Parameter Identification of Fixtures Based on Artificial Neural Networks and Regression Analysis for Ensuring the Efficient Machining  
Author Vitalii Ivanov

 

Co-Author(s) Ivan Pavlenko; Ivan Kuric; Vladyslav Andrusyshyn

 

Abstract The paper presents the scientific approach for the comprehensive use of computational intelligence systems, numerical simulation data, and mathematical modeling using regression procedures for identification parameters of the complicated mechanical system “machine tool – fixture – cutting tool – workpiece”. The developed methodology is realized in the example of a flexible fixture for fork-type parts machining. As a result, the dependencies between the system of clamping and cutting forces/moments and corresponding generalized displacements of points/surfaces are obtained. The proposed methodology allows evaluating the maximum value of displacements for a workpiece under the influence of clamping and cutting forces and moments. The developed methodology is clarified using an artificial neural network for the case of significant nonlinearities in contact interactions between functional elements of the considered system. Finally, the reliability of the developed approach is proved experimentally with the maximum relative error of about 1 %. Since the permissible tolerance is equal to 21 μm, the obtained result ensure the reliability of the proposed design scheme of the mechanical system “fixture – workpiece”.

 

Keywords computer-aided fixture design, mathematical modeling, numerical simulation, computational intelligence systems
   
    Article #:  DSIS19-84
 
Proceedings of ISSAT International Conference on Data Science & Intelligent Systems
August 1-3, 2019 - Las Vegas, NV, U.S.A.