Content area
Abstract
One of the problems faced from the manufacturing industry refers to the optimal positioning of the blank parts within the machine-tool. The reference frame associated to the blank part on the machine-tool should be set in order to ensure sufficiency of material all along the programmed paths, which are generated from a particular design reference frame associated to a nominal model. Many factors can produce such undercutting problems, including the inappropriate positioning of the part on the machine-tool, the size deviation of the resulting blank from its manufacturing process, or even the oversize specification of the part, which is often designed too tight.
The main concern of this thesis is to develop a computer assisted tool, which can optimally balance the blank parts in order to establish an adequate reference frame on the machine-tool. The proposed technique does not only answer, yes or no, to the question "Is there enough material to machine this nominal part from this blank?", but is interested in determining how the unavoidable lack of material can be oriented in order to minimize the production costs. Much of the work existing in the three-dimensional alignment field of application cannot compute an appropriate reference frame based on this basis, which involves the unfeasible domain of solution simultaneously with the feasible domain of solution. From the proposed approach or solution, a software called "BALPART" is developed and integrated to the CAD/CAM software "ACIS".
The method proposed in this thesis, is an iterative process improving the solution in a way such that the constraints are satisfied with a decreasing order of priority. This process includes two important steps, one which best align the data set with respect to the CAD model, and the other which search for a best feasible or infeasible solution regarding the constraints to satisfy. This former step solves, at each iteration, an unconstrained and non linear least squares optimization problem with the Newton-Raphson iterative method. The constrained process involves the minimization of an artificial penalty function, specially developed to successively satisfy each constraint under its priority level. Both, a logarithmic and a root mean square artificial function, designed on this scheme. are compared in terms of results, efficiency, and functionality of the alignment process. (Abstract shortened by UMI.)