Розробка методу оцінки технічних характеристик спеціалізованих ієрархічних систем з використанням штучного інтелекту

The object of the study is specialized hierarchical systems. The problem addressed in the research is improving the efficiency of evaluating hierarchical systems while ensuring a specified level of reliability regardless of the volume of data entering the system. The originality of the method lies in the use of additional enhanced procedures that allow to: – verify the topology and parameters of specialized hierarchical systems, considering the degree of uncertainty of the initial data known about them. The consideration of uncertainty is achieved through the application of corresponding correction coefficients; – perform a primary selection of individuals for tuning the convolutional artificial neural network using an improved genetic algorithm, which reduces solution search time and increases the reliability of obtained results; – explore the solution spaces of the problem of evaluating the state of specialized hierarchical systems described by atypical functions using an improved monkey swarm algorithm; – tune the memory of the convolutional artificial neural network via a memory training procedure, thereby reducing the estimation error of parameters of specialized hierarchical systems; – adjust the weights of the convolutional artificial neural network, which leads to increased accuracy in estimating parameters of specialized hierarchical systems; – employ additional mechanisms to adjust convolutional artificial neural network parameters by changing the membership function. An increase in decision-making efficiency of 15–18% was established due to the use of additional procedures, with the reliability of the decisions ensured at the level of 0.9