Distributed collaborative optimization scheduling method for multi-park integrated energy system

Abstract

The present invention provides a distributed collaborative optimization scheduling method for a multi-park integrated energy system. Based on a deterministic model and combined with the uncertainty of source and load, a two-stage robust optimization scheduling model for a multi-park integrated energy system is constructed. The interactive energy between parks is used as the decision variable of the first stage, and the source and load uncertainty set is brought into the second stage. The output of equipment within the park and the energy purchased from the external network are used as the decision variables of the second stage. The nested CCG algorithm is used to iteratively solve the two-stage robust model to solve the second-stage two-level mixed integer programming problem. Further collaborative optimization is performed between different parks. Based on the asymmetric Nash bargaining theory and through relevant transformations, the problem is split into a social benefit maximization sub-problem and a benefit distribution sub-problem. The parallel ADMM method is used to solve it and obtain the final scheduling solution. The present invention can achieve a balance between the economic operation and robustness of each park while taking into account the privacy and interests of each park.

Description

Technical Field

[0001] The present invention belongs to the technical field of integrated energy systems, and relates to a distributed collaborative optimization scheduling method and system for multi-park integrated energy systems, in particular to a distributed collaborative optimization scheduling method and system for park integrated energy systems under source and load uncertainty. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] In the context of energy crises and environmental degradation, integrated energy systems (IESs) couple multiple heterogeneous energy flows to achieve flexible energy utilization and improved efficiency, making them a powerful means of achieving energy transformation. Based on application and geographic scale, park-level IEs can effectively adapt to local resource and energy usage characteristics. Different parks have different resource allocations and energy usage characteristics. By coordinating and interconnecting these systems, a multi-park IE system can effectively improve energy supply efficiency and reliability. However, multi-park IE systems are characterized by multiple agents and strong coupling. They must consider not only the transformation of multiple energy flows within each park but also the interaction of multiple energy types between parks. Therefore, optimizing their scheduling presents significant challenges.

[0004] However, current research focuses on deterministic optimization scheduling, ignoring the impact of source-load uncertainty on the system. This makes existing deterministic optimization no longer applicable. Therefore, it is necessary to consider not only the economic efficiency of the system but also its robustness; both are indispensable. Furthermore, while ensuring both robustness and economic efficiency, there is little research addressing the reasonable redistribution of benefits under the collaborative efforts of multiple entities. Summary of the Invention

[0005] In order to solve the above problems, the present invention proposes a distributed collaborative optimization scheduling method for a multi-park integrated energy system. The present invention can achieve a balance between the economy and robustness of the overall system operation while taking into account the privacy and interests of each park.

[0006] According to some embodiments, the present invention adopts the following technical solutions:

[0007] A distributed collaborative optimization scheduling method for a multi-park integrated energy system includes the following steps:

[0008] Establish a deterministic model of the multi-park integrated energy system and impose constraints;

[0009] Determine the range of source and load uncertainty variables within the park's integrated energy system and construct an uncertainty set for characterization;

[0010] Based on the deterministic model and combined with the uncertainty of source and load, a two-stage robust optimization scheduling model for the integrated energy system of the park is constructed. The interactive energy between parks is used as the decision variable in the first stage. The source and load uncertainty set is brought into the second stage, and the output of internal equipment in the park and the energy purchased from the external network are used as the decision variables in the second stage.

[0011] The model is split into the outer main problem and sub-problems, and the outer sub-problems are split into the inner main and sub-problems. The inner main and sub-problems are iteratively solved until convergence, and then the outer main and sub-problems are iterated until the convergence conditions are met and the optimal solution is calculated;

[0012] Collaborative optimization is carried out between different parks. Based on the asymmetric Nash bargaining theory, it is divided into the social benefit maximization sub-problem and the benefit distribution sub-problem. The two sub-problems are solved to obtain the final scheduling plan.

[0013] As an optional implementation method, the specific process of establishing a deterministic model of a multi-park integrated energy system includes:

[0014] The objective function of the deterministic model is to minimize the sum of the operating costs of each park. The operating costs include energy purchase costs, equipment operation and maintenance costs, and carbon trading costs in the independent operation mode, and energy trading costs and network access fees with other parks in the interconnected operation mode.

[0015] The constraints include equipment output constraints, energy balance constraints and tie line transmission power constraints.

[0016] As an optional implementation method, the specific process of determining the range of variation of source-load uncertainty variables within the park's integrated energy system includes applying a box-type uncertainty set to describe wind power photovoltaic, cooling and heating power loads, and the temporal correlation between wind turbines and photovoltaics is represented by a time correlation constraint model of the Pearson correlation coefficient.

[0017] As an alternative implementation, based on a deterministic model and taking into account the uncertainty of sources and loads, a two-stage robust optimization scheduling model uses the first stage to optimize the inter-park interaction power value based on the network transmission costs of the park. The second stage of the model uses the interaction power values from the first stage as a basis to solve the optimal scheduling strategy within the park and the worst-case scenario. When the interaction value in the first stage is zero, the model becomes an independent robust model for each park under uncertainty. Constraints include interaction power constraints, internal equality constraints, and constraints on the non-negativity of internal decision variables.

[0018] As an optional implementation method, an improved nested CCG algorithm is used to solve the problem, and the model is split into an outer main problem and sub-problems. The outer main problem is solved to obtain the traded electric heat between parks and update the lower bound of the algorithm; the outer sub-problem is a max-min double-layer optimization problem, which is a double-layer mixed integer linear programming problem. The conventional method of converting the double layer into a single layer fails. Drawing on the idea of the CCG algorithm, the outer sub-problem is split into inner main and sub-problems, and the inner main and sub-problems are iteratively solved until convergence, thereby obtaining the solution to the max-min double-layer optimization problem.

[0019] As an optional implementation method, the specific process of collaborative optimization between different parks includes coupling between parks reflected in interconnection power and transaction price, and introducing consistency variables to achieve decoupling of coupling variables.

[0020] Bargaining power is quantified by constructing a contribution function based on the transaction volume of electric and thermal energy. The total energy provided and total energy obtained by each park during optimization are then determined. A contribution function is then constructed for each park and added to the Nash bargaining expression, resulting in an asymmetric Nash bargaining profit distribution model for multiple parks. This model is then transformed into a social benefit maximization subproblem and a profit distribution subproblem. The social benefit maximization subproblem aims to optimize overall benefits under the worst-case scenario of uncertain variables, namely, the sum of the operating objectives of each park. Its coupled variables are only the interactive electric and thermal power values. Each park is iteratively optimized until convergence conditions are met, resulting in an optimized scheduling plan and interactive electric and thermal output for each park. The profit distribution subproblem aims to balance and optimize the interests of all parties and achieve a reasonable distribution of cooperative benefits. Through iterative solutions for each park, the energy interaction price for each park is obtained, ultimately determining the costs of each park.

[0021] A distributed collaborative optimization scheduling method for a multi-park integrated energy system, comprising:

[0022] The deterministic model building module is configured to build a deterministic model of the multi-park integrated energy system and impose constraints;

[0023] The uncertainty consideration module is configured to determine the range of source and load uncertainty variables in the park's integrated energy system and construct an uncertainty set for characterization;

[0024] The optimization scheduling model construction module is configured to construct a two-stage robust optimization scheduling model for the park's integrated energy system based on a deterministic model and combined with the uncertainty of sources and loads. The interactive energy between parks is used as the decision variable in the first stage, and the source and load uncertainty set is brought into the second stage, with the output of internal equipment in the park and the energy purchased from the external network as the decision variables in the second stage.

[0025] The solution module is configured to split the model into an outer main problem and subproblems, and then split the outer subproblems into inner main and subproblems, iteratively solve the inner main and subproblems until convergence, and then iterate the outer main and subproblems until convergence conditions are met, and calculate the optimal solution;

[0026] The collaborative optimization module is configured to perform collaborative optimization between different parks. Based on the asymmetric Nash bargaining theory, it is divided into the social benefit maximization sub-problem and the benefit distribution sub-problem. The two sub-problems are solved to obtain the final scheduling plan.

[0027] A computer-readable storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor of a terminal device and executing the steps in the method.

[0028] A terminal device includes a processor and a computer-readable storage medium, wherein the processor is used to implement various instructions; the computer-readable storage medium is used to store multiple instructions, and the instructions are suitable for being loaded by the processor and executing the steps in the method.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] The present invention constructs a two-stage robust optimization model to cope with the uncertainty of source and load, and adopts a nested CCG algorithm to solve it, thereby overcoming the difficulty that the inner sub-problem cannot be directly converted into a single-layer sub-problem when it contains 0-1 variables; for the model of electric and thermal multi-energy interconnection between different parks, it is split into a social benefit maximization sub-problem and a benefit distribution sub-problem based on the asymmetric Nash bargaining theory, and can achieve a fully distributed solution, protecting the privacy and interests of all participating entities, and achieving a balance between the economy and robustness of the entire system. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0032] Figure 1 This is a structural diagram of a multi-park integrated energy system according to embodiment 1 of the present invention;

[0033] Figure 2 This is a flowchart of a nested CCG algorithm for solving a two-stage robust optimization of a park integrated energy system according to a first embodiment of the present invention;

[0034] Figure 3 This is a flow chart of distributed collaborative solution of a multi-park integrated energy system according to embodiment 1 of the present invention;

[0035] Figure 4 This is a graph showing the iterative solution of the distributed algorithm for each park in the first embodiment of the present invention;

[0036] Figure 5 This is an interactive electric heating power diagram between parks in Example 1 of the present invention;

[0037] Figure 6 This is the worst scenario diagram of the park in Example 1 of the present invention;

[0038] Figure 7 This is a price chart of electric and thermal energy traded between parks in Example 1 of the present invention;

[0039] Figure 8 This is a diagram of the electric energy dispatching strategy of the integrated energy system of each park in Example 1 of the present invention;

[0040] Figure 9 This is a diagram of a thermal energy scheduling strategy for a multi-park integrated energy system according to embodiment 1 of the present invention. DETAILED DESCRIPTION

[0041] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0042] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0043] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0044] Example 1

[0045] A two-stage robust distributed collaborative optimization scheduling method for a multi-park integrated energy system is proposed. The multi-park integrated energy system described in the paper consists of three park-level integrated energy systems. Electricity and thermal energy can be shared and traded between parks. Each park consists of renewable energy, energy conversion equipment, energy storage equipment, and external energy ports. The overall system architecture is as follows: Figure 1 The optimization scheduling method described in this embodiment realizes distributed solution under the condition of source-load uncertainty, obtains the scheduling strategy of each park, takes into account the robustness and economy of the model, and reasonably distributes the residual benefits of cooperation among the main bodies of multi-energy collaborative interaction in each park.

[0046] 1. Deterministic model of multi-park integrated energy system

[0047] 1.1 Objective Function

[0048] The objective function of the conventional deterministic model proposed in this embodiment is to optimize the overall cost, that is, to minimize the sum of the operating costs of each park. Its expression is as follows:

[0049]

[0050] Where N is the number of parks, F i CO The cost of interconnected operation of campus i.

[0051] The operating costs of each park are composed of energy purchase costs, equipment operation and maintenance costs, and carbon trading costs in the independent operation mode, and energy transaction costs and network fees between other parks in the interconnected operation mode. The expression is as follows:

[0052]

[0053] Where C i,buy is the energy purchase cost of park i, is the carbon trading cost of park i, C i,om is the operation and maintenance cost of each device in park i, C i,ex is the energy transaction cost of park i and C i,net is the network access fee of park i, F i NO is the independent operating cost of park i.

[0054] The energy purchase cost is the sum of the electricity purchase cost from the external power grid and the gas purchase cost from the external natural gas grid. The expression is as follows:

[0055]

[0056] in, are the electricity purchase price and gas purchase price at time t, are the electricity purchased power and gas purchased power at time t respectively.

[0057] Carbon trading consists of three parts: carbon emission quotas, actual carbon emissions and carbon emission rights trading. The emission sources are electricity purchased from the external grid, gas turbines and gas boilers.

[0058]

[0059] Where, To fix the carbon trading price, is the carbon emission quota for the day, The actual carbon emissions for the day.

[0060]

[0061]

[0062] Where λ e ,λ h are the carbon quota coefficients per unit of electricity and heat production, respectively. After electricity is converted into heat, the equivalent heat supply emissions of gas turbines and gas boilers are similar. eh is the conversion coefficient of electricity and heat; ε e , ε h are the carbon emission coefficients per unit of electricity and heat production, respectively.

[0063] The operation of each device in the park will cause certain wear and tear to the equipment, and the equipment also needs to be repaired and maintained, which will be uniformly converted into the operation and maintenance costs of each device.

[0064]

[0065] Among them, K Trans , K CHP , K PV , K WT , K GB , K EC , K AC , K BAT , K TES , are the unit operation and maintenance costs of off-grid purchased energy transformers, gas turbines, fans, photovoltaics, gas boilers, electric refrigerators, absorption refrigerators, batteries and heat storage devices, respectively. The unit is Yuan / kW, and the power is based on the input end of the equipment.

[0066] The cost of transactions between park i and other parks, including electricity and heat transactions.

[0067]

[0068] Where, Ω i is the collection of parks that are interconnected with park i, and is the amount of electric power and thermal power that park i expects to exchange with interconnected park j at time t. When its value is positive, it means that park i purchases energy from park j, and when its value is negative, it means that park i sells energy to park j. and are the corresponding interactive electricity and heat energy prices, respectively. It is worth noting that since energy transactions are only conducted internally, transaction costs offset each other and their sum is zero, i.e., internal balance. Therefore, transaction costs can be ignored when calculating the overall cost.

[0069] The process of energy transmission between parks is called network transmission, which requires a certain fee to be paid according to the transmission power and transmission distance, and is shared by both parties of the transaction.

[0070]

[0071] Where, and are the unit transmission costs of electric energy and thermal energy exchanged between park i and park j at time t, which are related to the transmission distance between the parks.

[0072] 1.2 Constraints

[0073] (1) Equipment output constraints:

[0074] (a) Gas turbine

[0075]

[0076]

[0077]

[0078]

[0079] Where, and are the electricity generation and heat generation of gas turbine i in the park at time t, is the amount of natural gas consumed by the gas turbine at time t, η i,CHP,e and η i,CHP,h are the relevant efficiencies of electricity and heat production, respectively. and They are the upper and lower limits of the power generation of gas turbine i in the park, and They are the upper and lower limits of the heat production power of gas turbine i in the park.

[0080] (b) Gas boiler

[0081]

[0082]

[0083] Where, and is the heating power and gas consumption power of gas boiler i in the park at time t, η i,GB,h is the heating efficiency of the gas boiler, and They are the maximum and minimum heating power of gas boiler i in the park respectively.

[0084] (c) Electric Refrigerator

[0085]

[0086]

[0087] Where, and are the cooling power and power consumption of the electric refrigerator at time t, η i,EC is the cooling coefficient of the electric refrigerator; and are the minimum and maximum cooling power of the electric refrigerator respectively.

[0088] (d) Absorption refrigeration machine

[0089]

[0090]

[0091] Where, and is the refrigeration coefficient of the absorption chiller, η i,AC is the refrigeration coefficient of the absorption chiller, and are the minimum and maximum cooling power of the absorption chiller respectively.

[0092] (e)Batteries

[0093]

[0094]

[0095]

[0096]

[0097]

[0098]

[0099] Where, is the battery storage energy at time t, δ i,e The self-consumption coefficient of the battery, and are the charging and discharging powers at time t, η i,e,ch and η i,e,dis are the charging and discharging efficiencies of the battery, and are the minimum and maximum energy storage capacities of the battery, and are the maximum power of charging and discharging respectively, and It is the charging status flag. and are the initial energy storage capacity and final energy storage capacity of the battery respectively.

[0100] (g) Heat storage device

[0101]

[0102]

[0103]

[0104]

[0105]

[0106]

[0107] Where, is the energy storage capacity of the thermal storage device at time t, δ i,h is the self-loss coefficient of the heat storage device, and are the charging and dissipating powers at time t, η i,h,ch and η i,h,dis are the charging and discharging efficiencies of the heat storage device, and are the minimum and maximum energy storage capacities of the thermal storage device, and are the maximum powers of charging and discharging heat, and It is the status flag of hot charging. and are the initial energy storage capacity and final energy storage capacity of the thermal storage device respectively.

[0108] (2) Transmission power constraints of tie lines

[0109] (a) External online shopping can be constrained

[0110]

[0111]

[0112] Where, and are the electricity power and gas power purchased from the outside by park i at time t, and are the maximum power of electricity and gas purchased from outside by park i, respectively.

[0113] (b) Inter-park interaction power constraints

[0114]

[0115]

[0116]

[0117]

[0118]

[0119]

[0120] Where, and is the electric energy and thermal energy exchanged between park i and park j. When its value is positive, it means park i purchases energy from park j. When its value is negative, it means park i sells energy to park j. The electric and thermal energy exchange quantities of the opposite entities are opposite to each other. and are the transaction prices of the corresponding electric and thermal energy respectively. The transaction prices of electricity and heat between parks are equal. and is the maximum exchange of electrical energy and thermal energy between park i and park j.

[0121] (3) Energy balance constraints:

[0122]

[0123]

[0124]

[0125] Where, and are the electricity, heating and cooling loads of park i at time t, respectively. is the amount of electricity traded between park i and other external parks, is the amount of heat energy traded between park i and other external parks.

[0126] So far, a deterministic low-carbon scheduling model for a multi-park integrated energy system has been established. However, the source-load forecast data is random, and it is necessary to consider optimal scheduling under uncertainty to make the system more robust.

[0127] 2. Two-stage robust optimization scheduling model for multi-park integrated energy systems

[0128] 2.1 Source-load uncertainty set

[0129] The output and load of renewable energy in the system have great uncertainty, making it difficult to reflect the impact of source-load uncertainty on the system. Taking into account the uncertainty of source load, a box-type uncertainty set is used to describe wind power, photovoltaic power, cooling and heating power loads. The specific expression is as follows:

[0130]

[0131]

[0132]

[0133]

[0134]

[0135] Among them, D i is the set of uncertain parameters of park i, and are the predicted output and actual output of renewable energy sources in park i, namely wind power and photovoltaic variables, respectively. and are the predicted and actual values of cooling, heating and electricity loads respectively, and are auxiliary variables, is the power range of source load fluctuation, Γ i,d is the uncertainty value, which describes the degree of uncertainty of the source load. The proportion of different uncertainties can be obtained by the number of uncertain periods to the total number of periods.

[0136] Generally, there is a certain temporal correlation between wind turbines and photovoltaics. The temporal correlation constraint model of the Pearson correlation coefficient is as follows:

[0137]

[0138]

[0139]

[0140]

[0141]

[0142]

[0143] Where, are the change signs of the deviations in adjacent periods, is the total uncertainty of time correlation. The smaller its value is, the greater the time correlation is and the lower the robustness is.

[0144] 2.2 Two-stage robust optimization model and solution of the park integrated energy system

[0145] For the established multi-park deterministic model, combined with the uncertainty of source and load, a two-stage robust optimization model for multi-park is established, and its objective function expression is as follows:

[0146]

[0147] For each park, it is a two-stage robust optimization model, namely F i CO The system target is the sum of the targets of each park under the condition of source and load uncertainty, and park i is used as an example to illustrate. The first stage of the two-stage robust model of the park is the network transmission fee of the park, which optimizes the interactive power value between parks; the second stage of the model is based on the interactive power value of the first stage to solve the optimal scheduling strategy within the park and the worst scenario. When the interaction amount in the first stage is 0, it is an independent robust model of each park under uncertainty, that is, F i NO .

[0148] For the convenience of description, the two-stage robust optimization model of the park is simplified to its matrix form:

[0149]

[0150] stCy i ≤g i

[0151] Ax i +Bd i +Dy i =c i

[0152] Ex i +Fz i ≤h i

[0153] x i ≥0

[0154] Where y i is the first-stage decision variable of park i, namely, the electric-heat interaction; d i are uncertain parameters, including wind power, photovoltaic, and cooling and heating power loads; x i is the decision variable of the second stage, which represents the energy scheduling strategy of the park. The first constraint is the interactive power constraint, the second is the internal equality constraint, and the third is the related inequality constraint, where z i is a 0-1 variable, indicating the charge and discharge status of the energy storage. The last inequality constraint limits the non-negativity of the internal decision variable.

[0155] 2.3 Two-stage robust optimization problem solution based on nested CCG algorithm

[0156] The two-stage robust optimization model for the park is a min-max-min three-level optimization problem, which is difficult to solve directly. An improved nested CCG algorithm is used to solve it, splitting the model into an outer main problem and subproblems. The outer main problem is expressed as follows:

[0157] mina T y i +γ i

[0158] stCy i ≤g i

[0159]

[0160]

[0161]

[0162]

[0163] Among them, ξ i is an auxiliary variable introduced to characterize the maximum value of the objective function in the second stage; k is the number of iterations, is the second-stage decision variable corresponding to the l-th iteration, The worst case scenario of the uncertain variables solved by the sub-problem at the lth iteration is a constant in the main problem. By solving the main problem, the inter-park transaction electric heat is obtained and the algorithm lower bound LB is updated. out =a T y i +γ.

[0164] The outer subproblem is a max-min two-level optimization problem, which is a mixed integer linear programming problem. Drawing on the traditional CCG algorithm idea, the outer subproblem is split into the inner main and subproblems. The inner main and subproblems are iteratively solved until convergence, thus obtaining the solution of the max-min two-level optimization problem. The specific expression is as follows:

[0165] The solution model of the inner master problem is as follows:

[0166] maxξ i

[0167]

[0168]

[0169]

[0170]

[0171]

[0172]

[0173]

[0174] Among them, ξ i is the auxiliary variable introduced, which represents the optimal cut target value of the inner sub-problem, k is the number of iterations so far, corresponding to k groups of 0-1 variable scenarios, are the decision variable values of the first stage passed to the second stage, is the decision variable of the inner master problem at the rth iteration, is the optimal solution of the inner subproblem at the rth iteration; and is the dual variable, and For the complementary relaxation condition in the formula, due to the existence of nonlinear terms, the large M method is used for equivalent linearization processing, and the transformation form is as follows:

[0175]

[0176]

[0177] Among them, M is a large positive number. A 0-1 auxiliary variable.

[0178] By solving the inner master problem, the value of the uncertain variable is obtained Update the upper bound UB of the inner subproblem in =θ, and bring the obtained uncertain scenario into the inner sub-problem. The specific solution model is as follows:

[0179] minb T x i

[0180]

[0181] Ex i +Fz i ≤h i

[0182] x i ≥0,z i ∈{0,1}

[0183] The inner sub-problems SPs solve the second-stage optimization problem under uncertain scenarios to obtain the optimal solution for the energy storage state. and decision variables And update the lower bound If UB in -LBin ≤ε in It means that the algorithm converges and stops iteration, and updates the lower bound of the objective value of the outer subproblem Otherwise, repeat the iterative solution of the main and sub-problems. When the algorithm converges and obtains the optimal solution When , the worst scenario in the second stage problem is obtained. The subsequent steps are consistent with the basic CCG algorithm, and the outer main subproblem is iterated. The upper and lower bounds are continuously approached until UB is satisfied. out -LB out ≤ε out .

[0184] The solution flow chart of the nested CCG algorithm is as follows Figure 2 shown.

[0185] 3. Multi-park distributed collaborative optimization solution based on two-stage robustness

[0186] 3.1 Coupling decoupling

[0187] The coupling between parks is reflected in the interconnected power and transaction price. The decoupling of the coupling variables is achieved by introducing consistency variables. The expression of the introduced consistency variables is as follows:

[0188]

[0189]

[0190]

[0191]

[0192] in, and are the consistency variables of the interactive electric energy and thermal energy between park i and park j, respectively. Since the interactive power has the characteristics of positive buying and negative selling, a negative sign is added at the opposite end; and are the consistency variables of electricity transaction price and thermal energy transaction price respectively.

[0193] 3.2 Distributed Collaborative Solution Based on Nash Bargaining in Multiple Parks

[0194] This embodiment uses the transaction volume of electric and thermal energy to construct a contribution function to quantify the bargaining power. First, the total energy value provided by each park during optimization is obtained. and total energy gained The specific expression is as follows:

[0195]

[0196]

[0197] Then, the contribution function of each park is constructed. The specific expression is as follows:

[0198]

[0199] in, and The negotiated value is always non-negative. In the absence of energy sharing, the negotiated value is 0, which means there is no benefit distribution link.

[0200] Finally, the contribution function is added to the Nash bargaining expression to obtain the multi-park asymmetric Nash bargaining income distribution model, which is expressed as follows:

[0201]

[0202]

[0203] Among them, F i CO is the utility function for participating in the cooperative game, is the bargaining strategy, N is the number of entities participating in the bargaining, that is, the total number of parks, F i NO In order to ensure the benefits of participating entities before cooperation, that is, the negotiation breakdown point in the cooperative game, it is necessary to meet the constraints of improving the benefits of each entity before and after cooperation.

[0204] When the above formula reaches its maximum value, the following formula is satisfied:

[0205]

[0206] Since the total transaction cost satisfies the internal balance, Therefore, transaction costs can be omitted. Substitute it into the equivalent expression and take its logarithm, then the objective function can be transformed into:

[0207]

[0208] Because F i NO The cost of independent operation of the participating entities can be regarded as a known quantity. At the same time, due to the strictly monotonically increasing characteristics of the natural logarithm function, It is always greater than 0. The original formula can be equivalently expressed and transformed as follows:

[0209]

[0210] The above equation is the social benefit maximization subproblem. The goal is to optimize the overall benefits under the worst-case scenario of uncertain variables, which is the sum of the operating objectives of each park. The only coupled variable is the interactive electric and thermal power value. The iterative optimization solution process for park i for the social benefit maximization subproblem is as follows:

[0211]

[0212]

[0213]

[0214]

[0215]

[0216] The objective function consists of a two-stage robust model of the park and a penalty term for coupled variables, k is the number of iterations, and are the dual variables corresponding to the interactive electrical energy and thermal energy coupling constraints at the kth iteration, and ρ1 is the penalty factor for the overall benefit optimization problem.

[0217] The termination condition of the iteration can also be determined independently by each park. The solution of the subproblem of each park needs to meet the requirements of its primal residual and dual residual:

[0218]

[0219]

[0220] The subproblems are solved iteratively until they meet the convergence accuracy, and the optimized scheduling plan and interactive electric and heat consumption of each park are obtained. After obtaining the optimal result of overall benefit, the logarithm of the original asymmetric objective function is taken to obtain the benefit distribution subproblem:

[0221]

[0222] In the formula, the decision variable is the transaction price of electricity and heat. This problem can also be solved using the distributed ADMM method to achieve a reasonable distribution of the transaction surplus value of each park. The iterative solution process is as follows:

[0223]

[0224]

[0225]

[0226]

[0227]

[0228] in, are the decision variables of the profit distribution subproblem, including the transaction electricity price and transaction heat price. ρ2 is the penalty factor corresponding to the profit distribution subproblem. Its iteration termination condition is consistent with that of the social benefit maximization subproblem.

[0229] The overall solution process of multi-park collaboration is as follows Figure 3 shown.

[0230] Thus, a two-stage robust multi-park distributed collaborative optimization model has been established, using a parallel ADMM algorithm and a nested CCG algorithm for solution. The internal problems within each park in the algorithm are solved using the CPLEX and MOSEK solvers. To verify the advantages of multi-energy interconnection in a multi-park integrated energy system, considering source and load uncertainty, and the benefits of improving and distributing benefits under interconnected parks, the following four scenarios are specifically set for analysis:

[0231] Scenario 1: Taking into account the uncertainty of source and load, each park operates electricity and heat together and uses a distributed solution.

[0232] Scenario 2: Taking into account the uncertainty of source and load, each park operates electricity and heat together, and the solution is centralized.

[0233] Scenario 3: Ignoring the uncertainty of source and load, each park operates electricity and heat together, and a distributed solution is used.

[0234] Scenario 4: Taking into account the uncertainty of source load, each park operates independently and solves the problem separately.

[0235] The overall optimization scheduling results of the system under different scenarios are shown in Table 1.

[0236] Table 1 Overall optimization scheduling results under different scenarios

[0237]

[0238] (1) Analysis of results of different solution methods

[0239] Comparing the optimization results for Scenario 1 and Scenario 2 shows that the total system costs obtained by the centralized and distributed methods are essentially the same. The distributed method's cost is slightly higher by 2.25 yuan, with an error of only 0.0026%, meeting the required accuracy. Taking Scenario 1 as an example, energy purchase costs are the primary cost item, accounting for 94.47%, while network fees and operation and maintenance costs contribute 0.24% and 6.60% respectively. Due to the introduction of a carbon trading mechanism, the system prioritizes natural gas to meet load requirements, ensuring environmental friendliness. Carbon emissions are 90.31 tons, and the carbon trading cost is negative, reducing total costs by 1.31%.

[0240] The centralized method needs to collect all the internal data of each park, while the distributed method only needs the interactive power information between parks, taking into account the differentiated interests and privacy protection of each subject. The distributed algorithm can meet the requirements of accuracy and privacy and has stronger rationality. The iterative solution process of the algorithm's social benefit maximization sub-problem is as follows: Figure 4 As shown in the figure, the electric and thermal energy interaction power between the parks is obtained as follows Figure 5 shown.

[0241] Depend on Figure 4 It can be seen that in the initial iterations, each park primarily optimized for cost, somewhat neglecting the coupling constraints between parks. As the number of iterations increased, the importance of the coupling constraint penalty gradually increased, encouraging the interaction power to meet the coupling constraints. The cost targets for each park gradually converged, meeting the convergence requirement at 76 iterations. The total cost of the distributed solution also converged, consistent with the results obtained by the centralized method. The different trends in cost values for each park also reflect the inter-park power dynamics.

[0242] Depend on Figure 5 As can be seen, the various parks can collaborate through electricity and heat trading, selling and purchasing electricity and heat at different times. Park 1 has a higher load during the day and a lower load at night, and its wind power output is also higher at night. During the day, it purchases energy from Parks 2 and 3 and sells it at night. The electricity exchange trends between Parks 2 and 3 are essentially the same, and opposite to Park 1. Park 2, with its high heat load during the day, purchases heat from other parks. At all other times, energy exchange between parks is optimized for optimal use. The sum of the exchange volume for the three types of energy at each time period is zero, meeting internal energy balance requirements.

[0243] (2) Whether to consider the uncertainty of source load in the result analysis

[0244] In order to verify the effectiveness of the constructed two-stage robust optimization model, the source-load data and load loss amount of scenario 1 and scenario 3 were compared and analyzed. By solving scenario 1, the worst source-load scenario of each park was obtained. Taking the result of park 1 as an example, the worst scenario of its two-stage robust scheduling is as follows: Figure 6 As shown in Figure 2, (a), (b), (c), and (d) represent the scenarios for electric load, thermal load, cooling load, and photovoltaic load, respectively. A box-type uncertainty set is used to process source and load parameters. The selection of uncertainty values will greatly affect the optimization results. 300 scenarios are randomly selected within the range of uncertainty variables. The mean of the overall load loss under each scenario is calculated as an evaluation indicator to represent system risk. The system operating cost and load loss under different uncertainty ratios are shown in Table 2.

[0245] Comparing the optimization results of Scenario 1 and Scenario 3 with the worst-case scenario for source and load, we can see that accounting for source and load uncertainty significantly increases total system cost. This is primarily due to increased external energy purchases, increased backup capacity, and enhanced reliability. PV reaches its minimum value during the 10-15 hour period, when power generation is high, while each load reaches its maximum value during peak prices and energy demand, consistent with the worst-case scenario.

[0246] Table 2 Sensitivity test results of uncertainty ratio

[0247]

[0248]

[0249] Table 2 shows that when the uncertainty ratio is 0, the results of the two-stage robust optimization and deterministic optimization are consistent. As the uncertainty ratio increases, more time periods of source and load conditions are considered in day-ahead scheduling, increasing operating costs but reducing risk, enhancing robustness, and making the scheduling plan more conservative. When the uncertainty ratio is 1, the load loss is 0, making the scheduling plan the most conservative. High reliability comes with high costs, but high risk can also bring high returns. In actual scheduling, decision makers should comprehensively consider economy and robustness, weigh trade-offs based on actual conditions, and set reasonable uncertainty values to account for source and load uncertainty.

[0250] (3) Analysis of benefit distribution results

[0251] By comparing scenario one and scenario four, it can be seen that the total cost of interconnected operation is 3583.02 yuan lower than that of independent operation, reaching 4.1% of the cost. Through the interconnection of electricity and heat, each park reduces the amount of energy purchased from the external power grid and natural gas network, reduces carbon emissions, and thus reduces energy purchase costs and carbon trading costs. The interconnection of multiple energy sources further improves the equipment utilization rate of each park, so its operation and maintenance costs have increased. Compared with independent operation, energy exchange between parks requires a certain amount of network transmission fee, but the value is small. Taking all costs into consideration, the total cost is reduced. It can be seen that cooperation and backup between parks can improve the overall energy consumption level. For each park, the inter-park electricity and heat transaction price obtained by the benefit distribution sub-problem is as follows: Figure 7 As shown in Table 3, the asymmetric Nash bargaining benefit distribution results are shown in Table 3.

[0252] Depend on Figure 7As can be seen, the negotiated price for electricity traded between parks is lower than the external grid price at all times, and the negotiated price for thermal energy is also relatively low. Therefore, when a park's internal energy supply is insufficient, it will first purchase energy from other parks with sufficient energy, and then from the higher-level energy network. This allows both parties to achieve better returns and reduce operating costs. Therefore, each park can determine the transaction cost based on the exchanged energy power and price, achieving a balanced distribution of benefits and final costs for each park.

[0253] Table 3 Distribution of benefits from multi-park interconnection Unit: Yuan

[0254]

[0255] As can be seen from Table 3, the transaction costs of each park are both positive and negative, and the sum of the transaction costs is 0, which meets the requirements of internal balance in energy transactions. The energy transaction cost of Park 1 is negative, which means that it sells more energy to other parks, and its final cost is reduced. The energy transaction cost of Park 2 is positive, which means that it purchases more energy from other parks, and its total cost has increased. The same is true for Park 3. Since the price of traded energy is lower than that of energy purchased from the external network, the multi-energy interaction and coordination between parks further optimizes the park scheduling strategy. The final cost of each park is lower than the cost of independent operation. The cost reduction ratios are 7.6%, 4.1% and 1.2% respectively. Each park can actively participate in the cooperation. The reduced costs are allocated based on the comprehensive contribution of traded electricity and heat energy. Compared with the standard form of Nash bargaining, the cost reduction of each park is consistent. This method can

[0256] Reflect the differences in the roles played by each park in the cooperation and achieve fair and reasonable distribution of benefits.

[0257] Taking scenario 1 as an example, the dispatching strategies of the integrated energy systems of each park obtained are analyzed in detail to further verify the rationality of the results.

[0258] Figure 8This is the power dispatch strategy for scenario three, where (a), (b), and (c) correspond to parks 1, 2, and 3, respectively. At any point in the dispatch cycle, each park maintains a balanced supply and demand for electricity. Each park's load is primarily met by renewable energy and gas turbines, with any shortfalls met by park-to-park power, power purchased from the grid, and batteries. The optimized dispatch strategy is primarily influenced by time-of-use electricity prices. Each park's grid power purchases are concentrated during low-price periods, allowing batteries to charge during these times and discharge during peak electricity prices and energy demand periods, effectively shaving peak loads and filling valleys. Taking Park 1 as an example, during the hours of midnight to 6:00 a.m. and 11:00 p.m. to midnight, when load and prices are low, Park 1 purchases electricity from the grid and uses energy storage, fully utilizing wind power during these times and selling electricity to other parks. During the hour of 11:00 a.m. to 4:00 p.m., when load levels are higher, Park 1 primarily generates electricity from its gas turbines, discharges its batteries, and purchases electricity from other parks. During the hour of 6:00 p.m. to 10:00 p.m., when load levels are lower and prices are higher, Park 1 fully utilizes energy storage and renewable energy, and sells electricity to other parks. The dispatch strategy analysis for other parks is similar to the above analysis and is not repeated here.

[0259] Figure 9 This is the thermal energy scheduling strategy under scenario three, where (a), (b), and (c) correspond to parks 1, 2, and 3, respectively. At any time within the scheduling cycle, each park meets the balance of thermal energy supply and demand, and the thermal load of each park is provided by gas turbines, gas boilers, heat storage devices, and interactive heat energy between parks. In park 1, from 1:00 to 6:00 and from 23:00 to 24:00, the thermal load is mainly met by gas boilers, and the heat storage device releases a certain amount of heat energy. At this time, the load level is low and heat can be sold to other parks; from 7:00 to 22:00, heat energy is mainly provided by gas turbines. Combined with the energy consumption differences of other parks, there are some interactive heat energy behaviors, and heat energy is provided to the absorption chiller. The scheduling strategies for other time periods and parks 2 and 3 are the same as the above analysis and will not be repeated here.

[0260] The cooling load energy supply form of each park is relatively simple. Based on the time-of-use electricity price, the cooling load is all met by absorption chillers during peak electricity price periods, and all met by electric chillers during normal and off-peak periods. Therefore, the cooling energy scheduling plan of each park is no longer analyzed separately.

[0261] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it 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 on the basis of the technical solution of the present invention without any creative work are still within the scope of protection of the present invention.

Claims

1. A distributed collaborative optimization scheduling method for a multi-park integrated energy system, characterized by: The following steps are involved: Establish a deterministic model of the multi-park integrated energy system and impose constraints; Determine the range of source and load uncertainty variables within the park's integrated energy system and construct an uncertainty set for characterization; Based on the deterministic model and combined with the uncertainty of source and load, a two-stage robust optimization scheduling model for the integrated energy system of the park is constructed. The interactive energy between parks is used as the decision variable in the first stage. The source and load uncertainty set is brought into the second stage, and the output of internal equipment in the park and the energy purchased from the external network are used as the decision variables in the second stage. The model is split into the outer main problem and sub-problems, and the outer sub-problems are split into the inner main and sub-problems. The inner main and sub-problems are iteratively solved until convergence, and then the outer main and sub-problems are iterated until the convergence conditions are met and the optimal solution is calculated; Collaborative optimization between different parks is carried out. Based on the asymmetric Nash bargaining theory, it is divided into the social benefit maximization sub-problem and the benefit distribution sub-problem. The two sub-problems are solved to obtain the final scheduling solution. The specific process of determining the range of source and load uncertainty variables in the park's integrated energy system includes applying boxed uncertainty sets to describe wind power, photovoltaic power, cooling and heating power loads, and the temporal correlation between wind turbines and photovoltaic power is represented by the time correlation constraint model of the Pearson correlation coefficient. Based on a deterministic model and taking into account the uncertainty of sources and loads, the first stage of the two-stage robust optimization scheduling model is to optimize the network transmission costs of the park and the interaction power value between parks. The second stage of the model uses the interaction power value of the first stage as a basis to solve the optimal scheduling strategy within the park and the worst-case scenario. When the interaction amount in the first stage is all zero, it becomes an independent robust model for each park under uncertainty. Constraints include interaction power constraints, internal equality constraints, and constraints limiting the non-negativity of internal decision variables.

2. A distributed collaborative optimization scheduling method for a multi-park integrated energy system according to claim 1, characterized in that: The specific process of establishing a deterministic model of a multi-park integrated energy system includes: The deterministic model takes the minimization of the sum of the operating costs of each park as the objective function. The operating costs include energy purchase costs, equipment operation and maintenance costs, and carbon trading costs in the independent operation mode, and energy trading costs and network access fees between other parks in the interconnected operation mode.

3. The distributed collaborative optimization scheduling method for a multi-park integrated energy system according to claim 1 is characterized in that: The constraints of the deterministic model include equipment output constraints, energy balance constraints and tie line transmission power constraints.

4. The distributed collaborative optimization scheduling method for a multi-park integrated energy system according to claim 1 is characterized in that: An improved nested CCG algorithm is used to solve the problem, splitting the model into an outer main problem and subproblems. The outer main problem is solved to obtain the inter-park transaction electric and heat amount, and the algorithm lower bound is updated. The outer subproblem is The bi-level optimization problem is a bi-level mixed integer linear programming problem. The conventional method of converting the bi-level into a single-level method fails. By referring to the CCG algorithm, the outer sub-problem is split into the inner main and sub-problems, and the inner main and sub-problems are iteratively solved until convergence, thus obtaining Solutions to the bilevel optimization problem.

5. The distributed collaborative optimization scheduling method for a multi-park integrated energy system according to claim 1 is characterized in that: The specific process of collaborative optimization between different parks includes coupling between parks reflected in interconnected power and transaction prices, and introducing consistency variables to achieve decoupling of coupling variables; The bargaining power is quantified by constructing a contribution function based on the transaction volume of electric and thermal energy, and the total energy provided and obtained by each park during optimization are obtained. The contribution function of each park is constructed and added to the Nash bargaining expression to obtain an asymmetric Nash bargaining benefit distribution model for multiple parks. After relevant transformation, it is equivalent to a social benefit maximization sub-problem and a benefit distribution sub-problem; the goal of the social benefit maximization sub-problem is to optimize the overall benefits of uncertain variables under the worst conditions, that is, the sum of the operating objectives of each park, and its coupling variables are only the interactive electric and thermal power values. Each park is iteratively optimized and solved until its convergence conditions are met, and the optimized scheduling plan and interactive electric and thermal amount of each park are obtained; the goal of the benefit distribution sub-problem is to balance and optimize the interests of all parties and realize the reasonable distribution of cooperation benefits. Through iterative solution of each park, the energy interaction price of each park is obtained, and finally the cost of each park is determined.

6. A distributed collaborative optimization scheduling system for multi-park integrated energy systems, characterized by: include: The deterministic model building module is configured to build a deterministic model of the multi-park integrated energy system and impose constraints; The uncertainty consideration module is configured to determine the range of source and load uncertainty variables in the park's integrated energy system and construct an uncertainty set for characterization; The optimization scheduling model construction module is configured to construct a two-stage robust optimization scheduling model for the park's integrated energy system based on a deterministic model and combined with the uncertainty of sources and loads. The interactive energy between parks is used as the decision variable in the first stage, and the source and load uncertainty set is brought into the second stage, with the output of internal equipment in the park and the energy purchased from the external network as the decision variables in the second stage. The solution module is configured to split the model into an outer main problem and subproblems, and then split the outer subproblems into inner main and subproblems, iteratively solve the inner main and subproblems until convergence, and then iterate the outer main and subproblems until convergence conditions are met, and calculate the optimal solution; The collaborative optimization module is configured to perform collaborative optimization between different parks. Based on the asymmetric Nash bargaining theory, it is divided into a social benefit maximization subproblem and a benefit distribution subproblem. The two subproblems are solved to obtain the final scheduling solution. The specific process of determining the range of source and load uncertainty variables in the park's integrated energy system includes applying boxed uncertainty sets to describe wind power, photovoltaic power, cooling and heating power loads, and the temporal correlation between wind turbines and photovoltaic power is represented by the time correlation constraint model of the Pearson correlation coefficient. Based on a deterministic model and taking into account the uncertainty of sources and loads, the first stage of the two-stage robust optimization scheduling model is to optimize the network transmission costs of the park and the interaction power value between parks. The second stage of the model uses the interaction power value of the first stage as a basis to solve the optimal scheduling strategy within the park and the worst-case scenario. When the interaction amount in the first stage is all zero, it becomes an independent robust model for each park under uncertainty. Constraints include interaction power constraints, internal equality constraints, and constraints limiting the non-negativity of internal decision variables.

7. A computer-readable storage medium, characterized in that: A plurality of instructions are stored therein, and the instructions are suitable for being loaded by a processor of a terminal device and executing the steps of the method according to any one of claims 1 to 5.

8. A terminal device, characterized in that: The method comprises a processor and a computer-readable storage medium, wherein the processor is used to implement various instructions; and the computer-readable storage medium is used to store multiple instructions, wherein the instructions are suitable for being loaded by the processor and executing the steps in the method according to any one of claims 1 to 5.

Images