关键词: B样条曲线拟合, 激光位移传感器, 节点配置, 克隆选择算法, 石油钻杆螺纹

Abstract: Considering the measurement difficulties caused by loading and unloading processes of drill pipes with long size and great weight, a on-machine precise measuring method is proposed for the internal and external thread contour parameters of drill pipe and has been employed on CNC repairing turning machine of petroleum thread successfully. The method takes advantage of NC cutter frame to drive 2D laser displacement sensor which gains data from a certain shaft section of thread. In order to amend sensor measurement error due to the inclination of the measuring point, a quantifiable inclination error model is built to compensate in real time. Compensative data fitting with B-splines is applied and the improved elitist clonal selection algorithm for knot adjustment of B-spline curves is adopted to resolve this multimodal nonlinear optimization problem. The Hannan-Quinn information criteria is adopted as the affinity measure to gain adequate trade-off between the quality of fitness and the complexity of the model, which also computes the number of interior knots automatically and accurately, thus the optimized knot vector is determined. Finally, the best fitting model with optimized knot vector is used to obtain the parameters of thread. The inclination error experiment indicate that the compensative model is accurate and efficient, which can improve data collection precision of the sensor. Simulation results show that the data reconstruction method with improved elitist clonal selection algorithm has high precision in terms of knots adjustment for B-spline curve and performs better than other intelligent algorithms. The measurement of NC50 drill pipe thread show that the above measure method can realize the on-machine measuring with high precision of drill pipe thread and the measurement accuracy is ±2 μm.

Key words: B-spline curves fitting, clonal selection algorithm, drill pipe thread, knot adjustment, laser displacement sensor