Research on Multi-sensor On-machine Measurement Method for Surfaces Based on Residual-Adaptive Sampling

Abstract On-machine measurement is a critical step for acquiring the true part state and enabling closed-loop control in the high-precision machining of curved thin-walled parts in the aerospace field. However, a single sensor struggles to achieve an effective balance between measurement efficiency and accuracy. Existing research on multi-sensor fusion for on-machine measurement generally overlooks the dynamic optimization between data acquisition and fusion algorithms. To address these challenges, this paper proposes a residual adaptive sampling method for multi-sensor on-machine measurement of curved surfaces. The method iteratively optimizes the sampling process by constructing a residual model and employing a residual adaptive sampling strategy. First, a surface model is constructed based on point cloud data from an initial measurement; Then, a residual model is established, and the sampling process is iteratively optimized using a residual-adaptive sampling strategy; Finally, a high-precision residual model is established to correct the surface model. Experimental validation demonstrates that the proposed method outperforms both single sensors and traditional single-sampling strategies in terms of measurement efficiency and accuracy.