Generals cheme of complex systems models of self-organization algorithms construction using evolutionary search

Purpose. There are occur tasks in the mathematical modeling, the object of study which are complex systems. Convection, heat transfer and other are the objects examples of study. In addition, such systems could be difficult for study. Because of that, number of small experimental data often obtained. Therefore, the main question is to find mathematical model constructing variant with small amount of input data. The self-organization of mathematical models designed by A.G. Ivahnenko is chose. Methodology. In comparison with other methods, such as analysis of variance, this method is relatively new. However, during its short stories, it has practical value already proved. The clarification of generating models gives by structural algorithm of self-organization that illustrated at Pic.1. “Partial description” has a special importance. It characterizes the complexity of the model and possibility of its using. Linear “private description” is using most commonly as it pointed in the article. However, considering the scope of the using this method, linear description cannot describe the object of study. Therefore, it proposed to use non-linear “partial description” for this method. In addition, there are examples of non-linear functions that could using during description of the object. Findings. There are methods of evolutionary search for self-organization of mathematical models are developed. This is search reviewed more detailed in the article and shown a block-diagram of the algorithm at Pic.2. It is suggest using “partial description” in a nonlinear form. There are description to this algorithm and shown the importance of “partial description” form for building object of study a mathematical model. Originality. There is developed construction of complex systems mathematical models general scheme by using self-organization with model parameters “partial description” evolutionary search. There is distinctive convergence with probability 1 process model parameters “partial description” evolutionary search. Practical value. Models self-organization general scheme using evolutionary solutions search has been realized programmatically. It is indicated the possibility of these methods more widely using and increasing the simulation accuracy in comparison with existing methods