Search

CN-122022115-A - Multi-agent cooperation micro-grid energy management platform integrating electric power market mechanism

CN122022115ACN 122022115 ACN122022115 ACN 122022115ACN-122022115-A

Abstract

The invention provides a multi-agent cooperative micro-grid energy management platform integrating an electric power market mechanism, which comprises a market interaction layer, a resource scheduling layer and a physical execution layer, wherein the market interaction layer is used for realizing dynamic electricity price prediction, multi-market bidding optimization and carbon emission cost accounting, the resource scheduling layer is used for realizing resource level optimization control by adopting a multi-agent reinforcement learning framework, and the physical execution layer is used for realizing high-frequency data communication and control response between devices based on an IEC 61850 protocol. The management platform integrates a multi-market mechanism and optimizes risk perception capability so as to solve the problems of inconsistent scheduling, unreasonable system revenue distribution and the like caused by the lack of autonomous decision making capability and a collaborative optimization mechanism of the existing platform.

Inventors

  • MENG XIANGQI

Assignees

  • 上海理工大学

Dates

Publication Date
20260512
Application Date
20251211

Claims (8)

  1. 1. The multi-agent cooperative micro-grid energy management platform integrating the electric power market mechanism is characterized by comprising a market interaction layer, a resource scheduling layer and a physical execution layer; The market interaction layer is used for realizing dynamic electricity price prediction, multi-market bidding optimization and carbon emission cost accounting, the resource scheduling layer is used for realizing resource level optimization control by adopting a multi-agent reinforcement learning framework, and the physical execution layer is used for realizing high-frequency data communication and control response between devices based on an IEC 61850 protocol.
  2. 2. The multi-agent collaborative micro-grid energy management platform incorporating an electric power market mechanism according to claim 1, wherein the market interaction layer employs an LSTM-CNN hybrid neural network to conduct short-term predictions of electricity prices, incorporating calendar information and meteorological factors.
  3. 3. The multi-agent collaborative micro-grid energy management platform integrating the power market mechanism according to claim 1, wherein the market interaction layer builds an electricity price probability distribution model and a multi-stage decision tree structure based on a random programming theory, comprehensively considers multi-time scale constraint and electricity price uncertainty of a market and a capacity market in the future, and realizes a combined quotation optimization strategy of energy storage equipment; the calculation formula of the connection quotation optimizing strategy is as follows: ; In the formula, The electricity price is the day-ahead electricity price; the bidding power; is a conditional risk value; The multi-agent collaborative micro-grid energy management platform incorporating an electrical market mechanism of claim 1, wherein the market interaction layer dynamically modifies operating costs based on real-time carbon trade market prices to output carbon constraint scheduling factors.
  4. 4. The multi-Agent collaborative micro-grid energy management platform incorporating an electrical market mechanism according to claim 1, wherein the resource scheduling layer includes a photovoltaic Agent module that adjusts the system's response to photovoltaic fluctuations by incorporating irradiance predictions with power confidence intervals.
  5. 5. The multi-Agent collaborative micro-grid energy management platform integrating the power market mechanism according to claim 1, wherein the resource scheduling layer comprises an energy storage Agent module, the energy storage Agent module constructs a double-target optimization strategy, balances economic arbitrage and cycle life, and supports peak-to-valley electricity price difference strategy automatic adaptation: the energy storage Agent objective function: ; wherein: revenue is a benefit weight coefficient, namely economic priority; Is a life weight coefficient, i.e., durability priority; is the capacity attenuation; is the rated capacity of energy storage; is the capacity retention, i.e., life health.
  6. 6. The multi-Agent collaborative micro-grid energy management platform integrating the power market mechanism according to claim 1, wherein the resource scheduling layer comprises a charging pile Agent module, the charging pile Agent module adjusts charging power according to load peak-valley time periods and market signals, guides end user behaviors, and improves load flexibility.
  7. 7. The multi-agent collaborative micro grid energy management platform incorporating the power market mechanism according to claim 1, wherein the resource scheduling layer includes a virtual power plant coordinator to enable cross-device, cross-market revenue coordination and response scheduling.
  8. 8. The multi-agent collaborative micro-grid energy management platform incorporating the power market mechanism according to claim 1, wherein the physical execution layer enables dynamic adjustment of energy storage system SOC (state of charge) based on seconds-level cycles, charging stake power settings support millisecond-level responses, adapt to fast load regulation scenarios, and support heterogeneous device access and plug-and-play configurations.

Description

Multi-agent cooperation micro-grid energy management platform integrating electric power market mechanism Technical Field The invention relates to the technical field of intelligent energy management, in particular to a multi-agent cooperative micro-grid energy management platform integrating an electric power market mechanism. Background Currently, a micro-grid system becomes an important technical form for converting a pushing energy structure and separating regional energy, and is widely deployed in scenes such as industrial and commercial parks, transportation hubs, remote islands and the like. With the continuous increase of the clean energy access proportion in overseas markets, the micro-grid is gradually bearing more market participation functions such as load adjustment, electricity price response, frequency support and the like. However, overseas power market mechanisms generally exhibit complex, multiple features, including multi-level, wide time scale operating logic for real-time electricity prices, day-ahead transactions, capacity markets, and auxiliary service markets. The existing micro-grid energy management platform is designed aiming at a fixed electricity price mechanism, is difficult to be compatible with frequently-changed market rules, and is generally lack of response elasticity and bidding strategy support especially when facing an overseas market dynamic bidding mechanism. In addition, the distributed renewable energy sources and the electric automobile loads are widely connected, so that the source-load volatility is obviously increased, the traditional centralized scheduling strategy is limited by response time delay and calculation force bottlenecks, and the practical requirement of minute-level dynamic decision is difficult to be satisfied. The related art mainly focuses on the local energy balance and cost minimization targets, and the integration of electric power market rule modeling and bidding mechanism is less considered, and meanwhile, the modeling capability of elements such as carbon transaction cost, capacity bidding constraint and the like is limited. In a distributed environment, the problems of inconsistent scheduling, unreasonable system revenue distribution and the like are still outstanding due to the general lack of autonomous decision making capability and collaborative optimization mechanism of various energy resource main bodies (such as photovoltaic inverters, energy storage equipment and charging piles). Disclosure of Invention The invention aims to provide a multi-agent collaborative micro-grid energy management platform which realizes the cooperative control of photovoltaic, energy storage and load equipment and improves the overall economy, safety and market response capability of a system by introducing market mechanism modeling, reinforcement learning optimization strategy and carbon emission cost accounting logic. In order to achieve the purpose, the invention provides a multi-agent cooperative micro-grid energy management platform integrating an electric power market mechanism, wherein the energy management platform comprises a market interaction layer, a resource scheduling layer and a physical execution layer; The market interaction layer realizes dynamic electricity price prediction, multi-market bidding optimization and carbon emission cost accounting, the resource scheduling layer adopts a multi-agent reinforcement learning framework to realize resource level optimization control, and the physical execution layer realizes high-frequency data communication and control response between devices based on an IEC 61850 protocol. Furthermore, the market interaction layer adopts an LSTM-CNN hybrid neural network to predict electricity prices in a short period, and combines calendar information and meteorological factors. Further, the market interaction layer builds an electricity price probability distribution model and a multi-stage decision tree structure based on a random programming theory, comprehensively considers multi-time scale constraint and electricity price uncertainty of the market and the capacity market in the future, and realizes a combined quotation optimization strategy of the energy storage device; The connection quotation optimization strategy calculation formula is as follows: ; In the formula, The electricity price is the day-ahead electricity price; the bidding power; is a conditional risk value; further, the market interaction layer dynamically corrects the running cost according to the real-time carbon trade market price and outputs a carbon constraint scheduling factor. Further, the resource scheduling layer comprises a photovoltaic Agent module, and the photovoltaic Agent module adjusts the response of the system to photovoltaic fluctuation by fusing irradiance prediction and a power confidence interval. Further, the resource scheduling layer comprises an energy storage Agent module, the energy storage Agent module construct