming at the actual needs of elevation cross-boundary detection in the process of substation reconstruction and expansion, a virtual electronic fence based on ultra-wideband(UWB) positioning technology is proposed to replace the traditional physical fences. UWB is essentially a wireless ranging-based positioning technology, and its positioning accuracy depends on the completeness and accuracy of the distance information between UWB nodes. However, in the complex scene of substation reconstruction and expansion, the distance information obtained by UWB nodes is often missing or noisy. For this reason, a noise-tolerant UWB localization algorithm based on mixture of exponential power distribution(MoEP) is further proposed. On the one hand, the algorithm introduces MoEP to fit the unknown complex noise, and on the other hand, it employs the low-rank matrix decomposition technology to fill in the missing distance information. Compared with other existing methods, the proposed noise-tolerant UWB localization algorithm can well reconstruct the distance information between UWB nodes, thus ensuring the feasibility of the elevation cross-boundary detection in complex environment.