Real-time optimal power distribution method for intelligent power grid system based on deep neural network

Abstract

The invention provides a real-time optimal power distribution method for an intelligent power grid system based on a deep neural network. The method comprises the steps: firstly obtaining power generation parameters of a power grid, and determining power distribution strategies for different total power demands by using a traditional optimization method based on gradient descent with the purposesof improving the power generation efficiency and reducing the total power generation cost; determining the parameters of the deep neural network according to the used traditional optimization method based on gradient descent; training a deep neural network model by using a complete training set until the loss value of the verification set tends to be stable; and acquiring a to-be-processed total power demand in real time, inputting the to-be-processed total power demand into the trained deep neural network for neural network calculation and outputting a result as a power generation load of each generator. The occupation of computing resources is reduced, the service delay is greatly reduced and the allocation rate is improved. The identification precision of the method can meet the requirements of practical application, and related parameters required by training can be obtained by calculation of 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