The Research Of NURBS Curve Interpolation Based On BP Neural Network

With the rapid development of automobile, electronic engineering and automaticcontrol technology, the traditional manufacturing techniques can not fully meet thepeoples’ need. So only to vigorously develop modern manufacturing technology can wemet the new requirements of production. Because of CNC technology is the key ofmodern manufacturing technology, so it has become a top priority to accelerate thedevelopment of the new open CNC system. The quality of interpolation function willdirectly affect the performance of the CNC system. Therefore, it is important to developthe interpolation technology in the numerical control systems, especially strengtheningthe research of curves and surfaces directly interpolation method. This is important toimprove the interpolation quality when the complex curve parts and surfaces aremachined.NURBS has good mathematical properties, so it can not only accurate and unifiedexpress an analytical curve and free curve, but also has good application in the CAD/CAM domain. So in the process of parts of complex curves and surfaces, NURBScurves and surfaces can be widely used for direct interpolation, which will effectivelysolve the problem of complex parts machining.This thesis will make NURBS curve directly interpolation method as a researchobject, whose purpose is to improve the interpolation precision and efficiency ofinterpolation, the specific contents are as follows:First, the theory of NURBS curves and surfaces will be analyzed, including thehistory of spline curves, the definition and nature of NURBS curves and surfaces. Thenfocuses on the implementation of NURBS curves reverse design, and an example willbe given that use CMM Mitutoyo Crysta-Apex C7106to measure a free curve which ison a Logitech mouse, then using the reverse algorithm to determine this NURBS curve.Then the principle of BP neural network, its composition and design techniques will bedescribed detailed, and apply it to a NURBS curve interpolation to complete thecalculation of the estimated parameter u, so the interpolation efficiency of NURBScurve will be improved. And to make the interpolation method can meet real-timerequirements of CNC machine tools and systems, also considering the impact of thebow high and other factors on the interpolation calculation, finally an instance will beused to test this improved interpolation method what have been mentioned above. VC ++and MATLAB software will be used to complete the design of the automaticinterpolation system of reverse NURBS curve.