The aim was to determine the reliability indicators of a power supply system using an artificial neural network model. A model for calculating technical reliability was developed using the following methods: an algorithm for calculating reliability indicators of power supply systems, the method of failure rate of a power supply system and a forecasting model using artificial neural networks. It was established that a power supply system is formed by an open radial power supply circuit. The failure rate of the power supply subsystem was determined by calculating the failure rate of i-th element of the subsystem. As a result of calculating the probability of failure-free operation of the subsystem for various conditions (5 time intervals), it was found that with an increase in the operating time from 100 to 500 h, a linear increase in the rate of system failures occurs from 0.0051 to 0.0073 1/h. A comparison of the obtained mean-to-failure values of the main and the same backup subsystem in the unloaded mode with an absolutely reliable switch (269.62 h) with the main and the same backup subsystem in the loaded mode (202.21 h) was carried out. The results differ by 67.41 h, which indicates a higher degree of reliability of the first method. The software package Prognoz_INS_2020 was developed. An acceptable accuracy of no more than 2.17% was obtained by comparing the results of the conventional calculation of the failure rate of power supply systems and using the Prognoz_INS_2020 software package. This indicates the efficiency of the proposed software package in reliability calculations at operating energy enterprises. The proposed methods for assessing technical reliability both using the conventional model and a model based on an artificial neural network made it possible to assess the state of power supply systems, which helps to prevent dangerous emergencies.
MODELING OF RELIABILITY INDICATORS OF ELEMENTS OF POWER SUPPLY SYSTEMS WITH UNSINUSOIDAL VOLTAGE
