Research On Registration For Machining And Inspection Of Complex Aircraft Parts

Aviation manufacturing is heading in the direction of integration of structure,large scale of parts and high-precision.Complex aircraft components are increasingly applied.Geometric accuracy directly affects the performance.Consequently,precis e machining and inspection is of signif icance to aviation manufacturing.Precise and efficient 3D registration can realize alignment and comparison between the measured model of complex aircraft components and its corresponding theoretical model.It is one of the key techniques of machining localization and precision inspection in measurement aided manufacturing(MAM).Based on points cloud data from electro-optical measuring methods,the registration problems in aircraft machining and inspection are deeply studied in this paper,including optimizing registration technology for allowance balance of the blank with r igorous allowance,differentiated registration method for complex parts inspection,minimum zone registration and registration stability and reliability analysis.The main innovations are summarized as follows.For preformed blanks with small and ununiform machining allowance,dimensional deformation and datum absence,an adaptive optimal envelope registration method is presented for machining localization.A uniform and smooth approximation to the objective function of the max-min registration is proposed based on the entropy optimization principle.Thereby,the solution of the registration model improves the precision and efficiency of the machining localization.Whether an envelope registration exists or not,the method can always efficiently achieve reasonable localization results without any prior knowledge and/or constraints.That is,for the blank parts that the material shortage is inevitable,the method can still confine the mater ial shortage to a least extent.Thus precise data support for the follow-on feasible soldering or blank process improvement is obtained by the adaptive optimization envelope registration.For the registration problem with complex constraints of process and design requirements,an optimization model with constraints of machining allowance,tolerance and wall thickness is established.Further more,a hierarchical strategy of data processing is proposed in the large-scale nonlinear and constrained optimization problems.The registration is carried out in a high efficiency.High requirement and regional differentiation of the precision in aircraft parts surfaces result from high performance,complicated structures and recombined multi-process in aviation manufacture.To address this inspection issue,a new registration in method is proposed.It features in that a novel dynamic assessment mechanism is introduced in the iterative optimization process to make the registration in favor of the surfaces with higher quality requirements,which is evaluated progressively by approximating of iterative registration.In addition,a distinct fast algor ithm on linearization is proposed to solve the registration optimization problem.This registration strategy based on iteratively known precision offers a flexible and flexible way to the precision verification and proves to be more reasonable for the quality evaluation of aircraft parts.In order to deal with the unfavorable effect on the registration of points with the large deviation,two hierarchical data registration methods are put forward.In the first one,deformed points with large errors were found firstly by the statistical analys is of the current pose.A hierarchical registration is presented based on current control points.The less and accurate control points lead to the result closer to the actual and the higher efficiency.With the strategy,the most common sheet metal forming problem is realized by solving optimization problems with conforming points on one side of the nominal model.Extra load is applied to sheet metal to simulate assembly stress.In the second one,M-estimation is employed for a robust registration method.The effect of the low precision measured data is weakened on the result by the M-estimation functions.The efficiency is improved by approximating deviation function in a closer position by the 1st-order Taylor expansion.Four functions with the different restraints are applicable to the complication of detection in aircraft parts.A unified registration model based on the minimum zone criterion is established,which conforms to the definition of form and position error in ISO standard.The model directly optimizes the minimum contour error.To deal with the computational intractability,an aggregate function is used to approximate uniformly the non-differentiable range function.An efficient limited-memory Quasi-Newton algorithm is prevalent for large-scale and differentiable unconstrained optimization problems.The proposed method is easy to implement,works well for dense measurement points,and has no special requirements of geometric property.An effective and reliable method of registration data layout is presented.The reference value of the optimal constraint strength and the critical of the constraint strength for registration are proposed from the quantity index based on experimental studies.The reference value of the optimal constraint strength for complex surface registration is the target of the r egistration data layout.The weakest direction of registration constraint is found by analyzing the information matrix,which is the pointed direction for registration data layout.