A Relative Pose Measurement Method for Multi-Shield TBM Adjacent Shields Based on Multi-Sensor Integration

In addressing the challenge of measuring the relative pose between adjacent shields in Multi-Shield tunnel boring machine (TBM), a Multi-Sensor integrated measurement method combined with artificial neural networks (ANNs) is proposed. The adjacent shields, along with the articulation cylinders, collectively form a parallel mechanism. While solving for the relative pose between the shields, the parallel arrangement of the articulation cylinders results in insufficient constraints provided by the lengths of the cylinders. After analyzing the sensitivities of changes in the articulation cylinder strokes and relative pose parameters, additional sensors, including a photosensitive target, a collimated laser, and two inclinometers are added. These are integrated with the articulation cylinder displacement sensors to form the sensing system. An ANN model is established to calculate the relative pose between adjacent shields based on data from these sensors. The required dataset for training the ANN is obtained through the kinematic inverse solution of the parallel mechanism. The proposed method was validated on the emulation test platform for pose measurement. Compared with the existing methods, it achieved higher accuracy and stronger environmental adaptability to dust and fog interference. The proposed method was also applied to a 6-m class double-shield TBM, it achieved an accuracy of 4 mm for relative positions and 1 mrad for relative attitudes, meeting the guidance requirements for Multi-Shield TBM.