Electricity-carbon collaborative scheduling optimization method and device for low-carbon park comprehensive energy system

Abstract

The present application relates to a kind of low-carbon park integrated energy system electric-carbon collaborative scheduling optimization method and device, adopt panoramic situation deduction model to carry out collaborative prediction to multiple state parameters, generate panoramic dynamic situation scene set;Establish the electric-carbon collaborative scheduling model considering carbon trading mechanism, the real-time carbon cost is deeply embedded in objective function to carry out Pareto optimization of economic cost and carbon emission cost;The electric-carbon collaborative scheduling model is converted into standard mixed integer linear programming model, using situation deduction-rolling correction hierarchical computing framework of day-ahead optimization, collaborative scheduling optimization problem is decomposed to different time scales to make decision, in day-ahead layer, based on panoramic dynamic situation, global optimization plan is made, in intraday layer, deviation is corrected online by rolling optimization;Comprehensive energy system executes the scheduling plan after correction.Compared with prior art, while guaranteeing the economy of system operation, the present application can significantly improve renewable energy consumption rate and effectively reduce carbon emissions.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of park integrated energy system scheduling optimization, in particular to a multi-type resource situation deduction and electricity-carbon collaborative scheduling optimization method for a low-carbon park integrated energy system. BACKGROUND

[0002] As a key carrier integrating various energy forms such as electricity, gas, heat and cold in the region, coupling multiple types of resources such as renewable power generation, energy storage and flexible load, the integrated energy system (IES) has become an important way to improve energy utilization efficiency, promote renewable energy consumption and realize low-carbon transformation of terminal energy users. However, the efficient and low-carbon operation of IES faces unprecedented complexity: on the one hand, the strong randomness and volatility of distributed photovoltaic and wind power in the system, as well as the time-varying characteristics of multi-element load demand, bring great challenges to the safe and stable operation of the system; on the other hand, with the increasing maturity of the national carbon trading market and the tightening of carbon constraints, the traditional scheduling mode with economic efficiency as the single target has been difficult to adapt to the development needs under the new situation, and how to coordinate economic operation and carbon emission reduction targets to achieve "electricity-carbon" collaborative optimization has become a core problem that needs to be solved in the operation decision of park IES.

[0003] Current research still has several bottlenecks when facing the above challenges. First, in the aspect of situation awareness, most research focuses on the independent prediction of a single type of resource (such as photovoltaic output or electrical load), and lacks the ability to uniformly, correlate and panoramically dynamically deduce the state of multiple types of resources (such as "source-load-storage-market") in the system, resulting in incomplete and limited prediction information dimension for optimization scheduling, weak decision-making foundation. Secondly, in the aspect of optimization modeling, the existing scheduling model mostly takes carbon emission reduction as a constraint condition or an ex post evaluation index, and fails to embed carbon cost as a core decision variable into the objective function for integrated trade-off and collaborative optimization with energy procurement cost and equipment operation cost, thus failing to truly stimulate the endogenous motivation of the system to achieve active emission reduction through multi-energy complementation. Finally, in the aspect of solution application, the high-dimensional nonlinear optimization model constructed is often inefficient in solving, or is excessively simplified and loses practical guiding significance, lacking an efficient solution framework that takes into account the calculation efficiency and engineering practicability. SUMMARY

[0004] The purpose of the present application is to provide a multi-type resource situation deduction and electricity-carbon collaborative scheduling optimization method and device for a low-carbon park integrated energy system to solve the problems of multi-dimensional uncertainty, multi-objective collaborative decision difficulty and model complexity and solution efficiency in the current park-level integrated energy system scheduling scheme.

[0005] The object of the application can be realized by the following technical solutions:

[0006] As a first aspect of the application, a low-carbon park comprehensive energy system electric-carbon collaborative scheduling optimization method is provided, and the steps include:

[0007] A panoramic situation deduction model based on multi-source information fusion and dynamic correlation analysis is combined with multiple learning algorithms to collaboratively predict multiple state variables and generate a panoramic dynamic situation scenario set;

[0008] An electric-carbon collaborative scheduling model considering the carbon trading mechanism is established, and real-time carbon costs are deeply embedded in the objective function for Pareto optimization of economic costs and carbon emission costs;

[0009] The nonlinear electric-carbon collaborative scheduling model is converted into a standard mixed integer linear programming model, and a situation deduction-day-ahead optimization-rolling correction hierarchical calculation framework is used to decompose the collaborative scheduling optimization problem into different time scales for decision-making. At the day-ahead level, a global optimization plan is made based on the panoramic dynamic situation, and at the intra-day level, the deviation is corrected online through rolling optimization; The comprehensive energy system executes the corrected electric-carbon collaborative scheduling plan.

[0010] As a preferred technical solution, the state variables to be predicted in the panoramic dynamic situation scenario set generation process include photovoltaic and wind power output, electricity, heat, and cold load values, adjustable potential values of flexible load, and grid electricity purchase price;

[0011] The panoramic situation deduction model simultaneously uses variables with coupling relationships as inputs and performs prediction through LSTM, specifically including: using future light intensity as a prediction feature for photovoltaic and wind power output; using environmental temperature as a prediction feature for heat and cold load; and using real-time electricity price as a prediction feature for adjustable potential of flexible load and electricity load;

[0012] For scenarios with static and non-time sequence relationships between features and outputs, an XGBoost model is used for supplementary prediction or feature importance analysis;

[0013] Through collaborative prediction, all state variables at all times in the future scheduling period are rolled and deduced to generate a deterministic panoramic situation scenario for system operation.

[0014] As a preferred technical solution, the objective function of the electric-carbon collaborative scheduling model considering the carbon trading mechanism takes the minimization of the total cost of the system within the scheduling period as the optimization objective, and the total cost includes energy purchase cost, carbon trading cost, equipment operation and maintenance cost, and start-stop cost.

[0015] As a preferred technical solution, the energy purchase cost includes the cost of purchasing electricity / selling electricity from the external grid and the cost of purchasing natural gas:

[0016]

[0017] wherein: , are the electricity price at time t for purchasing electricity from the grid and selling electricity to the grid, respectively; t , are the power at time t for purchasing electricity from the grid and selling electricity to the grid, respectively; t is the natural gas price at time t; t is the total natural gas consumption of the system at time t; t is the dispatch time interval; T is the dispatch period.

[0018] As a preferred technical solution, if the total carbon emissions of the integrated energy system exceed the free quota, additional quotas are purchased; if not, the remaining quotas are sold, and the carbon trading cost is represented as follows:

[0019]

[0020] wherein: is the carbon trading market price; is the total carbon emissions of the system in the dispatch period; is the free carbon quota obtained by the system in the dispatch period;

[0021] The total carbon emissions of the system in the dispatch period T come from purchased electricity and consumed natural gas:

[0022]

[0023] wherein: is the marginal carbon emission factor of the grid at time t; t is the purchased power from the grid at time t; is the carbon emission coefficient of the nth gas equipment; t is the natural gas consumption of the nth gas equipment at time t; is the total number of gas equipment. i As a preferred technical solution, the constraint conditions of the electricity-carbon collaborative scheduling model considering the carbon trading mechanism include: i t

[0024] As a preferred technical solution, the constraint conditions of the electricity-carbon collaborative scheduling model considering the carbon trading mechanism include:

[0025] ​​​​​​​​​Energy balance constraint: the sum of the purchased power from the grid, the output of photovoltaic and wind power, the power generation of gas turbine and the discharge power of energy storage is equal to the sum of the sold power to the grid, the electrical load, the power consumption of equipment and the charging power of energy storage; the sum of the heating power of gas boiler, gas turbine and heat pump and the heat release power of heat storage device is equal to the sum of the heat load, the charging power of heat storage device and the heat loss of heat network;

[0026] Equipment operation constraint, including gas turbine / gas boiler operation constraint, energy storage system operation constraint and energy conversion equipment constraint;

[0027] System safety constraint: the purchased and sold power of the system to the grid is between the minimum and maximum values of the grid interaction power.

[0028] As a preferred technical solution, the nonlinear electric-carbon collaborative scheduling model is converted into a standard mixed integer linear programming model, specifically as follows:

[0029] For the fuel consumption of equipment in the electric-carbon collaborative scheduling model and its output, piecewise linearization is used for approximation: multiple piecewise points are selected in the upper and lower limit interval of equipment output, and the fuel consumption corresponding to the piecewise is calculated; a continuous variable as the weight coefficient of the piecewise point and a binary variable for selecting the active piecewise are introduced for each piecewise, and a linearization constraint is established;

[0030] For the charging and discharging loss of energy storage, equivalent linear processing is adopted: two binary variables and are introduced to represent the charging and discharging states of energy storage respectively, and mutual exclusion constraints are applied; the nonlinear terms in the energy storage system operation constraint equation are replaced equivalently, and together with the additional constraints, a linear constraint set is formed, and the additional constraints convert the nonlinear charging and discharging efficiency multiplication relationship into a variable definition relationship under linear constraints by introducing equivalent power variables.

[0031] As a preferred technical solution, based on the hierarchical calculation framework of situation deduction-day-ahead optimization-rolling correction, the planning model is solved as follows:

[0032] Day-ahead panoramic situation deduction: on the day before scheduling, the deterministic prediction scenario of the future scheduling day is generated based on the panoramic situation deduction model;

[0033] Day-ahead electric-carbon collaborative optimization scheduling: on the day before scheduling, the generated deterministic prediction scenario is substituted into the electric-carbon collaborative scheduling model which has been converted into a mixed integer linear programming as a known parameter to solve, and the optimal start-stop plan and output plan of all equipment in the scheduling day are generated, i.e. day-ahead scheduling plan;

[0034] Intraday rolling optimization correction: the day-ahead scheduling plan is executed in real time on the scheduling day, and the rolling correction is performed on the scheduling plan based on the set period, and the specific process is as follows:

[0035] Obtain the latest actual operation state of the integrated energy system; detect the deviation between the actual value and the day-ahead predicted value of the operation state, and calculate the prediction error; update the short-term prediction data with the current actual operation state as the initial value, and re-optimize the future set short-period scheduling plan; execute the optimization result of the first time period in the rolling optimization period, and issue adjustment instructions to the equipment.

[0036] As a second aspect of the present application, an electric-carbon collaborative scheduling optimization device of an integrated energy system is provided, comprising a memory, a processor, and a program stored in the memory, wherein the processor implements the low-carbon park integrated energy system electric-carbon collaborative scheduling optimization method as described above when executing the program.

[0037] As a third aspect of the present application, a storage medium is provided, which stores a program, wherein the program implements the low-carbon park integrated energy system electric-carbon collaborative scheduling optimization method as described above when executed.

[0038] Compared with the prior art, the present application has the following beneficial effects:

[0039] 1) The multi-type resource situation deduction and electric-carbon collaborative scheduling optimization method of the integrated energy system proposed in the present application constructs a complete technical solution from situation awareness to optimization decision, effectively solving the multi-dimensional uncertainty modeling and multi-objective collaborative decision-making problems faced by the park IES in low-carbon economic operation. While ensuring the economic efficiency of the system operation, it can significantly improve the renewable energy consumption rate and effectively reduce carbon emissions, providing scientific theoretical support and feasible technical path for the low-carbon and intelligent operation of the park IES.

[0040] 2) The park IES multi-dimensional operation situation panoramic deduction technology based on multi-source information fusion and dynamic correlation analysis of the present application innovatively proposes a comprehensive situation deduction method for multi-type resources of the park IES, overcoming the limitations of the traditional independent prediction mode. By constructing a multi-source information fusion framework that integrates historical data, real-time monitoring data, and external environmental information, and using machine learning algorithms and time series analysis techniques, not only can the output of renewable energy sources such as photovoltaic and wind power be accurately predicted, but also the demand changes of electric, heat, and cold multi-element loads, the adjustment potential of flexible loads, and the fluctuation trend of energy market prices can be collaboratively deduced. The core lies in the introduction of a dynamic correlation analysis module, which deeply excavates the spatio-temporal coupling relationship and causal relationship between key state parameters such as "source-load-storage-market", thereby generating a continuous, correlated, and panoramic dynamic situation scene of the system operation in the future scheduling period. This technology provides a high-reliability, multi-dimensional decision-making data foundation for subsequent optimization scheduling, greatly improving the forward-looking and robustness of the scheduling scheme.

[0041] 3) The proposed electric-carbon two-way coupling collaborative optimization model considering carbon trading mechanism and multi-energy complementary characteristics deeply embeds carbon trading cost in the double-objective collaborative optimization method of park IES scheduling model. The total carbon emission of system operation is converted into explicit carbon cost through carbon trading market price, and is minimized in the unified objective function together with energy purchase cost, equipment operation and maintenance cost, etc. to realize two-way coupling. Firstly, the coupling of electric-carbon cost is realized. The model accurately quantifies the implicit carbon cost of each degree of purchased electricity through the real-time updated marginal carbon emission factor of power grid, and guides the system to balance price and carbon in the decision of purchasing electricity. Secondly, the coupling of multi-energy flow is realized. The model fully utilizes the electric-thermal / electric-cooling coupling characteristics of gas turbines, heat pumps and other energy conversion devices, introduces carbon cost variables in the whole link of energy production, conversion, storage and consumption, and encourages the system to actively adjust the operation strategy, so as to realize the Pareto optimization of economic benefit and carbon emission reduction benefit. The method provides a precise quantification and decision tool for park IES participating in carbon market.

[0042] 4) The present application provides an efficient solving strategy and systematic implementation framework based on mixed integer linear programming and hierarchical calculation, which is used for solving the high-dimensional, complex and multi-constrained electric-carbon collaborative scheduling model. According to the characteristics of containing a large number of integer variables and continuous variables in the model, the non-linear factors such as device efficiency and pipe network loss are reasonably segmented and linearized or equivalently processed through fine linear processing technology, so that the original problem is transformed into a standard mixed integer linear programming problem, thereby the mature commercial solver can be used for efficient solving, and the global optimal or high-quality feasible solution in acceptable time is ensured. Further, the present application provides a systematic hierarchical calculation framework of situation deduction-day-ahead optimization-rolling correction, which decomposes the complex decision-making process into sub-problems of different time scales. The coarse-grained plan is made based on the panoramic situation deduction in the day-ahead stage, and the prediction deviation is corrected by using rolling optimization in the intra-day stage, so as to balance the calculation complexity and the real-time decision-making. The strategy and framework ensure the engineering practicability and application value of the proposed method. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 The flow chart of the low-carbon park comprehensive energy system electric-carbon collaborative scheduling optimization method of the present application.

[0044] Figure 2 The collaborative prediction flow chart based on dynamic correlation analysis of the present application.

[0045] Figure 3 The hierarchical calculation framework of situation deduction-day-ahead optimization-rolling correction of the present application.

[0046] Figure 4 The time-of-use electricity price chart of the system in the embodiment of the present application.

[0047] Figure 5 Renewable energy output prediction result in the embodiment of the present application.

[0048] Figure 6 Multi-element load prediction result in the embodiment of the present application.

[0049] Figure 7 Prediction error in the embodiment of the present application.

[0050] Figure 8 Electric power balance optimization result in the embodiment of the present application.

[0051] Figure 9 Thermal power balance optimization result in the embodiment of the present application.

[0052] Figure 10 Energy storage system operating state in the embodiment of the present application.

[0053] Figure 11 Grid interactive power in the embodiment of the present application.

[0054] Figure 12 Carbon emission intensity in the embodiment of the present application.

[0055] Figure 13 Carbon emission cost of each period in the embodiment of the present application.

[0056] Figure 14 Operating cost composition analysis in the embodiment of the present application.

[0057] Figure 15 Rolling correction power balance effect schematic diagram in the embodiment of the present application. DETAILED DESCRIPTION

[0058] The present application will be described in detail below in conjunction with the accompanying drawings and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present application, and gives a detailed implementation manner and specific operation process, but the protection scope of the present application is not limited to the following embodiments.

[0059] Embodiment 1

[0060] The present application proposes an optimization method for low-carbon park comprehensive energy system, which integrates multi-type resource situation deduction and electric-carbon collaborative scheduling. For example, Figure 1As shown, first, a panoramic situation deduction model based on multi-source information fusion and dynamic correlation analysis is constructed, and a hybrid algorithm of LSTM and XGBoost is adopted to realize high-precision collaborative prediction of key state parameters of "source-load-storage-market". Second, an electricity-carbon collaborative scheduling model considering carbon trading mechanism is established, and by deeply embedding real-time carbon cost into the objective function, the Pareto optimization of economic cost and carbon emission cost is realized. Finally, an efficient solving strategy based on mixed integer linear programming (MILP) and a hierarchical calculation framework of "situation deduction-day-ahead optimization-rolling correction" is designed. This method can significantly improve the renewable energy consumption rate while ensuring the economic efficiency of the system, and effectively reduce carbon emissions, providing a scientific theoretical support and feasible technical path for the low-carbon and intelligent operation of the IES.

[0061] Step one, multi-dimensional running situation panoramic deduction of IES in park.

[0062] This step is the data basis and forward-looking section of the whole scheme. The idea is to break the limitations of traditional "chimney type" independent prediction and build a unified information framework integrating historical, real-time and external environmental data. On this basis, the concept of dynamic correlation analysis is introduced, and machine learning algorithms such as LSTM and XGBoost are used to not only predict single variables, but also to focus on mining and learning the spatio-temporal coupling and causal relationship between key parameters of "source-load-storage-market" (such as meteorological conditions affecting photovoltaic output and air conditioning load). The ultimate goal is to generate a set of high-precision panoramic dynamic situation scenarios that are coordinated and interrelated within the future scheduling period, providing reliable and multi-dimensional input data for subsequent optimal scheduling and fundamentally improving the forward-looking and robustness of decision-making.

[0063] Step two, modeling of electricity-carbon collaborative scheduling optimization considering carbon trading mechanism.

[0064] This step is the core model and decision-making section of the scheme. The idea is to innovatively transform the carbon trading cost from an external constraint into an internalized decision variable. By using real-time carbon price and marginal carbon emission factor of the power grid, the implicit carbon cost of each unit of purchased electricity and natural gas consumption is accurately quantified and included in the total objective function to be minimized together with the economic cost. This achieves a two-way coupling of "electricity-carbon" costs at the financial level, enabling the scheduling model to automatically weigh economic efficiency and low carbon when making each decision (such as purchasing electricity, starting and stopping equipment, and distributing loads). At the same time, the model fully utilizes the flexibility of multi-energy coupling devices such as gas turbines and heat pumps, and through the introduction of carbon cost incentives in all aspects of energy production, conversion, and storage, it guides the system to actively seek the optimal collaborative scheduling strategy of electricity, heat, and cold, thereby achieving Pareto improvement of economic and emission reduction benefits.

[0065] Step three, efficient solving strategy and systematic implementation framework.

[0066] The idea of this step is: for the characteristics of high dimension and nonlinearity of the electricity-carbon collaborative model, the fine linear modeling technology (such as piecewise linearization, large M method) is used to process the non-linear terms of device efficiency, pipe network loss, etc. The original problem is transformed into a standard mixed integer linear programming (MILP) model, so that it can be solved efficiently by using mature solvers such as Gurobi. To further solve the contradiction between model complexity and real-time requirements, a "situation deduction-day-ahead optimization-rolling correction" hierarchical calculation framework is designed. This framework decomposes the complex problem according to the time scale, the day-ahead layer formulates the global optimization plan based on the panoramic situation, and the intra-day layer corrects the deviation online through rolling optimization, ensuring the calculation efficiency and engineering practicability of the method.

[0067] I. Park IES multi-dimensional running situation panoramic deduction model

[0068] High-precision running situation awareness is the premise and foundation of park integrated energy system (IES) optimization and dispatching. Traditional prediction methods usually predict renewable energy output and load demand independently, ignoring the strong spatio-temporal coupling and causal relationship between multiple types of resources such as "source-load-storage" within the system, resulting in poor coordination of prediction results and difficulty in supporting system-level collaborative optimization decision-making. To overcome this limitation, this part aims to build a park IES multi-dimensional running situation panoramic deduction model based on multi-source information fusion and dynamic correlation analysis. The core idea of this model is to comprehensively utilize historical data, real-time monitoring data and external environmental data, to deeply mine the internal relationship between variables through machine learning algorithms, and simultaneously and collaboratively deduce the continuous change trajectory of key state parameters such as renewable energy output, multi-element load and market electricity price in the future dispatching period, generating a panoramic dynamic situation scenario for system operation, providing high reliability and consistency of input data for subsequent electricity-carbon collaborative optimization and dispatching.

[0069] 1.1. Multi-source information fusion framework

[0070] The running state of park IES is influenced by both internal characteristics and external environment. To achieve panoramic deduction, a unified multi-source information fusion framework needs to be built, integrating the following types of data:

[0071] Historical data set (H): including historical output data of photovoltaic and wind power, historical data of electricity, heat and cold load, and historical electricity price and gas price data, etc.

[0072] Real-time monitoring data set (S): including real-time output of photovoltaic and wind power, real-time values of various types of load, real-time state of charge (SOC) of energy storage system, and running state of key equipment, etc.

[0073] External environment dataset (E): covering meteorological forecast data (such as future light intensity, environmental temperature, wind speed), future time-of-use price signal, date type (workday / holiday) and the like.

[0074] By cleaning, aligning and normalizing the multi-source heterogeneous data, a feature matrix X suitable for the input of the machine learning model is constructed.

[0075] 1.2, the collaborative prediction model based on dynamic correlation analysis.

[0076] The application adopts a hybrid machine learning model mainly using long short-term memory network (LSTM) and supplemented by extreme gradient boosting (XGBoost) to collaboratively predict multiple variables, and the modeling process is as shown in Figure 2 .

[0077] 1.2.1 prediction variable definition, define t The state variable set to be predicted at time t is:

[0078]

[0079] In the formula: , Pv(t) and Pw(t) are the predicted output of photovoltaic and wind power at time t, respectively, and the unit is t ; kW , , Pd(t), Ph(t) and Pc(t) are the predicted values of the electric, heat and cold load at time t, respectively, and the unit is ; t Pf(t) is the adjustable potential prediction value of the flexible load at time t, and the unit is kW ; Pb(t) is the electricity purchase price of the power grid at time t, and the unit is yuan / t . kW 1.2.2 long short-term memory network (LSTM) model, LSTM effectively captures the long-term dependence of time series through its unique gating mechanism, and is an effective tool for processing energy time series prediction problems. Its core calculation process is as follows: t kWh Forget gate:

[0080]

[0081] Input gate:

[0082]

[0083] Output gate:

[0084] ​​​

[0085]

[0086] Cell state update:

[0087]

[0088] Output gate:

[0089]

[0090]

[0091] Final prediction:

[0092]

[0093] where: is the sigmoid activation function; represents the Hadamard product (element-wise multiplication); is t the input feature vector at time t, constructed by multi-source data fusion in Section 1.2; is t the hidden state at time t; is t the cell state at time t; , , , , are the weight matrices for each corresponding gate; , , , , are the bias vectors for each corresponding gate; is t the predicted output at time t.

[0094] 1.2.3 Implementation of dynamic correlation analysis, the innovation of the model lies in realizing dynamic correlation analysis through feature engineering. In constructing the input feature vector of LSTM at time t, variables with coupling relationship are simultaneously taken as inputs. For example:

[0095] the future light intensity is simultaneously taken as and the predicted features of air conditioning load (because air conditioning load is related to light).

[0096] obtain the real-time electricity price simultaneously as and predicted features.

[0097] obtain the real-time electricity price as and predicted features.

[0098] In this way, the model can automatically learn and quantify the complex nonlinear relationship between these features and multiple output variables during the training process, thereby realizing the coordinated deduction of the "source-load-storage-market" key state parameters, rather than independent single variable prediction.

[0099] For scenarios where the relationship between features and outputs is more static and non-sequential (such as the instantaneous influence of light on photovoltaics), XGBoost models can be used for supplementary prediction or feature importance analysis to enhance the interpretability and robustness of the model.

[0100] 1.3, panoramic dynamic situation scenario generation, through the above-mentioned collaborative prediction model, the state variables at all times within a future scheduling period T (such as 24 hours) are rolled and deduced, and finally a deterministic panoramic situation scenario Ω of system operation is generated:

[0101]

[0102] The scenario Ω constitutes a multi-dimensional, continuous, and internally correlated time series set, accurately depicting the expected future operation trajectory of the system. This scenario will serve as an input parameter for the second part of the electricity-carbon collaborative scheduling model, used to develop a day-ahead optimization scheduling plan.

[0103] Step two, electricity-carbon collaborative scheduling model considering carbon trading mechanism.

[0104] After obtaining the panoramic situation deduction scenario, how to develop an optimization scheduling scheme that takes into account both economic efficiency and low carbon is the core decision-making link of the present application. Traditional IES scheduling models either consider carbon emissions as a constraint or only consider energy purchase costs, and cannot quantify the trade-off relationship between carbon costs and economic costs, making it difficult to guide the system to actively reduce emissions. This part aims to build an electricity-carbon collaborative scheduling model considering the carbon trading mechanism. The core innovation lies in deeply embedding the carbon trading cost in the optimization objective, through the two-way interactive mechanism of "electricity-carbon" cost coupling and "multi-energy flow" operation coupling, converting carbon emissions into explicit costs, thereby realizing the Pareto optimization of economic and emission reduction benefits under a unified objective function. The model is a mixed integer programming problem considering multiple energy balance and device operation constraints.

[0105] 2.1 Objective Function: This model aims to minimize the total system cost within a scheduling period T. Total Cost This mainly includes energy purchase costs, carbon trading costs, equipment operation and maintenance costs, and start-up and shutdown costs.

[0106]

[0107] 2.1.1 Energy Purchase Cost

[0108] Energy purchase costs include the cost of purchasing / selling electricity from an external grid and the cost of purchasing natural gas.

[0109]

[0110] In the formula: , They are respectively t The electricity price for purchasing and selling electricity from the grid at any time is expressed in yuan / kWh, and its value is given by the situational deduction in step one; , They are respectively t The power purchased from and sold to the grid at any given time, measured in kW; for t Real-time natural gas price, in yuan / m³ 3 ; for t Total natural gas consumption of the system at any given time, in cubic meters. 3 / h; The time interval is the scheduling interval, measured in hours (h).

[0111] 2.1.2 Carbon Trading Cost

[0112] Based on the carbon trading market mechanism, if the system's total carbon emissions exceed the free allowances, additional allowances must be purchased; if not, the remaining allowances can be sold. This model directly incorporates carbon trading costs into the objective function.

[0113]

[0114] In the formula: The price is in the carbon trading market, expressed in yuan / tCO2. The total carbon emissions of the system during the scheduling period are expressed in tCO2. The free carbon allowances obtained by the system during the scheduling period are expressed in tCO2.

[0115] Total carbon emissions of the system Electricity is sourced from purchased electricity and natural gas consumption.

[0116]

[0117] wherein: is t the marginal carbon emission factor of the power grid at time t, unit: CO2 / MWh (or kgCO2 / kWh), which realizes the "electricity-carbon" cost coupling and converts the intangible power carbon emission into quantifiable cost; is the carbon emission factor of the i th gas equipment (such as gas turbine, gas boiler), unit: tCO2 / m 3 ; is the natural gas consumption of the i th gas equipment at time t, unit: m 3 / h; t is the total number of gas equipment.

[0118] 2.1.3 Operation & Maintenance Cost

[0119] The operation & maintenance cost of equipment is usually positively correlated with its output, which can be simplified as a linear function:

[0120]

[0121] wherein: is the unit output operation & maintenance cost of the j th equipment (such as CHP, heat pump, energy storage, etc.), unit: yuan / kWh; is the output electric power or thermal power of the j th equipment at time t, unit: kW; t is the total number of equipment.

[0122] 2.1.4 Start-up Cost

[0123] For equipment with high start-up cost such as gas turbine, the start-up consumption needs to be considered.

[0124]

[0125] wherein: is the start-up cost of the k th equipment, unit: yuan / time; is a 0-1 variable, indicating whether the k th equipment is started at time t, 1 if yes, 0 if no; t is the total number of start-up equipment.

[0126] 2.2 Constraints​​​

[0127] 2.2.1 Energy balance constraints

[0128] Electric power balance constraints:

[0129]

[0130] where: , are the electric power output of the photovoltaic and wind power plants, respectively, at time t, given by step 1 ; t is the electric power output of the gas turbine at time t; t , are the electric power output of the energy storage system at time t, given by step 1 ; t t , t are the electric power consumption of the heat pump and electric chiller at time t;

[0131] Other electric power consumption.

[0132]

[0133] where: , , are the heating power of the gas boiler, gas turbine (waste heat) and heat pump at time t; t , t are the heating and cooling power of the thermal storage device at time t; t

[0134] 2.2.2 Equipment operation constraints

[0135] Gas turbine / gas boiler operation constraints:

[0136]

[0137]

[0138] ​​​​​​​​​​​​

[0139] where: Poutis the output power of the device i is a 0-1 variable representing the on-off state of the device i , is the lower and upper bound of the ramp rate of the device i , is the gas consumption characteristic coefficient of the device i

[0140] Energy storage system operation constraints (for example, electrical energy storage):

[0141]

[0142]

[0143]

[0144]

[0145]

[0146] where: SoC(t) is the state of charge of the energy storage at time t t , Cch, Cdis the charging and discharging efficiency of the energy storage, respectively Cn is the nominal capacity of the energy storage , is a 0-1 variable representing the charging and discharging state of the energy storage

[0147] Energy conversion device constraints (for example, heat pump):

[0148]

[0149] where: COP is the coefficient of performance of the heat pump

[0150] 2.2.3 System safety constraints

[0151]

[0152]

[0153] where: , Pmin, Pmax are the minimum and maximum values of the power exchanged with the grid, usually limited by contracts or physical lines.​​​​​

[0154] 2.3 The embodiment of the electric-carbon bidirectional coupling mechanism

[0155] The above model embodies the innovative concept of "electric-carbon bidirectional coupling" in the following ways:

[0156] 2.3.1. Cost coupling: in the objective function , The terms and in the terms make the electricity purchase decision simultaneously affected by the electricity price and the carbon price. The system will tend to purchase electricity at a time when the sum of the electricity price and the carbon electricity price is low.

[0157] 2.3.2. Multi-energy flow coupling: the term in the objective function simultaneously contains the carbon cost of electricity purchase and the carbon cost of gas equipment. The model will automatically find the operation strategy with the lowest total cost (economic + environmental) by optimizing the equipment combination (for example, prefer to use a heat pump driven by low-carbon electricity instead of a gas boiler for heating when the carbon price is high) under the premise of meeting energy balance, realizing the coordinated optimization of multi-energy flow.

[0158] Step three, efficient solving strategy based on MILP and hierarchical calculation.

[0159] The park IES electric-carbon coordinated scheduling model is a complex optimization problem with high dimension, nonlinearity, and continuous and discrete variables. It is difficult to solve directly and time-consuming, and it is difficult to meet the real-time requirements of engineering applications. To ensure the practicality and solving efficiency of the model, the present invention proposes a systematic efficient solving strategy. The strategy contains two core contents: first, through the fine linear modeling technology, the original nonlinear model is transformed into a standard mixed integer linear programming (MILP) model, so that it can be efficiently solved by using mature commercial solvers; second, a "situation deduction-day-ahead optimization-rolling correction" hierarchical calculation framework is designed, which decomposes the complex problem into different time scales for decision-making, effectively balancing the global optimization accuracy and calculation real-time performance.

[0160] 3.1, model linearization and MILP transformation

[0161] The nonlinear terms (such as device efficiency curves, energy storage charging and discharging losses) in the original model are the main reason for the difficulty in solving. The innovation of the present invention is to linearize them by the following methods, and finally transform them into MILP problems.

[0162] 3.1.1 Piecewise linearization of device fuel consumption characteristics

[0163] The fuel consumption of gas turbines (GT), gas boilers (GB), etc. is related to its outputP Generally, the relationship is nonlinear. The present application adopts a piecewise linear approximation (PWA) method to accurately approximate it.

[0164] Select the piecewise points: within the upper and lower limits of the device output P min ,P max Select K +1 piecewise points , and calculate the corresponding fuel consumption .

[0165] Introduce auxiliary variables: for each piecewise k Introduce continuous variables and binary variables .

[0166] Construct linearization constraints: the output and gas consumption of the device at t time can be expressed as:

[0167]

[0168]

[0169] Constraint conditions:

[0170]

[0171]

[0172]

[0173]

[0174] In the formula: is the weight coefficient corresponding to the k th piecewise point, and is a continuous variable; is a 0-1 variable for selecting active segments, ensuring that the value can only fall within one segment interval.

[0175] Through the above processing, the nonlinear fuel consumption curve is accurately approximated by a series of linear segments, successfully converting the nonlinear relationship into linear constraints containing continuous variables and integer variables.

[0176] 3.1.2 Equivalent linear processing of energy storage charging and discharging loss

[0177] The state of charge (SOC) equation of the energy storage system is nonlinear due to the charging and discharging efficiency , The nonlinearity arises from the differences. This invention employs a method of equivalent linearization combined with mutual exclusion constraints of charging and discharging states to address this issue.

[0178] Introducing binary variables and Let represent the charging and discharging states of the energy storage at time t, respectively, and apply mutual exclusion constraints:

[0179]

[0180] At this point, the nonlinear terms in the SOC equation can be equivalently replaced and, together with other constraints, form a linear constraint set:

[0181]

[0182] Additional constraints:

[0183]

[0184]

[0185]

[0186]

[0187] In the formula: for t The equivalent charging power at any given moment (i.e., the actual energy stored). for t The equivalent discharge power at any given moment (i.e., the actual energy released).

[0188] By introducing an equivalent power variable, the nonlinear efficiency multiplication relationship is transformed into a variable definition relationship under linear constraints, thereby achieving model linearization.

[0189] Transformation Result: After the above series of processing steps, the original nonlinear optimization model is completely transformed into a mixed-integer linear programming (MILP) problem, whose standard form is:

[0190]

[0191]

[0192] In the formula: This is a vector of decision variables, containing continuous variables. (such as equipment output, energy storage SOC, and interactive power) and integer variables (e.g., equipment start / stop status) Piecewise linearization Energy storage charging and discharging status , ; is the objective function coefficient vector; , is the constraint matrix and vector, corresponding to all linearized equality and inequality constraints.

[0193] The MILP model can be solved efficiently by using mature commercial optimization solvers such as Gurobi, CPLEX, etc. These solvers have built-in advanced algorithms such as Branch-and-Bound, Cutting Plane, etc. which can obtain global optimal solutions or high-quality feasible solutions within an acceptable time.

[0194] 3.2, "Situation deduction-day-ahead optimization-rolling correction" hierarchical calculation framework

[0195] To solve the contradiction between large calculation amount of long-time scale optimization and the need to deal with uncertainty in real-time operation of the system, the invention designs a hierarchical calculation framework as shown in Figure 3 .

[0196] 3.2.1 First layer: panoramic situation deduction (day-ahead)

[0197] When executed on the dispatch day (D-1 day), the system generates a high-precision, multi-dimensional deterministic prediction scenario Ω for the next 24 hours (dispatch day D) based on the panoramic situation deduction model in step one, and outputs prediction data including renewable energy output, various types of load and electricity price, etc. This mechanism is executed once a day to ensure the timeliness and accuracy of the prediction results.

[0198] 3.2.2 Second layer: day-ahead electric-carbon collaborative optimization scheduling (day-ahead)

[0199] At night on the dispatch day (D-1 day), the system receives the scenario Ω output by the first layer as known parameters, and substitutes it into the electric-carbon collaborative scheduling model in step three which has been converted into MILP, and solves it through a commercial solver. This solving process will generate the optimal start-stop plan and output plan of all devices at 96 time points (15-minute intervals) throughout the day on the dispatch day D, i.e. the day-ahead scheduling plan. This mechanism is executed once a day to ensure the timeliness and accuracy of the scheduling scheme.

[0200] 3.2.3 Third layer: intra-day rolling optimization correction (intra-day)

[0201] The real-time execution of the dispatch day D adopts a minute / 15-minute level cycle rolling mechanism, and the specific process is as follows: first, the latest actual running state of the system (such as the energy storage SOC, the actual output of renewable energy) is obtained through state updating; then, deviation detection is performed, the actual value is compared with the day-ahead prediction value, and the prediction error is calculated; then, based on the updated short-term prediction data, the future 4-hour short cycle is re-optimized (the optimization model is the same as the day-ahead model, but the decision variables before the current time are fixed); finally, the optimization result of the first time period in the rolling optimization cycle is executed, and the adjustment instruction is issued to the device. The mechanism effectively eliminates the prediction error and uncertain disturbance through high-frequency rolling execution (such as every 15 minutes), ensures the real-time and robustness of the dispatch, and divides the complex global optimization problem into the coordinated operation of the day-ahead layer (plan) and the intra-day layer (real-time control), which not only maintains the consistency of the model but also realizes dynamic correction, and finally guarantees the engineering practicability and effectiveness of the whole method.

[0202] Embodiment 2

[0203] To verify the effectiveness of the low-carbon park integrated energy system multi-type resource situation deduction and electric-carbon collaborative scheduling optimization method proposed in the application, a typical park integrated energy system test case is also provided in this embodiment. The system configuration parameters are shown in Table 1:

[0204] Table 1: Park IES device parameter configuration

[0205]

[0206] The economic environment parameters of the system operation are set as follows: the basic electricity price is 0.8 yuan / kWh, the peak-valley flat time-of-use electricity price mechanism is adopted; the natural gas price is 3.5 yuan / m 3 ; the carbon trading price is 60 yuan / tCO2; the free carbon quota is 5 tCO2 / day; the power grid carbon emission factor is 0.8 tCO2 / MWh; the natural gas carbon emission factor is 2.2 tCO2 / thousand m 3 ; the time-of-use electricity price is shown in Table 2. Figure 4

[0207] 1. Analysis of multi-type resource situation deduction results

[0208] The renewable energy output prediction verification results are as follows:

[0209] Figure 5 The actual value and the predicted value of the photovoltaic and wind power output are shown. The photovoltaic output presents the typical characteristics of high in the daytime and zero at night, and the maximum output appears at noon, reaching 240 kW. The wind power output presents a certain volatility, and the standard deviation of the prediction error is about 10%. The prediction accuracy index is shown in Table 2.

[0210] ​Table 2 Renewable energy prediction accuracy indicators

[0211]

[0212] Table 2 shows the deviation between predicted and actual values. These errors are the main source of uncertainty in system operation. Photovoltaic prediction: its MAPE is 8.5%, RMSE is 18.2 kW, R 2 up to 0.94. This indicates that the model can extremely accurately capture the main trend of photovoltaic output (high R 2 ), and the absolute error is small (low RMSE and MAPE). For a photovoltaic system with a rated capacity of 300 kW, an average error of 8.5% is completely acceptable in engineering, providing a reliable basis for day-ahead scheduling.

[0213] Wind power prediction can be known, its prediction is more difficult than photovoltaic, MAPE is 12.3%, RMSE is 24.7 kW, R 2 0.89. Although the accuracy is slightly lower than photovoltaic prediction, its R 2 value still indicates that the model can explain most of the output fluctuations, and for a fan with a rated power of 200 kW, this accuracy is sufficient to support the system to make risk prediction and standby capacity arrangement.

[0214] The verification results of multi-element load demand prediction are as follows:

[0215] The MAPE of electric, heat, and cold load prediction in load prediction is 5.2%, which is excellent. Figure 6 The "double peak" electric load curve shown is highly consistent with the actual work schedule of the park, and the negative correlation between heat load and temperature change is also accurately captured, proving that the model has successfully learned the complex coupling relationship between weather, time, and load through "dynamic correlation analysis".

[0216] The electric load presents a clear "double peak" feature, with an early peak at 9:00-11:00 and a late peak at 18:00-21:00, consistent with the work schedule of the park on weekdays. The heat load is higher in the early and late periods, consistent with the heating demand time. The average absolute percentage error of load prediction is 5.2%, meeting the requirements of engineering application.

[0217] Figure 7 The deviation between predicted and actual values is shown. These errors are the main source of uncertainty in system operation. Although the paper handles these errors in the intra-day rolling layer, at the day-ahead level, the statistical characteristics of these errors (such as Table 2) are important inputs for evaluating the robustness of the scheduling scheme. For example, the standard deviation of wind power prediction error is about 10% (estimated based on RMSE), which means that the scheduling plan needs to have flexibility to cope with ±20 kW power fluctuations.

[0218] 2. Analysis of the results of the electric-carbon coordinated dispatch optimization

[0219] The results of the electric power balance optimization verification are as follows, Figure 8 The electric power balance in a 24-hour dispatch cycle is shown. The system makes full use of renewable energy, and in the daytime when photovoltaic output is sufficient, clean energy is used preferentially for power supply, and excess power is stored or sold to the grid. During the night load peak period, the gas turbine is reasonably started and stopped and the energy storage is discharged to reduce high-priced grid power purchase.

[0220] The analysis of the optimization strategy and effect of electric power balance shows that renewable energy is preferentially consumed (such as Figure 8 ), and photovoltaic is the core power source of the system during the day (about 6:00-18:00). The total operation cost of the system is 5,832 yuan, and the renewable energy consumption rate is as high as 87.3%, which is a very impressive indicator, far exceeding the consumption level of many traditional power grids, fully embodying the great advantage of IES in local consumption of renewable energy.

[0221] The results of the thermal power balance optimization verification are as follows: the heat supply adopts a multi-energy complementary strategy, as shown in Figure 9 . During the low electricity price period, the heat pump is preferentially used for heating to take advantage of its high energy efficiency; during the peak heat load period, the gas-fired boiler is started and the gas turbine waste heat is utilized to ensure the reliability of heat supply. The thermal energy storage system effectively smooths the heat supply fluctuations and improves the system flexibility.

[0222] During the electricity price valley and flat period, the system preferentially uses the heat pump (COP=3.3) for heating, which has a much higher heating efficiency than the gas-fired boiler (thermal efficiency 85%), realizes "electricity instead of gas", and is a key measure for low carbonization. During the peak heat load or peak electricity price period, the gas-fired boiler and the gas turbine waste heat become the main heat source. This multi-heat source coordination strategy not only guarantees the reliability of heat supply, but also realizes the minimization of the total cost (economic cost + carbon cost) through flexible switching of energy input. The investment in thermal energy storage further enhances the adjustment flexibility of the system and smooths the fluctuations of the heat source output.

[0223] The results of the verification of the operating characteristics of the energy storage system are as follows: the SOC changes of the electric energy storage and the thermal energy storage are shown in Figure 10 , which embodies the "low storage and high generation" operation strategy. The electric energy storage is charged during the low electricity price period and discharged during the peak period, which not only reduces the electricity cost, but also provides peak shaving service. The thermal energy storage flexibly adjusts the charging and discharging strategy according to the heat load change. Figure 11 and Figure 8The analysis shows that the electric energy storage charges at the early morning valley price (0.35 yuan / kWh) and discharges at the noon and evening peak price (0.85 yuan / kWh). This not only brings direct arbitrage benefits of 342 yuan (by selling electricity to the grid or reducing high-priced electricity purchases), but more importantly, it smooths the net load curve and reduces the dependence on the gas turbine and external grid during peak hours.

[0224] 3. Analysis of the electricity-carbon synergistic benefits

[0225] The carbon emission analysis is as follows: as Figure 12 The composition of the carbon emission sources of the system is shown. Through electricity-carbon synergistic optimization, the total carbon emission of the system is 7.2 tCO2, of which the carbon emission of grid electricity purchase accounts for 58.3%, and the carbon emission of natural gas consumption accounts for 41.7%. Compared with the dispatching scheme without considering the carbon cost, the carbon emission is reduced by 15.6%.

[0226] The carbon emission traceability table shows that among the total carbon emission of 7.2 tCO2 of the system, the carbon emission of grid electricity purchase accounts for 58.3%, which is the largest carbon source. This highlights the extreme importance of reducing dependence on external electricity and improving renewable energy self-sufficiency at the park level for emission reduction.

[0227] The endogenous driving of the carbon cost, Figure 13 The time sequence change of the carbon emission cost is shown, and the peak value basically coincides with the high peak period of the electricity price. This means that during the peak period, the system not only faces high electricity prices, but also has to pay higher carbon costs. The superimposed effect of this "electricity-carbon" cost in the time dimension is the core mechanism that drives the model to make low-carbon decisions. For example, it will prefer to use energy storage to discharge during this period, rather than increasing the output of the gas turbine or purchasing electricity from the grid.

[0228] The cost-benefit analysis is as follows: the total cost of system operation is 5,832 yuan, and the cost composition is shown in Figure 14 . Among them, the energy purchase cost accounts for the highest proportion (62.3%), and the carbon trading cost accounts for 12.8%, which reflects the important influence of carbon cost internalization on the economic operation of the system.

[0229] Compared with the traditional single economic target dispatching, the electricity-carbon synergistic dispatching method proposed in the invention achieves a good effect of reducing carbon emissions by 15.6% with an increase in cost of 8.7%, which reflects the effective trade-off between economy and low carbon.

[0230] Quantification of Synergistic Benefits: Compared to scheduling schemes that do not consider carbon costs, this method achieves a significant reduction in carbon emissions of 15.6% with only an 8.7% increase in costs. This data is extremely valuable, as it precisely quantifies the "cost" of "low carbon" and proves that through refined synergistic optimization, substantial environmental benefits can be obtained at a relatively low economic cost, achieving Pareto improvement. Carbon trading costs account for 12.8% of total costs, which is no longer a negligible item. It profoundly affects the system's operational logic, transforming emission reduction from an "external constraint" to an "endogenous driving force."

[0231] 4. Verification of intraday rolling correction effect

[0232] The effect of prediction error correction is verified as follows: (e.g.) Figure 15 The diagram illustrates the optimization effect of intraday rolling correction on power balance. Due to errors in renewable energy forecasting, power deficits and surpluses occurred in actual operation. Through rolling optimization correction, the system can adjust energy storage charging and discharging strategies and gas turbine output in real time, effectively addressing forecast uncertainties. Day-ahead optimization, based on high-precision situational simulation, formulates a globally optimal scheduling blueprint, solving the "planning" problem. Intraday rolling correction vividly demonstrates how rolling optimization "corrects deviations." When actual renewable energy output or load deviates from the forecast, the rolling layer quickly re-optimizes, adjusting the output of flexible resources such as energy storage and gas turbines to maintain the system's power balance in real time. This is the core means of addressing forecast uncertainties, ensuring that the scheduling scheme transforms from "theoretical optimality" to "practical feasibility."

[0233] Robustness analysis: The robustness of the method of this invention was tested under three scenarios with prediction errors of 5%, 10%, and 15%, and the results are shown in Table 3.

[0234] Table 3 Scheduling performance under different prediction errors

[0235]

[0236] The results show that as the prediction error increases from 5% to 15%, the total cost rises from RMB 5,745 to RMB 5,968 (an increase of approximately 3.9%), carbon emissions increase from 7.05 tCO2 to 7.42 tCO2 (an increase of approximately 5.2%), and the renewable energy utilization rate decreases from 88.9% to 85.6%. Even with a 15% prediction error, the system still maintains good economic operation and low-carbon performance, demonstrating the robustness of this method.

[0237] In summary, the method proposed in the application can systematically solve the core problems of multi-dimensional uncertainty, multi-objective coordination and model complexity in the operation optimization of low-carbon park comprehensive energy system. Through method innovation and integration, a complete technology system from "accurate perception" to "intelligent decision" to "efficient execution" is constructed. The panoramic situation deduction model proposed in the application breaks through the limitations of traditional independent prediction, providing a high-reliability and forward-looking data basis for optimal scheduling; the electricity-carbon collaborative optimization model innovatively creates a two-way coupling mechanism between carbon cost and economic cost, successfully transforming carbon emission reduction from an external constraint into an endogenous driving force, achieving the coordinated win-win of economic benefits and environmental benefits; the MILP transformation and hierarchical solving framework effectively overcomes the solving bottleneck of high-dimensional nonlinear models, ensuring the engineering practicability and robustness of the method. The example analysis fully verifies the superiority of the proposed method, which shows significant effect in improving energy efficiency, promoting renewable energy consumption and reducing carbon emissions.

[0238] The above functions, if realized in the form of software function units and sold or used as independent products, can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the application or the part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product, which is stored in a storage medium and includes instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in various embodiments of the application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0239] The preferred embodiments of the application are described in detail above. It should be understood that those skilled in the art can make many modifications and changes without creative labor based on the concept of the application. Therefore, any technical solution obtained by logical analysis, reasoning or limited experiment based on the prior art according to the concept of the application should be within the protection scope determined by the claims.

Claims

1. A method for optimizing the coordinated electricity-carbon scheduling of a low-carbon industrial park's integrated energy system, characterized by the following steps: include: A panoramic situational inference model based on multi-source information fusion and dynamic correlation analysis is used to collaboratively predict multiple state parameters by combining various learning algorithms to generate a panoramic dynamic situational scenario set. The state variables to be predicted during the generation of the panoramic dynamic situational scenario set include photovoltaic and wind power output, electricity, heat and cooling load values, the adjustable potential value of flexible loads and the grid purchase price of electricity. The panoramic situation simulation model takes variables with coupling relationships as inputs and makes predictions using LSTM. Specifically, it uses future solar intensity as a prediction feature for both photovoltaic and wind power output; ambient temperature as a prediction feature for both heat and cooling loads; and real-time electricity price as a prediction feature for both the adjustability potential of flexible loads and electrical loads. For scenarios where the relationship between features and output is static and non-temporal, the XGBoost model is used for supplementary prediction or feature importance analysis. Through collaborative prediction, the state variables at all times within a future scheduling cycle are continuously simulated to generate a deterministic panoramic situational scenario for system operation. Establish an electricity-carbon coordinated scheduling model that takes into account the carbon trading mechanism, deeply embed real-time carbon costs into the objective function, and perform Pareto optimization of economic costs and carbon emission costs. The nonlinear electricity-carbon coordinated scheduling model is transformed into a standard mixed-integer linear programming model. A hierarchical calculation framework of situation inference-day-ahead optimization-rolling correction is adopted to decompose the coordinated scheduling optimization problem into different time scales for decision-making. At the day-ahead level, a global optimization plan is formulated based on the panoramic dynamic situation. At the intraday level, deviations are corrected online through rolling optimization. The integrated energy system executes the corrected electricity-carbon coordinated scheduling plan. Based on the hierarchical calculation framework of situational inference-day-ahead optimization-rolling correction, the planning model is solved as follows: Pre-schedule panoramic situational simulation: Before the scheduling date, a deterministic prediction scenario for the future scheduling date is generated based on the panoramic situational simulation model; Electricity-carbon collaborative optimization scheduling: Before the scheduling date, the generated deterministic prediction scenario is used as known parameters and substituted into the electricity-carbon collaborative scheduling model, which has been transformed into a mixed integer linear programming model, to generate the optimal start-up and shutdown plan and output plan for all equipment on the scheduling day, i.e., the day-ahead scheduling plan; Intraday Rolling Optimization and Correction: The scheduling plan is executed in real-time on the scheduling day, and rolling corrections are performed on the scheduling plan based on a set period. The specific process is as follows: Obtain the latest actual operating status of the integrated energy system; detect the deviation between the actual operating status and the day-ahead forecast value, and calculate the forecast error; update the short-term forecast data with the current actual operating status as the initial value, and re-optimize the scheduling plan for the short period set in the future; execute the optimization results of the first time period in the rolling optimization cycle, and issue adjustment instructions to the equipment.

2. The method for electricity-carbon coordinated scheduling optimization of a low-carbon industrial park integrated energy system according to claim 1, characterized in that, The objective function of the electricity-carbon coordinated scheduling model that takes into account the carbon trading mechanism is to minimize the total system cost within the scheduling period. The total cost includes energy purchase cost, carbon trading cost, equipment operation and maintenance cost, and start-up and shutdown cost.

3. The method for electricity-carbon coordinated scheduling optimization of a low-carbon industrial park integrated energy system according to claim 2, characterized in that, The energy purchase cost includes the cost of purchasing / selling electricity from an external power grid and the cost of purchasing natural gas: In the formula: , They are respectively t The electricity price for purchasing and selling electricity from the grid at all times; , They are respectively t The power consumed from and sold to the grid at all times; for t Real-time natural gas prices; for t Total natural gas consumption of the system at any given time; The scheduling time interval; T The scheduling period is [number].

4. The method for electricity-carbon coordinated scheduling optimization of a low-carbon industrial park integrated energy system according to claim 2, characterized in that, If the total carbon emissions of the integrated energy system exceed the free allowance, additional allowances will be purchased; if not, the remaining allowances will be sold. The carbon trading costs are expressed as follows: In the formula: For carbon trading market prices; This represents the total carbon emissions of the system during the scheduling cycle. The free carbon allowances obtained by the system during the scheduling period; Scheduling cycle T Total carbon emissions from internal systems Electricity is sourced from purchased electricity and natural gas consumption. In the formula: for t Marginal carbon emission factor of the power grid at any given time; for t Power purchased from the power grid at all times; For the first i Carbon emission coefficient of a gas-fired appliance; For the first i Taiwan gas equipment t Natural gas consumption at any given time; This represents the total number of gas-fired appliances.

5. The method for electricity-carbon coordinated scheduling optimization of a low-carbon industrial park integrated energy system according to claim 1, characterized in that, The constraints of the electricity-carbon coordinated dispatch model taking into account the carbon trading mechanism include: Energy balance constraints: The sum of the power purchased by the system from the grid, the output of photovoltaic and wind power, the power generation of the gas turbine and the discharge power of the energy storage equals the sum of the power sold to the grid, the electrical load, the power consumption of the equipment and the charging power of the energy storage; the sum of the heating power of the gas boiler, gas turbine, heat pump and the heat release power of the thermal storage device equals the sum of the heat load, the charging power of the thermal storage device and the heat network loss. Equipment operation constraints include gas turbine / gas boiler operation constraints, energy storage system operation constraints, and energy conversion equipment constraints; System safety constraints: The system's power purchase and sale to the grid are between the minimum and maximum power of the grid interaction.

6. The method for electricity-carbon coordinated scheduling optimization of a low-carbon industrial park integrated energy system according to claim 1, characterized in that, The nonlinear electricity-carbon co-scheduling model is transformed into a standard mixed-integer linear programming model, as detailed below: In the electric-carbon coordinated scheduling model, the equipment fuel consumption and its output are approximated by piecewise linearization: multiple segment points are selected within the upper and lower limits of equipment output, and the fuel consumption of the corresponding segment is calculated; continuous variables as weight coefficients corresponding to the segment points and binary variables for selecting active segments are introduced for each segment to establish linearization constraints. The energy storage charging and discharging losses are treated with equivalent linearity: two binary variables are introduced to represent the charging and discharging states of the energy storage, respectively, and mutual exclusion constraints are applied; the nonlinear terms in the energy storage system operation constraint equations are replaced with equivalent terms, and together with the additional constraints, they form a linear constraint set. The additional constraints, by introducing equivalent power variables, transform the nonlinear charging and discharging efficiency multiplication relationship into a variable definition relationship under linear constraints.

7. A power-carbon coordinated scheduling optimization device for an integrated energy system, comprising a memory, a processor, and a program stored in the memory, characterized in that, When the processor executes the program, it implements the electricity-carbon coordinated scheduling optimization method for the integrated energy system of low-carbon parks as described in any one of claims 1-6.

8. A storage medium having a program stored thereon, characterized in that, When the program is executed, it implements the electricity-carbon coordinated scheduling optimization method for the integrated energy system of low-carbon parks as described in any one of claims 1-6.

Images