When remanufacturing complex surface parts, such as twisted blade, it is difficult to obtain an accurate model. An iterative Genetic-algorithm-based-surface reconstruction method for repair of twisted blade is presented. Genetic algorithm is applied in parametrizing data points and computing knot vectors. Then, the control points of the fitting B-spline surface are calculated by least-squares approximation through either SVD or LU methods. It shows that the accuracy of the method is improved significantly when three different twisted blades surfaces are verified by using the method.
Extending Neural Networks for B-Spline Surface Reconstruction
