ДОСЛІДЖЕННЯ МЕТОДІВ ПРОАКТИВНОГО УПРАВЛІННЯ ВІДХИЛЕННЯМИ НА ОСНОВІ НЕЙРОМЕРЕЖ В ІТ ПРОЕКТАХ

Abstract

This paper describes the results of a study of proposed methods of proactively managing key parameter deviations in complex projects based on the study of the effects of the external and internal environment of such projects. The methods of forecasting the level of changes in the results of project activity at any time during the execution of projects and depending on changes in the time parameters of the work of the projects and the study of the effects on changes in the cost of the work of the projects are proposed. Impact reactions on cost parameters and project timelines are investigated. An integrated information system has been developed to simulate the flow of changes to key IT project parameters using cloud data warehouses. In the process of modeling modern information technologies of project management of leading developers are involved and integrated. Modeling effects of the environment on project parameters based on models of deep learning neural networks are used as research tools. A model of deep learning of the neural network is proposed, through the experimental representation of the input and output data of numerical experiments. This model takes into account the optimistic and pessimistic distribution of the cost of each project when planning the projects and choosing their optimal configuration. The evaluation of the results of modeling the effects of changes on the timing and cost of performing work is based on the context of project characteristics, including resource allocations both in time and in project work, cost allocations, etc. Thus, the modeled indicators in the system indicate slight deviations within 10-15% of the set values under the influence of a wide range of values of environmental factors and their effects on changes in project work resources for the selected and unchanged technological configuration of the project model. Using proactive controls, in the re-simulation, it became possible to significantly reduce deviations in costs that do not exceed 10% of the deviation from the optimum values