A Real-time Optimal Power Allocation Method for Smart Grid System Based on Deep Neural Network

Abstract

A real-time optimal power distribution method for smart grid system based on deep neural network, first obtain the power generation parameters of the power grid, with the goal of improving the power generation efficiency and reducing the total power generation cost, using the traditional optimization method based on gradient descent to determine for different total power Power allocation strategy for demand size; parameters of deep neural network are determined according to the traditional optimization method based on gradient descent used. Use the complete training set to train the deep neural network model until the loss value of the validation set tends to be stable; obtain the total power demand to be processed in real time, input it into the trained deep neural network for neural network calculation, and output the results for each generator. power generation load. The invention reduces the occupation of computing resources, greatly reduces the service delay, and improves the allocation rate. The identification accuracy of the method can meet the requirements of practical applications, and the relevant parameters required for training can be obtained by calculating the traditional optimization method.

Description

technical field [0001] The invention belongs to the field of power grid optimization and machine learning, and specifically provides a real-time optimal power distribution method for a smart grid system based on a deep neural network, which can calculate real-time generator load distribution strategies under different power requirements and improve the service of the smart grid system efficacy. Background technique [0002] Power grid system problem operation and control is an important class of optimization problems. Its essence is to maximize the economy and real-time performance of the power grid system and reduce the losses caused by various links in the power generation process under the premise of satisfying the constraints of supply and demand power balance, line flow and generator power upper and lower limits. In recent years, the sources of electric power have increased dramatically, and human beings have attached great importance to environmental protection, so th...

AI Technical Summary

Problems solved by technology

[0004] In order to overcome the disadvantages of the existing technology, such as high computational complexity, long computational time, inability to calculate the optimal power allocation result in the first time, and difficulty in meeting the real-time scheduling requirements of the smart grid system, the present invention provides a deep neural network based The real-time optimal power allocation method of the smart grid system of the network, specifically, first obtain the power generation parameters of the power grid, aim at improving the power generation efficiency and reducing the total power generation cost, and use the traditional optimization method based on gradient descent to determine the different total power requirements size power allocation strategy

Images