A zero-carbon park can coordinate multi-objective optimization scheduling method and system

Abstract

The application provides a zero-carbon park energy-carbon collaborative multi-objective optimal scheduling method and system, which comprises the following steps: constructing a park comprehensive energy system topology model and a device operation characteristic model; based on the park comprehensive energy system topology model and the device operation characteristic model, a park dynamic carbon emission flow tracking model is constructed, the dynamic carbon potential of each node is recursively calculated, and a dynamic carbon emission factor is obtained; an energy-carbon collaborative multi-objective optimal scheduling model is constructed; an adaptive weight mechanism based on an entropy weight method and fuzzy reasoning fusion is used to dynamically weight and aggregate each optimization target; the energy-carbon collaborative multi-objective optimal scheduling model is processed by using a two-stage distributed robust optimization framework of day-ahead and day-ahead to deal with the uncertainty of source and load prediction, and a Pareto optimal solution set is obtained by using a column and constraint generation algorithm iteration solution, and a compromise optimal scheduling scheme is selected by using an approximate ideal solution sorting method, so that the reliability of the zero-carbon park energy-carbon collaborative multi-objective optimal scheduling is improved.

Description

Technical Field

[0001] This invention relates to the field of integrated energy system optimization scheduling, and in particular to a zero-carbon park energy-carbon coordinated multi-objective optimization scheduling method and system. Background Technology

[0002] Zero-carbon industrial parks, serving as crucial spatial carriers for energy transition and industrial upgrading, are being rapidly constructed nationwide. These parks typically integrate various energy conversion and storage devices, including distributed photovoltaic power, wind power, electrochemical energy storage, electric vehicle charging facilities, combined heat and power (CHP) units, electric heating equipment, and hydrogen energy storage, forming a typical integrated energy system that couples electricity, heat, gas, and hydrogen. One of the core technical challenges in achieving the zero-carbon goals of these parks is how to optimize and schedule such multi-energy coupled systems in an economical, low-carbon, and reliable manner.

[0003] Meanwhile, Energy Performance Contracting (EPC), as a market-oriented energy-saving service business model, is widely used in energy-saving renovations of industrial parks. Under the EPC model, energy service companies provide energy-saving equipment investment, renovation, and operation and maintenance services to the park, gradually recovering their investment and generating profits by sharing the energy-saving benefits generated after the renovation. Since the investment recovery and profit generation of the energy service provider depend on the continuous realization of actual energy savings after the renovation, the park's operation and scheduling must not only consider the optimization of the energy system itself but also the commercial constraints stipulated in the EPC agreement, such as the guaranteed minimum energy saving rate, the energy-saving benefit sharing ratio, and the investment recovery period. If the scheduling strategy deviates from the contract requirements, the energy service provider will face the risk of insufficient energy savings and difficulty in recovering its investment, thereby damaging the sustainability of the energy-saving renovation project.

[0004] A systematic analysis of existing energy dispatching technologies in industrial parks reveals the following main shortcomings.

[0005] First, most existing dispatching methods take minimizing operational economic costs as the sole optimization objective, often treating carbon emissions only as a post-hoc accounting indicator and failing to truly integrate them into the real-time optimization loop. Even when some methods attempt to incorporate carbon emissions into the optimization considerations, they generally employ a fixed average carbon emission factor for grid operation. However, the carbon emission intensity of the power grid fluctuates significantly with real-time changes in power supply structure and power flow distribution. A fixed carbon factor cannot reflect this dynamic characteristic, leading to distorted carbon emission calculation results. This makes it difficult to support refined low-carbon dispatching decisions in industrial parks, and even more difficult to guide energy consumption towards low-carbon practices at different times.

[0006] Secondly, existing multi-objective optimization scheduling methods mostly use fixed weights to linearly weight and aggregate economic and carbon emission targets. Once the weights are set, they remain unchanged throughout the entire scheduling cycle, making it impossible to adapt to real-time fluctuations in operating conditions such as electricity prices, carbon prices, and renewable energy output. In certain operating scenarios, fixed weights can cause the scheduling results to systematically deviate from the overall optimum, making it difficult to flexibly balance economic efficiency and low-carbon performance under different price signals and resource conditions.

[0007] Third, existing scheduling methods generally fail to incorporate the business model of Energy Performance Contracting (EPC) into their optimization considerations. Under the EPC model, scheduling strategies not only affect the park's own energy costs and carbon emission levels, but also directly determine whether the energy service provider can achieve the guaranteed minimum energy saving rate and investment recovery stipulated in the contract, thus affecting the overall sustainability of energy-saving renovation projects. Current technologies lack effective means to quantify and embed the commercial terms of EPC into scheduling optimization models, resulting in a disconnect between technical operational optimization and commercial revenue objectives, making it difficult to reliably realize energy-saving benefits.

[0008] Fourth, the high proportion of renewable energy access in zero-carbon industrial parks brings significant uncertainty to source-load forecasting. Existing deterministic optimization methods typically formulate scheduling plans based on predicted expected values. Once the actual output and load deviate from the predicted values, the scheduling plan may fail or even violate the system's safe operation constraints, resulting in insufficient robustness.

[0009] In summary, current technologies lack a zero-carbon park optimization scheduling method that can simultaneously integrate dynamic carbon emission tracking, commercial constraints of energy performance contracting, multi-objective adaptive coordination, and robust handling of source-load uncertainty. Therefore, a new technical solution is urgently needed to overcome these shortcomings. Summary of the Invention

[0010] To address the problems existing in the prior art, this invention innovatively proposes a multi-objective optimization scheduling method and system for energy and carbon coordination in zero-carbon industrial parks. This effectively solves the problem of low reliability in multi-objective optimization scheduling for energy and carbon coordination in zero-carbon industrial parks caused by existing technologies, and effectively improves the reliability of multi-objective optimization scheduling for energy and carbon coordination in zero-carbon industrial parks.

[0011] The first aspect of this invention provides a multi-objective optimization scheduling method for energy and carbon coordination in zero-carbon industrial parks, comprising: Collect multi-source operation data of the zero-carbon park and construct a topology model of the park's integrated energy system and a model of equipment operation characteristics. Based on the topology model and equipment operation characteristic model of the park's integrated energy system, a dynamic carbon emission flow tracking model of the park is constructed. The dynamic carbon potential of each node is calculated recursively according to time period to obtain the dynamic carbon emission factor coupled with the system operation structure. With the optimization objectives of minimizing the economic cost of park operation, minimizing the total carbon emissions of the park, and maximizing the comprehensive benefits of energy performance contracting, and embedding the minimum guaranteed energy saving rate constraint, revenue sharing constraint and investment payback period constraint of energy performance contracting, a multi-objective optimization scheduling model for energy and carbon synergy is constructed. An adaptive weighting mechanism based on the fusion of entropy weighting and fuzzy inference is adopted. Based on the real-time electricity-carbon price ratio and the confidence level of renewable energy output prediction, the optimization objectives are dynamically weighted and aggregated into a single objective. For the energy-carbon coordinated multi-objective optimization scheduling model, a two-stage biblical optimization framework of day-ahead and intraday is adopted to handle the uncertainty of source load prediction. The Pareto optimal solution set is obtained by iterative solution through column and constraint generation algorithm, and the compromise optimal scheduling scheme is selected by approximating the ideal solution sorting method.

[0012] A second aspect of this invention provides a zero-carbon park energy-carbon coordinated multi-objective optimization scheduling system, comprising: The data acquisition module collects multi-source operational data from the zero-carbon park and constructs a topology model of the park's integrated energy system and a model of equipment operation characteristics. The dynamic carbon flow tracking module, based on the topology model and equipment operation characteristic model of the park's integrated energy system, constructs a dynamic carbon emission flow tracking model for the park, recursively calculates the dynamic carbon potential of each node according to time period, and obtains the dynamic carbon emission factor coupled with the system operation structure. The multi-objective modeling module aims to minimize the economic cost of park operation, minimize the total carbon emissions of the park, and maximize the comprehensive benefits of energy performance contracting. It also incorporates the minimum guaranteed energy saving rate constraint, revenue sharing constraint, and investment payback period constraint of energy performance contracting to construct an energy-carbon coordinated multi-objective optimization scheduling model. The adaptive weighting module adopts an adaptive weighting mechanism based on the fusion of entropy weighting and fuzzy inference. It dynamically weights and aggregates each optimization objective according to the real-time electricity-carbon price ratio and the confidence level of renewable energy output prediction, thus aggregating multiple objectives into a single objective. The robust solution module employs a two-stage birch robust optimization framework (both day-ahead and intraday) to handle the uncertainty of source-load prediction for the energy-carbon coordinated multi-objective optimization scheduling model. It iteratively solves the problem using a column and constraint generation algorithm to obtain the Pareto optimal solution set, and then uses a sorting method to approximate the ideal solution to select the compromise optimal scheduling scheme.

[0013] The technical solution adopted in this invention has the following technical effects: 1. The technical solution of this invention constructs a topology model and equipment operation characteristic model of the integrated energy system of the park; based on the topology model and equipment operation characteristic model of the integrated energy system of the park, a dynamic carbon emission flow tracking model of the park is constructed, and the dynamic carbon potential of each node is recursively calculated according to the time period to obtain the dynamic carbon emission factor coupled with the system operation structure; a multi-objective optimization scheduling model for energy and carbon coordination is constructed; an adaptive weighting mechanism based on the fusion of entropy weighting method and fuzzy reasoning is adopted to dynamically weight and aggregate each optimization objective; for the multi-objective optimization scheduling model for energy and carbon coordination, a two-stage bibliometric optimization framework for day-ahead and intraday is used to handle the uncertainty of source load prediction, and the Pareto optimal solution set is obtained by iterative solution through column and constraint generation algorithm, and the optimal scheduling scheme is selected by approximating the ideal solution sorting method, which effectively solves the problem of low reliability of multi-objective optimization scheduling for energy and carbon coordination in zero-carbon parks caused by existing technology, and effectively improves the reliability of multi-objective optimization scheduling for energy and carbon coordination in zero-carbon parks.

[0014] 2. The technical solution of this invention adopts a dynamic carbon emission flow tracking model, which makes the carbon emission factor dynamically updated with the real-time operation structure changes of the system. Compared with the method of using a fixed carbon factor, this invention significantly reduces the carbon emission accounting error and can guide the scheduling to increase energy consumption during low-carbon periods and reduce electricity purchases during high-carbon periods, thereby reducing the carbon emissions of the park.

[0015] 3. The multi-objective collaboration in the technical solution of this invention is more flexible. The adaptive weighting mechanism that integrates entropy weight and fuzzy reasoning proposed in this invention can dynamically adjust the weights of each objective according to real-time electricity price, carbon price and renewable energy prediction confidence, so that the scheduling scheme can maintain overall optimality under different operating scenarios, thereby reducing the comprehensive energy cost of the park.

[0016] 4. The technical solution of this invention quantifies and embeds the guaranteed energy saving rate, revenue sharing and investment recovery period of contract energy management into the scheduling optimization kernel, so that scheduling optimization directly serves the reliable realization of energy saving benefits, ensures that the guaranteed energy saving rate and investment recovery requirements are met, solves the problem of the separation between technical optimization and commercial goals, and significantly improves the sustainability of contract energy management projects.

[0017] 5. This invention employs a two-stage bibliometric optimization framework to handle source load prediction uncertainties and uses a column and constraint generation algorithm for efficient solution. Compared with deterministic optimization methods, this invention can ensure the feasibility and economy of the scheduling scheme even when the source load output deviates significantly from the predicted value, significantly reducing the risk of scheduling plan failure, and achieving a good balance between conservatism and economy.

[0018] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the method in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the node carbon potential recursion process of the dynamic carbon emission flow tracking model in step S2 of embodiment one of the present invention. Figure 3 This is a schematic diagram of the structure of the energy-carbon coordinated multi-objective optimization scheduling model in step S3 of the method in Embodiment 1 of the present invention; Figure 4 This is a flowchart illustrating the fusion mechanism of adaptive weight entropy weight and fuzzy inference in step S4 of the method in embodiment one of the present invention. Figure 5 This is a schematic diagram of the iterative solution process of the two-stage sub-Bruker optimization column and constraint generation algorithm in step S5 of the method in embodiment one of the present invention. Figure 6 This is a schematic diagram comparing the park cost and carbon emissions before and after scheduling optimization in the method of Embodiment 1 of the present invention; Figure 7 This is a schematic diagram of the system structure in Embodiment 2 of the present invention. Detailed Implementation

[0021] To clearly illustrate the technical features of this solution, the invention will be described in detail below through specific embodiments and in conjunction with the accompanying drawings. The following disclosure provides many different embodiments or examples for implementing different structures of the invention. To simplify the disclosure of the invention, components and arrangements of specific examples are described below. Furthermore, reference numerals and / or letters may be repeated in different examples. This repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed. It should be noted that the components illustrated in the drawings are not necessarily drawn to scale. Descriptions of well-known components, processing techniques, and processes are omitted in this invention to avoid unnecessarily limiting the invention.

[0022] Example 1 like Figure 1 As shown, this invention provides a multi-objective optimization scheduling method for energy and carbon coordination in zero-carbon industrial parks, comprising: Step S1: Collect multi-source operation data of the zero-carbon park and construct a topology model and equipment operation characteristic model of the park's integrated energy system; Step S2: Based on the topology model and equipment operation characteristic model of the park's integrated energy system, construct a dynamic carbon emission flow tracking model for the park, recursively calculate the dynamic carbon potential of each node according to time period, and obtain the dynamic carbon emission factor coupled with the system operation structure. Step S3: With the optimization objectives of minimizing the economic cost of park operation, minimizing the total carbon emissions of the park, and maximizing the comprehensive benefits of energy performance contracting, and embedding the minimum guaranteed energy saving rate constraint, revenue sharing constraint and investment payback period constraint of energy performance contracting, a multi-objective optimization scheduling model for energy and carbon synergy is constructed. Step S4: An adaptive weighting mechanism based on the fusion of entropy weighting and fuzzy inference is adopted. Based on the real-time electricity-carbon price ratio and the confidence level of renewable energy output prediction, the optimization objectives are dynamically weighted and aggregated to aggregate multiple objectives into a single objective. Step S5: For the energy-carbon coordinated multi-objective optimization scheduling model, a two-stage biblical optimization framework of day-ahead and intraday is used to handle the uncertainty of source load prediction. The Pareto optimal solution set is obtained by iterative solution through column and constraint generation algorithm, and the optimal scheduling scheme is selected by approximating the ideal solution sorting method.

[0023] Preferably, the technical solution of the present invention also provides a zero-carbon park energy-carbon coordinated multi-objective optimization scheduling method, which further includes: Step S6: The optimal compromise scheduling scheme is parsed into energy storage charging and discharging commands, controllable unit output commands, and flexible load response commands, which are then sent to the park's equipment execution mechanism and continuously revised based on measured data throughout the day.

[0024] In step S1, multi-source data acquisition and system modeling are performed. Collect multi-source operational data on the zero-carbon park, including real-time output of distributed photovoltaic and wind power and corresponding meteorological data, operation data of combined heat and power units, state of charge of energy storage systems, electricity and heat consumption data of various loads, power exchange between the park and the external power grid, real-time carbon emission factors of purchased electricity, and real-time time-of-use electricity price and carbon price.

[0025] Based on the above data, a topology model and equipment operation characteristic model of the park's integrated energy system are constructed. The topology model is based on the nodes and branches of the park's power distribution network, depicting the connection relationships of various power sources, loads, energy storage, and conversion equipment. In the equipment operation characteristic model, the photovoltaic output model is represented as follows:

[0026] In the formula, To contribute to the output of photovoltaic modules; Photoelectric conversion efficiency; The area of ​​the photovoltaic module; Let be the irradiance during time period t; Let t be the ambient temperature during time period t; the coefficient 0.005 characterizes the rate of power degradation of the components as temperature increases. The energy storage state-of-charge model is expressed as follows:

[0027] In the formula, The energy storage state of charge for time period t+1; The energy storage state of charge during time period t; and These are the charging and discharging efficiencies, respectively. and These are the charging and discharging power, respectively. This refers to the rated capacity of the energy storage. The duration is denoted as . The combined heat and power (CHP) unit model is represented as:

[0028] In the formula, The power corresponding to the gas consumption during time period t; and These are power generation efficiency and heat production efficiency, respectively. and These refer to the generating capacity and heat output of the unit, respectively.

[0029] For industrial parks containing hydrogen energy storage systems, operational characteristic models for electrolyzers, hydrogen storage tanks, and fuel cells will be further established. The hydrogen production model for electrolyzers is represented as follows:

[0030] In the formula, This refers to the power consumption of the electrolytic cell; For the efficiency of hydrogen production by electrolysis; This represents the energy power corresponding to hydrogen production. The hydrogen energy storage model of the hydrogen storage tank is expressed as follows:

[0031] In the formula, This represents the hydrogen energy storage capacity in the hydrogen storage tank during the t+1 time period; The hydrogen energy storage capacity of the hydrogen storage tank during time period t; The hydrogen energy consumed by the fuel cell, For fuel cell power generation efficiency, Let be the hydrogen energy storage capacity in the hydrogen storage tank during time period t. The fuel cell power generation model is represented as follows: ; The power output of the fuel cell; For the power generation efficiency of fuel cells; The hydrogen energy consumed by the fuel cell.

[0032] In step S2, dynamic carbon emission flow tracking modeling is performed. Unlike existing technologies that use a fixed average carbon emission factor from the power grid, this step is based on carbon emission flow theory and power flow tracking to build a dynamic carbon emission flow tracking model for the park, so that the carbon emission intensity changes in real time with the system's operating structure.

[0033] Carbon emissions flow through the network along with electricity flow; carbon emissions generated on the generation side are attributed to the generated electricity and propagate to the load side via the power flow. To characterize this physical process, nodal carbon potential is defined. Carbon potential is the carbon emissions carried per unit of electricity flowing into node i, expressed in kgCO2 / kW. The carbon potential of a node is numerically equal to the actual carbon emission intensity of electricity used at that node.

[0034] For any node i in the park's power network, its nodal carbon potential can be solved according to the following recursive relationship:

[0035] in, Let i be the dynamic carbon potential of node i; Let i be the set of branches flowing into node i; This represents the active power flow from node j to node i in the branch. Let be the carbon potential of upstream node j; Let i be the set of power sources connected to node i. For the active power output of power source g, The carbon emission factor on the generation side of power source g; The power received by the node from the external power grid. The power grid carbon emission factor changes in real time with power grid dispatching; The discharge power of the energy storage node. This represents the potential for energy and carbon storage. When a node does not contain a certain type of component, the corresponding term is set to zero. This automatically degenerates into a regular node, ensuring model consistency.

[0036] Among them, energy storage and carbon storage potential The specific calculation method is as follows:

[0037] in, This represents the energy storage charge from the previous time period; This represents the carbon storage potential of the previous period. For charging power, For charging efficiency, The duration of the time period; This represents the dynamic carbon potential of the energy storage access node during that period; the current node's carbon potential is stored during charging and released during discharging, and the stored carbon potential is updated weighted according to the remaining electricity—thus achieving continuous tracking of carbon emissions across time periods in the energy storage process.

[0038] This recursive relationship indicates that the carbon potential of node i is equal to the sum of the carbon flow carried by all upstream branches flowing into that node and the carbon flow injected by the power source connected to that node, divided by the total active power flux flowing into that node. Physically, this reflects the conservation and mixing law of carbon emissions with the power flow. The recursion starts from the power source node and propagates step-by-step along the power flow direction to each load node, thus obtaining the dynamic carbon emission factor of each node at each time period, coupled with the actual operating structure of the system.

[0039] The dynamic carbon emission factor is numerically equal to the node carbon potential. Its time period Node location and the power flow distribution determined by scheduling With power output Real-time changes: Changes in energy structure cause fluctuations in carbon potential over time, differences in network location cause carbon potential distribution in space, and scheduling decisions change the flow direction, which drives the endogenous refresh of carbon potential. This achieves real-time coupling between carbon emission intensity and system operation structure, which is different from the existing technology that uses a fixed average carbon emission factor of the power grid.

[0040] To achieve efficient overall solution of the carbon potential of all nodes in the network, this invention further presents a matrix-based solution method. The active flux matrix of each node is defined. It is a diagonal matrix, where the diagonal elements are the sum of the active power flux flowing into each node; define the branch power flow distribution matrix. Its elements are the active power flow of the corresponding branch; define the generator injection distribution matrix. Vector of carbon emission intensity on the power generation side The carbon potential vector of all nodes in the network can then be solved globally using the following formula:

[0041] This matrix form is mathematically equivalent to the stepwise recursive form, but it has higher solution efficiency in large-scale network computing and is easier to embed into optimization models for simultaneous solution.

[0042] For the power exchanged with the external power grid, its carbon potential The carbon emission factor is determined by the real-time carbon emission factor of the external power grid and participates in the recursive calculation of the nodal carbon potential depending on whether the park purchases or sells electricity. When the park purchases electricity from the external power grid, the external power grid is considered an equivalent power source connected to the grid-connected node, and its carbon emission intensity is the real-time carbon emission factor of the external power grid. When the park sells electricity to the external power grid, the carbon potential of the sold electricity is determined by the nodal carbon potential of the grid-connected node. For photovoltaic and wind power, their generation-side carbon emission factor is taken as zero, and they are considered zero-carbon power sources. For gas-fired combined heat and power (CHP) units, their generation-side carbon emission factor is calculated based on the carbon emission intensity of gas combustion and the power generation efficiency, and is expressed as follows:

[0043] In the formula, This represents the carbon emission intensity per unit of fuel corresponding to the combustion of natural gas.

[0044] For parks containing hydrogen energy storage systems, the carbon potential of the electricity consumed in hydrogen electrolysis is determined according to the dynamic carbon potential of the electrolyzer access node and is recorded as the equivalent carbon potential of the stored hydrogen energy. When the fuel cell generates electricity, its carbon potential on the power generation side inherits the carbon potential of the electricity stored in the hydrogen production process, thereby realizing continuous tracking of carbon emissions in the energy storage process across time periods.

[0045] Definition of carbon emission factor on the power generation side:

[0046] In the formula, , , , It comprises four types of power sources: renewable energy, natural gas, grid energy, and hydrogen energy. Carbon emission intensity per unit of natural gas fuel. For the unit's power generation efficiency, Real-time carbon emission factor for external power grid This represents the dynamic carbon potential at the electrolyzer access node.

[0047] The carbon emission factors for all types of power sources in the park are determined according to a unified set of rules: renewable energy sources are set to zero; units containing fossil fuels are calculated based on fuel carbon emission intensity and power generation efficiency; external grid exchange power is assigned based on its real-time carbon emission factor, distinguishing between the direction of power purchase and sale; and energy storage conversion is transferred across time periods according to the principle of carbon potential inheritance, thus providing a unified framework for carbon emission factors for each energy type. In other words, all local primary energy sources are calculated based on their fuel carbon intensity; all external inputs are assigned based on the real-time carbon potential of their source; and all energy storage conversions are transferred according to the principle of inheritance. Therefore, the total real-time carbon emissions of the park during time period t are:

[0048] In the formula, Let be the load power of node i during time period t. Let represent the dynamic carbon potential of node i during time period t. The power corresponding to the direct consumption of natural gas. This represents the carbon emission factor for natural gas. This real-time total carbon emission will serve as the basis for calculating subsequent carbon emission optimization targets and carbon emission quota constraints.

[0049] In step S3, the energy-carbon coordinated multi-objective optimization scheduling model is constructed, which includes a multi-objective optimization scheduling model with three optimization objectives.

[0050] The first objective is to minimize the economic cost of operation, expressed as

[0051] In the formula, To determine the economic cost of operation; T1 represents the time-of-use electricity price for period t; T1 represents the total number of periods. This refers to the power exchanged between the park and the external power grid. For natural gas prices; This refers to the power generation capacity of the gas turbine unit. This refers to the unit price for equipment or unit operation and maintenance; The output of the k-th type of equipment or unit during time period t; This refers to the depreciation unit price for energy storage. These are the charging and discharging power of the energy storage, respectively. The second objective is to minimize total carbon emissions, expressed as

[0052] in, Total carbon emissions; This represents the total real-time carbon emissions of the park during time period t. Let be the load power of node i during time period t. Let represent the dynamic carbon potential of node i during time period t. The power corresponding to the direct consumption of natural gas. Carbon emission factors of natural gas; The objective is calculated using the dynamic carbon potential obtained in step S2, so that carbon emission optimization can accurately reflect the real-time operating structure of the system and guide the scheduling to increase energy consumption during low-carbon periods and reduce electricity purchases during high-carbon periods.

[0053] The third objective is to maximize the overall benefits of energy performance contracting. To unify the direction of this objective, it is transformed into a minimization form by taking its negative value. The original objective is expressed as follows:

[0054] In the formula, The baseline energy consumption was determined based on the energy consumption during the same period before the renovation. The actual energy consumption determined by the scheduling scheme. As a coefficient for sharing energy-saving benefits, The price is based on a comprehensive energy consumption factor. Assuming dynamic fluctuation based on operational scenarios, the energy-saving benefit sharing coefficient is dynamically adjusted from the baseline sharing ratio according to the actual energy-saving rate and carbon emission reduction contribution:

[0055] In the formula: The base profit-sharing coefficient stipulated in the contract; for Actual energy saving rate during the time period; for The contribution rate of carbon emission reduction during a given period, relative to the baseline carbon emission reduction; , This serves as an incentive adjustment factor for energy conservation and carbon reduction. The greater the energy conservation and the more significant the carbon reduction, the higher the profit-sharing coefficient. In other words, the profit-sharing ratio dynamically increases with the actual operational results. This not only aligns with the business logic of energy performance contracting—more energy conservation, more profit-sharing—but also incorporates carbon emission reduction into the profit-sharing incentive, achieving energy-carbon synergy.

[0056] in, The percentage decrease in carbon emission reduction contribution rate relative to baseline carbon emissions over a given period Determine using the following formula:

[0057] In the formula: This represents the baseline carbon emissions for the same period before the upgrade. The actual carbon emissions determined by the scheduling scheme are calculated according to the dynamic carbon emission factor in this scheme. Actual power consumption at each node The product sums are obtained, i.e. Actual carbon emissions are calculated using dynamic carbon potential, and the emission reduction rate is obtained by comparing it with a benchmark, thus directly coupling it with the aforementioned dynamic carbon potential model.

[0058] Preferably, , It does not need to be set to a fixed value; it can be dynamically allocated according to the price ratio of electricity to carbon, reflecting the principle of incentivizing whichever is more valuable.

[0059] In the formula Let t be the electricity price for period t; Let be the carbon price during time period t; The incentive adjustment factor for energy saving during time period t; This is the incentive adjustment factor for carbon reduction during time period t. When electricity prices are high, the focus is on energy-saving incentives; when carbon prices are high, the focus is on carbon reduction incentives. Ensure the overall incentive intensity remains stable to achieve dynamic synergy between energy conservation and carbon reduction.

[0060] The constraints of the energy-carbon coordinated multi-objective optimization scheduling model include: The power balance constraint requires that the power supply and demand in the park be balanced in each time period, that is, the sum of photovoltaic, wind power, combined heat and power, energy storage discharge, and purchased power equals the sum of various electrical loads, energy storage charging and electric heating, and the power consumption of electrolytic cells. The thermal power balance constraint requires that the supply and demand of thermal power in the park be balanced in each period, that is, the sum of the heat generated by cogeneration and the heat generated by electricity equals the demand for various types of heat loads.

[0061] Equipment operation constraints include upper and lower limits for the output of each piece of equipment, ramp-up rate constraints for the unit, upper and lower limits for the state of charge (SBC) of energy storage, and constraints ensuring the equality of the first and last states of energy storage. The SBC constraint is expressed as follows:

[0062] in, This represents the lower limit of the state of charge of energy storage. This represents the upper limit of the energy storage state of charge. This represents the initial value of the energy storage's state of charge. This is the final value of the energy storage state of charge.

[0063] The unit ramp rate constraint is expressed as

[0064] In the formula, and These are the limits for the unit's uphill and downhill ramp rates, respectively. The output of the k-th type of equipment or unit during time period t; This refers to the output of the k-th type of equipment or unit during time period t+1.

[0065] The key innovation of this invention lies in the revenue constraints of energy performance contracting, which include guaranteed minimum energy saving rate constraints, revenue sharing constraints, and investment payback period constraints. The guaranteed minimum energy saving rate constraint requires that the actual energy saving rate of the park in each time period or within the assessment period stipulated in the contract not be lower than the contractually agreed value. , represented as

[0066] in, The baseline energy consumption was determined based on the energy consumption during the same period before the renovation. The actual energy consumption is determined by the scheduling scheme.

[0067] Revenue sharing constraints characterize the relationship between the energy service provider's revenue and the actual energy savings, expressed as:

[0068] in, The revenue of the energy-saving service provider during time period t. The baseline energy consumption was determined based on the energy consumption during the same period before the renovation. The actual energy consumption determined by the scheduling scheme. The energy-saving benefits are divided into constraint coefficients. This refers to the comprehensive energy price.

[0069] The payback period constraint requires that the net present value (NPV) income of the energy service provider within the agreed payback period be no less than its investment amount. , represented as

[0070] in, t represents the total agreed payback period; t represents the agreed initial period; and r represents the discount rate.

[0071] Carbon emission quota constraints require the park to have a cumulative carbon emission limit during the scheduling cycle. Not exceeding its carbon emission quota , represented as

[0072] The key innovation of this invention lies in transforming the guaranteed energy saving rate, revenue sharing, and investment recovery period of contract energy management into mathematical constraints and objective function terms in the optimization model. This allows scheduling optimization to directly serve the realization of business objectives at the technical level, thereby solving the problem of the separation between technical optimization and business objectives.

[0073] In step S4, adaptive weighted multi-objective aggregation is performed. To transform the multi-objective optimization problem into a single-objective problem that can be solved efficiently, this step linearly aggregates the three optimization objectives into a single objective:

[0074] In the formula, , , These are the normalized values ​​for each objective. , , The adaptive weights for each objective are defined. Normalization uses range transformation to map the dimensions of each objective to the interval between zero and one, expressed as follows:

[0075] In the formula, and These are the minimum and maximum values ​​of the q-th objective in the solution set, respectively.

[0076] The weights are determined using a mechanism that combines entropy weighting and fuzzy reasoning. First, the objective weights are calculated based on the historical fluctuation data of each target using entropy weighting. For the q-th target, its information entropy is...

[0077] In the formula, For the first The normalized value of the q-th target in a given historical sample represents the proportion of that target's value in all samples, where m is the number of historical samples. A higher information entropy indicates smaller historical fluctuations and less discrepancy information carried by the target, thus it should be assigned a smaller weight. Therefore, the objective weight is...

[0078] Secondly, fuzzy reasoning is used to correct the objective weights in real time. This is based on the real-time electricity-carbon price ratio. confidence level of renewable energy output forecast For fuzzy inference input, where Let t be the carbon price at time period t. The two types of inputs are fuzzified into low, medium, and high-level fuzzy linguistic variables, and a fuzzy rule base is established. The fuzzy rule base includes the following typical rules: if the electricity-to-carbon price ratio is high and the prediction confidence is high, the weight of the economic objective is increased; if the carbon price is relatively high, the weight of the carbon emission objective is increased; if the investment payback period is near and the cumulative energy saving rate is low, the weight of the contract energy management revenue objective is increased. After fuzzification, fuzzy rule reasoning, and centroid method defuzzification, the weight adjustment factors for each objective are output. .

[0079] Finally, the objective weights and the weight adjustment factor are combined to obtain the adaptive weights:

[0080] The resulting weights not only reflect the historical objective differences of each objective, but can also be dynamically adjusted based on real-time price signals and resource conditions, thus ensuring that the scheduling scheme remains optimal overall under different operating scenarios.

[0081] In step S5, the two-stage Bruker bar optimization solution is performed. To address the uncertainty in source load forecasting, this step employs a two-stage sub-Bruker optimization framework, with day-ahead and intraday phases, and the model has a three-layer structure of min-max-min:

[0082] In the formula, x is the decision variable determined at this moment in the first stage, including the start-up and shutdown status and planned output of energy storage and controllable units; y is the waiting observation decision variable that is rolled over during the day based on actual observations in the second stage; P is the probability distribution of uncertain parameters such as photovoltaic, wind power and load. It is a set of uncertainties in the probability distribution; To achieve decision x in the first stage with uncertainty The feasible domain for the next phase of decision-making; and Let T be the cost coefficient vectors for the first and second stage decisions, respectively; T is the vector transpose. The meaning of this structure is: among all possible probability distributions, find the worst-case distribution that maximizes the expected correction cost of the second stage, and under this worst-case distribution, minimize the total cost of the first and second stages, thereby ensuring that the scheduling scheme is still feasible and economical in the worst-case scenario.

[0083] Uncertainty set This is achieved by constructing discrete scenarios and their probability distributions based on historical data. First, clustering methods are used to summarize historical source load data into several typical scenarios, and the frequency of occurrence of each scenario is used as the empirical probability. Second, norm 1 and norm constraints are applied to the probability distribution to limit its deviation from the empirical distribution.

[0084] In the formula, Let be the probability to be optimized for scenario s. Let be the empirical probability of scenario s. and These are the deviation limits corresponding to the 1-norm and the infinite norm, respectively, both determined by a given confidence level. The 1-norm constraint controls the degree to which the overall probability distribution deviates from the empirical distribution, while the infinite norm constraint controls the maximum deviation of the probability in a single scenario. Combining the two can achieve a balance between conservatism and economy.

[0085] The column and constraint generation algorithm is used to solve the above three-level optimization problem. This algorithm decomposes the original problem into a main problem and subproblems, which are solved iteratively and alternately. The main problem solves the first-stage decision under a given finite set of adverse distributions and provides a lower bound for the original problem; the subproblems, under the first-stage decision given by the main problem, solve for the uncertainty set... The internal search identifies the worst-case probability distribution with the highest correction cost in the second stage and provides an upper bound for the original problem. Each iteration adds the second-stage variables and constraints corresponding to the worst-case distributions identified in the subproblems to the main problem, thus gradually tightening the main problem. Iteration continues until the difference between the lower bound of the main problem and the upper bound of the subproblems converges to a given tolerance, at which point the solution terminates. The mathematical form of the main problem can be expressed as:

[0086] In the formula, As an auxiliary variable characterizing the expected adjustment cost in the second stage, The worst distribution identified in the l-th iteration. Let L be the second-stage decision variable in the corresponding scenario, and L be the current iteration number.

[0087] To generate a Pareto optimal solution set, this step involves solving the multi-objective problem multiple times by changing the combination of objective weights or using a constraint method. The constraint method transforms one or two objectives into constraints, uniformly selecting several levels within their value range, and solving the corresponding single-objective optimization problem one by one, thus obtaining a set of mutually independent Pareto optimal solutions.

[0088] Step S6: Decision on the optimal compromise solution and issuance of instructions For the obtained Pareto optimal solution set, this step uses the approximation ideal solution ranking method for compromise decision-making. First, a weighted normalized decision matrix is ​​constructed, normalizing the values ​​of each solution for each objective and multiplying them by the corresponding weights. Second, positive and negative ideal solutions are determined; positive ideal solutions consist of the optimal values ​​of each objective, while negative ideal solutions consist of the worst values ​​of each objective. Then, the weighted Euclidean distance from each solution to the positive ideal solution is calculated. Weighted Euclidean distance to the negative ideal solution :

[0089] In the formula, For the first The weighted normalized value of a solution on the q-th objective. and Let the positive and negative ideal values ​​be the values ​​of the q-th target, respectively; then calculate the relative closeness.

[0090] The greater the relative proximity, the closer the solution is to the positive ideal solution and the farther it is from the negative ideal solution. Therefore, the solution with the greatest relative proximity is selected as the optimal scheduling scheme.

[0091] Finally, the optimal compromise solution is analyzed into energy storage charging and discharging power commands, controllable unit output commands, and flexible load response commands, which are then issued to the park's equipment execution mechanisms. During intraday operation, based on measured photovoltaic output, wind power output, and load data, the second-stage decision is rolled over according to a set rolling cycle, thereby achieving coordination between the day-ahead plan and intraday adjustments.

[0092] To more clearly illustrate the technical solution of this invention, a specific example is provided below: This embodiment uses a zero-carbon industrial park as an example to describe in detail the implementation process of the method of the present invention. The park is equipped with a distributed photovoltaic installed capacity of 2 MW, a wind power installed capacity of 1 MW, a lithium iron phosphate energy storage capacity of 1 MWh with a rated power of 0.5 MW, a gas-fired combined heat and power unit capacity of 1 MW, and an electric heating equipment capacity of 500 kW. The park's average daily electricity consumption is approximately 25 MWh, and its average daily heat consumption is approximately 15 MWh, and it is connected to the external power grid via a 35 kV line. The park has implemented an energy-saving renovation project under a contract energy management system, with a guaranteed minimum energy saving rate of 15%, an energy saving benefit sharing coefficient of 0.6, an investment of 5 million yuan in energy-saving renovation, and a contracted investment payback period of 5 years.

[0093] Reference Figure 1 The scheduling optimization process in this embodiment is as follows.

[0094] In step S1, the aforementioned multi-source operational data is collected at a time resolution of 15 minutes through the data acquisition and monitoring control system of the park's energy management system and smart meters. Typical daily data collected includes: photovoltaic output curve, with a peak of approximately 1.6 MW occurring at noon; wind power output curve; load curves of each feeder; energy storage state-of-charge curve; real-time carbon emission factor of purchased electricity, approximately 0.85 kg CO2 / kWh at night and approximately 0.45 kg CO2 / kWh at midday; time-of-use electricity price, with peak hours at RMB 1.2 per kWh, normal hours at RMB 0.7 per kWh, and off-peak hours at RMB 0.35 per kWh; and carbon price, approximately RMB 80 per ton. Based on the above data, a topology model and equipment operation characteristic model of the park's integrated energy system are established, with the scheduling cycle set to 24 hours and divided into 96 time periods.

[0095] In step S2, refer to Figure 2 A dynamic carbon emission flow tracking model for the industrial park was constructed. Using the low-voltage busbar of the park's main transformer as the root node, the carbon potential of purchased electricity was taken from the real-time carbon emission factor of the external power grid. The carbon potential of photovoltaic and wind power was set to zero, considered as zero-carbon power sources. The carbon potential of gas-fired combined heat and power units was calculated as follows: Calculation. The dynamic carbon potential of each load node is calculated level by level along the power flow direction of the distribution network, based on the recursive relationship of node carbon potential. To improve calculation efficiency, this embodiment uses a matrix approach to solve the overall node carbon potential vector of the entire network. Calculation results show that during the noon period (12:00 noon), photovoltaic output is high, the self-consumption ratio of the park is high, and the amount of purchased electricity is low, with the average carbon potential of the park being approximately 0.32 kgCO2 / kWh. However, during the nighttime period (20:00), photovoltaic output is zero, and the park purchases a large amount of electricity, with the average carbon potential rising to approximately 0.78 kgCO2 / kWh. This demonstrates that dynamic carbon potential can accurately reflect the differences in carbon emission intensity at different times within the park, while traditional fixed carbon factors, such as 0.58 kgCO2 / kWh, cannot reflect this time-period difference, thus generating significant errors in low-carbon dispatching.

[0096] In step S3, refer to Figure 3 A multi-objective optimization scheduling model for energy and carbon coordination in the industrial park was constructed. The three optimization objectives are listed as minimizing operating economic costs, minimizing total carbon emissions, and maximizing the overall benefits of energy performance contracting. The benchmark energy consumption for energy performance contracting is specified in the model. The actual energy consumption was determined based on the energy consumption data from the same period before the renovation. The constraints are determined by the dispatching plan. These constraints include power balance constraints, thermal power balance constraints, upper and lower limits of equipment output constraints, unit ramp-up rate constraints, energy storage state of charge constraints, guaranteed minimum energy saving rate constraints under energy performance contracting, revenue sharing constraints, investment payback period constraints, and carbon emission quota constraints. Specifically, the energy storage state of charge must be maintained between 20% and 90%, and the initial and final states of charge must be equal; the daily energy saving rate under energy performance contracting must be no less than 15%.

[0097] In step S4, refer to Figure 4 Adaptive weight calculation is performed. Using the entropy weight method based on historical scheduling data from the past 30 days, objective weights are calculated, resulting in an objective weight of 0.40 for the economic target, 0.35 for the carbon emission target, and 0.25 for the contract energy management revenue target. Fuzzy inference uses the current electricity-carbon price ratio and the photovoltaic output prediction confidence level as inputs; in this embodiment, the photovoltaic output prediction confidence level is 0.88. Since this embodiment is in the third year of the contract energy management investment payback period, and the cumulative energy-saving benefits are slightly lower than expected, the fuzzy rules trigger a rule to increase the weight of the contract energy management revenue target, outputting weight adjustment factors of 0.95, 1.00, and 1.20, respectively. After fusion calculation, adaptive weights are obtained: an economic target weight of 0.37, a carbon emission target weight of 0.34, and a contract energy management revenue target weight of 0.29.

[0098] In step S5, refer to Figure 5 A two-stage Pareto optimal solution was performed. Based on historical data, K-means clustering was used to generate 10 typical scenarios and their empirical probabilities, constructing a probability distribution uncertainty set with a first-norm deviation limit of 0.05 and an infinite-norm deviation limit of 0.02. A column and constraint generation algorithm was used for iterative solution. Initially, the main problem only included empirically distributed scenarios, yielding the first-stage decision. Under this first-stage decision, the subproblems searched within the uncertainty set for the worst-case probability distribution that maximized the intraday correction cost, and the variables and constraints corresponding to this worst-case distribution were added to the main problem. After 5 iterations, the lower bound of the main problem converged to 48,200 yuan, and the upper bound of the subproblems converged to 48,900 yuan, with a convergence tolerance of less than 2%, at which point the solution terminated. Subsequently, the constraint method was used to solve the multi-objective problem multiple times, obtaining a Pareto optimal solution set containing 20 mutually non-dominant solutions.

[0099] In step S6, a compromise decision is made on the Pareto optimal solution set using the approximation ideal solution sorting method. The relative proximity of each solution is calculated, and the solution with the largest relative proximity is selected as the compromise optimal scheduling scheme. In this embodiment, the relative proximity of this solution is 0.83. The main scheduling strategy of this scheme is as follows: during off-peak hours, i.e., the early morning hours, energy storage is charged and electric heating and heat storage are moderately increased; during the midday peak photovoltaic power generation period, priority is given to realizing self-consumption of photovoltaic power generation and supplementing charging of energy storage; during peak hours and the high carbon period at night, energy storage is discharged, and the cogeneration unit operates at peak speed to reduce the purchase of external electricity during the high carbon period. This compromise optimal scheme is parsed into power commands for each device and issued to the execution mechanism, and is rolled over every 15 minutes during the day based on the measured photovoltaic output and load data.

[0100] Reference Figure 6After applying the method of this invention, compared with the traditional scheduling method that focuses solely on minimizing operating costs, the typical daily operating cost of the park in this embodiment decreased from RMB 53,100 to RMB 48,200, a reduction of 9.2%; carbon emissions decreased from 12.6 tons to 10.3 tons, a reduction of 18.3%; simultaneously, the daily energy saving rate under the contract energy management system reached 17.2%, meeting the minimum guaranteed energy saving rate requirement, and shortening the investment payback period by approximately 0.4 years compared to before optimization. These results verify the significant effects of this invention in terms of economic efficiency, low carbon emissions, and guaranteeing the benefits of contract energy management.

[0101] To further illustrate the implementation details of each step of this invention, the following supplementary explanation is provided regarding the matrix solution process for dynamic carbon emission flow tracking in step S2. Assume the power grid in the industrial park has n nodes. For each time period, the branch power flow distribution matrix and generator injection distribution matrix are determined based on the power flow calculation results for that time period. The diagonal element of the node active power flux matrix represents the sum of the active power flux flowing into the corresponding node, including the power flowing into the upstream branch and the power injected by the local power source. After obtaining these three matrices, the dynamic carbon potential of each node in the entire network for that time period is calculated in one step using the matrix solution formula. Combined with the load power of each node for that time period, the total real-time carbon emissions of the industrial park for that time period can be obtained. By performing the above calculations for each of the 96 time periods, the dynamic carbon potential sequence and carbon emission sequence for the entire scheduling cycle can be obtained. This sequence is used for both the calculation of carbon emission targets and the verification of carbon emission quota constraints, and also provides historical information on carbon emission fluctuations for the adaptive weighting module.

[0102] To further explain the implementation details of fuzzy inference in step S4, the following supplementary explanations are provided regarding the design and defuzzification process of the fuzzy rules. The electricity-carbon price ratio is divided into three fuzzy sets: low, medium, and high. Similarly, the confidence level of renewable energy output prediction is divided into three fuzzy sets, thus forming several fuzzy rules. The design of the fuzzy rules is based on the following economic and low-carbon logic: When the electricity-carbon price ratio is high, it indicates that the electricity price is relatively high compared to the carbon price, and the contribution of saving electricity purchase costs to the overall goal is more prominent; therefore, the weight of the economic goal should be appropriately increased. When the carbon price is relatively high, i.e., the electricity-carbon price ratio is low, carbon emission costs increase, and the weight of the carbon emission goal should be increased. When the park is nearing the contract investment recovery period and the cumulative energy saving rate is low, the urgency of realizing energy-saving benefits increases, and the weight of the contract energy management revenue goal should be increased. Fuzzy inference uses a composition operation that takes the smallest value and takes the largest value, and defuzzification uses the centroid method to output the weight adjustment factors for each goal. This mechanism allows the weights to transition smoothly with operating conditions, avoiding scheduling abrupt changes caused by fixed weights during scenario switching.

[0103] To further illustrate the convergence characteristics of the column and constraint generation algorithm in step S5, the following supplementary explanation is provided. In each iteration, the main problem is a mixed-integer linear programming problem, the size of which increases with the number of iterations; the subproblem is a maximization problem on a polyhedral uncertainty set, which can be transformed into a mixed-integer linear programming problem through dual transformation and the Big M method. Since the uncertainty set is the intersection of the probabilistic simplex and norm constraints on a finite discrete scenario, the optimal worst-case distribution of the subproblem must be obtained at a finite number of extreme points in this set. Therefore, the algorithm converges within a finite number of iterations. In this embodiment, the algorithm converges after five iterations, demonstrating the efficiency of this solution framework.

[0104] Preferably, this embodiment applies the method of the present invention to a comprehensive energy park containing hydrogen energy storage. Based on the existing equipment configuration, the park adds a hydrogen energy storage system consisting of an electrolyzer, a hydrogen storage tank, and a fuel cell, to achieve long-term, cross-period energy transfer and multi-energy coupling.

[0105] In this embodiment, hydrogen energy storage participates in cross-period energy dispatch as a long-cycle energy storage system. In the dynamic carbon emission tracking stage, the carbon potential of the electricity consumed by the electrolyzer for hydrogen production is determined according to the dynamic carbon potential of the electrolyzer's access node during the hydrogen production period and is recorded as the equivalent carbon potential of the stored hydrogen energy. When the fuel cell generates electricity in subsequent periods, its power generation side inherits the carbon potential of the electricity stored in the hydrogen production stage. Therefore, this invention achieves continuous tracking of carbon emissions across multiple time periods and stages in the hydrogen production, storage, and power generation process, enabling the carbon emission reduction contribution of hydrogen energy storage to be accurately quantified and incorporated into optimization objectives.

[0106] Simulation results show that, guided by the method of this invention, electrolyzers tend to produce and store hydrogen during low-carbon, low-cost periods, while fuel cells tend to generate electricity during high-carbon, high-cost periods, thus transferring surplus clean electricity from low-carbon periods to high-carbon periods. Compared to Example 1 without hydrogen storage, this example further reduces the park's carbon emissions by approximately five percentage points, while significantly reducing the demand for purchased electricity during high-carbon periods. These results verify the good adaptability of the method of this invention to various energy storage forms, including hydrogen storage, and multi-energy coupled systems, demonstrating that the dynamic carbon emission flow tracking model and energy-carbon collaborative multi-objective optimization scheduling framework proposed in this invention have strong versatility and scalability.

[0107] This invention constructs a topology model and equipment operation characteristic model of an integrated energy system in a park; based on the topology model and equipment operation characteristic model, it constructs a dynamic carbon emission flow tracking model for the park, recursively calculates the dynamic carbon potential of each node according to time periods, and obtains the dynamic carbon emission factor coupled with the system operation structure; it constructs an energy-carbon coordinated multi-objective optimization scheduling model; it adopts an adaptive weighting mechanism based on the fusion of entropy weighting and fuzzy reasoning to dynamically weight and aggregate each optimization objective; for the energy-carbon coordinated multi-objective optimization scheduling model, it uses a two-stage bibloc optimization framework of day-ahead and intraday to handle the uncertainty of source load prediction, it iteratively solves the problem by using a column and constraint generation algorithm to obtain the Pareto optimal solution set, and uses the approximation ideal solution sorting method to select the compromise optimal scheduling scheme, effectively solving the problem of low reliability of energy-carbon coordinated multi-objective optimization scheduling in zero-carbon parks caused by existing technologies, and effectively improving the reliability of energy-carbon coordinated multi-objective optimization scheduling in zero-carbon parks.

[0108] The technical solution of this invention adopts a dynamic carbon emission flow tracking model, which makes the carbon emission factor dynamically updated as the system's real-time operating structure changes. Compared with the method of using a fixed carbon factor, this invention significantly reduces the carbon emission accounting error and can guide the scheduling to increase energy consumption during low-carbon periods and reduce electricity purchases during high-carbon periods, thereby reducing the carbon emissions of the park.

[0109] The multi-objective collaboration in the technical solution of this invention is more flexible. The adaptive weighting mechanism that integrates entropy weight and fuzzy reasoning proposed in this invention can dynamically adjust the weights of each objective according to real-time electricity price, carbon price and renewable energy prediction confidence, so that the scheduling scheme can maintain overall optimality under different operating scenarios, thereby reducing the comprehensive energy cost of the park.

[0110] The technical solution of this invention quantifies and embeds commercial terms such as guaranteed energy saving rate, revenue sharing and investment recovery period of energy performance contracting into the scheduling optimization kernel, so that scheduling optimization directly serves the reliable realization of energy saving benefits, ensures that the guaranteed energy saving rate and investment recovery requirements are met, solves the problem of the separation between technical optimization and commercial goals, and significantly improves the sustainability of energy performance contracting projects.

[0111] This invention employs a two-stage bibliometric optimization framework to handle source load prediction uncertainties and uses a column and constraint generation algorithm for efficient solution. Compared with deterministic optimization methods, this invention can ensure the feasibility and economy of the scheduling scheme even when the source load output deviates significantly from the predicted value, significantly reducing the risk of scheduling plan failure, and achieving a good balance between conservatism and economy.

[0112] Example 2 like Figure 7 As shown, the present invention also provides a zero-carbon park energy-carbon coordinated multi-objective optimization scheduling system, comprising: The data acquisition module collects multi-source operational data from the zero-carbon park and constructs a topology model of the park's integrated energy system and a model of equipment operation characteristics. The dynamic carbon flow tracking module, based on the topology model and equipment operation characteristic model of the park's integrated energy system, constructs a dynamic carbon emission flow tracking model for the park, recursively calculates the dynamic carbon potential of each node according to time period, and obtains the dynamic carbon emission factor coupled with the system operation structure. The multi-objective modeling module aims to minimize the economic cost of park operation, minimize the total carbon emissions of the park, and maximize the comprehensive benefits of energy performance contracting. It also incorporates the minimum guaranteed energy saving rate constraint, revenue sharing constraint, and investment payback period constraint of energy performance contracting to construct an energy-carbon coordinated multi-objective optimization scheduling model. The adaptive weighting module adopts an adaptive weighting mechanism based on the fusion of entropy weighting and fuzzy inference. It dynamically weights and aggregates each optimization objective according to the real-time electricity-carbon price ratio and the confidence level of renewable energy output prediction, thus aggregating multiple objectives into a single objective. The robust solution module employs a two-stage birch robust optimization framework (both day-ahead and intraday) to handle the uncertainty of source-load prediction for the energy-carbon coordinated multi-objective optimization scheduling model. It iteratively solves the problem using a column and constraint generation algorithm to obtain the Pareto optimal solution set, and then uses a sorting method to approximate the ideal solution to select the compromise optimal scheduling scheme.

[0113] The data acquisition module is used to collect multi-source operation data of the zero-carbon park, including sources, grid, load, storage, and carbon, and to construct a topology model and equipment operation characteristic model of the park's integrated energy system, corresponding to step S1 of the execution method. The dynamic carbon flow tracking module is used to construct a dynamic carbon emission flow tracking model based on carbon emission flow theory and real-time power flow distribution, recursively solving for the dynamic carbon potential of each node in time-period or in matrix form, corresponding to step S2 of the execution method. The multi-objective modeling module is used to construct an energy-carbon coordinated multi-objective optimization scheduling model with the objectives of minimizing operating economic costs, minimizing total carbon emissions, and maximizing the comprehensive benefits of energy performance contracting, while embedding energy performance contracting benefit constraints, corresponding to step S3 of the execution method. The adaptive weighting module is used to calculate the adaptive weights of each objective using an entropy weighting method and a fuzzy inference fusion mechanism, and to aggregate multiple objectives into a single objective, corresponding to step S4 of the execution method. The robust solution module is used to solve the model using a two-stage bibliometric optimization framework and a column and constraint generation algorithm, and to select the optimal scheduling scheme using a sorting method that approximates the ideal solution, corresponding to step S5 of the execution method. The instruction issuing module is used to parse the optimal compromise solution into energy storage charging and discharging instructions, controllable unit output instructions, and flexible load response instructions, and issue them for execution, corresponding to step S6 of the execution method.

[0114] The modules described above are sequentially connected via data interfaces, with the output of one module serving as the input of the next, thus forming a complete closed loop for coordinated energy and carbon optimization scheduling. This system can be deployed on the park's energy management platform, interfacing with data acquisition and monitoring control systems, smart meters, and various equipment actuators to achieve automated operation of daily planning and rolling adjustments.

[0115] This invention constructs a topology model and equipment operation characteristic model of an integrated energy system in a park. Based on these models, a dynamic carbon emission flow tracking model is built, recursively calculating the dynamic carbon potential of each node over time periods to obtain a dynamic carbon emission factor coupled with the system's operational structure. A multi-objective optimization scheduling model for energy and carbon coordination is constructed. An adaptive weighting mechanism based on entropy weighting and fuzzy reasoning is used to dynamically weight and aggregate each optimization objective. For the multi-objective optimization scheduling model, a two-stage bibloc optimization framework (day-ahead and intraday) is used to handle source-load prediction uncertainties. An iterative solution using a column and constraint generation algorithm is employed to obtain a Pareto optimal solution set. Finally, a sorting method for approximating ideal solutions is used to select the optimal scheduling scheme. This effectively solves the problem of low reliability in multi-objective optimization scheduling of energy and carbon coordination in zero-carbon parks caused by existing technologies, and effectively improves the reliability of multi-objective optimization scheduling of energy and carbon coordination in zero-carbon parks.

[0116] The technical solution of this invention adopts a dynamic carbon emission flow tracking model, which makes the carbon emission factor dynamically updated as the system's real-time operating structure changes. Compared with the method of using a fixed carbon factor, this invention significantly reduces the carbon emission accounting error and can guide the scheduling to increase energy consumption during low-carbon periods and reduce electricity purchases during high-carbon periods, thereby reducing the carbon emissions of the park.

[0117] The multi-objective collaboration in the technical solution of this invention is more flexible. The adaptive weighting mechanism that integrates entropy weight and fuzzy reasoning proposed in this invention can dynamically adjust the weights of each objective according to real-time electricity price, carbon price and renewable energy prediction confidence, so that the scheduling scheme can maintain overall optimality under different operating scenarios, thereby reducing the comprehensive energy cost of the park by 8% to 15%.

[0118] The technical solution of this invention quantifies and embeds commercial terms such as guaranteed energy saving rate, revenue sharing and investment recovery period of energy performance contracting into the scheduling optimization kernel, so that scheduling optimization directly serves the reliable realization of energy saving benefits, ensures that the guaranteed energy saving rate and investment recovery requirements are met, solves the problem of the separation between technical optimization and commercial goals, and significantly improves the sustainability of energy performance contracting projects.

[0119] This invention employs a two-stage bibliometric optimization framework to handle source load prediction uncertainties and uses a column and constraint generation algorithm for efficient solution. Compared with deterministic optimization methods, this invention can ensure the feasibility and economy of the scheduling scheme even when the source load output deviates significantly from the predicted value, significantly reducing the risk of scheduling plan failure, and achieving a good balance between conservatism and economy.

[0120] Example 3 This embodiment provides an electronic device and a computer-readable storage medium. The electronic device includes a memory, a processor, and a computer program stored on the memory and executable on the processor. The memory may include random access memory or non-volatile memory, such as at least one disk storage device. The processor may be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits (ASICs), or field-programmable gate arrays (FPGAs). When the processor executes the computer program, it implements steps S1 to S6 of the method described in the foregoing embodiments.

[0121] The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the methods described in the foregoing embodiments. The computer-readable storage medium may be a read-only memory, random access memory, magnetic disk, optical disk, or other medium capable of storing program code. Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by hardware related to computer program instructions. The computer program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the above method embodiments.

[0122] It should be noted that the specific values ​​of park equipment configuration parameters, price parameters, carbon emission factor parameters, contract terms parameters, convergence tolerance, iteration number, etc., in the above embodiments are merely examples. Those skilled in the art can make corresponding adjustments based on the actual park's equipment composition, market environment, and contractual agreements without departing from the technical concept of this invention. For example, for a pure electric park without heat load, the heat power balance constraint and cogeneration unit model can be omitted; for a park with multiple parallel energy performance contracting projects, a minimum guaranteed energy saving rate constraint, revenue sharing constraint, and investment payback period constraint can be established for each project, and the revenue of each project can be weighted and summed in the objective function. All the above changes fall within the protection scope of this invention.

[0123] In summary, this invention achieves accurate carbon emission calculation and optimization through a dynamic carbon emission flow tracking model, realizes flexible multi-objective coordination through an adaptive weighting mechanism that integrates entropy weighting and fuzzy inference, achieves integrated technology and business coordination by quantifying and embedding the commercial terms of energy performance contracting into the scheduling model, and achieves robust scheduling under source-load uncertainty through a two-stage biblical optimization and column and constraint generation algorithm. This invention overcomes the shortcomings of existing technologies, such as distorted carbon emission calculation, rigid fixed weights, separation of technology and business, and insufficient robustness. While ensuring the reliable realization of energy-saving benefits from energy performance contracting, it significantly reduces the overall energy cost and carbon emission level of industrial parks, demonstrating good engineering application value and promising prospects for promotion.

[0124] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A multi-objective optimization scheduling method for energy and carbon coordination in zero-carbon industrial parks, characterized in that, include: Collect multi-source operation data of the zero-carbon park and construct a topology model of the park's integrated energy system and an equipment operation characteristic model; Based on the topology model and equipment operation characteristic model of the park's integrated energy system, a dynamic carbon emission flow tracking model of the park is constructed. The dynamic carbon potential of each node is calculated recursively according to time period to obtain the dynamic carbon emission factor coupled with the system operation structure. With the optimization objectives of minimizing the economic cost of park operation, minimizing the total carbon emissions of the park, and maximizing the comprehensive benefits of energy performance contracting, and embedding the minimum guaranteed energy saving rate constraint, revenue sharing constraint and investment payback period constraint of energy performance contracting, a multi-objective optimization scheduling model for energy and carbon synergy is constructed. An adaptive weighting mechanism based on the fusion of entropy weighting and fuzzy inference is adopted. Based on the real-time electricity-carbon price ratio and the confidence level of renewable energy output prediction, the optimization objectives are dynamically weighted and aggregated into a single objective. For the energy-carbon coordinated multi-objective optimization scheduling model, a two-stage biblical optimization framework of day-ahead and intraday is adopted to handle the uncertainty of source load prediction. The Pareto optimal solution set is obtained by iterative solution through column and constraint generation algorithm, and the compromise optimal scheduling scheme is selected by approximating the ideal solution sorting method.

2. The zero-carbon park energy-carbon coordinated multi-objective optimization scheduling method according to claim 1, characterized in that, Also includes: The optimal compromise scheduling scheme is parsed into energy storage charging and discharging commands, controllable unit output commands, and flexible load response commands, which are then sent to the park's equipment execution mechanisms and continuously revised based on measured data throughout the day.

3. The zero-carbon park energy-carbon coordinated multi-objective optimization scheduling method according to claim 1, characterized in that, The multi-source operational data includes real-time output of distributed photovoltaic and wind power and corresponding meteorological data, operational data of combined heat and power units, state of charge of energy storage systems, electricity and heat consumption data of various loads, exchange power between the park and the external power grid, real-time carbon emission factors of purchased electricity, and real-time time-of-use electricity prices and carbon prices. The park's integrated energy system topology model is based on the nodes and branches of the park's power distribution network, depicting the access relationships of various power sources, loads, energy storage, and conversion equipment. The equipment operation characteristic models include photovoltaic output models, energy storage state of charge models, combined heat and power unit models, electrolyzer hydrogen production models, hydrogen energy storage tank hydrogen energy inventory models, and fuel cell power generation models.

4. The zero-carbon park energy-carbon coordinated multi-objective optimization scheduling method according to claim 1, characterized in that, The specific method for calculating the dynamic carbon potential of a node is as follows: in, Let i be the dynamic carbon potential of node i; Let i be the set of branches flowing into node i; This represents the active power flow from node j to node i in the branch. Let be the carbon potential of upstream node j; Let i be the set of power sources connected to node i. For the active power output of power source g, The carbon emission factor on the generation side of power source g; The power received by the node from the external power grid. The power grid carbon emission factor changes in real time with power grid dispatching; The discharge power of the energy storage node. For energy storage and carbon storage potential.

5. The zero-carbon park energy-carbon coordinated multi-objective optimization scheduling method according to claim 4, characterized in that, Energy storage and carbon storage potential The specific calculation method is as follows: in, This represents the energy storage charge from the previous time period; This represents the carbon storage potential of the previous period. For charging power, For charging efficiency, The duration of the time period; This represents the dynamic carbon potential of the energy storage access node during this period. Carbon emission factor of power generation side of power source g The specific calculation method is as follows: In the formula, , , , These are renewable power generation, gas power, power purchased from the external power grid, and hydrogen power. Carbon emission intensity per unit of natural gas fuel. For the unit's power generation efficiency, Real-time carbon emission factor for external power grid This represents the dynamic carbon potential of the electrolyzer access node during this period.

6. The zero-carbon park energy-carbon coordinated multi-objective optimization scheduling method according to claim 1, characterized in that, The multi-objective optimization scheduling model for energy and carbon coordination includes minimizing operating economic costs, minimizing total carbon emissions, and maximizing the overall benefits of energy performance contracting; specifically, the operating economic costs are: In the formula, To determine the economic cost of operation; T1 represents the time-of-use electricity price for period t; T1 represents the total number of periods. This refers to the power exchanged between the park and the external power grid. For natural gas prices; This refers to the power generation capacity of the gas turbine unit. This refers to the unit price for equipment or unit operation and maintenance; For the output of the k-th type of equipment or unit during time period t, This refers to the depreciation unit price for energy storage. These are the charging and discharging power of the energy storage, respectively. The total carbon emissions are as follows: in, Total carbon emissions; This represents the total real-time carbon emissions of the park during time period t. Let be the load power of node i during time period t. Let represent the dynamic carbon potential of node i during time period t. The power corresponding to the direct consumption of natural gas. Carbon emission factors of natural gas; The comprehensive benefits of energy performance contracting are as follows: in, For comprehensive benefits from energy performance contracting; The baseline energy consumption was determined based on the energy consumption during the same period before the renovation. The actual energy consumption determined by the scheduling scheme. As a coefficient for sharing energy-saving benefits, This refers to the comprehensive energy price.

7. A zero-carbon park energy-carbon coordinated multi-objective optimization scheduling method according to claim 6, characterized in that, Energy saving benefit sharing coefficient Specifically: In the formula: The base profit-sharing coefficient stipulated in the contract; for Actual energy saving rate during the time period; for The carbon emission reduction contribution rate of the time period relative to the baseline carbon emission reduction rate; , These are the incentive adjustment factors for energy conservation and carbon reduction, respectively.

8. The zero-carbon park energy-carbon coordinated multi-objective optimization scheduling method according to claim 1, characterized in that, The constraints of the energy-carbon coordinated multi-objective optimization scheduling model include electrical power balance constraints, thermal power balance constraints, equipment operation constraints, and energy performance contracting revenue constraints. The energy performance contracting revenue constraints include guaranteed minimum energy saving rate constraints, revenue sharing constraints, investment payback period constraints, and carbon emission quota constraints. The guaranteed minimum energy saving rate constraint requires that the actual energy saving rate of the park in each time period or within the contractually agreed assessment period is not lower than the contractually agreed value. The investment payback period constraint requires that the net present value return of the energy service provider within the agreed payback period is not lower than its investment amount. The carbon emission quota constraint requires that the cumulative carbon emissions of the park within the scheduling period do not exceed its carbon emission quota. The revenue sharing constraint is the relationship between the energy service provider's revenue and the actual energy saving, specifically: in, The revenue of the energy-saving service provider during time period t. The baseline energy consumption was determined based on the energy consumption during the same period before the renovation. The actual energy consumption determined by the scheduling scheme. The energy-saving benefits are divided into constraint coefficients. This refers to the comprehensive energy price.

9. A zero-carbon park energy-carbon coordinated multi-objective optimization scheduling method according to claim 1, characterized in that, For the energy-carbon coordinated multi-objective optimization scheduling model, a two-stage birusk optimization framework (day-ahead and intraday) is adopted to handle the uncertainty of source-load prediction. The Pareto optimal solution set is obtained through iterative solution using a column and constraint generation algorithm. Employing a two-stage sub-Bruker optimization framework (day-ahead and intraday), the energy-carbon coordinated multi-objective optimization scheduling model is divided into a three-layer structure: In the formula, x is the decision variable for time period t determined in the first stage before the day, including the start-up and shutdown status and planned output of energy storage and controllable units; y is the waiting observation decision variable that is rolled over during the day based on actual observations in the second stage; P is the probability distribution of the uncertain parameter; It is a set of uncertainties in the probability distribution; To achieve decision x in the first stage with uncertainty The feasible domain for the next phase of decision-making; and Let T be the cost coefficient vectors for the first and second stage decisions, respectively; T is the vector transpose; among all possible probability distributions, find the worst-case distribution that maximizes the expected correction cost of the second stage, and minimize the total cost of the first and second stages under this worst-case distribution; The historical source load data is summarized into several typical scenarios using a clustering method, and the frequency of occurrence of each scenario is used as the empirical probability. Secondly, the probability distribution is constrained by the first norm and the infinite norm to limit its deviation from the empirical distribution. The column and constraint generation algorithm is used to solve the above three-level optimization problem; the original problem is decomposed into a main problem and sub-problems and solved iteratively; the main problem is solved for the first stage decision under a given finite set of adverse distributions, and the lower bound of the original problem is given; The subproblems, under the first-stage decision given by the main problem, are within a set of uncertainties. The internal search yields the worst-case probability distribution that maximizes the cost of the second-stage correction and provides an upper bound for the original problem. By using the constraint method to solve the multi-objective problem multiple times, a set of non-dominated Pareto optimal solutions can be obtained.

10. A zero-carbon park energy-carbon coordinated multi-objective optimization scheduling system, characterized in that, include: The data acquisition module collects multi-source operational data from the zero-carbon park and constructs a topology model of the park's integrated energy system and a model of equipment operation characteristics. The dynamic carbon flow tracking module, based on the topology model and equipment operation characteristic model of the park's integrated energy system, constructs a dynamic carbon emission flow tracking model for the park, recursively calculates the dynamic carbon potential of each node according to time period, and obtains the dynamic carbon emission factor coupled with the system operation structure. The multi-objective modeling module aims to minimize the economic cost of park operation, minimize the total carbon emissions of the park, and maximize the comprehensive benefits of energy performance contracting. It also incorporates the minimum guaranteed energy saving rate constraint, revenue sharing constraint, and investment payback period constraint of energy performance contracting to construct an energy-carbon coordinated multi-objective optimization scheduling model. The adaptive weighting module adopts an adaptive weighting mechanism based on the fusion of entropy weighting and fuzzy inference. It dynamically weights and aggregates each optimization objective according to the real-time electricity-carbon price ratio and the confidence level of renewable energy output prediction, thus aggregating multiple objectives into a single objective. The robust solution module employs a two-stage birch robust optimization framework (both day-ahead and intraday) to handle the uncertainty of source-load prediction for the energy-carbon coordinated multi-objective optimization scheduling model. It iteratively solves the problem using a column and constraint generation algorithm to obtain the Pareto optimal solution set, and then uses a sorting method to approximate the ideal solution to select the compromise optimal scheduling scheme.

Images