The Design And Optimization Of Locator Arrangement And Measuring Points For Measuring System Of Complex Thin-wall Parts

With the development of the automobile market and increasingcompetition, the quality control of auto parts is becoming an importanttopic of the automobile enterprises. Auto parts manufacturing sizedeviation directly affect the quality of the products. In order to slove thedetection and quality control difficulty of the auto manufacturingenterprise, the optimization of the locator arrangement for the complicatedthin-walled parts and the design and optimization of the measuring pointsfor the parts are studied.In the aspect of the optimization of the locator arrangement,optimization methods are presented to determine the locator arrangementfor the measurement fixture. Based on the “N-2-1” locating principle, thefinite element method (FEM) is used to calculate the Maximumdeformation of the work piece which is used to be objective function of theoptimization. The locator optimization system is proposed based on NXNastran and Isight, using the Isight software integration NX Nastran toimprove the efficiency of the optimization.Two optimization methods are presented in this paper. In one method，the Particle Swarm Optimization with fine performance of globalconvergence is adopted as an optimization solver to seek for the optima.One case is used to illustrate the flowchart of this method. The resultshows that the method has good performance；In the other method, Latinhypercube design is used as DOE method for sample collection, SupportVector Regression is used to get the surrogate model; Particle Swarm Optimization is used to get the optima of the surrogate model. NX Nastranis used to calculate the simulation result of the optima. The same case isused to illustrate the flowchart of this method.The result shows thatoptimal value in the second method is nearly the same with that optimalvalue in the first method, but the efficiency is improved a lot.In the aspect of the design and optimization of the measuring points,this paper puts forward the concept of prediction matrix. Prediction matrixis used to discriminate the different contribution of the measuring points.The contribution is the standard for the collection. Some measuring pointsare removed for their low contribution data to improve the efficiency andreduce the cost in the measuring process.Through the optimization of the locator arrangement and measuringpoints, the influence of the locator arrangement and measuring points iseffectively reduced in the detection process for the test results. Thedetection accuracy gets improved and the efficiency gets better.