Voltage control method for active distribution network driven by data knowledge

Abstract

The invention discloses an active distribution network voltage control method driven by data knowledge. Firstly, a reactive power optimization model of active distribution network based on the second-order cone optimal power flow is established; secondly, based on the established reactive power optimization model, a large data simulation is performed to form a data expansion strategy, and the historical data and solution expansion strategy constraints are screened out And combined, and establish policy constraints through the data model, form accumulative data knowledge, and establish a reactive power voltage control strategy knowledge graph with distribution network network data and control strategies as the core; finally, based on the constructed distribution network reactive power The voltage control knowledge map matches the similar state in the knowledge map with the current network state, adopts an improved time series similar data retrieval algorithm, and uses similarity calculation, segment verification, and slope verification to match the control strategy and parameter results for safety calibration. It can solve the problem by testing and optimizing, and generate the control strategy and update the state strategy in the knowledge graph.

Description

technical field [0001] The invention belongs to the electric power industry and relates to distribution network voltage control, in particular to an active distribution network voltage control method driven by data knowledge. Background technique [0002] At present, the penetration rate of distributed power generation in the distribution network continues to increase. In 2050, non-fossil energy will account for more than 50% of primary energy consumption in my country, which poses new challenges to the system voltage stability and the complexity of regulation and operation. At the same time, ensuring a higher standard of power quality is an inevitable trend in the development of electric power, and it is also an objective demand for promoting social stability and development. Overvoltage and low voltage will not only reduce energy conversion efficiency, but also easily cause equipment inrush and harmonics to damage user equipment and even lead to safety accidents. At prese...

AI Technical Summary

Problems solved by technology

In optimization-based methods, due to the discreteness of transformer tap settings and shunt capacitors and the nonlinearity of power flow equations, it is generally transformed into solving mixed-integer nonlinear programming problems, and second-order cone relaxation has high accuracy in solving reactive-voltage control problems. However, with the increase of the distribution network scale and controllable equipment, the complexity of the solution increases exponentially, and the calculation time is long, which makes it difficult to meet the real-time control requirements

At the same time, there is a convergence problem and may fall into a local optimum

In terms of reactive power control based on the reinforcement learning method, it can meet the real-time control requirements in terms of speed, and has a certain heuristic and robustness, but the interpretability is insufficient, and it is easy to fall into the "dimension disaster". At the same time, free action exploration is prone to occur during learning. In extreme policy situations, security cannot be guaranteed

Images