Parametric tolerance synthesis as multiple inverse problem of providing complex engineering system operability

• Datum: 2014
• Hauptverfasser: Горошко, А. В., Ройзман, В. П.
• Format: Artikel
• Sprache: Ukrainian
• Veröffentlicht: Інститут енергетичних машин і систем ім. А. М. Підгорного Національної академії наук України 2014
• Schlagworte: tolerances; parametric synthesis; operability; inverse problem
• Online Zugang: https://journals.uran.ua/jme/article/view/35905
• Назва журналу: Journal of Mechanical Engineering

The paper considers the current state of the problem of parametric tolerances synthesis for complex technical systems. Given the shortcomings of the existing methods to search the area of efficiency, using the Monte Carlo method, in particular a significant amount of computation required and the lack of information on patterns of random processes, variations of system parameters. A formalization of parametric synthesis of tolerances as the inverse ensure efficiency of complex technical systems. This task should be carried out with the use of modern approaches to the solution of inverse problems in general and the particular parametric synthesis. Requested to provide tolerances as hyper parallelepiped. To construct the field of technical efficiency of the system suggested to solve the problem of incorporating sound tolerances as hyper parallelepiped to performance using criterion value, solving vector optimization problems with constraints. The requirement for a minimum cost of manufacturing the product replaced in some sense equivalent to the requirement of maximizing all tolerances. Based recommendations on the choice of the optimization method. It is shown that the approximation to clarify the area of efficiency field tolerances effectively use irregular grids as a modification of the method of matrix test and parametric synthesis method according to the criterion of efficiency of stock. Introduced in the approach to the parametric synthesis of tolerances provides a region of system performance at lower cost and is not limited to the normal distribution changes the output parameters.