Scheduling optimization method for active power distribution system in industrial parks considering the demand of power grid peak regulation

Abstract

The invention discloses a scheduling optimization method for an active power distribution system in an industrial park considering the peak-shaving demand of the power grid. The photovoltaic power generation unit PV, the all-vanadium liquid flow battery energy storage system VRB and the industrial load are used as industrial loads to participate in considering the peak-shaving demand of the power grid. The dispatching unit of the active power distribution system in the park establishes the physical model of each dispatching unit; the dynamic dispatching optimization problem of the active power distribution system in the industrial park considering the peak-shaving demand of the power grid is established as a DTMDP model; the Q-learning algorithm based on simulated annealing is used to carry out the DTMDP model Strategy solution, the obtained optimization control strategy is used to guide the system to choose a reasonable action plan in the actual operation process, and realize the system scheduling optimization. The invention can effectively solve the peak-shaving demand problem of the power grid under the consideration of the randomness on both sides of the source and load, promote the on-site consumption of new energy, and improve the operation economy of the active power distribution system and the operation stability of the power system.

Description

technical field [0001] The invention relates to the field of optimal dispatching of active power distribution systems, and specifically relates to a scheduling optimization method for active power distribution systems in industrial parks in consideration of the randomness of both sides of the source and load, considering the peak regulation requirements of the power grid. Background technique [0002] With the increasing penetration rate of new energy and the scale of controllable load, the situation of power grid peak shaving is becoming more and more severe, and the distributed characteristics of distribution network are becoming more and more obvious. Active power distribution system with active control and active management of flexible load and energy storage. Effective use of the interaction between distributed energy and loads in the active power distribution system, on the one hand, can realize the safe and economical operation of the system itself, and on the other h...

AI Technical Summary

Problems solved by technology

[0003] The existing optimal dispatching methods for active power distribution systems are mainly divided into two categories. One is to use genetic algorithm, biogeographic optimization algorithm, particle swarm optimization algorithm and other optimization algorithms without considering the randomness of new energy output and load demand. The optimal scheduling problem of the active power distribution system is solved strategically, but the uncertainty of new energy output and load demand is ubiquitous and unavoidable. It will be difficult to guarantee the long-term safe and economic operation of the active power distribution system if the optimal control strategy is used as the benchmark value; the other is to predict the uncertainty of new energy output and load demand through time series, neural network, combined forecasting and other deterministic forecasting methods. Forecasting, transforming the uncertain optimal scheduling problem into a deterministic problem to solve the corresponding strategy, this type of method is to predict the new energy output and load demand at a certain moment, without taking into account the external environment and user demand. Strong randomness, the forecast of new energy output and load demand within a day can be regarded as a multi-period dynamic change process in essence, and has a strong temporal correlation between each time period. At the same time, the optimization strategy obtained by this type of method does not Taking into account the impact of unexpected uncertainties of the environment and users

Images