THE DIAGNOSTICS OF THE ENGINEERING ENTERPRISE'S FINANCIAL CONDITION BASED ON THE USE OF NEURAL NETWORK MODELING

Abstract

The purpose of the paper consists in diagnosing the level of the financial condition of the engineering enterprise using the neural network approach and providing a forecast of its level for the future. The paper emphasizes the importance of diagnosing the financial condition of Ukrainian enterprises under modern conditions. Methods of traditional financial analysis are considered. The necessity of using modelling to improve the quality and accuracy of the analysis is emphasized. An analysis of existing models for assessing the financial condition and bankruptcy of enterprises is carried out. Different types of models developed by domestic and foreign authors are considered: models built on the basis of multivariate discriminant analysis, based on fuzzy logic methods, and others. The use of neural network modelling for assessing the financial condition is substantiated. A neural network model of financial state diagnostics is built based on the financial data of an engineering enterprise. To decrease the domain of input data, the use of the "centre of gravity" method is proposed, with the help of which the number of input variables of the model is reduced to five. A model based on a multilayer per-ceptron is built with the help of a powerful neural network modelling tool (SSN). The neural network was trained by the backpropagation method. An assessment of the financial condition of the engineering enterprise PJSC NKMZ is made with the help of the model for 10 future periods. The proposed method of diagnosing the financial state allows the management of the engineering enterprise to predict the onset of a crisis state and develop a financial recovery plan.