Mathematical Modeling of the Data Processing Problems of Heat Experiments Based on Multiprocessor Computing Complexes

Abstract

The research aims at solving problems associated with identifying parallel structures of algorithms and programs, and their reflection in computer architecture. This paper highlights the modeling processes of multidimensional non-stationary problems based on multiprocessor computing systems. An algorithm is developed for solving the coefficient inverse heat conduction problem. In the above studies, the inverse heat conduction problem is interpreted as the optimal control problem. The problem of choosing a regularization parameter is discussed. The regularization parameter is selected so that the residual of the approximate solution could be comparable in magnitude with the accuracy degree of the problem’s initial data. Such a choice of the regularization parameter is easily realized when modeling computations by a multiprocessor computing system. It is manifested in iterative methods, which allows obtaining results close in accuracy to optimal. It is noted that the parallel algorithm basis of numerical minimization is the procedure for establishing the minimum function of many variables. The test modeling results of problems using multiprocessor computing systems are presented.