Multimodel Cascade-Based Full-Field Displacement Online Measurement Method of the Aircraft Tooling Positioner

Aircraft tooling positioners (ATPs) ensure that aircraft components are precisely aligned and positioned during assembly. Online measuring of the full-field displacement can better monitor the performance of the tooling and ensure assembly quality. However, due to significant visual obstructions and limited measurement space, only the displacement of a few points that are in the line of sight and visible can be monitored. Data-driven displacement reconstruction techniques offer effective full-field displacement states from discrete point measurements. However, the complex structure of the positioners makes it challenging for data-driven models with low-dimensional displacement measurements as inputs and high-dimensional full-field displacement states as outputs to balance the reconstruction accuracy, training cost, prediction efficiency, and storage space. To address the aforementioned challenges, a multimodel cascaded method for real-time full-field displacement reconstruction is introduced. This method leverages finite displacement data captured by a binocular vision sensor in constrained environments to deduce the full-field displacement state of the tooling positioners. First, the data-driven techniques are employed to establish prediction models that correlate finite sensing points to auxiliary control points (ACPs), extending a few measurable data to broader global envelope points. A high-dimensional interpolation algorithm swiftly reconstructs the full-field deformation using the defined sparse control points. The full-field displacement is subsequently derived by superimposing rigid displacement. In addition, for enhanced precision and efficiency, a resampling method, accelerated by a greedy algorithm, is proposed to optimize the placement and quantity of ACPs. Experimental results indicate that the average relative error of this method is below 7.56%. Moreover, while ensuring precision, it reduces model construction costs, enhances prediction efficiency, and minimizes resource occupation. It provides an innovative idea for real-time and high-precision measurement of full-field displacement.