An industrial load management method, system, device and medium

Abstract

The present application belongs to the field of power system demand response and carbon emission monitoring, and discloses an industrial load management method, system, device and medium, which constructs a power-carbon coupling topological model of transformer substation-feeder-pilot area-user level, generates a user dynamic carbon potential image in real time, then establishes an industrial load flexible adjustment potential evaluation model based on carbon efficiency ratio to divide user priority, executes the control instruction in cooperation relying on the local layer-station layer-main station layer three-layer control architecture, and finally collects actual carbon reduction data for iterative optimization by reinforcement learning. The present application improves the carbon perception accuracy and control pertinence of load management, stimulates the internal motivation of users to actively participate in green power consumption and demand response, and realizes the precise quantitative evaluation of carbon reduction effect, thereby effectively supporting the green, efficient and safe operation of the new power system.

Description

Technical Field

[0001] This invention belongs to the field of power system demand response and carbon emission monitoring technology, and particularly relates to an industrial load management method, system, equipment and medium. Background Technology

[0002] Currently, the volatility and uncertainty of new energy power generation pose significant challenges to the safe and stable operation of the power grid, while also placing higher demands on the load-side's flexible adjustment capabilities. Demand response, as an important adjustment tool, guides users to change their electricity consumption behavior, thereby achieving peak shaving and valley filling, and ensuring a balance between power supply and demand.

[0003] However, existing load management technologies still have many shortcomings and are difficult to adapt to the development needs of new power systems: First, the control methods are relatively crude and lack a carbon perspective. Traditional orderly electricity use and demand response mainly focus on reducing load volume, ignoring the differences in carbon emission intensity of electricity from different time periods and sources. This may result in the reduction of load for absorbing clean energy such as distributed photovoltaic power, reducing emission reduction efficiency, and failing to accurately identify high-carbon emission loads, making it difficult to quantify the carbon reduction effect of control measures. Second, the carbon responsibility of users' electricity consumption is not clearly defined. Users cannot perceive the carbon emissions corresponding to their own electricity consumption behavior in real time, lacking the intrinsic motivation to actively participate in green electricity consumption or demand response. At the same time, existing incentive policies are mostly based on load volume and fail to reflect the "carbon" value, making it difficult to fully tap the adjustment potential on the user side. Third, the evaluation of control effects lacks precise quantitative means. When formulating control strategies, there is a lack of evaluation models that can integrate user load characteristics, electricity consumption behavior, and dynamic carbon emission factors, making it impossible to accurately profile and classify users' carbon adjustment potential. Consequently, after the control instructions are issued, the actual carbon reduction effect is difficult to effectively quantify and evaluate.

[0004] It is evident that existing load management technologies, due to a lack of an electricity-carbon integration perspective, insufficient user participation, and inaccurate control and assessment, cannot meet the needs of green, efficient, and safe operation of the new power system. Summary of the Invention

[0005] This invention provides an industrial load management method, system, equipment, and medium. This method can effectively solve the problems of existing load management technologies, such as lack of an electricity-carbon integration perspective, insufficient user participation, and inaccurate control and evaluation, and can meet the needs of green, efficient, and safe operation of new power systems.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: An industrial load control method includes: Based on the collected power grid topology, source carbon emission factors, and fine-grained user load data, a substation-feeder-transformer area-user level electric carbon coupling topology model is constructed. Based on the electric carbon coupling topology model, a dynamic carbon potential profile is generated for each user in real time. The dynamic carbon potential profile includes load size and real-time carbon emission factors. Based on dynamic carbon potential profiles and historical load data of users, a model for assessing the flexible adjustment potential of industrial load is established by introducing carbon efficiency ratio. The model is used to match users with corresponding priorities. The priorities are divided into high-carbon and low-efficiency type, high-carbon and high-efficiency type, and low-carbon and flexible type. A three-layer control architecture is constructed, consisting of a local layer, a station layer, and a main station layer. The main station layer is used to generate dynamic carbon quota curves by region and industry. The station layer is used to aggregate and sort users under its jurisdiction based on user priority to output adjustment instructions. The local layer is used to execute flexible industrial load adjustment operations according to the received adjustment instructions. We collect actual carbon reduction data after flexible adjustment of industrial load, and iteratively optimize the industrial load flexible adjustment potential assessment model and dynamic carbon quota curve based on reinforcement learning algorithm combined with actual carbon reduction data.

[0007] Furthermore, the method of generating a dynamic carbon potential profile for each user in real time based on the electro-carbon coupling topology model includes: Based on the access transformer identifier in the user profile, match the user's upstream feeder node in the power grid topology database and subscribe to the real-time carbon emission factor data stream of this node, and time-align the user's time-sharing load data with the corresponding upstream feeder carbon emission factor. Based on the mapping relationship between topological nodes and carbon emission factors established by the electric carbon coupling topology model, the instantaneous carbon emissions and dynamic carbon potential of users are calculated by combining the total user load and the real-time carbon emission factors of the upstream feeder, and the user carbon footprint curve is generated. By integrating user static attribute information with user carbon footprint curves, a complete dynamic carbon potential profile is constructed. In the user carbon footprint curve, the total load of the user at time t is... The instantaneous carbon emissions (Emission(t)) and dynamic carbon potential corresponding to the total load for:

[0008]

[0009] in, The real-time carbon emission factor of the upstream feeder at time t; For time intervals.

[0010] Furthermore, based on dynamic carbon potential profiles and historical user load data, a carbon efficiency ratio is introduced to establish an industrial load flexible adjustment potential assessment model. This model is used to match users with corresponding priorities, including: Based on dynamic carbon potential profiles and historical user load data, a model for assessing the flexible adjustment potential of industrial load is established by introducing the carbon efficiency ratio. The carbon efficiency ratio is the ratio of the economic added value generated per unit of user load to the carbon emission intensity of their electricity consumption, calculated using the following formula:

[0011] In the formula, The economic value added per unit of time for the user. For average electrical load, This represents the average carbon emission factor for the corresponding time period; Carbon efficiency ratio; Based on historical load data of users, the time-shiftable characteristics, interruptible characteristics and power-adjustable characteristics of user load are analyzed, and key parameters are extracted to output the results of the adjustability analysis; the key parameters include adjustable load capacity, adjustment response time and maximum continuous adjustment time. Cluster analysis was performed on the results of carbon efficiency ratio and adjustability analysis to prioritize users into high-carbon low-efficiency, high-carbon high-efficiency, and low-carbon flexible types.

[0012] Furthermore, the high-carbon low-efficiency type, the high-carbon high-efficiency type, and the low-carbon flexible type are specifically classified as follows: High-carbon, low-efficiency type: has the highest adjustment priority, low carbon efficiency ratio, and a large proportion of interruptible / transferable loads; High-carbon and high-efficiency type: with secondary adjustment priority, high carbon efficiency ratio and rigid production, and strong incentives to guide corresponding users to actively avoid peaks during high-carbon periods of the power grid; Low-carbon flexible type: encourages priority use of electricity, with high carbon efficiency and a large proportion of flexible loads, and encourages corresponding users to use more electricity during low-carbon periods of the power grid.

[0013] Furthermore, the construction of the three-layer control architecture—local layer, station layer, and main station layer—includes: A three-tier control architecture is established: local layer - station layer - main station layer. The specific deployment is as follows: The main station layer is deployed in the upper load control room, integrating the electricity-carbon joint dispatching cockpit, and is responsible for generating dynamic carbon quota curves based on the power grid operation status and carbon potential prediction, and then sending them down to the station layer. The station layer is deployed at the edge computing nodes of power supply stations or substations. It is responsible for receiving dynamic carbon quota curves, completing user aggregation sorting and optimal adjustment combination calculations, and outputting adjustment instructions. The local layer is deployed on the user-side edge computing terminal and is responsible for real-time monitoring of user load and carbon potential, receiving and executing adjustment commands, and completing flexible adjustment operations of industrial load.

[0014] Furthermore, the local layer performs flexible industrial load adjustment operations in two modes: Invitation Mode: Targeting high-carbon and high-efficiency users, carbon reduction invitations and incentive programs are pushed out, and users can choose whether to take the adjustment action themselves; Direct control mode: Designed for users with high carbon emissions and low efficiency, it enables automatic closed-loop control of pre-authorized flexible loads, achieving adjustment response at the second or minute level.

[0015] Furthermore, the actual carbon reduction data collected after flexible adjustment of industrial load is used to iteratively optimize the industrial load flexible adjustment potential assessment model and dynamic carbon quota curve based on reinforcement learning algorithms and the actual carbon reduction data, including: By comparing the actual load curves of users with the carbon potential curves before and after flexible industrial load adjustment, the actual carbon reduction data is calculated; the specific calculation formula is as follows:

[0016] In the formula, For the user's baseline load; This is the actual load. This represents the actual carbon potential. This is actual carbon reduction data; The actual carbon emission reduction and user production impact feedback information are used as inputs to the reinforcement learning algorithm to iteratively optimize the weight parameters and dynamic carbon quota curve of the industrial load flexible adjustment potential assessment model.

[0017] An industrial load management system, comprising: The carbon potential profile generation module is used to construct a substation-feeder-transformer-user level electric carbon coupling topology model based on the collected power grid topology, source carbon emission factors and fine-grained user load data, and to generate a dynamic carbon potential profile for each user in real time based on the electric carbon coupling topology model; wherein, the dynamic carbon potential profile includes load size and real-time carbon emission factors. The priority classification module is used to establish an industrial load flexibility adjustment potential assessment model based on dynamic carbon potential profiles and user historical load data, and to match users with corresponding priorities through the industrial load flexibility adjustment potential assessment model; wherein, the priorities are classified into high carbon low efficiency type, high carbon high efficiency type and low carbon flexible type. The architecture construction module is used to build a three-layer control architecture: local layer, station layer, and main station layer. The main station layer is used to generate dynamic carbon quota curves by region and industry. The station layer is used to aggregate and sort the users under its jurisdiction based on user priority to output adjustment instructions. The local layer is used to execute flexible adjustment operations of industrial load according to the received adjustment instructions. The optimization module is used to collect actual carbon reduction data after flexible adjustment of industrial load, and to iteratively optimize the industrial load flexible adjustment potential assessment model and dynamic carbon quota curve based on reinforcement learning algorithm combined with actual carbon reduction data.

[0018] An industrial load control device, comprising: Memory, used to store computer programs; A processor is used to implement the aforementioned industrial load control method when executing the computer program.

[0019] A computer-readable storage medium storing a computer program, which, when executed by a processor, is used to implement the aforementioned industrial load control method.

[0020] Compared with the prior art, the present invention has the following beneficial effects: This invention provides an industrial load management method. It constructs an electric-carbon coupling topology model at the substation-feeder-transformer area-user level to generate real-time dynamic carbon potential profiles for users. Then, based on carbon efficiency ratio, it establishes an industrial load flexible regulation potential assessment model to prioritize users. Finally, it relies on a three-layer control architecture—local layer, substation layer, and master station layer—to collaboratively execute regulation commands. Finally, it collects actual carbon reduction data and uses reinforcement learning for iterative optimization. In this method, the electric-carbon coupling model integrates grid topology and dynamic carbon emission factors to achieve accurate source tracing and responsibility identification of carbon emissions; the carbon efficiency ratio assessment quantifies users' carbon regulation potential, ensuring precise stratification of regulation targets; the three-layer architecture achieves closed-loop coordination from carbon quota generation to load regulation; and reinforcement learning continuously and adaptively optimizes the model and strategy through feedback data. This method improves the carbon perception accuracy and regulation targeting of load management, stimulates users' intrinsic motivation to actively participate in green electricity consumption and demand response, and achieves accurate quantitative assessment of carbon reduction effects, thereby effectively supporting the green, efficient, and safe operation of the new power system. Attached Figure Description

[0021] Figure 1 A flowchart illustrating the implementation of an industrial load control method according to an embodiment of the present invention; Figure 2 A framework diagram of an industrial load control system provided in an embodiment of the present invention; Figure 3 This is a core flowchart of an industrial load control method provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of an industrial load control system provided in an embodiment of the present invention. Detailed Implementation

[0022] To further understand the content of this invention, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments are merely illustrative and not limiting of the invention.

[0023] For example, this embodiment provides an industrial load management method. By introducing dynamic carbon potential and constructing a user "carbon efficiency ratio" model, it can achieve accurate identification and flexible adjustment of high carbon loads of industrial users, solving the problems of traditional load control methods such as extensive load management, lack of carbon perspective, low user participation, and difficulty in quantifying and evaluating carbon reduction effects.

[0024] For example, such as Figure 3 As shown, this embodiment provides an industrial load control method, including: Based on the collected power grid topology, source carbon emission factors, and fine-grained user load data, a substation-feeder-transformer area-user level electric carbon coupling topology model is constructed. Based on the electric carbon coupling topology model, a dynamic carbon potential profile is generated for each user in real time. The dynamic carbon potential profile includes load size and real-time carbon emission factors. Based on dynamic carbon potential profiles and historical load data of users, a model for assessing the flexible adjustment potential of industrial load is established by introducing carbon efficiency ratio. The model is used to match users with corresponding priorities. The priorities are divided into high-carbon and low-efficiency type, high-carbon and high-efficiency type, and low-carbon and flexible type. A three-layer control architecture is constructed, consisting of a local layer, a station layer, and a main station layer. The main station layer is used to generate dynamic carbon quota curves by region and industry. The station layer is used to aggregate and sort users under its jurisdiction based on user priority to output adjustment instructions. The local layer is used to execute flexible industrial load adjustment operations according to the received adjustment instructions. We collect actual carbon reduction data after flexible adjustment of industrial load, and iteratively optimize the industrial load flexible adjustment potential assessment model and dynamic carbon quota curve based on reinforcement learning algorithm combined with actual carbon reduction data.

[0025] The control method provided in this embodiment will be described in more detail below: For example, such as Figure 1 As shown, this embodiment provides an industrial load management method, including but not limited to the following steps: Step 1, Multi-source data acquisition and dynamic carbon potential profile construction: Collect grid topology, source carbon emission factors and large user load fine-grained data, construct an electric carbon coupling topology model of "substation-feeder-transformer area-user", and generate a "dynamic carbon potential profile" for each user in real time, including load size and real-time carbon emission factors. Step 2, User-level regulation potential assessment: Based on the user's dynamic carbon potential profile and historical load data, the concept of carbon efficiency ratio is introduced to establish a regulation potential assessment model, classifying users into high-carbon low-efficiency, high-carbon high-efficiency, and low-carbon flexible types. Step 3, Layered Coordination and Control Strategy Formulation: Construct a three-layer control architecture of local-station-master station. The master station layer generates dynamic carbon quota curves by region and industry. The station layer aggregates and sorts according to user priority. The edge terminal of the local layer receives signals and automatically executes or invites users to execute adjustment instructions. Step 4, Carbon Reduction Effect Feedback and Model Correction: Collect adjusted actual carbon reduction data and use reinforcement learning algorithms to iteratively optimize the potential assessment model and carbon quota curve.

[0026] Preferably, the construction of the dynamic carbon potential profile in step 1 includes: Electricity-carbon coupling topology modeling: Based on the household-transformer relationship data, a refined topological connection relationship from the substation to the end user is constructed, and a mapping relationship between topological nodes and carbon emission factors is established; Real-time carbon emission factor calculation: Combining day-ahead / intra-day renewable energy output forecasts and external power import plans, a time-of-use carbon flow calculation method based on power flow tracking or proportional sharing principles is adopted to dynamically calculate the real-time carbon emission factor (unit: kgCO2 / kWh) flowing into each feeder, transformer area, and even user. Fine-grained user load acquisition: Active power data of each production line of large industrial users is collected at a frequency of 15 minutes or higher. Dynamic carbon potential profile generation: The user's real-time load data is multiplied by the corresponding real-time carbon emission factor to generate the user's dynamic carbon footprint curve (user carbon footprint curve). This curve serves as the core content of the user's "dynamic carbon potential profile" and intuitively reflects the intensity of the user's electricity consumption carbon emissions at different times.

[0027] In this embodiment, the user's carbon footprint curve represents the user's total load at time t. The instantaneous carbon emissions (Emission(t)) and dynamic carbon potential corresponding to the total load for:

[0028]

[0029] in, The real-time carbon emission factor of the upstream feeder at time t; For time intervals.

[0030] For each production line within the user's organization, the load can also be based on its branch load P. i (t), combined with a unified The time-sharing carbon emissions of each production line are calculated to form a more detailed "production line-level carbon potential profile".

[0031] Preferably, the user dynamic carbon potential profile in step 1 also includes a user basic information profile, including industry category, main production process, interruptible load ratio, and self-owned photovoltaic / energy storage capacity, to provide static attribute support for subsequent potential assessment.

[0032] Specifically, information such as the user's industry category, main products, annual output value, and transformer capacity is extracted from the marketing business system; information such as the user's historical maximum load, typical daily load curve, and interruptible load agreement capacity is extracted from the load management system; and the user's own distributed photovoltaic and energy storage installed capacity and real-time output data are accessed from the new energy cloud platform. This static information and dynamic carbon footprint curve are integrated to form a complete "dynamic carbon potential profile" data model for the user, which is then stored in a graph database or time-series database.

[0033] Furthermore, the adjustment potential assessment model in step 2 includes: Carbon efficiency ratio calculation: Defined as the ratio of the economic value added per unit of user load to the carbon emission intensity of the electricity consumed, the formula is:

[0034] in, The economic value added per unit of time for the user. For average electrical load, This represents the average carbon emission factor for the corresponding time period; Adjustability analysis: Based on the user's historical load curve, analyze the time shiftability, interruptibility and power adjustability of the load, and extract key parameters such as adjustable load capacity, adjustment response time and maximum continuous adjustment time; User classification and grading: Cluster analysis was performed on carbon efficiency ratio and adjustability to divide users into three categories: High-carbon, low-efficiency type: has the highest adjustment priority, low carbon efficiency ratio, and a large proportion of interruptible / transferable loads; High-carbon and high-efficiency type: It guides short-term peak avoidance, has a high carbon efficiency ratio but rigid production, and needs to be strongly incentivized to guide it to actively avoid peaks during high-carbon periods of the power grid. Low-carbon flexible type: Encourage priority use of electricity, high carbon efficiency ratio and large proportion of flexible load, such as enterprises that cooperate with distributed photovoltaic production, and encourage them to use more electricity during low-carbon periods of the power grid.

[0035] Furthermore, the hierarchical coordination and control strategy in step 3 is specifically as follows: Master station level: Deployed in the provincial company's load control room, integrating the "electricity-carbon" joint dispatching dashboard. The master station level does not directly control equipment, but rather generates dynamic "carbon quota curves" or "carbon reduction targets" for different power supply areas or industries based on the overall operation of the power grid and future carbon potential predictions, and distributes them to each station level.

[0036] Station / Substation Layer: Edge computing nodes deployed in power supply stations or substations. This node receives carbon quota targets from the master station layer, and calculates the optimal adjustment combination based on the "carbon efficiency ratio" ranking and real-time carbon potential of users within its jurisdiction. It then generates specific adjustment instructions and sends them to the corresponding local layer terminals.

[0037] Local Layer: Edge computing terminals deployed on the user side. These terminals possess bidirectional communication and edge computing capabilities. On one hand, they monitor user load and carbon potential in real time and receive instructions from the station layer; on the other hand, based on preset local policies, they can automatically send "carbon reduction invitations" to the user's central control system or directly and automatically adjust pre-authorized temperature-controlled loads, some flexible loads such as machining equipment, etc.

[0038] Furthermore, the local terminal in step 3 supports two modes when executing instructions: Invitation Mode: For high-carbon and high-efficiency users, after receiving the instruction, the terminal will push carbon reduction invitations and incentive programs to users through APP, SMS and other means, and users can choose whether to implement them and how to implement them.

[0039] Direct control mode: For high-carbon and low-efficiency users who have signed relevant agreements, the terminal can automatically perform closed-loop control on non-critical production auxiliary equipment or pre-modified flexible loads on the user side according to the instructions issued by the station level, with a response speed of seconds or minutes.

[0040] Furthermore, step 4, the feedback and model correction regarding carbon reduction effects, includes: Actual carbon reduction calculation: By comparing the actual load curves of users before and after the adjustment with the carbon potential curve, the actual carbon emission reduction brought about by this regulation action can be accurately calculated. Model parameter optimization: Using actual carbon reduction effects and user feedback information (such as production impact and acceptance) as inputs, reinforcement learning algorithms are used to iteratively optimize the weight parameters of the potential assessment model in step 2 and the carbon quota curve generation model in step 3 online or offline, continuously improving the accuracy and user-friendliness of the regulation strategy.

[0041] To implement the above-mentioned industrial load control methods, such as Figure 2As shown, this embodiment provides the overall hardware framework of an industrial load management system. This system adopts a distributed architecture that integrates cloud, edge, and terminal technologies, and includes four core modules: a data acquisition and carbon potential profiling module, a regulation potential assessment module, a hierarchical coordination and control module, and an effect evaluation and model correction module. The specific functions are as follows: Data Acquisition and Carbon Potential Profiling Module: Deployed in the data center, this module is responsible for executing step 1 above. It includes a data interface unit responsible for interfacing with the energy management system, data management system, marketing system, and user acquisition system; an electricity-carbon topology management unit responsible for building and maintaining the power grid topology; a carbon flow calculation unit responsible for real-time calculation of carbon emission factors at each node; and a user profile generation unit responsible for fusing multi-source data to generate a dynamic carbon potential profile for each user. The output consists of a user carbon potential profile database and a real-time carbon potential data stream.

[0042] Adjustment potential assessment module: Deployed in the data center, it is responsible for performing step 2 above. This module includes a carbon efficiency ratio calculation unit, an adjustability analysis unit, and a user classification and grading unit. It periodically reads the user profile database and massive historical data in an offline or near-line manner, runs cluster analysis algorithms, outputs the latest user classification and grading labels, and updates them back to the user profile database.

[0043] Layered Coordination and Control Module: This is a module that is logically deployed in layers.

[0044] Master station strategy generation unit: Deployed at the provincial company's load control room master station, responsible for executing step 3.1 above. It integrates the "electricity-carbon" joint dispatching dashboard, providing a human-machine interface for dispatchers to set global goals and monitor grid operation status and carbon potential.

[0045] The substation aggregation and optimization unit is deployed as a virtual machine on the edge server of each power supply station / substation. It is responsible for executing step 3.2 above and interacting with the local terminal in real time.

[0046] Local Terminal Execution Unit: Deployed in the user's power distribution room as a hardware device. It is responsible for executing step 3.3 above and has edge computing, protocol conversion, and local automatic control functions.

[0047] Effect Evaluation and Model Correction Module: Deployed in the data center, this module is responsible for executing step 4 above. It includes a carbon reduction accounting unit, responsible for post-event quantitative analysis of each regulation event; and a model self-learning unit, responsible for continuously iteratively optimizing the regulation potential assessment model and the main station carbon quota decomposition model using reinforcement learning algorithms. The optimized model parameters are pushed back to the regulation potential assessment module and the main station strategy generation unit, forming a closed-loop business process.

[0048] This embodiment also provides a specific application of the above-mentioned control methods, and the implementation process is as follows: Step 1, Multi-source data acquisition and dynamic carbon potential profile construction: Step 1.1, Electro-carbon coupling topology modeling: The system includes the main wiring diagram and operation mode data of the 220kV / 110kV substation connected to the power grid; the geographical information and electrical connection relationships of the 10kV / 20kV feeders, switching stations, and distribution transformers; and information such as the transformer relationships, electricity addresses, and contracted capacity of dedicated transformer users.

[0049] Through data cleaning and mapping, a refined topology connection database is established, showing the path from "220kV substation busbar -> 110kV line -> 110kV substation busbar -> 10kV feeder -> distribution transformer -> low-voltage outgoing line -> dedicated transformer user". A unique ID is assigned to each topology node, and a parent-child node association table is created for subsequent carbon flow tracing.

[0050] Step 1.2, Real-time carbon emission factor calculation: Step 1.2.1, Obtaining source-level carbon emission factors: It accesses day-ahead and intraday ultra-short-term renewable energy (wind power, photovoltaic) output forecast data, external power import plan curves, and the average carbon emission factors of the corresponding source regions. Simultaneously, it collects real-time output data and the design / measured carbon emission intensity of provincial grid-dispatched thermal power units.

[0051] Step 1.2.2, Time-sharing carbon flow calculation: Carbon flow calculation is performed using a proportional sharing principle based on power flow tracing. The provincial power grid with a voltage level of 220kV and above is used as the calculation boundary. The output of each generator unit, including thermal power, hydropower, wind power, photovoltaic power, and equivalent external power, is used as the carbon flow injection source, and the load of each lower-level power grid (110kV / 10kV) is used as the carbon flow outflow point.

[0052] For a given calculation period T, by solving the power flow tracking matrix, the proportion of electricity flowing into the target node from different energy sources is calculated. Then, the average carbon emission factor of that node within period T is:

[0053] in, Let be the amount of electricity flowing from power source i into node node during period T. Let be the carbon emission factor of power source i during period T. This represents the total amount of electricity flowing through node during period T.

[0054] For all key nodes in the entire network, such as 110kV substation busbars and 10kV feeder outlets, recursive calculations are performed to obtain a dynamically updated spatiotemporal distribution map of the power grid carbon emission factor. The calculation results are published in real time via a message queue through the dedicated power network.

[0055] Step 1.3, Fine-grained collection of user load: Through the main station of the electricity consumption information collection system, high-frequency data collection tasks are set up for large industrial users participating in demand response within the jurisdiction. Collection points include metering at the user's control point, as well as metering points for key internal production lines and non-productive loads (air conditioning, lighting). Collected data items include: positive active energy readings and active power of phases A / B / C. The collection frequency is set to once every 15 minutes, which can be increased to once every 1 minute for important users or directly controlled users.

[0056] The collected data is transmitted to the provincial company's data platform in real time via a message queue.

[0057] Step 1.4, Dynamic carbon potential profile generation: Step 1.4.1, Carbon potential matching at the user access point: Based on the access transformer ID in the user profile, the upstream feeder node is searched in reverse in the power grid topology database, and the real-time carbon emission factor data stream published by that node is subscribed to. The user's load data every minute / 15 minutes is time-aligned with the corresponding upstream feeder carbon emission factor.

[0058] Step 1.4.2, generate the carbon footprint curve: For the total load of users at time t The corresponding instantaneous carbon emissions (Emission(t)) and dynamic carbon potential for:

[0059]

[0060] in, Let be the real-time carbon emission factor of the upstream feeder at time t. For time intervals.

[0061] For each production line within the user's organization, the load can also be based on its branch load P. i (t), combined with a unified The time-sharing carbon emissions of each production line are calculated to form a more detailed "production line-level carbon potential profile".

[0062] Step 1.4.3, Construction of Comprehensive User Information Profile: Information such as the user's industry category, main products, annual output value, and transformer capacity is extracted from the marketing business system. Information such as the user's historical maximum load, typical daily load curve, and interruptible load agreement capacity is extracted from the load management system. The user's own distributed photovoltaic and energy storage installed capacity and real-time output data are accessed from the new energy cloud platform. This static information is integrated with the dynamic carbon footprint curve generated in step 1.4.2 to form a complete "dynamic carbon potential profile" data model for the user, which is stored in a graph database or time series database.

[0063] Step 2, User-level adjustment potential assessment: Step 2.1, Carbon efficiency ratio calculation: Estimate the previous year's monthly / quarterly industrial added value based on the unit electricity consumption output value of the user's industry. Calculate the user's total electricity consumption during a typical time period. and weighted average carbon emission factor .

[0064] Average carbon efficiency ratio of users during this period for:

[0065] To further evaluate a user's carbon efficiency performance at different times, their dynamic carbon efficiency ratio can be calculated during peak, flat, and low periods.

[0066] Step 2.2, Adjustability Analysis Based on at least one year of historical load data from users, with a sampling interval of 15 minutes, load characteristic mining is performed: Interruptible load identification: Analyze the instantaneous drop characteristics of the load curve when the user receives a demand response command or when load control occurs to identify the capacity and response speed of interruptible loads. If no historical control data is available, set an upper limit for the proportion of interruptible loads based on the user's typical industry experience and production process research.

[0067] Load time shiftability analysis: Analyze the shape of the user's daily load curve, calculate indicators such as load peak-to-valley difference and load factor, and combine them with the user's production shift information, such as whether production is continuous and whether there are maintenance windows, to assess the potential for the user to shift some of the peak load to the off-peak period.

[0068] Power adjustability analysis: For users' large temperature-controlled loads and energy storage equipment such as industrial air conditioners, cold storage, and electric kilns, analyze their ability to continuously adjust power within a certain range.

[0069] Step 2.3, User Classification and Grading: The K-means clustering algorithm (an unsupervised learning algorithm) was used to cluster all users into 5 categories using their carbon efficiency ratio, interruptible load ratio, and peak-valley load difference as feature vectors. Based on business experience, the clustering results were labeled as "high-carbon, low-efficiency," "high-carbon, high-efficiency," and "low-carbon, flexible," and assigned a priority level of 1-5, with higher numbers indicating higher priority. The evaluation results were updated weekly or monthly and stored in a user profile database.

[0070] Step 3, Development of a hierarchical coordination and control strategy: Step 3.1, Main station layer carbon quota generation: Step 3.1.1, Grid carbon potential prediction: The provincial company's load control room main station system receives the next day's new energy output forecast and load forecast data. Combined with the carbon flow calculation model in step 1.2, it predicts the carbon emission factors of the zones and feeders for the next 24 hours (96 points) and generates the "grid carbon potential prediction curve".

[0071] Step 3.1.2, Dynamic Carbon Quota Decomposition: The main station system sets a global carbon reduction target, such as a 5% reduction in total carbon emissions from the city's industrial load between 14:00 and 15:00 tomorrow. Based on the load characteristics and carbon efficiency levels of each power supply area, the overall carbon reduction target is decomposed into carbon quota curves or carbon reduction potential indicators for each area, and then distributed to the edge computing nodes at the station level of each area through the dedicated power grid.

[0072] Step 3.2, Station-level aggregation optimization: Each power supply station / substation's edge computing node stores dynamic carbon potential profiles and the latest adjustment priorities for all users within its jurisdiction.

[0073] Upon receiving the regional carbon quota curve from the main station, the station-level nodes initiate aggregate optimization calculations: Objective function: To minimize the impact on user production and maximize user participation while meeting regional carbon quota constraints; Constraints: Adjustment capacity, adjustment rate, and maximum adjustment duration for each user; Optimization variables: Selecting which users participate in the adjustment, and their respective adjustment amounts.

[0074] A heuristic algorithm is used to solve this optimization problem, resulting in an optimal combination of user-load-regulation instructions. For example: Instruction 1 notifies user A (priority 1) to reduce the load on production line 2 from 800kW to 300kW between 14:00 and 15:00; Instruction 2 sends an invitation to user B (priority 2), suggesting that they activate energy storage discharge at a power of 500kW between 14:00 and 15:00.

[0075] Step 3.3, execute the following on the local terminal: The edge computing terminal deployed on the user side listens to the commands from the station layer in real time. The terminal has a local control policy library set up for the user, which includes two types of execution logic: direct control and invitation control.

[0076] Invitation Mode Execution Flow: After receiving the invitation instruction, the terminal parses the instruction content, including time, adjustment amount, and incentive price. Based on the user's preset acceptance threshold, such as an incentive price > 3 yuan / kW and no impact on core production, it automatically determines whether to accept. If accepted, it automatically sends an adjustment signal to the user's central control system, or reminds the on-duty personnel to manually confirm via SMS / mobile terminal application. After user confirmation, the terminal records the start and end times of execution and sends the execution results back.

[0077] Direct control mode execution process: After receiving the direct control command, the terminal performs a safety check, such as whether the limit is exceeded or whether it is in a prohibited adjustment period. After the check is passed, the terminal directly sends start / stop or adjustment commands to non-production air conditioners, some auxiliary motors, and other equipment that have been pre-connected to the control loop via RS485 (serial communication interface standard), thereby realizing closed-loop automatic control of the load.

[0078] Step 4, Feedback on carbon reduction effect and model correction: Step 4.1, Calculation of actual carbon reduction: After each adjustment event, the station-level nodes automatically retrieve the actual load curves of participating users before and after the adjustment period and the corresponding real-time carbon emission factor data from the data platform. The specific calculation formula is as follows:

[0079] in, The baseline load for users is typically taken as the load of similar days at the same time before adjustment. This is the actual load. This represents the actual carbon potential. The calculation results serve as the basis for evaluating the effectiveness of regulation and calculating incentive costs.

[0080] Step 4.2, Model self-learning optimization: Each regulation event is recorded, including participating users, regulation amount, actual carbon reduction effect, and user feedback, and used as a training sample. Reinforcement learning algorithms are used to train and iteratively update the classification threshold of the potential assessment model in step 2 and the carbon quota decomposition model in step 3.1 offline.

[0081] If the model predicts a user has a regulation potential of 500kW, but only 200kW is actually responded to, and the feedback impacts production, the model will adjust the user's interruptible load proportion weight parameter, reducing its potential value in the next assessment to better reflect reality. Through continuous closed-loop feedback, the prediction accuracy and the effectiveness of the regulation strategy of the entire system will be continuously improved.

[0082] like Figure 4 As shown, this embodiment also provides an industrial load management system, including: a carbon potential profile generation module, used to construct a substation-feeder-transformer area-user level electric carbon coupling topology model based on the collected power grid topology, source carbon emission factors and user load fine-grained data, and generate a dynamic carbon potential profile for each user in real time based on the electric carbon coupling topology model; wherein, the dynamic carbon potential profile includes load size and real-time carbon emission factors; The priority classification module is used to establish an industrial load flexibility adjustment potential assessment model based on dynamic carbon potential profiles and user historical load data, and to match users with corresponding priorities through the industrial load flexibility adjustment potential assessment model; wherein, the priorities are classified into high carbon low efficiency type, high carbon high efficiency type and low carbon flexible type. The architecture construction module is used to build a three-layer control architecture: local layer, station layer, and main station layer. The main station layer is used to generate dynamic carbon quota curves by region and industry. The station layer is used to aggregate and sort the users under its jurisdiction based on user priority to output adjustment instructions. The local layer is used to execute flexible adjustment operations of industrial load according to the received adjustment instructions. The optimization module is used to collect actual carbon reduction data after flexible adjustment of industrial load, and to iteratively optimize the industrial load flexible adjustment potential assessment model and dynamic carbon quota curve based on reinforcement learning algorithm combined with actual carbon reduction data.

[0083] The present invention also provides an industrial load management device, comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the industrial load management method.

[0084] The present invention also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the industrial load control method.

[0085] When the processor executes the computer program, it implements the above-mentioned industrial load management steps, for example: based on the collected power grid topology, source carbon emission factors and user load fine-grained data, it constructs a substation-feeder-transformer area-user level electric carbon coupling topology model, and generates a dynamic carbon potential profile for each user in real time based on the electric carbon coupling topology model; wherein, the dynamic carbon potential profile includes load size and real-time carbon emission factors. Based on dynamic carbon potential profiles and historical load data of users, a model for assessing the flexible adjustment potential of industrial load is established by introducing carbon efficiency ratio. The model is used to match users with corresponding priorities. The priorities are divided into high-carbon and low-efficiency type, high-carbon and high-efficiency type, and low-carbon and flexible type. A three-layer control architecture is constructed, consisting of a local layer, a station layer, and a main station layer. The main station layer is used to generate dynamic carbon quota curves by region and industry. The station layer is used to aggregate and sort users under its jurisdiction based on user priority to output adjustment instructions. The local layer is used to execute flexible industrial load adjustment operations according to the received adjustment instructions. We collect actual carbon reduction data after flexible adjustment of industrial load, and iteratively optimize the industrial load flexible adjustment potential assessment model and dynamic carbon quota curve based on reinforcement learning algorithm combined with actual carbon reduction data.

[0086] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing preset functions, the instruction segments describing the execution process of the computer program in the industrial load management equipment. For example, the computer program can be divided into a carbon potential profile generation module, a priority allocation module, an architecture construction module, and an optimization module; the specific functions are as follows: The carbon potential profile generation module is used to construct a substation-feeder-transformer area-user level electric carbon coupling topology model based on the collected power grid topology, source-end carbon emission factors, and fine-grained user load data, and to generate a dynamic carbon potential profile for each user in real time based on the electric carbon coupling topology model; wherein, the dynamic carbon potential profile includes load size and real-time carbon emission factors; The priority allocation module is used to establish an industrial load flexible adjustment potential assessment model based on the dynamic carbon potential profile and user historical load data, introducing carbon efficiency ratio, and assessing the industrial load flexible adjustment potential. The model matches users with corresponding priorities; these priorities are categorized into high-carbon, low-efficiency, high-carbon, high-efficiency, and low-carbon, flexible types. An architecture construction module is used to build a three-layer control architecture: local layer, station layer, and main station layer. The main station layer generates dynamic carbon quota curves by region and industry. The station layer aggregates and sorts users under its jurisdiction based on user priorities to output adjustment instructions. The local layer executes flexible industrial load adjustment operations according to the received adjustment instructions. An optimization module collects actual carbon reduction data after flexible industrial load adjustment and iteratively optimizes the industrial load flexible adjustment potential assessment model and dynamic carbon quota curves based on reinforcement learning algorithms combined with the actual carbon reduction data. The industrial load management equipment can be a desktop computer, laptop, handheld computer, or cloud server, etc. The industrial load management equipment may include, but is not limited to, processors and memory. Those skilled in the art will understand that the above are examples of industrial load management equipment and do not constitute a limitation on industrial load management equipment. It may include more components than described above, or combine certain components, or use different components. For example, the industrial load management equipment may also include input / output devices, network access devices, buses, etc.

[0087] The processor referred to can 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. The general-purpose processor can be a microprocessor, or any conventional processor, etc. The processor is the control center of the industrial load management system, connecting various parts of the entire industrial load management equipment through various interfaces and lines.

[0088] The memory can be used to store the computer program and / or modules. The processor implements various functions of the industrial load control equipment by running or executing the computer program and / or modules stored in the memory and by calling the data stored in the memory.

[0089] The memory may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function (such as sound playback, image playback, etc.). The data storage area may store data created based on the use of the mobile phone (such as audio data, phonebook, etc.). Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart media cards (SMC), secure digital cards (SD cards), flash cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0090] The present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the industrial load control method described above.

[0091] If the modules / units integrated in the industrial load management system are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0092] Based on this understanding, the present invention can implement all or part of the processes in the above-mentioned industrial load control method, or it can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the above-mentioned industrial load control method. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or a preset intermediate form, etc.

[0093] The computer-readable storage medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0094] It should be noted that the content contained in the computer-readable storage medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0095] Compared with existing control measures, this invention provides an industrial load control method, system, equipment, and medium, which have the following advantages: First, the regulation is highly precise: by introducing the concepts of dynamic carbon potential and carbon efficiency ratio, "carbon" is used as a new dimension of load regulation, realizing the transformation from "single load regulation" to "electricity-carbon joint regulation". It can accurately identify and prioritize the regulation of high carbon emission loads, and significantly improve the carbon reduction effect.

[0096] Second, clear user profiles: A dynamic profile including carbon efficiency level and regulation potential has been built for users, enabling them to clearly understand the "carbon content" of their electricity consumption behavior and the level of green production, effectively stimulating users' enthusiasm for participating in green electricity consumption and demand response.

[0097] Third, the system boasts excellent flexibility: its pioneering three-tiered control architecture of "master station-station-local" enables cloud-edge collaboration. The master station sets targets, the stations perform aggregation, and the local system responds quickly, ensuring both the global optimality of the control strategy and the real-time performance and reliability of local responses, resulting in strong system scalability.

[0098] Fourth, significant economic and social benefits: For power grid companies, it has improved the capacity for renewable energy consumption and the level of safe operation of the power grid; for energy-consuming enterprises, it has reduced energy costs and product carbon footprint by optimizing electricity consumption behavior, thereby enhancing market competitiveness.

[0099] The above embodiments are merely one of the implementation methods for achieving the technical solution of the present invention. The scope of protection claimed by the present invention is not limited to this embodiment, but also includes any variations, substitutions and other implementation methods that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention.

[0100] 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 protection scope of the present invention.

Claims

1. An industrial load management method, characterized by, include: Based on the collected power grid topology, source carbon emission factors, and fine-grained user load data, a substation-feeder-transformer area-user level electric carbon coupling topology model is constructed. Based on the electric carbon coupling topology model, a dynamic carbon potential profile is generated for each user in real time. The dynamic carbon potential profile includes load size and real-time carbon emission factors. Based on dynamic carbon potential profiles and historical load data of users, a model for assessing the flexible adjustment potential of industrial load is established by introducing carbon efficiency ratio. The model is used to match users with corresponding priorities. The priorities are divided into high-carbon and low-efficiency type, high-carbon and high-efficiency type, and low-carbon and flexible type. A three-layer control architecture is constructed, consisting of a local layer, a station layer, and a main station layer. The main station layer is used to generate dynamic carbon quota curves by region and industry. The station layer is used to aggregate and sort users under its jurisdiction based on user priority to output adjustment instructions. The local layer is used to execute flexible industrial load adjustment operations according to the received adjustment instructions. We collect actual carbon reduction data after flexible adjustment of industrial load, and iteratively optimize the industrial load flexible adjustment potential assessment model and dynamic carbon quota curve based on reinforcement learning algorithm combined with actual carbon reduction data.

2. The industrial load management method of claim 1, wherein, The method of generating a dynamic carbon potential profile for each user in real time based on the electro-carbon coupling topology model includes: Based on the access transformer identifier in the user profile, match the user's upstream feeder node in the power grid topology database and subscribe to the real-time carbon emission factor data stream of this node, and time-align the user's time-sharing load data with the corresponding upstream feeder carbon emission factor. Based on the mapping relationship between topological nodes and carbon emission factors established by the electric carbon coupling topology model, the instantaneous carbon emissions and dynamic carbon potential of users are calculated by combining the total user load and the real-time carbon emission factors of the upstream feeder, and the user carbon footprint curve is generated. By integrating user static attribute information with user carbon footprint curves, a complete dynamic carbon potential profile is constructed. in the user carbon footprint curve, for the total load of the user at time instant t the instantaneous carbon emission Emission(t) and the dynamic carbon potential corresponding to the total load are: wherein, is the real-time carbon emission factor of the upper feeder line at time t; is the time interval.

3. The industrial load management method of claim 1, wherein, The aforementioned model, based on dynamic carbon potential profiles and historical user load data, introduces carbon efficiency ratio to establish an industrial load flexibility adjustment potential assessment model. This model then matches users with corresponding priorities, including: Based on dynamic carbon potential profiles and historical user load data, a model for assessing the flexible adjustment potential of industrial load is established by introducing the carbon efficiency ratio. The carbon efficiency ratio is the ratio of the economic added value generated per unit of user load to the carbon emission intensity of their electricity consumption, calculated using the following formula: In the formula, The economic value added per unit of time for the user. For average electrical load, This represents the average carbon emission factor for the corresponding time period; Carbon efficiency ratio; Based on historical load data of users, the time-shiftable characteristics, interruptible characteristics and power-adjustable characteristics of user load are analyzed, and key parameters are extracted to output the results of the adjustability analysis; the key parameters include adjustable load capacity, adjustment response time and maximum continuous adjustment time. Cluster analysis was performed on the results of carbon efficiency ratio and adjustability analysis to prioritize users into high-carbon low-efficiency, high-carbon high-efficiency, and low-carbon flexible types.

4. The industrial load control method according to claim 1, characterized in that, The high-carbon, low-efficiency type, the high-carbon, high-efficiency type, and the low-carbon, flexible type are specifically classified as follows: High-carbon, low-efficiency type: has the highest adjustment priority, low carbon efficiency ratio, and a large proportion of interruptible / transferable loads; High-carbon and high-efficiency type: with secondary adjustment priority, high carbon efficiency ratio and rigid production, and strong incentives to guide corresponding users to actively avoid peaks during high-carbon periods of the power grid; Low-carbon flexible type: encourages priority use of electricity, with high carbon efficiency and a large proportion of flexible loads, and encourages corresponding users to use more electricity during low-carbon periods of the power grid.

5. The industrial load control method according to claim 1, characterized in that, The construction of the three-layer control architecture—local layer, station layer, and main station layer—includes: A three-tier control architecture is established: local layer - station layer - main station layer. The specific deployment is as follows: The main station layer is deployed in the upper load control room, integrating the electricity-carbon joint dispatching cockpit, and is responsible for generating dynamic carbon quota curves based on the power grid operation status and carbon potential prediction, and then sending them down to the station layer. The station layer is deployed at the edge computing nodes of power supply stations or substations. It is responsible for receiving dynamic carbon quota curves, completing user aggregation sorting and optimal adjustment combination calculations, and outputting adjustment instructions. The local layer is deployed on the user-side edge computing terminal and is responsible for real-time monitoring of user load and carbon potential, receiving and executing adjustment commands, and completing flexible adjustment operations of industrial load.

6. The industrial load management method of claim 5, wherein, The local layer performs flexible industrial load regulation operations in two modes: Invitation Mode: Targeting high-carbon and high-efficiency users, carbon reduction invitations and incentive programs are pushed out, and users can choose whether to take the adjustment action themselves; Direct control mode: Designed for users with high carbon emissions and low efficiency, it enables automatic closed-loop control of pre-authorized flexible loads, achieving adjustment response at the second or minute level.

7. The industrial load management method of claim 1, wherein, The actual carbon reduction data collected after flexible adjustment of industrial load is used to iteratively optimize the industrial load flexible adjustment potential assessment model and dynamic carbon quota curve based on reinforcement learning algorithm combined with actual carbon reduction data, including: By comparing the actual load curves of users with the carbon potential curves before and after flexible industrial load adjustment, the actual carbon reduction data is calculated; the specific calculation formula is as follows: wherein is the user's baseline load; is the actual load, is the actual carbon potential; is the actual decarburization data; The actual carbon emission reduction and user production impact feedback information are used as inputs to the reinforcement learning algorithm to iteratively optimize the weight parameters and dynamic carbon quota curve of the industrial load flexible adjustment potential assessment model.

8. An industrial load management system, characterized by, include: The carbon potential profile generation module is used to construct a substation-feeder-transformer-user level electric carbon coupling topology model based on the collected power grid topology, source carbon emission factors and fine-grained user load data, and to generate a dynamic carbon potential profile for each user in real time based on the electric carbon coupling topology model; wherein, the dynamic carbon potential profile includes load size and real-time carbon emission factors. The priority classification module is used to establish an industrial load flexibility adjustment potential assessment model based on dynamic carbon potential profiles and user historical load data, and to match users with corresponding priorities through the industrial load flexibility adjustment potential assessment model; wherein, the priorities are classified into high carbon low efficiency type, high carbon high efficiency type and low carbon flexible type. The architecture construction module is used to build a three-layer control architecture: local layer, station layer, and main station layer. The main station layer is used to generate dynamic carbon quota curves by region and industry. The station layer is used to aggregate and sort the users under its jurisdiction based on user priority to output adjustment instructions. The local layer is used to execute flexible adjustment operations of industrial load according to the received adjustment instructions. The optimization module is used to collect actual carbon reduction data after flexible adjustment of industrial load, and to iteratively optimize the industrial load flexible adjustment potential assessment model and dynamic carbon quota curve based on reinforcement learning algorithm combined with actual carbon reduction data.

9. An industrial load management device, characterized by, include: Memory, used to store computer programs; A processor for executing the computer program to implement the industrial load control method according to any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 9. When the computer program is executed by the processor, it is used to implement the industrial load control method according to any one of claims 1-7.

Images