Application of neural networks for optimization mode reactive power in an extensive electric network

Abstract

Increase prices for electricity as well as the poor state of electrical networks force to develop and implement modern methods to reduce power consumption, as well as reduce losses in the power supply. The solution of this problem is primarily due to the optimization of the production process: separation of the most powerful power consumers, limiting their idling, reducing consumption for its own needs, using equipment with higher efficiency and optimization of reactive power mode. Compensating devices are installed today at many enterprises, which are used to reduce reactive power factor to the economic level in the point of common couple. However, it is not considered that the power flows in the complex network with non-optimal placement of the compensating devices and incorrect installation of the power compensation can reach high values, which causes an increase of losses in the network. Program implementing the prediction algorithms has been created with the help of neural networks based on both full and uncomplete data on the values of the electrical loads, as well as further optimization of reactive power mode. The influence of placement of compensating devices in the distributive network on the cost also analyzed. It was finally achieved a significant reduction in the amount of payment for reactive power flows, as well as the loss of active energy in the elements of electrical network.