Green electricity collaborative scheduling method for one-source multi-load scenario and related device

Abstract

The present application belongs to the field of new energy power generation and power demand side response technology, and discloses a green electricity collaborative scheduling method for one-source multi-load scenario and related devices, which comprises obtaining the electricity consumption history data of each user and performing load characteristic mining to obtain the load characteristic mining results of each user; obtaining power grid state data, green electricity output prediction data and electricity consumption prediction data of each user, solving a multi-objective optimization model in combination with the load characteristic mining results of each user to obtain the electricity consumption plan of each user, green electricity distribution scheme and power grid electricity purchase plan, wherein the maximum power supply fairness maximizes the average of the comprehensive fairness scores of each user, and the comprehensive fairness scores of each user are determined according to a preset multi-dimensional fairness index system. The present application realizes precise matching of green electricity output in the green electricity direct supply scenario of a single green electricity project to multiple industrial users, guarantees the fairness of green electricity distribution among multiple users and coordinates interaction with the external power grid, thereby improving the green electricity consumption rate and system economy.

Description

Technical Field

[0001] This invention belongs to the field of new energy power generation and electricity demand-side response technology, and relates to a green electricity collaborative scheduling method and related devices for a single source and multiple load scenario. Background Technology

[0002] With the rapid development and widespread adoption of new energy power generation technologies, the proportion of green electricity consumed by industrial users is continuously increasing. In the traditional one-to-one dedicated power supply model, the power source and load have a one-to-one correspondence, making dispatch and control relatively simple, mainly involving power balance between a single power source and a single user. However, in actual industrial parks, there are often scenarios where a single green power project simultaneously supplies power to multiple industrial users, the so-called "one source, multiple load" scenario. In this scenario, a single green power project needs to serve multiple industrial users simultaneously, leading to an exponential increase in system complexity.

[0003] While existing microgrid energy management systems can achieve coordinated dispatch of power sources, loads, and energy storage, they primarily focus on resource management within a single industrial park and lack coordination mechanisms among multiple users. This makes them unable to support the allocation and fairness of green electricity to multiple users in a single-source, multi-load scenario. Virtual power plant aggregation and control technology, although capable of aggregating distributed resources to participate in grid interaction, emphasizes grid-side dispatch demand response, with a many-to-many resource aggregation method, failing to address one-to-many direct green electricity supply scenarios. Regarding fair allocation, existing methods often employ simple proportional allocation or first-come-first-served mechanisms, failing to comprehensively consider multi-dimensional factors such as electricity consumption history, real-time demand, and regulatory contributions. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a green electricity collaborative scheduling method and related device for a single source and multiple load scenario.

[0005] To achieve the above objectives, the present invention employs the following technical solution: In a first aspect, this invention provides a green energy collaborative scheduling method for a single-source, multi-load scenario, comprising: acquiring historical electricity consumption data of each user and performing load characteristic mining to obtain load characteristic mining results for each user; wherein the load characteristic mining results include load allocation results, adjustable load characteristic parameter models, user-level adjustable capacity, and user-level adjustment cost functions; acquiring grid status data, green energy output prediction data, and electricity consumption prediction data of each user, and solving a preset multi-objective optimization model in conjunction with the load characteristic mining results of each user to obtain the electricity consumption plan, green energy allocation scheme, and grid power purchase plan for each user; wherein the multi-objective optimization model takes maximizing green energy absorption rate, minimizing total system operating cost, and maximizing power supply fairness as optimization objectives; wherein maximizing power supply fairness is to maximize the average comprehensive fairness score of each user, and the comprehensive fairness score of each user is determined based on the historical green energy allocation scheme of each user and in conjunction with the load characteristic mining results of each user, through a preset multi-dimensional fairness index system; wherein the multi-dimensional fairness index system includes historical fairness, real-time demand fairness, adjustment contribution fairness, and cost fairness.

[0006] Optionally, the historical electricity consumption data includes historical active power, historical reactive power, historical voltage, historical current, historical power factor, and sampling interval; the power grid status data includes power grid frequency, node voltage, and sampling interval; the green power output prediction data includes predicted photovoltaic power output, predicted wind power output, and sampling interval; the electricity consumption prediction data includes predicted active power, predicted reactive power, predicted voltage, predicted current, predicted power factor, and sampling interval; before performing load feature mining, the method further includes: performing outlier detection, missing value imputation, time alignment, and data aggregation on the historical electricity consumption data of each user.

[0007] Optionally, the load feature mining includes: for each user, extracting feature vectors based on historical electricity consumption data, and performing cluster analysis on the extracted feature vectors into four categories, which are respectively mapped to rigid load, transferable load, reduceable load, and adjustable load; constructing characteristic parameters, cost functions, and constraints for transferable load, reduceable load, and adjustable load to obtain an adjustable load characteristic parameter model; and performing user-level aggregation based on the adjustable load characteristic parameter model to obtain user-level adjustable capacity and user-level adjustment cost function.

[0008] Optionally, the feature vector includes daily load factor, daily peak-to-valley difference rate, load duration rate, nighttime load percentage, fluctuation coefficient, and load time autocorrelation coefficient; the cluster analysis of the extracted feature vector includes K-means cluster analysis; the construction of characteristic parameters, cost functions, and constraints for transferable load, reduceable load, and adjustable load to obtain an adjustable load characteristic parameter model includes: for users Transferable load Characteristic parameters include: transferable start time Transferable end time Maximum transfer power , To transfer electrical energy; the cost function is in quadratic form; the constraints are:

[0009]

[0010]

[0011] in, For users time Transfer power of transferable load.

[0012] For users Reduced load Characteristic parameters include: maximum power reduction Single continuous reduction of the time limit Maximum number of daily reductions and minimum recovery interval The cost function is in quadratic form; the constraints are:

[0013]

[0014]

[0015]

[0016] in, For counting days; For users No. The number of days reduced; For users The time interval between two consecutive cuts; For users time Power reduction that can reduce load.

[0017] For users Adjustable load Characteristic parameters include: minimum regulating power Maximum adjustable power and response time constant The cost function is in quadratic form; the constraints are:

[0018]

[0019] in, For users time Adjustable load operating power, For users Adjustable load The maximum regulation rate; For users Adjustable load operating power.

[0020] Adjustable loads also include dynamic response characteristics:

[0021] in, To adjust the command power.

[0022] The step of performing user-level aggregation based on the adjustable load characteristic parameter model to obtain the user-level adjustable capacity and user-level adjustment cost functions includes: user User-level adjustable capacity for:

[0023] user User-level adjustment cost function for:

[0024] in, The cost function of transferable loads, The cost function for load reduction, This is the cost function for adjustable loads.

[0025] Optionally, obtaining green power output prediction data and electricity consumption prediction data for each user includes: using an LSTM-Attention-based green power output prediction model to obtain green power output prediction data, and using an LSTM-Attention-based electricity consumption prediction model to obtain electricity consumption prediction data for each user; wherein, the input features of the green power output prediction model include numerical weather forecasts, historical power output data, and day type identifiers; the input features of the electricity consumption prediction model include historical electricity consumption data and production scheduling information.

[0026] Optionally, the objective function of the preset multi-objective optimization model is:

[0027]

[0028]

[0029] in, To maximize the green electricity consumption rate, To minimize the total operating cost of the system, To maximize fairness in power supply, For the scheduling period, for The actual amount of green electricity consumed at any given time. for Green electricity is always available. For time step, for Real-time electricity purchase price for Power purchase capacity of the power grid at any time For users The user-level adjustment cost function, For users User-level adjustable capacity, The price for penalties for power curtailment for Power curtailment at any given time For the number of users, For users The overall fairness score.

[0030] The pre-defined multi-objective optimization model's constraints include power balance constraints, green electricity output constraints, electricity demand constraints for each user, green electricity allocation constraints for each user, load regulation constraints, grid power purchase constraints, and fair allocation ratio constraints; among which, the green electricity allocation constraints for each user are:

[0031]

[0032] The fairness distribution ratio constraint is:

[0033]

[0034] in, For users time The allocated green electricity power, For users time The proportion of green electricity allocation For a moment The total green power actually absorbed by the system For users The lower limit of the green electricity allocation ratio For users The upper limit of green electricity allocation ratio, For users At any moment The proportion of green electricity allocation This represents the maximum permissible variation in the green electricity allocation ratio between adjacent time points.

[0035] Optionally, solving the preset multi-objective optimization model includes: using an improved multi-objective particle swarm optimization algorithm to solve the preset multi-objective optimization model, and when there are multiple non-dominated solutions in the Pareto solution set, using the TOPSIS method to select the final solution to obtain the electricity consumption plan, green electricity allocation scheme, and power grid purchase plan for each user; wherein, the improved multi-objective particle swarm optimization algorithm is obtained by improving the multi-objective particle swarm optimization algorithm as follows: improving the inertial weight of the multi-objective particle swarm optimization algorithm to an adaptive inertial weight:

[0036] in, For the first Inertia weights in the next iteration This represents the maximum value of the inertia weight. This represents the minimum value of the inertia weight. The maximum number of iterations, It is a non-linear adjustment factor.

[0037] The mutation strategy of the multi-objective particle swarm optimization algorithm is improved by using a Logistic mapping to generate a chaotic sequence to mutate the particle positions.

[0038] Optionally, the step of acquiring grid status data, green electricity output prediction data, and electricity consumption prediction data of each user, and solving a preset multi-objective optimization model in combination with the load characteristic mining results of each user to obtain the electricity consumption plan, green electricity allocation scheme, and grid power purchase plan of each user includes: adopting a rolling optimization strategy to solve the multi-objective optimization model once at preset intervals; and generating an emergency dispatch trigger command when any of the following trigger conditions are detected: the actual green electricity output deviates from the prediction by more than 20%, the user load changes by more than 30%, and the grid issues an emergency dispatch command.

[0039] Optionally, the comprehensive fairness score is a weighted sum of historical fairness, real-time demand fairness, adjustment contribution fairness, and cost fairness; wherein, historical fairness is: ,

[0040] in, For users Historical fairness, For users in the past The actual green electricity generated within a day For users The green electricity they deserve For users Historical total electricity consumption Total historical electricity consumption for all users. This will absorb the total amount of green electricity in history.

[0041] Real-time demand fairness is:

[0042] in, For users Real-time demand fairness, For users At any moment rigid load power, For users At any moment Total power consumption.

[0043] Adjusting the fairness of contributions is as follows:

[0044]

[0045] in, For users Adjusting the fairness of contributions, For users The regulating energy provided during the statistical period For users Total electricity consumption within the statistical period For the scheduling period, For users time Transfer power of transferable load, For users time Power reduction that can reduce load For users time Adjustable load operating power, For users time Reference power, For time step.

[0046] Cost fairness is:

[0047]

[0048]

[0049] in, For users Cost fairness For users Total electricity cost for Real-time electricity purchase price For users exist Power purchased by the power grid at any given time For green electricity prices, For users exist The green electricity power obtained at all times The average electricity cost for all users; For the number of users.

[0050] Optionally, this also includes: determining the green electricity allocation weight for each user based on their overall fairness score.

[0051] in, For users The overall fairness score, For the number of users, For users Green electricity allocation weight; For users The overall fairness score; The user is obtained through the following formula. time Fair green electricity allocation ratio And perform normalization:

[0052]

[0053]

[0054] in, For users Lower limit for fair green electricity allocation ratio For users Daily average green electricity allocation weight For users Upper limit on the proportion of fair green electricity allocation.

[0055] According to user time Fair green electricity allocation ratio Revise the green electricity allocation plan for each user.

[0056] Optionally, it also includes: obtaining the fairness deviation for each user using the following formula:

[0057]

[0058] in, For users time Fairness bias, For users time The overall fairness score, For all users at any time The average score of overall fairness, For the number of users.

[0059] when When this happens, a Level 1 warning is issued, triggering adjustments to the green electricity allocation plan, which are then corrected in the next scheduling cycle; when When this happens, a level-two early warning is issued, triggering adjustments to the green electricity allocation plan and correcting it within two scheduling cycles; when users Daily cumulative fairness deviation At that time, a three-level early warning will be issued, and the compensation will be included in the next day's dispatch plan.

[0060] Optionally, it also includes: receiving dispatch instructions issued by the power grid; and generating a peak-shaving response strategy when the dispatch instruction is a peak-shaving instruction.

[0061]

[0062]

[0063] in, The total system response power, For users time Adjustable power, For a moment The target response power, For users User-level adjustable capacity, This represents the total adjustable capacity of the system. For users time Adjustable power for transferable loads, For users time It can reduce the regulating power of the load. For users time Adjustable power for adjustable loads.

[0064] When the scheduling instruction is a frequency modulation instruction, a frequency modulation response strategy is generated:

[0065] in, The frequency regulation target power issued by the power grid, For users The response time constant, For users The response time constant, For the number of users.

[0066] In a second aspect, this invention provides a green electricity collaborative scheduling system for a single-source, multi-load scenario, comprising: a feature mining module, used to acquire historical electricity consumption data of each user and perform load feature mining to obtain load characteristic mining results for each user; wherein the load characteristic mining results include load allocation results, adjustable load characteristic parameter models, user-level adjustable capacity, and user-level adjustment cost functions; and an optimization module, used to acquire grid status data, green electricity output prediction data, and electricity consumption prediction data of each user, and solve a preset multi-objective optimization model in conjunction with the load characteristic mining results of each user to obtain the electricity consumption plan, green electricity allocation scheme, and grid power purchase plan for each user; wherein the multi-objective optimization model takes maximizing green electricity absorption rate, minimizing total system operating cost, and maximizing power supply fairness as optimization objectives; wherein maximizing power supply fairness is to maximize the average comprehensive fairness score of each user, and the comprehensive fairness score of each user is determined based on the historical green electricity allocation scheme of each user and in conjunction with the load characteristic mining results of each user, through a preset multi-dimensional fairness index system; wherein the multi-dimensional fairness index system includes historical fairness, real-time demand fairness, adjustment contribution fairness, and cost fairness.

[0067] In a third aspect, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the green electricity collaborative scheduling method for a single-source, multi-load scenario.

[0068] On the fourth aspect of this invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the steps of the green electricity collaborative scheduling method for a single-source, multi-load scenario.

[0069] Compared with the prior art, the present invention has the following beneficial effects: This invention presents a green energy collaborative scheduling method for single-source, multi-load scenarios. By acquiring historical electricity consumption data from each user and mining load characteristics, it can accurately identify the load type, adjustable potential, and adjustment cost of each user, providing a refined and quantitative data foundation for subsequent collaborative scheduling. This effectively overcomes the shortcomings of traditional methods, such as insufficient depth of load characteristic mining and inability to support multi-user collaborative scheduling. By constructing a multi-objective optimization model with the optimization objectives of maximizing green energy absorption rate, minimizing total system operating cost, and maximizing power supply fairness, this method can simultaneously achieve optimal allocation of green energy resources, improve system economy, and ensure fair allocation among users while ensuring the electricity needs of each user. The collaborative optimization of these three objectives overcomes the limitations of traditional single-objective optimization, which cannot take into account multi-dimensional needs. This significantly improves the overall performance of the system, achieves accurate matching and optimized scheduling of green energy and multi-user loads, effectively increases the green energy absorption rate, and realizes dynamic matching and coordinated scheduling of green energy output and multi-user loads in single-source, multi-load scenarios. This provides core technical support for a new model of direct green energy supply from a single green energy project to multiple industrial users. Furthermore, by constructing a multi-dimensional fairness indicator system that includes historical fairness, real-time demand fairness, adjustment contribution fairness, and cost fairness, the comprehensive fairness score of each user is determined, achieving fair, efficient, and transparent allocation of green electricity among multiple users. The fairness indicators can be stably maintained at a high level, avoiding disputes and conflicts. Attached Figure Description

[0070] Figure 1 This is a flowchart of a green electricity collaborative scheduling method for a single-source, multi-load scenario according to an embodiment of the present invention.

[0071] Figure 2 This is a block diagram of a green electricity collaborative scheduling system for a single-source, multi-load scenario, according to an embodiment of the present invention. Detailed Implementation

[0072] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0073] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0074] As introduced in the background section, in the traditional one-to-one dedicated power supply mode, the power supply and load have a one-to-one correspondence, and the dispatch and control are relatively simple, mainly involving the power balance between a single power supply and a single user. However, in the one-source-multiple-load mode, a single green power project needs to serve N industrial users (N≥2) at the same time, and the system complexity increases exponentially, facing the following three core technical challenges.

[0075] Challenge 1: It is difficult to accurately match the fluctuations in green electricity output with the diversity of multi-user loads.

[0076] Green electricity (solar / wind power) output is highly volatile, intermittent, and uncertain due to weather conditions. Solar power output exhibits a bell-shaped daily variation, fluctuating by more than 30% within 5 minutes due to cloud cover. Wind power output is even more random, with daily fluctuations reaching up to 80% of rated capacity. Meanwhile, different industrial users exhibit significantly different load characteristics: Different production processes result in different load types: steel mills operate on a continuous and stable model (load rate 85-95%), chemical companies experience batch-based fluctuations (day-night peak-valley difference 30-50%), machinery manufacturing companies have a concentrated daytime load (load rate 30-90%), and electronic component companies maintain a high-precision, constant load (voltage fluctuation tolerance <±2%). Production schedules also differ: three-shift companies use electricity continuously 24 hours a day, single-shift companies only use electricity during the day (8:00-18:00), and seasonal companies experience fluctuations based on orders. The load composition differs: steel companies have an adjustable load ratio of 30-40% (electric furnaces and rolling mills can operate at reduced power for short periods), electronics companies only 10-15% (precision equipment cannot be powered off), and chemical companies have a transferable load ratio of 20-30% (reactor heating can be staggered). The adjustment response speed also differs: electric arc furnaces have response times in the order of seconds, electric boilers in the order of minutes, and refrigeration systems in the order of 10-30 minutes.

[0077] This multi-dimensional and multi-scale complexity makes the dynamic balance of green electricity supply and demand a typical high-dimensional nonlinear optimization problem, which is completely unsuitable for traditional single-user scheduling methods.

[0078] Challenge 2: It is difficult to scientifically guarantee the fairness of green electricity allocation among multiple users.

[0079] When green electricity output is limited (e.g., on cloudy or windless days), fairly allocating scarce green electricity resources among N users presents an unprecedented technical challenge. This challenge manifests in several ways: 1. The "free-rider" problem: Some users enjoy the benefits of green electricity but do not participate in load regulation, continuously occupying green electricity quotas without contributing regulation resources, leading to a decrease in the overall green electricity consumption rate. 2. Unfair allocation: Simply allocating according to load ratio ignores differences in user characteristics. Users who contribute significantly to regulation (making room for green electricity in the system's flexible adjustments) receive less green electricity, severely discouraging user participation in regulation. 3. Traceability issues: The lack of a transparent allocation mechanism and traceable records makes it difficult to provide reasonable explanations when users question the allocation results, easily leading to disputes and conflicts. 4. Dynamic changes: Users' load demands change in real time; a fixed allocation ratio cannot adapt to dynamic scenarios and requires real-time dynamic adjustments. Existing methods often employ simple proportional allocation (allocation based on user capacity share) or a first-come, first-served mechanism (prioritizing users who apply first), completely ignoring multi-dimensional factors such as differences in user electricity consumption characteristics, historical fairness of electricity consumption, urgency of real-time demand, and contribution of regulation. These methods are neither reasonable nor fair, and are also impossible to trace.

[0080] Challenge 3: Lack of dedicated Green Energy Control Center (GCC) hardware and software systems.

[0081] Existing microgrid control systems are primarily designed for single-industry or single-user scenarios, with simple functional architectures. They cannot achieve complex functions such as multi-user collaborative scheduling, fairness assurance and transparent allocation, grid coordination and interaction, real-time monitoring and decision support in scenarios with multiple loads from a single source. While Virtual Power Plant (VPP) technology can aggregate distributed resources, it focuses on grid-side dispatch demand response, with a many-to-many resource aggregation method. It does not address one-to-many green electricity direct supply scenarios, nor does it design user-side resource collaboration and fair allocation mechanisms from the perspective of green electricity direct supply. Therefore, there is an urgent need to construct a new control system architecture and a complete set of hardware and software devices.

[0082] In summary, existing technical solutions in single-source, multi-load scenarios have the following main shortcomings: (1) Traditional microgrid energy management system (EMS) technology. Although this technology can achieve coordinated scheduling of power source, load and energy storage, it is mainly aimed at resource management within a single park. The system architecture is a centralized design, lacking a coordination mechanism among multiple users, and cannot support the allocation of green electricity and fairness guarantee for multiple users in a single source and multiple load scenario. Its scheduling algorithm is based on single objective optimization (usually the lowest operating cost) and does not consider multi-objective coordinated optimization (green electricity consumption rate, user cost, fairness). In addition, the system does not have the ability to collect and integrate multi-user data, cannot obtain real-time load data of multiple users, and is difficult to support multi-user coordinated scheduling.

[0083] (2) Virtual Power Plant (VPP) aggregation and control technology. Although this technology can aggregate distributed resources to participate in grid interaction, it focuses on grid-side dispatch demand response. The resource aggregation method is many-to-many (multiple distributed power sources aggregate to respond to grid dispatch), and it does not involve one-to-many green electricity direct supply scenarios (a single green electricity project is allocated to multiple users). Its dispatch strategy is based on electricity price signals or incentive signals, and it does not design a user-side resource coordination and fair allocation mechanism from the perspective of green electricity direct supply. In addition, the VPP system does not have green electricity output prediction and multi-user load matching optimization functions, and cannot achieve accurate matching between green electricity and multi-user loads.

[0084] (3) Industrial load characteristic mining and regulation potential assessment technology. Most of these methods target single users or single-type loads, employing statistical analysis or simple machine learning, resulting in insufficient depth of analysis. For example, loads are only divided into two main categories: adjustable and non-adjustable, without further subdivision into subcategories such as transferable, reduceable, and adjustable. Furthermore, there is a lack of precise quantitative methods for key parameters such as regulation capacity, regulation cost, and response speed for various adjustable load types. In addition, there is a lack of quantitative methods and models for the coordinated regulation potential of multi-user, multi-type industrial loads.

[0085] (4) Fairness allocation and scheduling technology. Existing fairness allocation methods are mostly applied to areas such as grid ancillary service allocation and carbon quota allocation, using methods such as Shapley value method and Data Envelopment Analysis (DEA). However, the Shapley value method has a computational complexity of O(N!), and the computational load increases dramatically as the number of users increases, making it unsuitable for real-time scheduling scenarios. The DEA method requires a large amount of historical data to construct the efficiency frontier and has high requirements for data quality. More importantly, these methods have not designed dedicated fairness indicators and allocation mechanisms for direct green electricity supply scenarios with multiple sources and loads, and cannot comprehensively consider multi-dimensional factors such as electricity consumption history, real-time demand, regulation contribution, and cost.

[0086] (5) Lack of dedicated control center hardware and software system design schemes for single-source, multi-load scenarios. Currently, there is no complete technical solution for dedicated control center system architecture, hardware devices, software functional modules, and control strategies for single-source, multi-load green electricity direct supply scenarios. Existing microgrid controllers, VPP aggregation platforms, and demand response management systems cannot meet the special needs of single-source, multi-load scenarios.

[0087] In summary, existing technologies cannot support the technical requirements of the new form of direct green electricity supply from a single source to multiple loads in terms of system architecture, control methods, fairness mechanisms, and hardware and software implementation. There is an urgent need for a green electricity collaborative scheduling method and related devices for single-source, multi-load scenarios to achieve accurate matching, fair allocation, and coordinated interaction between green electricity and multi-user loads.

[0088] The present invention will now be described in further detail with reference to the accompanying drawings: See Figure 1In one embodiment of the present invention, a green energy collaborative scheduling method for a single-source, multi-load scenario is provided to overcome some of the problems described above and achieve efficient green energy scheduling in a single-source, multi-load scenario.

[0089] Specifically, the green electricity collaborative scheduling method for a single-source, multi-load scenario of the present invention includes the following steps: S1: Obtain historical electricity consumption data for each user and perform load characteristic mining to obtain load characteristic mining results for each user; wherein, the load characteristic mining results include load division results, adjustable load characteristic parameter models, user-level adjustable capacity, and user-level adjustment cost functions.

[0090] Specifically, step S1 is used to perform in-depth analysis and feature extraction of the electricity consumption behavior of various industrial users participating in the green electricity direct supply. In a single-source, multi-load scenario, the load characteristics of different industrial users differ significantly. These differences stem from various factors such as production processes, production scheduling, load composition, and regulation response speed. Historical electricity consumption data can be obtained from data acquisition terminals deployed on each user side. These terminals include smart meters and power transmitters, with a data acquisition frequency of once per second. The load segmentation results in the load characteristic mining are used to classify user loads into categories with different regulation characteristics. The adjustable load characteristic parameter model is used to describe the regulation capacity boundary and economic cost of various adjustable loads. The user-level adjustable capacity and user-level regulation cost function are used for constraints and cost calculations in the subsequent multi-objective optimization model.

[0091] S2: Obtain power grid status data, green electricity output prediction data, and electricity consumption prediction data for each user, and solve the preset multi-objective optimization model by combining the load characteristic mining results of each user to obtain the electricity consumption plan, green electricity allocation scheme, and power grid purchase plan for each user; wherein, the multi-objective optimization model takes maximizing the green electricity absorption rate, minimizing the total system operating cost, and maximizing the fairness of power supply as optimization objectives.

[0092] Specifically, step S2 is used to make optimization decisions based on the load characteristics of each user and external environmental information. Green electricity output is highly volatile due to weather conditions, requiring the acquisition of green electricity output forecast data and user electricity consumption forecast data to predict the supply and demand situation in the future. Grid status data reflects the operating status of the external grid and supports coordinated interaction with the grid. The three optimization objectives of the multi-objective optimization model correspond to three dimensions: green electricity absorption efficiency, system operating economy, and fairness of distribution among users. By solving this model, we can obtain the electricity consumption plan of each user at each moment in the future, how green electricity will be distributed among users, and how much electricity needs to be purchased from the grid.

[0093] Specifically, in the optimization objective of maximizing power supply fairness, maximizing power supply fairness means maximizing the average comprehensive fairness score of each user. The comprehensive fairness score of each user is determined based on their historical green electricity allocation schemes and the results of load characteristic mining, through a pre-set multi-dimensional fairness index system. This multi-dimensional fairness index system includes historical fairness, real-time demand fairness, regulation contribution fairness, and cost fairness. Specifically, historical fairness reflects the ratio of green electricity a user receives to the green electricity they deserve within a historical period, used to achieve long-term allocation compensation; real-time demand fairness reflects the proportion of a user's current rigid load demand in total electricity demand, used to prioritize users with a high proportion of rigid load; regulation contribution fairness reflects the degree of a user's contribution to system regulation, used to incentivize users to actively participate in load regulation; and cost fairness reflects the deviation between a user's electricity cost and the system's average cost, used to balance cost differences between users with different electricity consumption structures. Through the comprehensive evaluation of these four dimensions, the share that each user should receive in green electricity allocation can be comprehensively and objectively measured, effectively ensuring the fairness and traceability of the allocation.

[0094] This invention presents a green energy collaborative scheduling method for single-source, multi-load scenarios. By acquiring historical electricity consumption data from each user and mining load characteristics, it can accurately identify the load type, adjustable potential, and adjustment cost of each user, providing a refined and quantitative data foundation for subsequent collaborative scheduling. This effectively overcomes the shortcomings of traditional methods, such as insufficient depth of load characteristic mining and inability to support multi-user collaborative scheduling. By constructing a multi-objective optimization model with the optimization objectives of maximizing green energy absorption rate, minimizing total system operating cost, and maximizing power supply fairness, this method can simultaneously achieve optimal allocation of green energy resources, improve system economy, and ensure fair allocation among users while ensuring the electricity needs of each user. The collaborative optimization of these three objectives overcomes the limitations of traditional single-objective optimization, which cannot take into account multi-dimensional needs. This significantly improves the overall performance of the system, achieves accurate matching and optimized scheduling of green energy and multi-user loads, effectively increases the green energy absorption rate, and realizes dynamic matching and coordinated scheduling of green energy output and multi-user loads in single-source, multi-load scenarios. This provides core technical support for a new model of direct green energy supply from a single green energy project to multiple industrial users. Furthermore, by constructing a multi-dimensional fairness indicator system that includes historical fairness, real-time demand fairness, adjustment contribution fairness, and cost fairness, the comprehensive fairness score of each user is determined, achieving fair, efficient, and transparent allocation of green electricity among multiple users. The fairness indicators can be stably maintained at a high level, avoiding disputes and conflicts.

[0095] In one possible implementation, the historical electricity consumption data includes historical active power, historical reactive power, historical voltage, historical current, historical power factor, and sampling interval; the grid status data includes grid frequency, node voltage, and sampling interval; the green power output prediction data includes predicted photovoltaic power output, predicted wind power output, and sampling interval; the electricity consumption prediction data includes predicted active power, predicted reactive power, predicted voltage, predicted current, predicted power factor, and sampling interval; before performing load feature mining, the method further includes: performing outlier detection, missing value imputation, time alignment, and data aggregation on the historical electricity consumption data of each user.

[0096] For example, outlier detection uses the 3σ criterion to remove outliers. Missing value imputation uses linear interpolation, which estimates the missing value using valid data from adjacent time points to ensure the continuity of the time series. Time alignment unifies data from different user sides and green electricity sides to a 1-second time base, as clocks at different data acquisition terminals may deviate, requiring synchronization to ensure consistent timestamps. Data aggregation aggregates the 1-second sampled data into 15-minute average data for subsequent scheduling. Through the above preprocessing steps, a high-quality, consistent, standardized dataset can be obtained, providing a reliable data foundation for subsequent load feature mining and predictive model training.

[0097] Explained, in this embodiment, by performing outlier detection, missing value imputation, time alignment, and data aggregation on historical electricity consumption data, standardized processing of multi-source heterogeneous data is achieved, which effectively improves data quality and avoids the adverse effects of outliers and missing values ​​in the original data on the accuracy of subsequent feature mining.

[0098] In one possible implementation, the load feature mining includes: for each user, extracting feature vectors based on historical electricity consumption data, and performing cluster analysis on the extracted feature vectors into four categories, which are respectively mapped to rigid load, transferable load, reduceable load, and adjustable load; constructing characteristic parameters, cost functions, and constraints for transferable load, reduceable load, and adjustable load to obtain an adjustable load characteristic parameter model; and performing user-level aggregation based on the adjustable load characteristic parameter model to obtain user-level adjustable capacity and user-level adjustment cost function.

[0099] In one possible implementation, the feature vector includes daily load factor, daily peak-to-valley difference rate, load duration rate, nighttime load percentage, fluctuation coefficient, and load time autocorrelation coefficient; the clustering analysis of the extracted feature vector includes performing K-means clustering analysis on the extracted feature vector.

[0100] For example, collecting each user's past... Historical electricity consumption data for days, building user... Load time series: ;in, For users At any moment Active power, in kW; For data length, (96 refers to 96 15-minute time points per day).

[0101] The following feature vector is extracted from the load time series: Daily load rate: This reflects the stability of the load. Daily peak-valley difference rate: This reflects the magnitude of load fluctuations. Load duty cycle: This reflects the percentage of high load duration. Nighttime load percentage: This reflects the proportion of electricity consumption at night. Fluctuation coefficient: This reflects load fluctuations. Load time autocorrelation coefficient: This reflects the time-dependent nature of the load. Among them, and users respectively Maximum and minimum load during the day; This represents the daily average load. and users respectively Standard deviation and mean of the load series; and These are covariance and variance, respectively. The time lag order ( ); This indicates the number of sampling points that meet the condition.

[0102] For example, K-means clustering analysis is performed on the extracted feature vectors. The load of each user at each time period is divided into 4 categories: ;in, For the first A cluster, For the first Cluster centers of each cluster For the first Feature vectors for each load period The initial center selection for K-means clustering uses the K-means++ algorithm to avoid local optima caused by random initialization. After clustering, the four clusters are mapped to the following load types: Cluster 1 corresponds to rigid loads: non-adjustable base loads, such as lighting, control power, and core production equipment. Cluster 2 corresponds to transferable loads: loads whose power consumption can be transferred within a specified time window, such as electric boiler heating and electric furnace smelting. Cluster 3 corresponds to loads that can be reduced in a short period of time, such as air conditioning and non-critical lighting. Cluster 4 corresponds to adjustable loads: loads whose power can be continuously adjusted within a certain range, such as variable frequency water pumps and adjustable speed fans. Then, using the K-means clustering results as labels, a random forest classifier is trained for real-time online classification of newly collected load data. Random forest model parameters: number of decision trees. Maximum depth Minimum number of leaf node samples The classification accuracy reached over 95% after cross-validation.

[0103] In one possible implementation, the construction of characteristic parameters, cost functions, and constraints for transferable loads, reduceable loads, and adjustable loads to obtain an adjustable load characteristic parameter model includes: For users Transferable load Characteristic parameters include: transferable start time Transferable end time Maximum transfer power , To transfer electrical energy; the cost function is in quadratic form; the constraints are:

[0104]

[0105]

[0106] in, For users time Transfer power of transferable load.

[0107] Specifically, the first constraint indicates that the power within the non-transfer window is 0; the second constraint indicates that the transferred power does not exceed the upper limit; and the third constraint indicates that the total transferred power meets the demand.

[0108] For users Reduced load Characteristic parameters include: maximum power reduction Single continuous reduction of the time limit Maximum number of daily reductions and minimum recovery interval The cost function is in quadratic form; the constraints are:

[0109]

[0110]

[0111]

[0112] in, For counting days; For users No. The number of days reduced; For users The time interval between two consecutive cuts; For users time Power reduction that can reduce load.

[0113] Specifically, the first constraint is the power reduction constraint; the second constraint is the single duration constraint; the third constraint is the cumulative number of times constraint; and the fourth constraint is the minimum recovery interval constraint.

[0114] For users Adjustable load Characteristic parameters include: minimum regulating power Maximum adjustable power and response time constant The cost function is in quadratic form; the constraints are:

[0115]

[0116] in, For users time Adjustable load operating power, For users Adjustable load The maximum regulation rate; For users Adjustable load operating power.

[0117] Specifically, the first constraint is the adjustment range constraint; the second constraint is the adjustment rate constraint.

[0118] An example, the transfer cost function: ;in, This is the secondary cost coefficient (yuan / kW²). The primary cost coefficient is (yuan / kW). The fixed cost is expressed in yuan. The coefficients were determined through regression fitting of historical data.

[0119] Adjustable loads also include dynamic response characteristics:

[0120] in, To adjust the command power.

[0121] For example, the dynamic response characteristics of adjustable loads are described using a first-order inertial model. The response time constant of electric arc furnaces in steel enterprises is 1~5s, the response time constant of electric boilers is 30~60s, and the response time constant of refrigeration systems is 300~600s.

[0122] In one possible implementation, the step of performing user-level aggregation based on the adjustable load characteristic parameter model to obtain the user-level adjustable capacity and user-level adjustment cost function includes: user User-level adjustable capacity for:

[0123] user User-level adjustment cost function for:

[0124] in, The cost function of transferable loads, The cost function for load reduction, This is the cost function for adjustable loads.

[0125] Meanwhile, the total adjustable capacity at the system level is: .

[0126] System-level regulation capacity must meet the following requirements: That is, the total regulation capacity of the system shall not be less than 15% of the rated installed capacity of green electricity, so as to ensure that the system has sufficient regulation margin.

[0127] In one possible implementation, obtaining green power output forecast data and electricity consumption forecast data for each user includes: using an LSTM-Attention-based green power output forecast model to obtain green power output forecast data, and using an LSTM-Attention-based electricity consumption forecast model to obtain electricity consumption forecast data for each user; wherein, the input features of the green power output forecast model include numerical weather forecasts, historical power output data, and day type identifiers; the input features of the electricity consumption forecast model include historical electricity consumption data and production scheduling information.

[0128] Interpretive analysis employs an LSTM-Attention-based green energy output prediction model, with input features including: numerical weather prediction (irradiance). Wind speed ,temperature Based on historical power output data and daily power type indicators (sunny / partly cloudy / overcast / rainy), the model predicts green power output for the next 24 hours with a time resolution of 15 minutes (96 time points). Prediction model structure: ,in, This is the feature vector of numerical weather prediction. Contributing to the sequence of history This is a one-hot encoding for the Japanese type.

[0129] Meanwhile, the electricity consumption forecast data for each user adopts a similar structure, with additional input of user production scheduling information (weekdays / rest days / overtime days), and the forecast accuracy MAPE < 8%.

[0130] In one possible implementation, the objective function of the preset multi-objective optimization model is:

[0131]

[0132]

[0133] in, To maximize the green electricity consumption rate, To minimize the total operating cost of the system, To maximize fairness in power supply, For the scheduling period, for The actual amount of green electricity consumed at any given time. for Green electricity is always available. For time step, for Real-time electricity purchase price for Power purchase capacity of the power grid at any time For users The user-level adjustment cost function, For users User-level adjustable capacity, The price for penalties for power curtailment for Power curtailment at any given time For the number of users, For users The overall fairness score.

[0134] objective function This measure assesses the utilization efficiency of green electricity. The numerator is the total green electricity actually consumed within the dispatch cycle, and the denominator is the total available green electricity within the dispatch cycle. A ratio closer to 1 indicates more efficient utilization of green electricity. Objective function The system's economics were measured, including the cost of purchasing electricity from the grid, the cost of load regulation for each user, and the cost of curtailment penalties. The curtailment penalty price reflects the resource waste and potential policy penalties resulting from the inability to absorb green electricity. Objective function This measure assesses the fairness of green electricity allocation among users. The value is the average of the overall fairness scores of each user, with a higher value indicating a fairer allocation.

[0135] Explaining this, in a single-source, multi-load scenario, green electricity output is highly volatile due to weather conditions, while the load characteristics of each user differ significantly and change in real time. This leads to complex coupling and contradictions among the three objective functions: for example, excessively pursuing maximizing green electricity absorption may require some users to frequently adjust their loads, increasing their adjustment costs; excessively pursuing cost minimization may result in green electricity allocation favoring users with lower adjustment costs, compromising allocation fairness. Therefore, a multi-objective optimization model is constructed, and a multi-objective optimization algorithm seeks the optimal compromise solution among the three objectives. In this implementation, by setting the aforementioned three objective functions, the performance of the green electricity dispatch scheme in a single-source, multi-load scenario can be comprehensively measured across three dimensions: resource utilization efficiency, economic cost, and allocation fairness, providing a clear direction for multi-objective optimization.

[0136] The pre-defined multi-objective optimization model's constraints include power balance constraints, green electricity output constraints, electricity demand constraints for each user, green electricity allocation constraints for each user, load regulation constraints, grid power purchase constraints, and fair allocation ratio constraints; among which, the green electricity allocation constraints for each user are:

[0137]

[0138] The fairness distribution ratio constraint is:

[0139]

[0140] in, For users time The allocated green electricity power, For users time The proportion of green electricity allocation For a moment The total green power actually absorbed by the system For users The lower limit of the green electricity allocation ratio For users The upper limit of green electricity allocation ratio, For users At any moment The proportion of green electricity allocation This represents the maximum permissible variation in the green electricity allocation ratio between adjacent time points.

[0141] Specifically, time step The time interval is 0.25 hours, the scheduling cycle is 96 time points, and the power unit is kW.

[0142] Specifically, power balance constraints: , indicating any time The sum of green electricity consumption capacity and grid-purchased electricity capacity equals the sum of all user electricity consumption capacity. Green electricity output constraint: Electricity demand constraints for each user: .in, For users At any moment The minimum power consumption (rigid load requirement). This represents the maximum power consumption. Constraints on green electricity allocation for each user: , Load regulation constraints: Each user's load regulation must meet the constraints of transferable load, load reductionable load, and adjustable load. Power grid purchase constraints: ; This represents the maximum power purchase capacity for the dedicated power line. Fairness allocation ratio constraints: , The first formula ensures that all green electricity is allocated to all users; the second formula limits the variation in allocation ratios between adjacent time periods. Meanwhile, to avoid frequent and drastic adjustments, The value is usually 0.1.

[0143] Explained, the aforementioned constraints collectively constitute the feasible region boundary of the multi-objective optimization model. Power balance constraints are fundamental physical constraints on power system operation; green electricity output constraints reflect the availability limitations of green electricity resources; user electricity demand constraints guarantee the production electricity needs of each user; green electricity allocation constraints for each user specify the adjustable range of allocation ratios for each user; load regulation constraints ensure that regulation operations are carried out within the physical limits of various load types; grid power purchase constraints reflect the physical limitations of dedicated line transmission capacity; and fair allocation ratio constraints limit the variation range between adjacent time periods from the perspective of allocation stability. These constraints work together to ensure the practical feasibility of the optimization solution.

[0144] In one possible implementation, solving the preset multi-objective optimization model includes: solving the preset multi-objective optimization model using an improved multi-objective particle swarm optimization algorithm, and when multiple non-dominated solutions exist in the Pareto solution set, using the TOPSIS method to select the final solution to obtain the electricity consumption plan, green electricity allocation scheme, and grid power purchase plan for each user; wherein, the improved multi-objective particle swarm optimization algorithm is obtained by improving the multi-objective particle swarm optimization algorithm as follows: the inertial weight of the multi-objective particle swarm optimization algorithm is improved to an adaptive inertial weight:

[0145] in, For the first Inertia weights in the next iteration This represents the maximum value of the inertia weight. This represents the minimum value of the inertia weight. The maximum number of iterations, It is a non-linear adjustment factor.

[0146] The mutation strategy of the multi-objective particle swarm optimization algorithm is improved by using a Logistic mapping to generate a chaotic sequence to mutate the particle positions.

[0147] Specifically, the standard MOPSO algorithm suffers from problems such as being prone to getting trapped in local optima, slow convergence speed, and insufficient Pareto solution set diversity when solving multi-objective optimization problems. This implementation proposes three improvement strategies: Improvement Strategy 1: Adaptive Inertia Weights

[0148] Example, general settings , , ).when In the early stages, the weight decreases rapidly, which is conducive to rapid convergence; in the later stages, the weight decreases more slowly, which is conducive to refined search.

[0149] Improvement Strategy 2: Chaotic Mutation Operator: A chaotic sequence is generated using a Logistic mapping, and mutation operations are performed on particle positions to enhance global search capabilities.

[0150] in, Here is the chaos parameter (taken as 3.9). It is a chaotic variable.

[0151] When random number (probability of mutation) When ), perform chaotic mutation on the particle position: ,in The time is variable.

[0152] Improvement Strategy 3: TOPSIS-based Pareto Solution Set Selection: When multiple non-dominated solutions exist in the Pareto solution set, the TOPSIS method is used to select the final compromise solution.

[0153] in, To solve Distance to the ideal solution To solve Distance to the negative ideal solution For relative similarity, select The largest solution is used as the final scheduling scheme.

[0154] For example, IMOPSO algorithm parameter settings: number of particles Maximum number of iterations External archive size Crossover probability Probability of mutation .

[0155] In one possible implementation, the acquisition of power grid status data, green electricity output prediction data, and electricity consumption prediction data of each user, and the solution of a preset multi-objective optimization model based on the load characteristic mining results of each user to obtain the electricity consumption plan, green electricity allocation scheme, and power grid purchase plan of each user, includes: adopting a rolling optimization strategy to solve the multi-objective optimization model once at preset intervals; and generating an emergency dispatch trigger command when any of the following triggering conditions are detected: the actual green electricity output deviates from the prediction by more than 20%, the user load change exceeds 30%, and the power grid issues an emergency dispatch command.

[0156] Specifically, conventional scheduling employs a rolling optimization strategy, each The optimization model is resolved every few minutes to update the scheduling plan for the next 24 hours. During the update, the latest green energy output forecast and the actual electricity consumption data of each user are used as inputs to correct for accumulated prediction errors.

[0157] Emergency dispatch mechanism: Emergency dispatch will be activated when any of the following triggering conditions are detected: 1. The actual output of green electricity deviates from the forecast by more than 20%; 2. User load changes by more than 30% during sudden changes: 3. The power grid issues an emergency dispatch order. Under the emergency dispatch mode, a simplified optimization model is adopted (considering only power balance constraints and minimum electricity demand constraints of each user) to generate an emergency dispatch plan within 5 minutes, prioritizing the supply of rigid loads of each user, and then maximizing the consumption of green electricity.

[0158] In one possible implementation, the comprehensive fairness score is a weighted sum of historical fairness, real-time demand fairness, adjustment contribution fairness, and cost fairness; wherein, historical fairness is: ,

[0159] in, For users Historical fairness, For users in the past The actual green electricity generated within a day For users The green electricity they deserve For users Historical total electricity consumption Total historical electricity consumption for all users. This will absorb the total amount of green electricity in history.

[0160] Specifically, when When, it indicates the user The green electricity obtained is exactly equal to the green electricity it is entitled to according to its electricity consumption ratio; Indicates user Those who received less green electricity in historical allocations should be compensated in subsequent allocations; Indicates user If a large proportion of green electricity has been allocated in the past, the proportion should be appropriately reduced in subsequent allocations.

[0161] Real-time demand fairness is:

[0162] in, For users Real-time demand fairness, For users At any moment rigid load power, For users At any moment Total power consumption. Specifically, The larger the value, the higher the proportion of rigid load for users, and the more urgent their demand for green electricity, which should be prioritized.

[0163] Adjusting the fairness of contributions is as follows:

[0164]

[0165] in, For users Adjusting the fairness of contributions, For users The regulating energy provided during the statistical period For users Total electricity consumption within the statistical period For the scheduling period, For users time Transfer power of transferable load, For users time Power reduction that can reduce load For users time Adjustable load operating power, For users time Reference power, For the time step. Specifically, The larger the value, the more likely the user is to be affected. The greater the contribution to system regulation, the more green electricity quotas should be awarded as an incentive.

[0166] Cost fairness is:

[0167]

[0168]

[0169] in, For users Cost fairness For users Total electricity cost for Real-time electricity purchase price For users exist Power purchased by the power grid at any given time For green electricity prices, For users exist The green electricity power obtained at all times The average electricity cost for all users; For the number of users.

[0170] Example of calculating the overall fairness score: ,in, and For the weighting coefficients, satisfying The default value is , , , Among these, real-time demand and adjustment contribution have higher weights because they directly reflect the fairness of current scheduling; historical fairness has a moderate weight and is used for long-term compensation mechanisms; cost fairness has a lower weight because cost differences partly stem from users' own electricity consumption patterns. The weighting coefficients can be adjusted according to the actual situation of the project using the Analytic Hierarchy Process (AHP) or expert scoring.

[0171] In one possible implementation, the method further includes: determining the green electricity allocation weight for each user based on their overall fairness score.

[0172] in, For users The overall fairness score, For the number of users, For users Green electricity allocation weight; For users The overall fairness score.

[0173] The user is obtained through the following formula. time Fair green electricity allocation ratio And perform normalization:

[0174]

[0175]

[0176] in, For users Lower limit for fair green electricity allocation ratio For users Daily average green electricity allocation weight For users Upper limit on the proportion of fair green electricity allocation.

[0177] According to user time Fair green electricity allocation ratio Revise the green electricity allocation plan for each user.

[0178] Specifically, the green electricity allocation weight is determined proportionally based on each user's comprehensive fairness score, and upper and lower limits are added to the green electricity allocation ratio based on the weight. Ensure that any user receives at least 50% or 5% of their average share (whichever is greater). Ensure that any user receives at most 200% or 50% of their average allocation (whichever is smaller). Normalization is performed:

[0179] in, For normalization .

[0180] In one possible implementation, the method further includes obtaining the fairness deviation for each user using the following formula:

[0181]

[0182] in, For users time Fairness bias, For users time The overall fairness score, For all users at any time The average score of overall fairness, For the number of users.

[0183] when When this happens, a Level 1 warning is issued, triggering adjustments to the green electricity allocation plan, which are then corrected in the next scheduling cycle; when When this happens, a level-two early warning is issued, triggering adjustments to the green electricity allocation plan and correcting it within two scheduling cycles; when users Daily cumulative fairness deviation At that time, a three-level early warning will be issued, and the compensation will be included in the next day's dispatch plan.

[0184] Specifically, real-time monitoring and early warning of fairness are implemented: a Level 1 warning immediately triggers an adjustment to the allocation ratio, which is corrected within the next scheduling cycle (15 minutes); a Level 2 warning is corrected within two scheduling cycles; and a Level 3 warning is included in the next day's scheduling plan compensation. By setting fairness deviation indicators and a three-level early warning mechanism, real-time monitoring and tiered response to the fairness of green electricity allocation among multiple users are achieved. This ensures that fairness deviations can be detected and corrected in a timely manner, preventing any user from being in an unfair situation for a long period of time, and enhancing the system's fairness assurance capabilities.

[0185] In one possible implementation, the method further includes: receiving a dispatch instruction issued by the power grid; and when the dispatch instruction is a peak-shaving instruction, generating a peak-shaving response strategy.

[0186]

[0187]

[0188] in, The total system response power, For users time Adjustable power, For a moment The target response power, For users User-level adjustable capacity, This represents the total adjustable capacity of the system. For users time Adjustable power for transferable loads, For users time It can reduce the regulating power of the load. For users time Adjustable power for adjustable loads.

[0189] When the scheduling instruction is a frequency modulation instruction, a frequency modulation response strategy is generated:

[0190] in, The frequency regulation target power issued by the power grid, For users The response time constant, For users The response time constant, For the number of users.

[0191] Specifically, it interacts with the power grid dispatch center through the IEC61850 communication protocol and receives the following signals: Peak shaving command: including peak shaving direction (upward / downward), peak shaving capacity (MW), and duration (min); Frequency regulation command: including frequency regulation direction, regulating power (MW), and response time requirement (s); Standby command: including standby capacity (MW) and standby type (cold standby / hot standby / spinning standby).

[0192] Upon receiving a peak shaving command, a response strategy is generated within 5 minutes; upon receiving a frequency shaving command, a response strategy is generated within 30 seconds. Frequency shaving responses prioritize the use of load resources with faster response times.

[0193] For example, the emergency response process includes the following steps: Step 1: 0s: Emergency triggering conditions are detected; Step 2: 5s: GCC starts the emergency dispatch module and freezes the current dispatch plan; Step 3: 10s: Assess the current system status and calculate available adjustment resources; Step 4: 30s: Generate an emergency dispatch plan (simplified model: only considers power balance + rigid load demand constraints); Step 5: 60s: Issue emergency dispatch instructions to each user-side control terminal; Step 6: 120s: Each user executes the emergency dispatch plan to restore system power balance; Step 7: 300s: The system returns to stable operation, exits emergency mode, and resumes normal rolling optimization dispatch.

[0194] In another embodiment of the present invention, a green industrial park is used as an example for illustration.

[0195] Park Overview: This park is one of the first batch of national green and low-carbon demonstration parks, covering an area of ​​5 square kilometers and housing 12 enterprises, including 4 industrial users participating in direct green electricity supply.

[0196] System Configuration: Green Energy Project: 50MW photovoltaic power station (monocrystalline silicon modules, conversion efficiency 22%, annual equivalent utilization hours 1200h) + 20MW wind farm (direct-drive permanent magnet units, annual equivalent utilization hours 2100h), totaling 70MW of green energy installed capacity. Industrial Users: User A - Steel rolling enterprise (peak load 15MW, annual electricity consumption 95 million kWh), User B - Chemical enterprise (peak load 12MW, annual electricity consumption 78 million kWh), User C - Machinery manufacturing enterprise (peak load 8MW, annual electricity consumption 42 million kWh), User D - Electronic component enterprise (peak load 5MW, annual electricity consumption 28 million kWh). Dedicated Line Configuration: 10kV dedicated lines from the green energy project to each user, with a total length of approximately 15km, and the transmission capacity meets the maximum load requirements of each user. GCC system: Deployed in the park's dispatch center, server cluster (2×8 core CPU / 64GB memory / 2TB SSD, dual-machine hot standby), monitoring screen (4×255-inch LED splicing screen).

[0197] Load characteristic mining results: User A (Steel): Rigid load accounts for 60% (electric furnace base load + lighting + control), transferable load accounts for 15% (electric furnace heating can be shifted to nighttime), load that can be reduced accounts for 10% (auxiliary lighting + air conditioning), and adjustable load accounts for 15% (rolling mill can reduce power by 5%). =5.25MW. User B (Chemical): Rigid load accounts for 55% (reactor + pump station + control), transferable load accounts for 20% (heating process can be staggered), load that can be reduced accounts for 10% (cooling system can reduce power for a short time), and adjustable load accounts for 15% (variable frequency pump can be adjusted ±20%). =4.2MW. User C (Mechanical): Rigid load accounts for 50% (core processing equipment), transferable load accounts for 15% (heat treatment can be staggered), load that can be reduced accounts for 20% (lighting + air conditioning + ventilation), and adjustable load accounts for 15% (CNC machine tools can reduce power). =2.8MW. User D (Electronics): Rigid load accounts for 80% (precision equipment cannot be powered off), transferable load accounts for 5% (some testing procedures), reduceable load accounts for 5% (lighting + air conditioning), and adjustable load accounts for 10% (cooling system adjustable ±10%). =0.75MW. System level =13.0MW, which meets the requirements. The requirement is ≥0.15×70MW=10.5MW.

[0198] System performance (6-month trial operation data): Green electricity consumption rate: 96.2% daily, an increase of 18.5 percentage points compared to the original independent power supply model, resulting in an annual increase of approximately 45 million kWh of green electricity consumption. User electricity costs: The average electricity cost for the four users decreased by 12.3%, with User D experiencing a cost reduction of 18.2% (due to its largest contribution to regulation but also its highest proportion of rigid load, thus receiving more green electricity compensation), saving approximately 3.8 million yuan in electricity costs annually. Fairness index: The daily average F_fair value is 0.93, with fairness deviation among users <5%, 100% user satisfaction, and zero complaints. Grid interaction: An average of 15 grid peak-shaving commands are responded to per month, with an average frequency regulation response time of 25 seconds, generating approximately 2.5 million yuan in ancillary service revenue annually. Emergency response: In one instance, a sudden equipment failure at User B caused a 35% load drop (-4.2MW). The system redistributed the load of other users within 45 seconds, ensuring full green electricity consumption and preventing approximately 1200 kWh of wind and solar curtailment.

[0199] Expansion verification was conducted. Two new industrial users were added: User E - a forging enterprise (peak load 3MW) and User F - a rolling enterprise (peak load 4MW).

[0200] System Expansion Process: T+0 min: Smart meters and edge computing nodes for new users E and F are online. T+15 min: GCC automatically identifies new users and initiates load characteristic mining (requires 7 days of historical data accumulation). T+7 days: Load characteristic mining for users E and F is completed, and they are automatically incorporated into the collaborative scheduling model. T+7 days + 15 min: The 6-user collaborative scheduling model runs for the first time, generating a scheduling plan.

[0201] The total expansion time is approximately 7 days (mainly due to waiting for historical data accumulation), with technical deployment time of less than 2 hours. After the expansion, the system's green energy consumption rate remains above 94%, and the fairness index is 0.91.

[0202] The following are embodiments of the apparatus of the present invention, which can be used to execute embodiments of the method of the present invention. For details not disclosed in the apparatus embodiments, please refer to the embodiments of the method of the present invention.

[0203] See Figure 2 In another embodiment of the present invention, a green energy collaborative scheduling system for a single-source, multi-load scenario is provided, which can be used to implement the above-mentioned green energy collaborative scheduling method for a single-source, multi-load scenario. Specifically, the green energy collaborative scheduling system for a single-source, multi-load scenario includes a feature mining module and an optimization module.

[0204] The feature mining module acquires historical electricity consumption data from each user and performs load feature mining to obtain load characteristic mining results for each user. These results include load allocation results, adjustable load characteristic parameter models, user-level adjustable capacity, and user-level adjustment cost functions. The optimization module acquires grid status data, green electricity output prediction data, and electricity consumption prediction data from each user. It then combines these with the load characteristic mining results to solve a pre-defined multi-objective optimization model, resulting in each user's electricity consumption plan, green electricity allocation scheme, and grid power purchase plan. The multi-objective optimization model aims to maximize green electricity absorption rate, minimize total system operating cost, and maximize power supply fairness. Maximizing power supply fairness involves maximizing the average comprehensive fairness score of each user. This comprehensive fairness score is determined based on each user's historical green electricity allocation scheme and the load characteristic mining results, using a pre-defined multi-dimensional fairness index system. This multi-dimensional fairness index system includes historical fairness, real-time demand fairness, adjustment contribution fairness, and cost fairness.

[0205] All relevant content of each step involved in the aforementioned embodiments of the green electricity collaborative scheduling method for a single source and multiple load scenarios can be referenced to the functional description of the corresponding functional module of the green electricity collaborative scheduling system for a single source and multiple load scenarios in the embodiments of the present invention, and will not be repeated here.

[0206] The module division in this embodiment of the invention is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional modules in the various embodiments of the invention can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0207] The present invention provides a green energy collaborative scheduling method and system for single-source, multi-load scenarios, which has the following beneficial effects: (1) Filling the technological gap in the single-source multi-load scenario. For the first time, a green power control center architecture for the single-source multi-load scenario is proposed, which helps to build a dedicated control system integrating software and hardware, filling the technological gap in this field.

[0208] (2) Achieve intelligent and precise matching between green electricity and multi-user loads, significantly improving the green electricity consumption rate. Establish a multi-user load characteristic mining model and a collaborative scheduling optimization model, which can accurately identify the load regulation characteristics of each user and achieve dynamic matching between green electricity output and multi-user loads. Through simulation calculation and pilot verification, the green electricity consumption rate can reach more than 95%, which is 15-20 percentage points higher than the traditional proportional allocation method. It can increase the green electricity consumption of 100MW-level green electricity projects by about 30 million kWh per year, which is equivalent to reducing standard coal consumption by about 9,000 tons and CO2 emissions by about 24,000 tons.

[0209] (3) Innovative multi-dimensional fairness indicators and allocation mechanisms were proposed to effectively ensure the fairness of green electricity allocation among multiple users. A four-dimensional fairness indicator system (historical fairness, real-time demand fairness, adjustment contribution fairness, and cost fairness) and a comprehensive fairness allocation mechanism were designed to comprehensively consider multiple factors and effectively ensure the fairness, transparency, and traceability of green electricity allocation among multiple users. The fairness indicator was stable at above 0.9 (1.0 is completely fair), user satisfaction was significantly improved, and the user complaint rate was zero.

[0210] (4) Reduce electricity costs for users and improve economic efficiency. Through green electricity-multi-user collaborative optimization scheduling, the green electricity allocation scheme is optimized, and the electricity costs for each user are significantly reduced. According to calculations, the average electricity cost for users is reduced by 10% to 15%, with users who make greater contributions to regulation experiencing more significant cost reductions, up to 18%. Taking a 100MW green electricity project and 4 industrial users as an example, the annual electricity cost savings are approximately RMB 5 million to 8 million.

[0211] (5) Supporting the safe and stable operation of the power grid and providing flexible adjustment resources. It can respond to the needs of ancillary services such as peak shaving, frequency regulation, and reserve, and provide flexible adjustment resources. The peak shaving response time is ≤5 minutes, and the frequency regulation response time is ≤30 seconds, effectively supporting the safe and stable operation of the power grid. Calculations show that it can provide adjustment capacity equivalent to 10-15% of the green electricity installed capacity to participate in power grid ancillary services, generating approximately 2-3 million yuan in ancillary service revenue annually.

[0212] (6) It has good scalability and adaptability, and is suitable for various scenarios. The system architecture is flexible, adopts modular design and standardized interfaces, and can be adapted to various scenarios with 2 to 20 industrial users and green power scales of 1MW to 100MW. Different scale projects can be adapted through parameter configuration and model optimization without redesigning the system architecture. The system expansion time is <2 hours and the expansion cost is <50,000 yuan.

[0213] In another embodiment of the present invention, a computer device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to achieve corresponding method flows or corresponding functions. The processor described in this embodiment of the present invention can be used for the operation of a green electricity collaborative scheduling method for a single-source, multi-load scenario.

[0214] In another embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, the storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the green electricity collaborative scheduling method for a single-source, multi-load scenario in the above embodiments.

[0215] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0216] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0217] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0218] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0219] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A green energy collaborative scheduling method for a single-source, multi-load scenario, characterized in that, include: Historical electricity consumption data of each user is acquired and load characteristic mining is performed to obtain load characteristic mining results for each user; wherein, the load characteristic mining results include load division results, adjustable load characteristic parameter model, user-level adjustable capacity, and user-level adjustment cost function; The system acquires grid status data, green electricity output forecast data, and electricity consumption forecast data for each user. It then combines the load characteristic mining results of each user to solve a pre-set multi-objective optimization model, thereby obtaining each user's electricity consumption plan, green electricity allocation scheme, and grid power purchase plan. The multi-objective optimization model aims to maximize the green electricity absorption rate, minimize the total system operating cost, and maximize power supply fairness. Among them, maximizing power supply fairness means maximizing the average comprehensive fairness score of each user. The comprehensive fairness score of each user is determined by a preset multi-dimensional fairness index system based on each user's historical green electricity allocation scheme and the results of load characteristic mining of each user. The multi-dimensional fairness index system includes historical fairness, real-time demand fairness, regulation contribution fairness, and cost fairness.

2. The green electricity collaborative scheduling method for a single-source, multi-load scenario as described in claim 1, characterized in that, The historical electricity consumption data includes historical active power, historical reactive power, historical voltage, historical current, historical power factor, and sampling interval; the power grid status data includes power grid frequency, node voltage, and sampling interval; the green power output prediction data includes predicted photovoltaic power output, predicted wind power output, and sampling interval; the electricity consumption prediction data includes predicted active power, predicted reactive power, predicted voltage, predicted current, predicted power factor, and sampling interval; before performing load feature mining, the process also includes: outlier detection, missing value imputation, time alignment, and data aggregation of the historical electricity consumption data of each user.

3. The green electricity collaborative scheduling method for a single-source, multi-load scenario as described in claim 1, characterized in that, The process of mining load features includes: For each user, feature vectors are extracted based on historical electricity consumption data, and the extracted feature vectors are clustered into four categories, which are then mapped to rigid loads, transferable loads, loads that can be reduced, and adjustable loads, respectively. The characteristic parameters, cost functions, and constraints of transferable load, reduceable load, and adjustable load are constructed to obtain the adjustable load characteristic parameter model; and user-level aggregation is performed based on the adjustable load characteristic parameter model to obtain the user-level adjustable capacity and user-level adjustment cost function.

4. The green electricity collaborative scheduling method for a single-source, multi-load scenario as described in claim 3, characterized in that, The feature vector includes daily load factor, daily peak-to-valley difference rate, load duration rate, nighttime load percentage, fluctuation coefficient, and load time autocorrelation coefficient; the cluster analysis of the extracted feature vector includes K-means cluster analysis of the extracted feature vector. The process of constructing the characteristic parameters, cost functions, and constraints of transferable, reduceable, and adjustable loads to obtain the adjustable load characteristic parameter model includes: For users Transferable load Characteristic parameters include: transferable start time Transferable end time Maximum transfer power , To transfer electrical energy; the cost function is in quadratic form; the constraints are: in, For users time Transfer power of transferable load; For users Reduced load Characteristic parameters include: maximum power reduction Single continuous reduction of the time limit Maximum number of daily reductions and minimum recovery interval The cost function is in quadratic form; the constraints are: in, For counting days; For users No. The number of days reduced; For users The time interval between two consecutive cuts; For users time Power reduction that can reduce load; For users Adjustable load Characteristic parameters include: minimum regulating power Maximum adjustable power and response time constant The cost function is in quadratic form; the constraints are: in, For users time Adjustable load operating power, For users Adjustable load The maximum regulation rate; For users Adjustable load operating power; Adjustable loads also include dynamic response characteristics: in, To adjust the command power; The step of performing user-level aggregation based on the adjustable load characteristic parameter model to obtain the user-level adjustable capacity and user-level adjustment cost functions includes: user User-level adjustable capacity for: user User-level adjustment cost function for: in, The cost function of transferable loads, The cost function for load reduction, This is the cost function for adjustable loads.

5. The green electricity collaborative scheduling method for a single-source, multi-load scenario according to claim 1, characterized in that, The acquisition of green electricity output forecast data and electricity consumption forecast data for each user includes: A green energy output prediction model based on LSTM-Attention is used to obtain green energy output prediction data, and an electricity consumption prediction model based on LSTM-Attention is used to obtain electricity consumption prediction data for each user. The input features of the green power output prediction model include numerical weather forecasts, historical power output data, and daily type identifiers; the input features of the electricity consumption data prediction model include historical electricity consumption data and production scheduling information.

6. The green electricity collaborative scheduling method for a single-source, multi-load scenario according to claim 1, characterized in that, The objective function of the preset multi-objective optimization model is: in, To maximize the green electricity consumption rate, To minimize the total operating cost of the system, To maximize fairness in power supply, For the scheduling period, for The actual amount of green electricity consumed at any given time. for Green electricity is always available. For time step, for Real-time electricity purchase price for Power purchase capacity of the power grid at any time For users The user-level adjustment cost function, For users User-level adjustable capacity, The price for penalties for power curtailment for Power curtailment at any given time For the number of users, For users The overall fairness score; The pre-defined multi-objective optimization model's constraints include power balance constraints, green electricity output constraints, electricity demand constraints for each user, green electricity allocation constraints for each user, load regulation constraints, grid power purchase constraints, and fair allocation ratio constraints; among which, the green electricity allocation constraints for each user are: The fairness distribution ratio constraint is: in, For users time The allocated green electricity power, For users time The proportion of green electricity allocation For a moment The total green power actually absorbed by the system For users The lower limit of the green electricity allocation ratio For users The upper limit of green electricity allocation ratio, For users At any moment The proportion of green electricity allocation This represents the maximum permissible variation in the green electricity allocation ratio between adjacent time points.

7. The green electricity collaborative scheduling method for a single-source, multi-load scenario according to claim 1, characterized in that, The solution to the preset multi-objective optimization model includes: An improved multi-objective particle swarm optimization algorithm is used to solve the pre-defined multi-objective optimization model. When multiple non-dominated solutions exist in the Pareto solution set, the TOPSIS method is used to select the final solution, thereby obtaining the electricity consumption plan, green electricity allocation scheme, and grid power purchase plan for each user. The improved multi-objective particle swarm optimization algorithm is obtained by making the following improvements to the existing multi-objective particle swarm optimization algorithm: The inertia weights in the multi-objective particle swarm optimization algorithm are improved to adaptive inertia weights: in, For the first Inertia weights in the next iteration This represents the maximum value of the inertia weight. This represents the minimum value of the inertia weight. The maximum number of iterations, It is a non-linear adjustment factor; The mutation strategy of the multi-objective particle swarm optimization algorithm is improved by using a Logistic mapping to generate a chaotic sequence to mutate the particle positions.

8. The green electricity collaborative scheduling method for a single-source, multi-load scenario according to claim 1, characterized in that, The process of acquiring grid status data, green electricity output forecast data, and electricity consumption forecast data for each user, and combining this with the load characteristic mining results of each user to solve a preset multi-objective optimization model, yields each user's electricity consumption plan, green electricity allocation scheme, and grid power purchase plan, including: A rolling optimization strategy is adopted, and the multi-objective optimization model is solved once at a preset time interval; and an emergency dispatch trigger command is generated when any of the following trigger conditions are detected: the actual output of green electricity deviates from the prediction by more than 20%, the user load changes by more than 30%, and the power grid issues an emergency dispatch command.

9. The green electricity collaborative scheduling method for a single-source, multi-load scenario according to claim 1, characterized in that, The overall fairness score is a weighted average of historical fairness, real-time demand fairness, adjustment contribution fairness, and cost fairness; where historical fairness is: ， in, For users Historical fairness, For users in the past The actual green electricity generated within a day For users The green electricity they deserve For users Historical total electricity consumption Total historical electricity consumption for all users. This represents the total green energy absorbed in history. Real-time demand fairness is: in, For users Real-time demand fairness, For users At any moment rigid load power, For users At any moment Total power consumption; Adjusting the fairness of contributions is as follows: in, For users Adjusting the fairness of contributions, For users The regulating energy provided during the statistical period For users Total electricity consumption within the statistical period For the scheduling period, For users time Transfer power of transferable load, For users time Power reduction that can reduce load For users time Adjustable load operating power, For users time Reference power, For time step; Cost fairness is: in, For users Cost fairness For users Total electricity cost for Real-time electricity purchase price For users exist Power purchased by the power grid at any given time For green electricity prices, For users exist The green electricity power obtained at all times The average electricity cost for all users; For the number of users.

10. The green electricity collaborative scheduling method for a single-source, multi-load scenario according to claim 1, characterized in that, Also includes: The green electricity allocation weight for each user is determined based on their overall fairness score: in, For users The overall fairness score, For the number of users, For users Green electricity allocation weight; For users The overall fairness score; The user is obtained through the following formula. time Fair green electricity allocation ratio And perform normalization: in, For users Lower limit for fair green electricity allocation ratio For users Daily average green electricity allocation weight For users Upper limit on the proportion of fair green electricity allocation; According to user time Fair green electricity allocation ratio Revise the green electricity allocation plan for each user.

11. The green electricity collaborative scheduling method for a single-source, multi-load scenario according to claim 1, characterized in that, Also includes: The fairness deviation for each user is obtained using the following formula: in, For users time Fairness bias, For users time The overall fairness score, For all users at any time The average score of overall fairness, For the number of users; when When this happens, a Level 1 warning is issued, triggering adjustments to the green electricity allocation plan, which are then corrected in the next scheduling cycle; when When this happens, a level-two early warning is issued, triggering adjustments to the green electricity allocation plan and correcting it within two scheduling cycles; when users Daily cumulative fairness deviation At that time, a three-level early warning will be issued, and the compensation will be included in the next day's dispatch plan.

12. The green electricity collaborative scheduling method for a single-source, multi-load scenario according to claim 1, characterized in that, Also includes: Receive dispatch instructions issued by the power grid; When the scheduling instruction is a peak shaving instruction, a peak shaving response strategy is generated: in, The total system response power, For users time Adjustable power, For a moment The target response power, For users User-level adjustable capacity, This represents the total adjustable capacity of the system. For users time Adjustable power for transferable loads, For users time It can reduce the regulating power of the load. For users time Adjustable power for adjustable loads; When the scheduling instruction is a frequency modulation instruction, a frequency modulation response strategy is generated: in, The frequency regulation target power issued by the power grid, For users The response time constant, For users The response time constant, For the number of users.

13. A green electricity collaborative scheduling system for a single-source, multi-load scenario, characterized in that, include: The feature mining module is used to acquire the electricity consumption history data of each user and perform load feature mining to obtain the load characteristic mining results of each user; wherein, the load characteristic mining results include load division results, adjustable load characteristic parameter models, user-level adjustable capacity, and user-level adjustment cost functions. The optimization module is used to acquire power grid status data, green electricity output prediction data, and electricity consumption prediction data of each user, and solve the preset multi-objective optimization model by combining the load characteristic mining results of each user to obtain the electricity consumption plan, green electricity allocation scheme, and power grid purchase plan of each user; wherein, the multi-objective optimization model takes maximizing the green electricity absorption rate, minimizing the total system operating cost, and maximizing the fairness of power supply as optimization objectives; Among them, maximizing power supply fairness means maximizing the average comprehensive fairness score of each user. The comprehensive fairness score of each user is determined by a preset multi-dimensional fairness index system based on each user's historical green electricity allocation scheme and the results of load characteristic mining of each user. The multi-dimensional fairness index system includes historical fairness, real-time demand fairness, regulation contribution fairness, and cost fairness.

14. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the green electricity collaborative scheduling method for a single-source, multi-load scenario as described in any one of claims 1 to 12.

15. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the green electricity collaborative scheduling method for a single-source, multi-load scenario as described in any one of claims 1 to 12.

Images