CN-121981439-A - Multi-energy scene generation park comprehensive energy random optimization scheduling method and device

CN121981439ACN 121981439 ACN121981439 ACN 121981439ACN-121981439-A

Abstract

The application discloses a comprehensive energy random optimization scheduling method for a multi-energy scene generation park, which comprises the steps of constructing a multi-energy coupling condition feature database based on historical source load multi-energy prediction data, inputting day-ahead prediction information in the multi-energy coupling condition feature database, constructing an improved condition generation countermeasure network model, generating a day-ahead random scene set with multi-energy coupling characteristics based on the countermeasure network model, carrying out clustering treatment on the day-ahead random scene set to obtain a typical scene set representing multi-energy uncertainty characteristics and probability distribution thereof, constructing a comprehensive energy system structure of the park based on an energy hub model, establishing a day-ahead multi-energy scene random optimization scheduling model by combining the typical scene set and the probability distribution thereof, and solving the random optimization scheduling model by adopting an optimization algorithm to obtain a robust optimization scheduling scheme of system equipment. By the method, a multi-energy coupling scene is generated, scheduling resource waste caused by prediction uncertainty is reduced, and robustness of optimal scheduling of the system is improved.

Inventors

LIU FANG
LIANG HAIPENG
LIU QIANYI
Ban Guihua
LIANG WEI
PENG HUI

Assignees

湘江实验室

Dates

Publication Date: 20260505
Application Date: 20251225

Claims (10)

1. A random optimization scheduling method for comprehensive energy of a multi-energy scene generation park is characterized by comprising the following steps: Constructing a multi-energy coupling condition characteristic database based on the history source load multi-energy prediction data; Inputting day-ahead prediction information in the multi-energy coupling condition feature database, and constructing an improved condition generation countermeasure network model; Generating a set of random future scenes with a multi-energy coupling characteristic based on the countermeasure network model; Clustering the random scene set before the day to obtain a typical scene set representing the characteristic of the multi-energy uncertainty and probability distribution thereof; Building a park comprehensive energy system structure based on an energy hub model, and building a day-ahead multi-energy scene random optimization scheduling model by combining the typical scene set and probability distribution thereof; and solving the random optimal scheduling model by adopting an optimization algorithm to obtain a robust optimal scheduling scheme of the system equipment.
2. The method of claim 1, wherein said inputting the day-ahead prediction information in the multi-energy coupling condition feature database, constructing an improved condition generation countermeasure network model, comprises: inputting deterministic day-ahead prediction information and interval prediction upper and lower limit information in the multi-energy coupling condition feature database; And judging the authenticity of the input scene by using a judging device network model and taking the deterministic future prediction information and the interval prediction upper and lower limit information as conditions.
3. The method of claim 2, wherein the training objectives of the generator network model comprise a discriminant loss term, a data fidelity term, and a boundary penalty term, and wherein the training objectives of the discriminant network model comprise bulldozer distance and gradient penalty terms.
4. The method of claim 1, wherein clustering the set of pre-day random scenes to obtain a set of typical scenes and their probability distribution characterizing a multi-energy uncertainty feature, comprises: Selecting an initial cluster center from the random scene set before the day, performing iterative clustering based on a distance minimization principle, and updating the scene cluster center; And calculating the occurrence probability of the typical scene set according to the scene quantity duty ratio in each scene cluster, and determining the probability distribution of the typical scene set.
5. The method of claim 1, wherein the energy hub model comprises an energy input layer, an energy conversion layer, an energy storage layer, and an energy output layer, and wherein the collaborative distribution and conversion relationship of the multi-energy flows is characterized by a coupling matrix.
6. The method of claim 1, wherein the day-ahead multi-energy scenario stochastic optimization scheduling model targets system aggregate cost minimization and comprises system operational constraints, device constraints, and energy storage constraints.
7. The method of claim 1, wherein the obtaining a robust optimized scheduling scheme for a system device comprises: and obtaining a day-ahead start-stop plan of the system equipment and an equipment output scheme under a maximum probability scene.
8. A multi-energy scene generation park comprehensive energy random optimization scheduling device, characterized in that the device comprises: The first construction module is used for constructing a multi-energy coupling condition characteristic database based on the history source load multi-energy prediction data; the second construction module is used for inputting day-ahead prediction information in the multi-energy coupling condition feature database and constructing an improved condition generation countermeasure network model; the generation module is used for generating a day-ahead random scene set with a multi-energy coupling characteristic based on the countermeasure network model; The processing module is used for carrying out clustering processing on the day-ahead random scene set to obtain a typical scene set representing the characteristic of the multi-energy uncertainty and probability distribution thereof; the third construction module is used for constructing a park comprehensive energy system structure based on the energy hub model, and establishing a day-ahead multi-energy scene random optimization scheduling model by combining the typical scene set and probability distribution thereof; And the obtaining module is used for obtaining a robust optimal scheduling scheme of the system equipment.
9. A storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform the steps of the method according to any one of claims 1 to 7.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the method according to any one of claims 1 to 7 when the program is executed.

Description

Multi-energy scene generation park comprehensive energy random optimization scheduling method and device Technical Field The application relates to the technical field of operation and analysis of a park comprehensive energy system, in particular to a method and a device for generating park comprehensive energy random optimization scheduling by a multi-energy scene. Background With the continuous promotion of energy internet construction, a park comprehensive energy system becomes a key carrier for improving renewable energy consumption capability and energy utilization efficiency, the system needs to coordinate the fluctuation of photovoltaic and wind power with the dynamic demands of electricity, heat and cold loads, and the integration of multiple energy sources increases the complexity of system operation and brings new challenges to optimal scheduling. At present, a large prediction error still exists for wind-solar and multi-energy load day-ahead prediction, a traditional scheduling method directly utilizes prediction information to determine a day-ahead scheduling plan, a large number of rotation standby precaution emergency situations are needed, the plan is too conservative and causes a large number of scheduling resource waste, a scene method quantifies uncertainty of random variables of a system through discrete representative scenes, the random optimization problem is converted into a large-scale deterministic optimization problem, the method is one of main methods for carrying out uncertainty analysis at present, a scene generation method usually only considers a single random variable of the system, generated scenes are obtained through direct splicing, multi-energy coupling characteristics of various random park comprehensive energy systems cannot be fully described, and secondly, the quality of scene generation of the scene method based on the random optimization scheduling method directly determines effectiveness of a scheduling strategy, how to construct a limited number of scene sets, and fully delineate the uncertainty feasible region of the wind-solar and multi-energy system according with actual conditions is a problem to be solved. Disclosure of Invention Based on the problems, the embodiment of the application provides a method and a device for randomly optimizing and scheduling comprehensive energy sources of a multi-energy scene generation park, which can effectively improve the daily front source load multi-energy scene generation quality of the comprehensive energy source system of the park by considering the source load multi-energy coupling characteristic of the comprehensive energy source system of the park and improving a condition generation countermeasure network model, and can obviously reduce the scheduling resource waste caused by a multi-energy prediction error by considering the source load multi-energy uncertainty of the random optimizing and scheduling method based on a scene method. The embodiment of the application provides a method for generating a park comprehensive energy random optimization scheduling by a multi-energy scene, which comprises the steps of constructing a multi-energy coupling condition feature database based on historical source load multi-energy prediction data, inputting day-ahead prediction information in the multi-energy coupling condition feature database, constructing an improved condition generation countermeasure network model, generating a day-ahead random scene set with multi-energy coupling characteristics based on the countermeasure network model, carrying out clustering treatment on the day-ahead random scene set to obtain a typical scene set representing multi-energy uncertainty characteristics and probability distribution thereof, constructing a park comprehensive energy system structure based on an energy hub model, establishing a day-ahead multi-energy scene random optimization scheduling model by combining the typical scene set and the probability distribution thereof, and solving the random optimization scheduling model by adopting an optimization algorithm to obtain a robust optimization scheduling scheme of system equipment. In some possible embodiments, inputting the day-ahead prediction information in the multi-energy coupling condition feature database, constructing the improved condition generation countermeasure network model includes: inputting deterministic day-ahead prediction information and interval prediction upper and lower limit information in a multi-energy coupling condition feature database; and judging the authenticity of the input scene by using the identifier network model and taking the deterministic day-ahead prediction information and the interval prediction upper and lower limit information as conditions. In some possible embodiments, the training objectives of the generator network model include a discriminant loss term, a data fidelity term, and a boundary penalty term, and the training objectives of