Comprehensive optimization method for power grid in high-permeability region of distributed power supply

Abstract

The invention discloses a comprehensive optimization method for a power grid in a high-permeability region of a distributed power supply, and the method comprises the following steps: S1, building a mathematic model for the comprehensive optimization of a power transmission network according to the operation safety constraint of a system through taking the minimization of power generation cost asa target; S2, modeling a decision process of comprehensive optimization of the power transmission network by using a Markov process, and defining a corresponding state space, an action space and a reward function; S3, employing a neural network fitting value function, employing parameter freezing, experience playback and progressive greedy methods for training, and initializing neural network parameters and control parameters; S4, performing simulation training locally to obtain a trained evaluation value network; S5, performing online operation control by using the trained evaluation value network. Through an efficient deep reinforcement learning algorithm, the optimal control strategy is learned in interaction with the simulation model, the online decision time is shortened, and a powergrid comprehensive optimization scheme is quickly obtained.

Description

technical field [0001] The invention relates to the field of smart grids, and more specifically, to a comprehensive optimization method for power grids in areas with high penetration rates of distributed power sources. Background technique [0002] In order to alleviate the dual pressure of energy shortage and environmental pollution, the construction of distributed new energy generating units represented by wind power and photovoltaic is the key direction of my country's energy structure reform. The 2019 National Renewable Energy Power Development Monitoring and Evaluation Report pointed out that the installed capacity of renewable energy in the country has reached nearly 800 million kilowatts, accounting for nearly 40% of the total installed capacity. Affected by the distribution of natural resources, distributed new energy power generation has strong randomness, and it is difficult to achieve precise control of the output of generating units. The output varies significant...

AI Technical Summary

Problems solved by technology

[0004] However, the comprehensive optimization of the transmission network considering network structure switching is a highly non-convex mixed integer nonlinear programming problem. It is not possible to efficiently solve the problem

The traditional linearized DC power flow model ignores the resistance of the transmission line and the phase angle difference at both ends in the approximation process, and considers that the per unit value of all node voltages is 1, which cannot accurately approximate the voltage and reactive power, and it is difficult to guarantee Feasibility and optimality of the solution

Images