CN-121998365-A - Green direct-connection energy supply optimization method and system considering uncertainty and system toughness

CN121998365ACN 121998365 ACN121998365 ACN 121998365ACN-121998365-A

Abstract

The invention relates to the technical field of energy supply scheme optimization, and provides a green electricity direct-connection energy supply optimization method and system taking uncertainty and system toughness into consideration, wherein the method comprises the steps of constructing a time fusion-based transducer model, acquiring operation data of a green electricity direct-connection power supply system, carrying out load prediction and upstream green electricity supply quantity prediction, and calculating uncertainty; the method comprises the steps of constructing a multi-objective optimization model by taking minimum total operation cost, minimum equipment power fluctuation and maximum system energy supply toughness as optimization targets, solving a pareto optimal scheduling scheme set by adopting a swarm adaptive collaborative algorithm aiming at the multi-objective optimization model to regulate and control the operation scheduling of a green direct-connected power supply system, introducing an IES-CACA algorithm of a multi-sub-population collaborative evolution and adaptive evolution strategy, and improving the global optimizing capacity and the system toughness of the multi-objective optimization scheduling in a green direct-connected scene by combining an uncertainty perception regulation mechanism.

Inventors

MAO YUDONG
Luan Zhaotao
YU MINGZHI
LIU JIYING
YANG KAIMIN
ZHOU SHIYU
ZHU LIJUAN

Assignees

山东建筑大学

Dates

Publication Date: 20260508
Application Date: 20260129

Claims (10)

1. The green direct-connection energy supply optimization method considering uncertainty and system toughness is characterized by comprising the following steps of: Constructing a time fusion-based converter model, acquiring operation data of a green electricity direct-connection power supply system, carrying out load prediction and upstream green electricity supply quantity prediction, and calculating uncertainty; constructing a multi-objective optimization model by taking the minimization of the total running cost, the minimization of the equipment power fluctuation and the maximization of the system energy supply toughness as optimization targets; aiming at the multi-objective optimization model, solving a pareto optimal scheduling scheme set by adopting a swarm adaptive cooperative algorithm so as to regulate and control the operation scheduling of the green direct-connected power supply system; In the solving process of adopting a group-dwelling self-adaptive collaborative algorithm, in an initialization stage, a corresponding number of sub-populations and an external archive are constructed according to the number of targets of a multi-target optimization model, each sub-population is used for optimizing one target, an initial sub-population is generated based on a mixed initialization strategy matched with a historical scene, and in a population evolution stage, a self-adaptive evolution strategy is adopted, and evolution opportunities are dynamically distributed according to optimization contributions of all sub-populations and predicted uncertainty information.
2. The method for optimizing green direct-connection energy supply by considering uncertainty and system toughness according to claim 1, wherein the method for constructing a time-fusion-based transducer model, carrying out load prediction and upstream green supply quantity prediction and calculating uncertainty comprises the following steps: Carrying out cleaning and normalization processing on the data, and carrying out covariate division on the processed data to obtain static covariates, known future covariates and historical observation covariates; For a static covariate, encoding the static covariate into a plurality of context vectors by a static encoder consisting of a plurality of GRNs; Aiming at the historical observation covariates and the known future covariates, GRN transformation, feature splicing, weight calculation and weighting are sequentially carried out respectively, and final historical feature characterization and known future features are obtained; Inputting the processed historical characteristic representation into a sequence encoder for time sequence characteristic extraction to obtain the encoded characteristic; Aiming at the coded characteristics, performing interpretable multi-head attention operation, performing one-dimensional linear transformation and time sequence fusion, and fusing with known future characteristics to realize decoding to obtain the hidden state of each future time step; mapping the obtained hidden state of each future time step into a final predicted value through a fully connected output layer, and further obtaining a predicted load and a bit number result of each period of predicted green electricity output; and calculating an uncertainty index based on the obtained quantile results of each period of different prediction targets.
3. The green direct-connection energy supply optimization method considering uncertainty and system toughness according to claim 1 is characterized in that for a constructed multi-objective function, optimization solution is carried out by adopting a swarm adaptive cooperative algorithm, and the method comprises the following steps: constructing a corresponding number of sub-populations and an external archive according to the number of targets of the multi-target function, wherein each sub-population is used for optimizing one target, and generating an initial sub-population based on a mixed initialization strategy matched with a historical scene; dynamically allocating evolution opportunities according to optimization contributions of each sub-population and the predicted uncertainty information so as to guide mutation operation of individuals; Every set iteration times, firstly evaluating the state of the sub-population and pairing good and bad sub-populations based on comprehensive sequencing, and then exchanging information by a bidirectional individual migration method of transferring high-quality genes from the good population to introduce diversity from the bad population to obtain the population after individual migration operation; combining all individuals after each generation of evolution is finished to form a candidate solution set, and screening non-dominant solutions through non-dominant sequencing; And iterating the steps until the iteration cut-off condition is met, obtaining the pareto optimal scheduling scheme set, and outputting the pareto optimal scheduling scheme set through external archiving.
4. The green direct power supply optimization method considering uncertainty and system toughness according to claim 3, wherein the process of generating an initial sub-population based on a mixed initialization strategy of historical scene matching comprises the following steps: acquiring historical scene data and extracting historical scene characteristics; Extracting current scene characteristics aiming at a current scene to obtain a characteristic vector of a current scheduling period; Calculating the similarity between the current scene characteristics and the historical scene characteristics to obtain the most similar historical scenes with a set number, and obtaining a similar scene set; The mixed initialization generation sub-population comprises knowledge guiding individuals aiming at corresponding optimization targets based on similar scene guiding generation and random generation individuals, wherein the knowledge guiding individuals and the random generation individuals are subjected to constraint verification to form the initialization sub-population.
5. The green direct-connection energy supply optimization method considering uncertainty and system toughness according to claim 3, wherein: dynamically allocating evolution opportunities according to optimization contributions of various sub-populations and predicted uncertainty information to guide mutation operation of individuals, wherein the method comprises the following steps of: based on the load prediction and the fractional number prediction result of the upstream green electricity supply quantity prediction, obtaining a prediction uncertainty index of each period in the scheduling period; based on the uncertainty index and the forward distance, an evolution probability is calculated: the method comprises the steps of screening high uncertainty time periods, calculating enhanced mutation factors aiming at the screened high uncertainty time periods, updating individual mutation operation, and carrying out individual advancing distance calculation, population advancing distance calculation and roulette selection on mutated populations to obtain updated populations.
6. The green direct current energy supply optimization method considering uncertainty and system toughness of claim 3, wherein the method for generating the population after the individual migration operation comprises the following steps: for each sub-population, arranging all individuals in the sub-population according to the target ascending order, and calculating a standardized median standardized quartile range; Calculating comprehensive sequencing of each sub population, carrying out community collaborative pairing, and pairing the optimal and the worst according to the sequencing value, and pairing the suboptimal and the inferior; And for a pair of paired superior sub-population Pi and inferior sub-population Pj, migrating the superior individuals in the superior sub-population Pi to the inferior sub-population Pj, and migrating the same number of individuals in the inferior sub-population Pj to the superior sub-population Pi after migrating.
7. The green electricity direct connection energy supply optimization method considering uncertainty and system toughness according to claim 1 is characterized in that aiming at the obtained pareto optimal scheduling scheme set, the optimal scheduling scheme is screened out based on dynamic preference driving and multi-attribute collaborative projection decision to regulate and control the operation of a comprehensive energy system, and the method comprises the following steps of: Acquiring a scheme of the pareto optimal scheduling scheme set, and constructing a candidate scheduling scheme set; generating weight vectors for different targets based on the acquired state features; Combining the dynamic preference weight, calculating the basic adaptation degree, the equilibrium punishment and the extreme solution punishment of the scheduling scheme, and finally obtaining a comprehensive score; and determining a final optimal scheduling scheme based on the comprehensive score and the manual fine adjustment.
8. Green electric direct-connection energy supply optimizing system taking uncertainty and system toughness into consideration is characterized by comprising: The prediction module is configured to construct a time fusion-based converter model, acquire the operation data of the green electricity direct-connection power supply system, conduct load prediction and upstream green electricity supply quantity prediction, and calculate uncertainty; The model construction module is configured to construct a multi-objective optimization model with minimum total running cost, minimum equipment power fluctuation and maximum system energy supply toughness as optimization targets; The solving module is configured to solve the pareto optimal scheduling scheme set by adopting a swarm adaptive cooperative algorithm aiming at the multi-objective optimization model so as to regulate and control the operation scheduling of the green direct-connected power supply system; In the solving process of adopting a group-dwelling self-adaptive collaborative algorithm, in an initialization stage, a corresponding number of sub-populations and an external archive are constructed according to the number of targets of a multi-target optimization model, each sub-population is used for optimizing one target, an initial sub-population is generated based on a mixed initialization strategy matched with a historical scene, and in a population evolution stage, a self-adaptive evolution strategy is adopted, and evolution opportunities are dynamically distributed according to optimization contributions of all sub-populations and predicted uncertainty information.
9. An electronic device comprising a memory and a processor and computer instructions stored on the memory and running on the processor, which when executed by the processor, perform the steps in the green direct power optimization method of any one of claims 1-7, taking into account uncertainty and system toughness.
10. A computer readable storage medium storing computer instructions which, when executed by a processor, perform the steps of the green direct energy delivery optimization method of any one of claims 1-7, taking into account uncertainty and system toughness.

Description

Green direct-connection energy supply optimization method and system considering uncertainty and system toughness Technical Field The invention relates to the technical field related to energy supply scheme optimization, in particular to a green direct-connection energy supply optimization method and system considering uncertainty and system toughness. Background The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art. The building field is as an important component of energy consumption and carbon emission, and is facing urgent demands for transformation to green low-carbon, high-efficiency and intelligent directions. The office building has large energy load, obvious peak-valley change and high operation stability requirement, and becomes one of key scenes for optimizing the comprehensive energy system. In recent years, a green electricity direct connection mode is widely focused in the building field, namely, clean energy is consumed in situ by directly coupling an office building to a nearby distributed wind power or photovoltaic power generation system, and a transfer link of a traditional power grid is avoided. The mode is favorable for reducing power transmission and distribution loss, improving energy utilization efficiency and providing an effective path for realizing near-zero carbon operation of office buildings. However, in the green direct connection mode, the two-sided uncertainty of building load and renewable energy output increases rapidly, and system scheduling faces more complex constraint and target conflict, so that an optimization method which takes multi-target performance, scheduling stability and operation toughness into consideration is needed to realize comprehensive coordination of the office building energy system in green, economic, reliable and other dimensions. However, the existing optimization method still has multiple technical bottlenecks when dealing with the high-dimensional nonlinear modeling and dynamic operation scheduling problems in the environment of 'green direct connection' of an office building: First, there is a discrepancy between model solving efficiency and accuracy. The optimization problem of the comprehensive energy system of the office building generally has the characteristics of high dimensionality, strong nonlinearity and multiple constraints, the traditional deterministic optimization algorithm is easy to sink into a local optimal solution and is difficult to search in a complex solution space, and the general meta-heuristic algorithm has global optimization capability, but is difficult to balance between convergence speed and solution quality, and cannot meet the dual requirements of building energy system scheduling on instantaneity and execution. Second, existing optimization strategies have difficulty achieving efficient coordination between solution set diversity and convergence. The traditional archiving updating mechanism mainly uses distance or density indexes, and is difficult to maintain the balanced distribution of the pareto solution set in the solution space, so that the generated optimization scheme has limited coverage range and insufficient flexibility in practical application. In addition, the collaborative optimization capability among multiple targets is obviously insufficient. Under the green electricity direct connection scene, economy, operation stability and system toughness often conflict with each other (for example, high toughness can lead to the increase of operation cost), a traditional optimization model is multi-biased and single-target or simple weight weighting, and a fully-balanced pareto front solution set is difficult to form, so that the strategy space of dynamic scheduling is limited. More seriously, most of the existing algorithms lack modeling and adapting mechanisms for source load uncertainty, and cannot fully consider the influence of wind-light output fluctuation and building load uncertainty on a modulation result, so that the robustness of an optimization scheme in actual execution is insufficient, the stability is poor, and the energy supply safety of a key load is seriously influenced. Disclosure of Invention In order to solve the problems, the invention provides a green electricity direct-connection energy supply optimization method and system taking uncertainty and system toughness into consideration, an IES-CACA algorithm of a multi-sub population collaborative evolution and self-adaptive evolution strategy is introduced, and an uncertainty perception regulation and control mechanism is combined, so that global optimizing capacity and system toughness of multi-objective optimization scheduling under a green electricity direct-connection scene are improved. In order to achieve the above purpose, the present invention adopts the following technical scheme: one or more embodiments provide a green direct power sup