Distributed model predictive control system for collaborative optimization of multi-energy carbon emission in industrial park

Abstract

The invention discloses a distributed model prediction control system for industrial park multi-energy carbon emission collaborative optimization, and relates to the technical field of intelligent energy system optimization control. According to the design, the efficiency is improved, tedious tasks are automated through technical means, manual operation is reduced, time and resources are saved, the working efficiency is improved, and the technology can help to optimize an existing working process, reduce invalid links and improve the overall operation efficiency; the design improves the accuracy and precision, the technical application can reduce human intervention and errors, and especially in complex tasks and data analysis, the technology can provide more accurate data processing and analysis results through more advanced algorithms and tools to help make more scientific decisions.

Description

Technical Field

[0001] The present invention relates to the technical field of smart energy system optimization and control, and specifically to a distributed model predictive control system for collaborative optimization of multi-energy carbon emissions in industrial parks. Background Art

[0002] Carbon emissions monitoring technology is already widely used in the industrial and energy sectors, using sensors, flow meters, and other monitoring equipment to collect real-time data. This data can be calculated to reveal the carbon flow during energy conversion.

[0003] Currently, existing technologies usually adopt static carbon emission models or carbon intensity calculation models based on historical data.

[0004] The shortcomings of these models are:

[0005] Centralized optimization limitations: Traditional centralized MPC requires access to system-wide data, and computational complexity increases exponentially with scale, making it difficult to apply to large campuses.

[0006] Extensive carbon emission measurement: Existing methods use an average carbon emission coefficient and do not consider the real-time carbon flow transmission in the energy conversion chain (such as the carbon emission reduction effect of waste heat utilization in cogeneration units);

[0007] Response lag: The dispatch cycle is greater than 15 minutes, making it impossible to track changes in carbon emission intensity caused by fluctuations in wind and solar power.

[0008] Conflict of interest: Each entity (PV operator, energy storage power station, gas heating supplier) independently optimizes its goals and lacks a Pareto optimal coordination mechanism.

[0009] Therefore, a new solution to the above problems needs to be proposed. Summary of the Invention

[0010] The purpose of the present invention is to provide a distributed model predictive control system for collaborative optimization of multi-energy carbon emissions in industrial parks, which solves the technical problems raised in the background technology.

[0011] To achieve the above objectives, the present invention provides the following technical solutions: a distributed model predictive control system for collaborative optimization of multi-energy carbon emissions in industrial parks, comprising at least a carbon flow dynamic tracking engine, a multi-time scale distributed MPC, a carbon-energy efficiency game coordination mechanism, and a data privacy protection module;

[0012] The carbon flow dynamic tracking engine provides refined carbon flow modeling. Through the dynamic carbon flow inheritance algorithm of the energy conversion link, the precision of the carbon flow model has been significantly improved, and the accuracy has been improved.

[0013] Multi-time-scale distributed MPC is an MPC framework based on ADMM decomposition, which distributes computing tasks to each agent for local optimization, thereby improving computing efficiency;

[0014] The carbon-energy efficiency game coordination mechanism is a dynamic pricing of carbon emission quotas. It combines carbon futures prices and real-time consumption contribution to generate floating quotas, which improves the market adaptability and flexibility of carbon emission quota allocation.

[0015] The data privacy protection module is used to prevent the leakage of sensitive data.

[0016] Furthermore, the carbon flow dynamic tracking engine includes a device-level carbon flow model and an algorithm for calculating the carbon intensity of energy conversion links in real time;

[0017] The equipment-level carbon flow model not only considers the carbon intensity of input energy, but also the conversion efficiency of the equipment and the impact of carbon recovery technology;

[0018] The basic formula of the carbon flow model is:

[0019] Carbon_equipment = ∑(input energy carbon flow × conversion efficiency) - carbon recovery

[0020] Among them: input energy carbon flow represents the carbon emissions brought by each energy when entering the system; conversion efficiency is the efficiency of the equipment in the energy conversion process, which affects carbon emissions; carbon recovery refers to whether the equipment uses carbon capture technology to reduce emissions;

[0021] The algorithm for calculating the carbon intensity of the energy conversion link in real time is a method for calculating the carbon intensity of the conversion process of different energy sources and should be determined based on real-time data;

[0022] Taking gas-fired power generation as an example, the carbon intensity calculation of this link is based on the real-time calorific value of the gas flow meter and the efficiency curve of the generator set. By real-time monitoring of the gas flow and unit operating status, the carbon emission intensity of the power generation process can be dynamically calculated. The calculation formula is as follows:

[0023]

[0024] At this time, heat is supplied through waste heat recovery. During the waste heat recovery process, the carbon flow inherits the carbon emissions from the power generation link, see the following formula:

[0025] Carbon_Intensity heat =0.4×C gas_to_power

[0026] If the carbon emissions from power generation are C gas_to_power , the carbon intensity of the heating link is 40% of that.

[0027] Furthermore, the multi-timescale distributed MPC contributes to the energy efficiency and carbon emission management of multi-objective and complex systems through real-time prediction and optimization of control strategies, dividing the system into different time periods for optimization, and ensuring effective carbon emission management at different time scales;

[0028] The multi-timescale distributed MPC includes a fast cycle layer and a slow cycle layer;

[0029] At the 1-second fast cycle layer, a federated learning architecture is used to allow each participant (such as the power grid, power generators, etc.) to perform data processing and target optimization locally;

[0030] The objective function of each agent includes operating cost, predicted carbon emissions, and power deviation penalty;

[0031] Each entity optimizes based on local data without directly sharing sensitive information;

[0032] Objective function form:

[0033] min[α·operating cost + β·predicted carbon emissions + γ·power deviation penalty]

[0034] Where: α, β, and γ are weight coefficients that control the relative importance of different objectives; operating costs include at least energy procurement and equipment operation; predicted carbon emissions are carbon emissions forecast based on the energy consumption of each entity; power deviation penalty is to control the deviation between power output and demand to avoid grid instability;

[0035] The optimization cycle of the slow cycle layer is longer, at 15 minutes, and is mainly used for the dynamic allocation of carbon emission quotas;

[0036] The blockchain coordinator allocates carbon emission quotas through an improved Shapley value algorithm to ensure that carbon emission responsibilities are fairly shared among all entities;

[0037] Quota calculation formula:

[0038] quota i = Baseline quota + λ·(wind and solar power consumption contribution - load fluctuation rate)

[0039] Among them: the baseline quota is the initial quota based on historical data or government regulations; the wind and solar power consumption contribution is a measure of the contribution of renewable energy (such as wind power and solar power) to total power generation; the load fluctuation rate is a measure of the impact of load fluctuations on carbon emissions;

[0040] The Shapley value algorithm ensures that each entity is allocated carbon emission quota according to its contribution to the overall system.

[0041] Furthermore, the carbon-energy efficiency game coordination mechanism is used to ensure optimal coordination between carbon emissions and energy efficiency, and a model based on non-cooperative game is constructed through the carbon-energy efficiency game coordination mechanism;

[0042] The entities of the non-cooperative game-based model include at least new energy operators, energy storage companies, and traditional power generators;

[0043] The strategy set of the non-cooperative game-based model is the optimization action space of each subject under the MPC framework, and the strategy set includes at least power generation strategy, energy storage strategy and load regulation strategy;

[0044] The payment function of the non-cooperative game-based model is that the payment of each subject consists of carbon emission costs, energy benefits and default penalties, and the goal of each subject is to maximize its payment;

[0045] The payment function is:

[0046] U i =L i -M i -O i

[0047] Where: L i Represents energy income, which is the income obtained by the subject through selling electricity and storing energy; M i Carbon emission cost is the carbon emission fee that the entity needs to pay based on its carbon emissions; i It represents the penalty for breach of contract, which is the penalty that the entity needs to pay if it fails to meet the carbon emission or energy efficiency targets as agreed;

[0048] Lyapunov optimization theory is used to prove that the model based on non-cooperative game has Nash equilibrium under appropriate conditions and can converge to a stable solution. Lyapunov function is used to analyze the stability of the system to ensure that the optimization strategy of the participants can reach equilibrium after long-term operation.

[0049] Furthermore, the data privacy protection module uses homomorphic encryption technology to protect the privacy of local data, ensuring that the coordinator can only receive gradient information.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] 1. The design of this invention improves efficiency by automating tedious tasks through technical means, reducing manual operations, saving time and resources, and improving work efficiency. The technology can also help optimize existing workflows, reduce ineffective links, and improve overall operational efficiency.

[0052] 2. The design of this invention improves accuracy and precision. The application of technology can reduce human intervention and reduce the occurrence of errors, especially in complex tasks and data analysis. The technology can provide more accurate data processing and analysis results through more advanced algorithms and tools, helping to make more scientific decisions.

[0053] 3. The design of the present invention reduces costs. The efficient technology can reduce the waste of resources, such as energy, time, and manpower, thereby helping to reduce operating costs and reduce repetitive work. Unnecessary manual intervention and repetitive work are reduced through automation and intelligent systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0055] Figure 1 It is a schematic structural diagram of the present invention as a whole. DETAILED DESCRIPTION

[0056] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0057] See also Figure 1 , a distributed model predictive control system for the coordinated optimization of multi-energy carbon emissions in industrial parks, including at least a carbon flow dynamic tracking engine, multi-time scale distributed MPC, a carbon-energy efficiency game coordination mechanism, and a data privacy protection module;

[0058] The carbon flow dynamic tracking engine provides refined carbon flow modeling. Through the dynamic carbon flow inheritance algorithm of the energy conversion link, the precision of the carbon flow model has been significantly improved, and the accuracy has been improved;

[0059] Multi-time-scale distributed MPC is an MPC framework based on ADMM decomposition, which distributes computing tasks to each agent for local optimization, thereby improving computing efficiency;

[0060] The carbon-energy efficiency game coordination mechanism is a dynamic pricing of carbon emission quotas. It combines carbon futures prices and real-time consumption contribution to generate floating quotas, which improves the market adaptability and flexibility of carbon emission quota allocation.

[0061] The data privacy protection module is used to prevent the leakage of sensitive data.

[0062] The carbon flow dynamic tracking engine includes a device-level carbon flow model and an algorithm for calculating the carbon intensity of energy conversion links in real time;

[0063] The equipment-level carbon flow model not only considers the carbon intensity of the input energy, but also the conversion efficiency of the equipment and the impact of carbon recovery technology;

[0064] The basic formula of the carbon flow model is:

[0065] Carbon_equipment = ∑(input energy carbon flow × conversion efficiency) - carbon recovery

[0066] Among them: input energy carbon flow represents the carbon emissions brought by each energy when entering the system; conversion efficiency is the efficiency of the equipment in the energy conversion process, which affects carbon emissions; carbon recovery refers to whether the equipment uses carbon capture technology to reduce emissions;

[0067] The real-time calculation algorithm of the carbon intensity of the energy conversion link is the calculation method of the carbon intensity of the conversion process of different energy sources, which should be determined based on real-time data;

[0068] Taking gas-fired power generation as an example, the carbon intensity calculation of this link is based on the real-time calorific value of the gas flow meter and the efficiency curve of the generator set. By real-time monitoring of the gas flow and unit operating status, the carbon emission intensity of the power generation process can be dynamically calculated. The calculation formula is as follows:

[0069]

[0070] At this time, heat is supplied through waste heat recovery. During the waste heat recovery process, the carbon flow inherits the carbon emissions from the power generation link, see the following formula:

[0071] Carbon_Intensity heat =0.4×C gas_to_power

[0072] If the carbon emissions from power generation are C gas_to_power , the carbon intensity of the heating link is 40% of that.

[0073] Multi-time-scale distributed MPC helps manage energy efficiency and carbon emissions in multi-objective, complex systems by predicting and optimizing control strategies in real time. It divides the system into different time periods for optimization, ensuring effective carbon emissions management at different time scales.

[0074] Multi-timescale distributed MPC includes a fast-cycle layer and a slow-cycle layer;

[0075] At the 1-second fast cycle layer, a federated learning architecture is used to allow each participant (such as the power grid, power generators, etc.) to perform data processing and target optimization locally;

[0076] The objective function of each agent includes operating cost, predicted carbon emissions, and power deviation penalty;

[0077] Each entity optimizes based on local data without directly sharing sensitive information;

[0078] Objective function form:

[0079] min[α·operating cost + β·predicted carbon emissions + γ·power deviation penalty]

[0080] Where: α, β, and γ are weight coefficients that control the relative importance of different objectives; operating costs include at least energy procurement and equipment operation; predicted carbon emissions are carbon emissions forecast based on the energy consumption of each entity; power deviation penalty is to control the deviation between power output and demand to avoid grid instability;

[0081] The optimization cycle of the slow cycle layer is longer, at 15 minutes, and is mainly used for the dynamic allocation of carbon emission quotas;

[0082] The blockchain coordinator allocates carbon emission quotas through an improved Shapley value algorithm to ensure that carbon emission responsibilities are fairly shared among all entities;

[0083] Quota calculation formula:

[0084] quota i = Baseline quota + λ·(wind and solar power consumption contribution - load fluctuation rate)

[0085] Among them: the baseline quota is the initial quota based on historical data or government regulations; the wind and solar power consumption contribution is a measure of the contribution of renewable energy (such as wind power and solar power) to total power generation; the load fluctuation rate is a measure of the impact of load fluctuations on carbon emissions;

[0086] The Shapley value algorithm ensures that each entity is allocated carbon emission quota according to its contribution to the overall system.

[0087] The carbon-energy efficiency game coordination mechanism is used to ensure the optimal coordination between carbon emissions and energy efficiency. A model based on non-cooperative game is constructed through the carbon-energy efficiency game coordination mechanism.

[0088] The main players in the non-cooperative game-based model include at least new energy operators, energy storage companies, and traditional power generators;

[0089] The strategy set of the non-cooperative game-based model is the optimized action space of each agent under the MPC framework. The strategy set includes at least power generation strategy, energy storage strategy and load regulation strategy.

[0090] The payment function of the model based on non-cooperative game is that each subject’s payment consists of carbon emission cost, energy benefit and default penalty, and each subject’s goal is to maximize its payment;

[0091] The payment function is:

[0092] U i =L i -M i -O i

[0093] Where: L i Represents energy income, which is the income obtained by the subject through selling electricity and storing energy; M i Carbon emission cost is the carbon emission fee that the entity needs to pay based on its carbon emissions; i It represents the penalty for breach of contract, which is the penalty that the entity needs to pay if it fails to meet the carbon emission or energy efficiency targets as agreed;

[0094] Lyapunov optimization theory is used to prove that the model based on non-cooperative game has Nash equilibrium under appropriate conditions and can converge to a stable solution. Lyapunov function is used to analyze the stability of the system to ensure that the optimization strategy of the participants can reach equilibrium after long-term operation.

[0095] The data privacy protection module uses homomorphic encryption technology to protect the privacy of local data, ensuring that the coordinator can only receive gradient information.

[0096] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A distributed model predictive control system for collaborative optimization of multi-energy carbon emissions in industrial parks, characterized by: It includes at least a carbon flow dynamic tracking engine, multi-time scale distributed MPC, a carbon-energy efficiency game coordination mechanism, and a data privacy protection module; The carbon flow dynamic tracking engine provides refined carbon flow modeling. Through the dynamic carbon flow inheritance algorithm of the energy conversion link, the precision of the carbon flow model has been significantly improved, and the accuracy has been improved. Multi-time-scale distributed MPC is an MPC framework based on ADMM decomposition, which distributes computing tasks to each agent for local optimization, thereby improving computing efficiency; The carbon-energy efficiency game coordination mechanism is a dynamic pricing of carbon emission quotas. It combines carbon futures prices and real-time consumption contribution to generate floating quotas, which improves the market adaptability and flexibility of carbon emission quota allocation. The data privacy protection module is used to prevent the leakage of sensitive data.

2. The distributed model predictive control system for coordinated optimization of multi-energy carbon emissions in industrial parks according to claim 1 is characterized by: The carbon flow dynamic tracking engine includes a device-level carbon flow model and an algorithm for calculating the carbon intensity of energy conversion links in real time; The equipment-level carbon flow model not only considers the carbon intensity of input energy, but also the conversion efficiency of the equipment and the impact of carbon recovery technology; The basic formula of the carbon flow model is: Carbon_equipment = ∑(input energy carbon flow × conversion efficiency) - carbon recovery Among them: input energy carbon flow represents the carbon emissions brought by each energy when entering the system; conversion efficiency is the efficiency of the equipment in the energy conversion process, which affects carbon emissions; carbon recovery refers to whether the equipment uses carbon capture technology to reduce emissions; The algorithm for calculating the carbon intensity of the energy conversion link in real time is a method for calculating the carbon intensity of the conversion process of different energy sources and should be determined based on real-time data; Taking gas-fired power generation as an example, the carbon intensity calculation of this link is based on the real-time calorific value of the gas flow meter and the efficiency curve of the generator set. By real-time monitoring of the gas flow and unit operating status, the carbon emission intensity of the power generation process can be dynamically calculated. The calculation formula is as follows: At this time, heat is supplied through waste heat recovery. During the waste heat recovery process, the carbon flow inherits the carbon emissions from the power generation link, see the following formula: Carbon_Intensity heat =0.4×C gas_to_power If the carbon emissions from power generation are C gas_to_power , the carbon intensity of the heating link is 40% of that.

3. The distributed model predictive control system for coordinated optimization of multi-energy carbon emissions in industrial parks according to claim 1 is characterized by: The multi-time-scale distributed MPC helps manage energy efficiency and carbon emissions in multi-objective, complex systems by predicting and optimizing control strategies in real time. It divides the system into different time periods for optimization, ensuring effective carbon emission management at different time scales. The multi-timescale distributed MPC includes a fast cycle layer and a slow cycle layer; At the 1-second fast cycle layer, a federated learning architecture is used to allow each participant to perform data processing and target optimization locally; The objective function of each agent includes operating cost, predicted carbon emissions, and power deviation penalty; Each entity optimizes based on local data without directly sharing sensitive information; Objective function form: min[α·operating cost + β·predicted carbon emissions + γ·power deviation penalty] Where: α, β, and γ are weight coefficients that control the relative importance of different objectives; operating costs include at least energy procurement and equipment operation; predicted carbon emissions are carbon emissions forecast based on the energy consumption of each entity; power deviation penalty is to control the deviation between power output and demand to avoid grid instability; The optimization cycle of the slow cycle layer is longer, at 15 minutes, and is mainly used for the dynamic allocation of carbon emission quotas; The blockchain coordinator allocates carbon emission quotas through an improved Shapley value algorithm to ensure that carbon emission responsibilities are fairly shared among all entities; Quota calculation formula: quota i = Baseline quota + λ·(wind and solar power consumption contribution - load fluctuation rate) Among them: the benchmark quota is the initial quota based on historical data or government regulations; the wind and solar power consumption contribution is a measure of the contribution of renewable energy in total power generation; the load fluctuation rate is a measure of the impact of load fluctuations on carbon emissions; The Shapley value algorithm ensures that each entity is allocated carbon emission quota according to its contribution to the overall system.

4. The distributed model predictive control system for coordinated optimization of multi-energy carbon emissions in industrial parks according to claim 1 is characterized by: The carbon-energy efficiency game coordination mechanism is used to ensure the optimal coordination between carbon emissions and energy efficiency. A model based on non-cooperative game is constructed through the carbon-energy efficiency game coordination mechanism. The entities of the non-cooperative game-based model include at least new energy operators, energy storage companies, and traditional power generators; The strategy set of the non-cooperative game-based model is the optimization action space of each subject under the MPC framework, and the strategy set includes at least power generation strategy, energy storage strategy and load regulation strategy; The payment function of the non-cooperative game-based model is that the payment of each subject consists of carbon emission costs, energy benefits and default penalties, and the goal of each subject is to maximize its payment; The payment function is: U i =L i -M i -O i Where: L i Represents energy income, which is the income obtained by the subject through selling electricity and storing energy; M i Carbon emission cost is the carbon emission fee that the entity needs to pay based on its carbon emissions; i It represents the penalty for breach of contract, which is the penalty that the entity needs to pay if it fails to meet the carbon emission or energy efficiency targets as agreed; Lyapunov optimization theory is used to prove that the model based on non-cooperative game has Nash equilibrium under appropriate conditions and can converge to a stable solution. Lyapunov function is used to analyze the stability of the system to ensure that the optimization strategy of the participants can reach equilibrium after long-term operation.

5. The distributed model predictive control system for collaborative optimization of multi-energy carbon emissions in industrial parks according to claim 1 is characterized by: The data privacy protection module uses homomorphic encryption technology to protect the privacy of local data, ensuring that the coordinator can only receive gradient information.

Images