信号盲区下基于SLAM的石化施工现场人员定位方法研究

孔 婷

摘  要

石化行业是国民经济的支柱产业之一，同时也是一个高危行业。石化施工现场作业面狭窄、交叉作业频繁、人员高度密集，安全事故频发。工人的实时定位管理被认为是保障石化现场安全最直接有效的途径之一。但当前的研究主要是采用以RFID、蓝牙等Outside-in（由外向内）定位技术，这类技术完全依赖定位信号的强度及覆盖率，在金属构件密集、大型设备广泛存在的这种强干扰、遮挡严重的石化场景中普遍存在定位信号弱、定位不连续或定位失效等信号盲区，直接造成定位数据的缺失，影响安全管控的质量。因此，如何解决石化现场信号盲区下的人员实时定位，成为石化工程建设中亟待解决的难题。

为此，本文引入了基于SLAM的工人定位方法。首先，结合石化施工现场特点，分析了现有人员定位系统的不足并列出信号盲区所对应的具体石化场景，从而提出了信号盲区下定位技术选型。然后，建立了基于ORB-SLAM的工人定位模型，通过不同影响因素下的对比实验，分析出SLAM对弱光或快速运动表现出极强的敏感性，定位效果急剧下降；在转弯时存在明显的定位丢失。为了提高SLAM定位的鲁棒性，本文引入IMU建立了ORB-SLAM与IMU融合的工人定位模型，并在同样的实验环境下进行性能比较。研究结果表明，该融合定位方法解决了SLAM由于工人快速移动等带来的瞬时定位数据丢失问题，弥补了其在复杂环境中定位的局限性，在不同光照、移动速度下，均能完整的进行运动轨迹跟踪，定位误差基本上控制在1-2%。

最后，以武汉某石化烷基化项目为背景验证了该融合定位方法的有效性和稳定性。结论表明，本文所提出基于SLAM与IMU融合的定位方法可以实现石化现场信号盲区下的工人实时定位，克服了现有Outside-in定位技术的不足，为信号盲区下的定位管理提供了可靠的技术支持。通过该定位方法可以有效地降低石化工程施工安全事故发生的概率，对保证现场安全、落实石化HSSE安全管理规定具有重大意义。

关键词：石化施工；现场安全管理；施工人员定位；信号盲区；SLAM；惯性测量单元

Abstract

Petrochemical industry is one of the primary industries for the national economy, and it is also regarded as one of the most high-risk industries. With the narrow construction space, a large number of cross-cutting operations and highly densely population, safety accidents frequently occurred during the construction process. Real-time tracking and location of people has proven to be an effective way for safety management in petrochemical sites. Currently, researches mainly focus on Outsdie-in positioning technology(eg., RFID and UWB) for real-time location management of people. However, this type of positioning technology completely relys on the strength and coverage of the positioning signal, which cannot be affordable in complex petrochemical sites. Serious siginal occlusion and signal interference caused by the number of metal components, tanks, and other components makes a seriers of signal blind areas in this scenario and cannot output sustainable positioning data, which lead to the issue of robustness and stability. Therefore, how to achieve efficient tracking of workers in the petrochemical site under the signal blind zone becomes an urgent problem.

In order to trackle this problem, a SLAM-based positioning method was introduced in this paper. Firstly, this paper combines the characteristics of the construction site of petrochemical engineering and the current research status of the positioning system, analyzes the shortcomings of the existing personnel positioning system, and proposes a comparison of construction personnel positioning technology under the signal blind zone. Then, according to the SLAM positioning principle and its main influencing factors, an ORB-SLAM on-site worker positioning model is established, and its positioning robustness is not strong through experimental analysis. Moreover, in order to further improve the robustness of SLAM positioning and combine the characteristics of IMU positioning, a new method combining ORB-SLAM and IMU is proposed. This proposed method considers the problem of positioning data loss due to instantaneous SLAM motion. The results showed that the fusion method can make up the tracking lost in SLAM mode. It could completely record the whole trajectory and accuracy can be limited in 1-2%.

Finally, taking a petrochemical project as an example, the effectiveness and robustness of the method are verified in the signal-blind areas of the petrochemical field. The results presented in this research show that the fusion positioning method proposed in this paper can realize the real-time tracking of people in signal-blind areas. We suggest that the method can be used for site managers to effectively reduce the probability of safety accidents in the construction of petrochemical projects and ensure site safety.

Keywords: Petrochemical project; On-site construction management; Worker Positioning; Signal-blind Area; Simultaneous localization and mapping (SLAM); Inertial measurement unit (IMU)