A Multi-Objective Robust Optimization Allocation Method for Integrated Energy Systems

Abstract

This invention discloses a multi-objective robust optimization configuration method for integrated energy systems. The method comprises the following steps: First, establishing a multivariate load uncertainty model; then, establishing a multi-objective robust bi-layer joint optimization configuration model for the integrated energy system: the upper-layer model is the master model, and the lower-layer model is the sub-model. Through iterative optimization of the upper and lower layers, the optimal configuration strategy for the integrated energy system is obtained; finally, an intelligent optimization algorithm is used to solve the problem. For the multivariate load uncertainty model of the lower-layer model, a multi-scenario technique and a minimum-maximum regret criterion are used for joint solution, ultimately obtaining the Pareto solution set of the multi-objective robust configuration scheme for the integrated energy system. This invention proposes a multi-objective robust bi-layer joint optimization configuration method for integrated energy systems, which can provide optimized configuration schemes for integrated energy systems, improve energy utilization efficiency, reduce total costs during the planning period, and reduce pollutant emissions.

Description

Technical Field

[0001] This invention relates to the field of integrated energy system optimization configuration, and more specifically, to a multi-objective robust optimization configuration method for integrated energy systems. Background Technology

[0002] To address global warming and the scarcity of fossil fuel resources, and to achieve sustainable economic development, the social energy supply system needs to be integrated to promote coordinated development. To achieve the goals of reliable and efficient comprehensive utilization of social energy and energy conservation and emission reduction, it is essential to integrate and optimize various energy supply systems to form an integrated energy system (IES).

[0003] With the integration of a large number of distributed energy sources, energy storage devices, electric vehicles, charging stations, etc. into the power system, and the increasing coupling between the power system and the natural gas system and the heating (cooling) system, the optimization and configuration technology of integrated energy systems is becoming increasingly difficult.

[0004] Most existing integrated energy system optimization methods take economic efficiency as a single objective. When considering multiple objectives of economic efficiency and environmental protection, the economic and environmental objectives are weighted and converted into a single objective problem by using weight coefficients. Only a certain configuration scheme can be obtained by adjusting the weight coefficients. This will lose the diversity and comprehensiveness of the solution, and cannot provide more alternative schemes, thus failing to meet the multi-faceted needs of decision-makers.

[0005] The purpose of integrated energy system optimization is to provide construction plans for integrated energy systems that meet the future electricity, heat, cooling, and gas loads of industrial, commercial, and residential users within the system. Integrated energy systems involve diverse load types, making precise load forecasting impossible. The randomness and uncertainty of electricity, heat, cooling, and gas loads from industrial, commercial, and residential users significantly impact the results of integrated energy system optimization, and may even lead to optimized configuration plans failing to meet actual load demands.

[0006] In conclusion, considering both economic and environmental factors, as well as the uncertainty of load forecasting, conducting research on multi-objective robust optimization configuration of integrated energy systems is of great significance. Summary of the Invention

[0007] To address the aforementioned problems, this invention provides a multi-objective robust optimization configuration method for integrated energy systems. This invention simultaneously considers economic efficiency and environmental friendliness, taking into account the uncertainties of electricity, heat, cooling, and gas loads from industry, commerce, and residential sectors. A multi-variable load uncertainty model is established, with the objectives of minimizing the total cost of optimal configuration and the minimum emissions of pollutants within the planning period. A multi-objective robust two-layer joint optimization configuration model for integrated energy systems is established, employing multi-scenario technology to represent multi-variable load uncertainties. The robust Pareto optimal solution set of the configuration scheme is obtained using intelligent optimization algorithms and the minimum-maximum regret criterion.

[0008] This invention discloses a multi-objective robust optimization configuration method for integrated energy systems, which is implemented through the following technical solution.

[0009] 1. Establish a multivariate load uncertainty model:

[0010] The uncertainty model of multiple loads in an integrated energy system is as follows:

[0011]

[0012]

[0013] In the formula, J represents the number of load objects applicable to the integrated energy system, K represents the load type, and S represents the number of typical daily scenarios per year, including 3 or 4 typical daily scenarios. For a typical daily scenario s, the actual load value of the j-th load user and the k-th load type at the i-th time; L s,j,k,i The baseline predicted load value for the j-th load user and the k-th load type at the i-th time in a typical daily scenario s; ω represents the maximum load fluctuation value at time i for the j-th load user and the k-th load type in a typical daily scenario; s,j,k,i Let ω be the load fluctuation coefficient of the j-th load user and the k-th load type at the i-th time in a typical daily scenario s. s,j,k,i ∈[-1,1];

[0014] 2. Multi-objective robust two-layer joint optimization configuration model for integrated energy systems:

[0015] The upper-level model is the main model, which is the equipment selection and capacity optimization model. The lower-level model is the sub-model, which is the operation scheduling optimization model. The equipment configuration scheme determined by the upper-level model is used as the boundary condition.

[0016] The upper-level model optimizes the configuration strategies and planned capacities of various equipment and passes them to the lower-level model. The lower-level model optimizes the scheduling and operation of the integrated energy system and returns the optimal operation plan and total operating cost to the upper-level model. Through the optimization iteration between the upper and lower levels, the optimal configuration strategy of the integrated energy system is obtained.

[0017] 2.1 Upper-level model

[0018] 2.1.1 Objective Function

[0019] The objective function of the upper-level model is to minimize the total cost and the amount of pollutant emissions within the planning period, as shown in the following formula:

[0020] minF(x,y)=[f c (x,y),f e (x,y)] T

[0021] Where F(x,y) is the multi-objective programming optimization objective function of the integrated energy system; f c (x,y) represents the objective function for minimizing the total cost of optimal allocation of the integrated energy system; f e (x,y) represents the objective function for minimizing pollutant emissions from the integrated energy system; x represents the equipment configuration decision variable; and y represents the system operation optimization decision variable.

[0022] (1) Economic objectives

[0023] The total cost of joint optimization configuration of the integrated energy system is as follows:

[0024]

[0025] Where T is the planning period, i.e., the total planning period; r is the discount rate. Let t be the total investment cost in year t; Let t be the total operating cost in year t.

[0026] Total investment cost in year t:

[0027]

[0028] In the formula, i is the energy supply production equipment number; Ω1 is the set of energy supply production equipment; The unit capacity investment cost for the i-th energy supply production equipment; Ω2 represents the installed capacity of the i-th energy supply production equipment in year t; j is the energy storage equipment number; Ω2 is the set of energy storage equipment. Let $ be the unit capacity investment cost of the j-th energy storage device. Let the installed capacity of the j-th energy storage device in year t be denoted as .

[0029] (2) Environmental protection goals

[0030] The pollutant emissions from the integrated energy system are configured in a coordinated and optimized manner as follows:

[0031]

[0032]

[0033]

[0034]

[0035]

[0036] In the formula, E t,CO E represents the CO emissions in year t. t,CO2 E represents the CO2 emissions in year t. t,SO2 E represents the SO2 emissions in year t. t,NOx P represents the NOx emissions in year t. e,t,s,h Ω3 represents the electricity purchase amount during the h-hour period of a typical day in year t; Ω3 represents the set of energy supply production equipment consuming natural gas; G g,i,t,s,h α represents the amount of natural gas purchased by the energy supply production equipment consuming natural gas during the h-th time period of a typical day in year t; e,CO α e,CO2 α e,SO2 and α e,NOx These are the emission coefficients for CO, CO2, SO2, and NOx generated from the consumption of electrical energy, respectively; α g,i,CO α g,i,CO2 α g,i,SO2 and α g,i,NOx Let G be the emission coefficients of CO, CO2, SO2, and NOx generated by the i-th energy supply production equipment consuming natural gas; gL,o,t,s,h For the typical daily scenario h-hour period in year t, other natural gas load power; α g,o,CO α g,o,CO2 α g,o,SO2 and α g,o,NOx These are the emission coefficients for CO, CO2, SO2, and NOx generated from natural gas consumption other than in energy supply production equipment that consumes natural gas; 2.1.2 Constraints

[0037] (1) Upper limit constraint of equipment planning capacity

[0038] The upper limit constraint on the planned capacity of equipment includes energy production equipment and energy storage equipment. Due to limitations such as geographical location, land area, space, geographical environment and natural environment, each type of equipment has a maximum planned capacity, and the actual planned capacity must not exceed the upper limit.

[0039]

[0040]

[0041] In the formula, These are the planned capacity limits for energy production equipment and energy storage equipment, respectively.

[0042] (2) Equipment commissioning planning constraints

[0043] Ensure that the external energy supply capacity, the total capacity of energy supply equipment, and the total discharge power of energy storage equipment are greater than or equal to the peak load during the h-hour period of a typical day in year t.

[0044]

[0045]

[0046]

[0047]

[0048] In the formula, These refer to the maximum external power supply and gas supply capacity, respectively. These are the rated capacities of the electrical, thermal, cold, and natural gas production equipment in year t, respectively, with the distributed power source considering the maximum output based on the reliable capacity. These represent the maximum energy release power of the energy storage, thermal storage, cold storage, and gas storage devices in year t, respectively. These represent the peak values ​​of the actual electricity load, heat load, cooling load, and natural gas load in year t, respectively, including the energy supply production equipment that consumes the corresponding energy; H g η is the calorific value of natural gas. P2G To improve the energy conversion efficiency of the electro-gas conversion equipment, The upper limit of the input electrical power of the electro-gas conversion device;

[0049] 2.2 Lower-level model

[0050] 2.2.1 Objective Function

[0051] The objective function of the lower-level model is to minimize the total annual operating cost, as shown in the following formula:

[0052]

[0053] In the formula, The equipment operation and maintenance cost in year t; Let t be the energy purchase cost of the system in year t.

[0054]

[0055] In the formula, The annual operating and maintenance cost per unit capacity of the i-th energy supply production equipment; Let $ be the annual operating and maintenance cost per unit capacity of the $j$-th energy storage device.

[0056] The energy purchase cost of an integrated energy system includes the cost of purchasing electricity and the cost of purchasing natural gas, as shown in the following formula:

[0057]

[0058] In the formula, N s H represents the number of days in a typical daily scene s throughout the year; H represents the total number of time periods in a day; C represents the number of days in a typical daily scene s throughout the year. e,t,s,h C g,t,s,h The electricity purchase price and natural gas purchase price per unit area for a typical day in year t, during the h-hour period; P e,t,s,h G g,t,s,h These represent the electricity and natural gas purchases during the h-hour period of a typical day in year t;

[0059] 2.1.2 Constraints

[0060] (1) Power balance constraint

[0061]

[0062] In the formula, Ω4 represents the set of energy supply production equipment that generates electrical energy; Ω5 represents the set of energy storage devices; Ω6 represents the set of energy supply production equipment that consumes electrical energy; P e,i,t,s,h P represents the electrical power generated by the i-th energy supply production equipment during time period h in a typical day of year t; e,j,ES-dis,t,s,h P represents the discharge power of the j-th energy storage device during time period h in a typical daily scenario of year t; e,k,t,s,h The electrical power consumed by the k-th energy supply device during the h-hour period of a typical day in year t; For other actual electrical load power during the h-hour period of a typical day in year t; P e,j,ES-ch,t,s,h The charging power of the j-th energy storage device during time period h in a typical day scenario of year t;

[0063] (2) Thermal power balance constraint

[0064]

[0065] In the formula, Ω7 represents the set of energy supply production equipment that generates heat; Ω8 represents the set of thermal storage equipment; Ω9 represents the set of energy supply production equipment that consumes heat; Q q,i,t,s,h Q represents the heat output of the i-th energy supply production equipment that generates heat during time period h in a typical day of year t; q,j,HS-dis,t,s,h Q represents the heat release power of the j-th thermal storage device during the h-hour period of a typical day in year t; q,k,t,s,h The heat load of the energy supply production equipment consuming heat energy during the h-th time period of a typical day in year t is given. Q represents the actual heat load power during the h-hour period of a typical day in year t; q,j,HS-ch,t,s,h The charging power of the j-th thermal storage device during the h-hour period of a typical day in year t;

[0066] (3) Cooling power balance constraint

[0067]

[0068] In the formula, Ω 10 A collection of energy supply and production equipment for generating cold energy; Ω 11 A collection of cold storage equipment; C c,i,t,s,h C represents the cooling power generated by the i-th energy supply production equipment during time period h in a typical day of year t; c,j,CS-dis,t,s,h The cooling power of the j-th cold storage device during the h-hour period of a typical day in year t; C represents the actual cooling load power during the h-hour period of a typical day in year t; c,j,CS-ch,t,s,h The charging power of the j-th cold storage device during the h-hour period of a typical day in year t;

[0069] (4) Natural gas power balance constraints

[0070]

[0071] In the formula, Ω3 represents the collection of energy supply and production equipment that consumes natural gas; Ω 12 A collection of gas storage devices; G g,t,s,h G represents the natural gas power during the h-hour period of a typical day in year t; g,P2G,t,s,h G represents the natural gas power generated by the electro-gas conversion equipment during the h-hour period of a typical day in year t; g,i,GS-dis,t,s,h G represents the gas release power of the j-th gas storage device during the h-hour period of a typical day in year t; g,i,t,s,h The natural gas power of the energy supply production equipment consuming natural gas during the time period h in a typical day of year t; For other actual natural gas load power during the h-hour period of a typical day in year t; G g,i,GS-ch,t,s,h The gas charging power of the j-th gas storage device during the h-hour period of a typical day in year t;

[0072] (5) Operational constraints of energy production and conversion equipment

[0073]

[0074] In the formula, P represents the lower and upper limits of the operating power of the i-th energy production and conversion equipment in year t, respectively. i,t,s,h The operating power of the energy production and conversion equipment s during a typical daily scenario h in year t;

[0075] (6) Operational constraints of energy storage devices

[0076]

[0077]

[0078]

[0079]

[0080] P k,t,s,h,ESS-dis P k,t,s,h,ESS-ch =0

[0081] E k,t,s,0,ESS =E k,t,s,H,ESS

[0082] In the formula, k represents the energy storage type; P k,ESS-ch,t,s,h The charging power during the Δh period; P k,ESS-dis,t,s,h τ represents the energy released during the time interval Δh; k,ES η is the self-discharge rate of the energy storage device. k,ESch The charging efficiency during the Δh time period; η k,ESdis E represents the energy release efficiency during the Δh time period. k,t,s,h,ESS The energy storage status of the energy storage device; This refers to the rated capacity of the energy storage device. These are the minimum and maximum values ​​of the allowable ratio of stored energy to stored capacity, respectively; P k,t,s,h,ESS-ch P k,t,s,h,ESS-dis These refer to the charging and discharging power of the energy storage device, respectively. These are the maximum energy storage capacity and energy release capacity of the energy storage device, respectively; E k,t,s,0,ESS E k,t,s,H,ESS These represent the energy stored by the energy storage device at the beginning and end of the day, respectively.

[0083] (7) Power exchange constraints with the superior network

[0084] Due to network architecture limitations, the maximum power delivered from the upper-level network to the integrated energy system is restricted. Therefore, the power exchange between the integrated energy system and the upper-level network cannot exceed the upper limit.

[0085]

[0086]

[0087] In the formula, These are the minimum and maximum power purchase capacities, respectively. These are the minimum and maximum gas purchasing power, respectively;

[0088] 3. Solution Method

[0089] The upper-level model passes the configuration scheme to the lower-level model, which uses the configuration scheme as boundary conditions, considers multivariate load uncertainties, optimizes to obtain the optimal operating strategy, and returns the optimal operating strategy and total operating cost to the upper-level model. From the perspective of both the upper and lower-level models, intelligent optimization algorithms are needed for solving the problem. For the multivariate load uncertainty model of the lower-level model, a multi-scenario technique and a minimum-maximum regret criterion are used for joint solution, ultimately obtaining the Pareto solution set for the multi-objective robust configuration scheme of the integrated energy system.

[0090] For the lower-level model, firstly, a multi-scenario technique is used to describe the uncertainty of multi-variable loads; secondly, the particle swarm optimization algorithm is used to obtain the optimal solution for each scenario; finally, the minimum-maximum regret criterion is used to obtain the Pareto solution set of the final robust configuration scheme of the integrated energy system.

[0091] 3.1 Multi-scenario technical description of multi-load uncertainty

[0092] To accurately represent the uncertainty of multivariate loads, a sufficient number of multivariate load fluctuation scenarios are generated using the Latin hypercube sampling method. However, too many samples will slow down the solution speed. To improve computational efficiency, a technique must be used to reduce the sample size while maintaining the maximum sample fitting accuracy. A backward scenario reduction technique is used to reduce the sample size. Let the number of sampled samples be Z, and the target number of samples be z. Taking the j-th load user and k-th load type in a typical daily scenario s as an example, the specific process is explained below.

[0093] (1) Latin hypercube sampling

[0094] 1) Set the sampling size to Z;

[0095] 2) The fluctuation coefficient ω s,j,k,i The interval [-1, 1] is divided into Z subintervals on average;

[0096] 3) Generate a Latin hypercube matrix of dimension Z×H, where H is the total number of time periods in a day. Each column of the matrix is ​​a random sequence of integers from 1 to Z, and each column corresponds to an uncertain decision variable. Each row corresponds to a sample. At this point, all Z samples have been obtained.

[0097] (2) Scene reduction

[0098] After generating Z samples, a backward scene reduction technique is used to reduce the multivariate load samples. The Z samples are represented by λ. r (i = 1, 2, ..., Z) represents that λ r The probability of occurrence is denoted by Pr(r), and the specific process is as follows:

[0099] 1) Let Ψ be the initial set of scenes, and ψ be the set of scenes removed from set Ψ. Initially, set ψ is empty. Calculate the distance between any two scenes;

[0100] D r,u =d(λ r ,λ u r,u=1,...,Z

[0101]

[0102] 2) For each scenario λ r Determine the scene λ that has the smallest distance to it. q A multi-dimensional load maximum allowable fluctuation constraint variable Δω is introduced. As the number of scenarios is reduced, Δω gradually increases to ensure that some scenarios that are far away are not deleted, thus ensuring scenario diversity.

[0103] D r,q =min{D r,u :r,u∈Ψ,r≠u}

[0104]

[0105] 3) Calculate PD r,q =Pr(r)×D r,q r∈Ψ, and obtain PD d,q =min{PD r,q d :r∈Ψ}.

[0106] 4) Delete scene Ψ from the initial scene set Ψ d And update each set and its corresponding scenario λ. q The probability of;

[0107] Ψ=Ψ-Ψ d

[0108] ψ=ψ+Ψ d

[0109] Pr(q) = Pr(q) + Pr(d)

[0110] 5) Determine if the target number of scenarios z has been reached. If so, return to step 2); otherwise, proceed to step 6. If no scenario in this round of reduction meets the maximum allowable fluctuation constraint of multi-dimensional load, update Δω to the smallest deviation among the existing scenarios.

[0111] 6) Scene reduction operation completed:

[0112] In all typical daily scenarios, various load types of industrial, commercial, and residential load users are obtained using the above method to acquire scenarios with a target sample size of z. The obtained scenarios are combined one by one to obtain the multi-variable load uncertainty scenario set Π of the integrated energy system.

[0113] 3.2 Minimum and Maximum Regret Criterion for Selecting the Final Execution Strategy

[0114] The minimum maximum regret criterion is an effective method for obtaining the final decision from multiple scenarios. Considering the probability of each multivariate load scenario, the minimum maximum regret criterion is calculated as follows;

[0115]

[0116]

[0117] In the formula, g (x,y),r To determine the objective function value of strategy y in scenario r under configuration scheme x; g p,r Y represents the minimum objective function of the running strategy in scenario r under configuration scheme x; Y is the set of running strategies under the scenario uncertainty set under configuration scheme x.

[0118] 3.3 Multi-objective robust optimization model for integrated energy systems based on multiple agents

[0119] Using multi-scenario technology to handle the uncertainty of diverse loads will inevitably lead to a significant increase in computational load. Therefore, it is necessary to reduce the dimensionality of the optimization problem and reduce computational complexity. For the lower-level model, all time periods of a typical day are set as the general agent. Based on the characteristics of diverse loads, a time period sub-agent partitioning method is proposed to divide the model into multiple time period sub-agents according to the time period, thereby reducing the dimensionality of variables.

[0120] 3.3.1 Time-based Sub-agent Partitioning Method Based on Multi-variable Load Characteristics

[0121] When dividing time-segment sub-agents, the optimization time-segment cannot be divided across time-segments. All time-segments in each sub-agent must be consecutive. Set the number of time-segment sub-agents to M. The first and last moments of each time-segment sub-agent are the boundary moments. The method for dividing time-segment sub-agents is as follows.

[0122] (1) Calculate load fluctuation. Calculate the load fluctuation within the sub-agent of the k-th load quality of the j-th load user in the m-th time period. The calculation formula is as follows:

[0123]

[0124]

[0125] In the formula: N represents the load fluctuation of the j-th load user, the k-th load type, and the m-th time period; m This represents the number of moments in the load curve within the m-th time period. The load value at time i within the m-th time period for the j-th load user and the k-th load type;

[0126] (2) Normalization: Due to the large differences in load values ​​between different load applications and different load types, directly calculating the load fluctuation can easily lead to significant errors. Therefore, it is necessary to normalize the load fluctuation. The calculation formula is as follows:

[0127]

[0128] In the formula, This is the normalized load fluctuation. The maximum load value of the j-th load user, the k-th load type, within the m-th time period;

[0129] (3) Calculate the total load fluctuation: Calculate the sum of the load fluctuation of J load users, K load types, and M time period sub-agents. Determine the time period sub-agent division strategy with the goal of minimizing the load fluctuation. Therefore, optimizing the time period sub-agent division can be described as the following optimization problem.

[0130]

[0131] (4) Constraints: The sum of the number of moments in all sub-agents is the total optimization period H; the optimization period cannot be divided across time periods, and all time periods in each sub-agent must be continuous; the load fluctuation between any two moments in a sub-agent does not exceed the maximum threshold.

[0132]

[0133]

[0134]

[0135] In the formula, U m Let be the set of times within the m-th time period, and ΔL be the maximum threshold for load fluctuation;

[0136] The optimization problem of the time-period sub-agent partitioning method based on multi-variable load characteristics described above is a nonlinear discrete optimization problem, which is solved using the particle swarm optimization algorithm. The solution process is as follows:

[0137] 1) Parameter initialization: Input the load values ​​of electricity, heat, cooling and gas for each time period of the representative daily industrial load, commercial load and residential load, as well as parameters such as the number of iterations of the intelligent optimization algorithm;

[0138] 2) Set up a time-segment sub-agent M;

[0139] 3) Optimize variable initialization, specifically the number of time periods within each sub-agent;

[0140] 4) Calculate the load fluctuation of the j-th load user, the k-th load type, within the m-th time period (m-th time period sub-agent);

[0141] 5) Normalization process;

[0142] 6) Calculate the total load fluctuation;

[0143] 7) Determine whether the termination condition is met. If it is met, terminate the iteration and jump to step (8); if it is not met, the optimization variable is moved according to the particle swarm optimization algorithm to obtain the next optimization variable and jump to step (4).

[0144] 8) Output the time-segment sub-agent partitioning results;

[0145] 3.3.2 Lower-level model multi-agent computation model

[0146] A time-segment general agent and time-segment sub-agent model was established. The calculation process was decomposed into general agent calculation and time-segment sub-agent calculation. The two parts of the calculation are related by the boundary time of the sub-agent. The decision variables at each boundary time can be obtained through the general agent calculation. These are also the boundary conditions in the time-segment sub-agent calculation. After each time-segment sub-agent calculation is completed, the obtained running strategy is passed to the general agent in sequence as the initial condition for the next iteration calculation until the calculation ends.

[0147] Since the calculation process is decomposed into general agent calculation and time-segment sub-agent calculation, the lower-level model needs to be revised. The mathematical models of these two parts of the calculation are analyzed below.

[0148] (1) General Agent Calculation Mathematical Model

[0149] The purpose of the general agent calculation is to obtain the decision variables at each boundary time. Its objective function is the operation and maintenance cost and energy purchase cost corresponding to the decision variables at each boundary time, as shown in the following formula:

[0150]

[0151]

[0152]

[0153] In the formula, C G2 Calculate the objective function for the general agent; C G2,m The objective function computed by the sub-agent in the m-th time period; t1 mt2 represents the minimum boundary time of the m-th time-segment sub-agent; m Let be the maximum boundary time of the m-th time period sub-agent; if the m-th time period sub-agent has only one moment, then the sub-agent is a boundary sub-agent;

[0154] The constraints for the general agent's calculations are the same as those for the lower-level model mentioned above;

[0155] (2) Time-based sub-agent computational mathematical model

[0156] The purpose of time-segment sub-agent computation is to obtain the decision variables for each time-segment sub-agent. Its objective function is the operation and maintenance cost and energy purchase cost corresponding to the decision variables at other times except the boundary time. The objective function is shown in the following formula:

[0157]

[0158]

[0159]

[0160] In the formula, C G3 Calculate the objective function for the time-phase sub-agent; C G3,m The sum of the objective functions of the subordinate time steps within the m-th time-period sub-agent; t3 m For the m-th time period sub-agent, the decision variables and objective function at the boundary time are not calculated repeatedly, but directly inherit the calculation results of the general agent, while the decision variables at the boundary time are transformed as known quantities into boundary conditions or initial conditions for the calculation of the time period sub-agent.

[0161] The constraints for time-phase sub-agent computation are the same as those for the lower-level model mentioned above;

[0162] 3.4 Solution Process of Multi-Objective Robust Optimization Model for Integrated Energy System Based on Multi-Agent

[0163] The solution steps for the multi-objective robust optimization model of a multi-agent integrated energy system are as follows:

[0164] (1) Parameter initialization, including equipment parameters, load data, energy price policy, emission coefficient, planning cycle, iteration number, particle number, and optimization variables; (2) Using Latin hypercube sampling and backward scenario reduction techniques to obtain the sampling scenario; (3) Solving the sub-agent partitioning of the time period to obtain the partitioning result; (4) Optimization begins, iteration number n = 1; (5) Determine whether the iteration number is greater than the maximum value. If it is greater than the maximum value, jump to step (15), otherwise jump to step (6); (6) Calculate the objective function of the upper-level model, the economic objective and the environmental objective; (7) Optimize the algorithm to generate the decision variable x; (8) Calculate the lower-level objective function, the total operating cost of the scheduling operation optimization sub-problem; (9) Optimize the general agent to obtain the boundary time operation strategy. (10) The sub-agents optimize the internal time-of-day operation strategy of the sub-agents; (11) The sub-agents of each time period upload the optimization results to the general agent, and the general agent performs combined solution; (12) Determine whether all sub-agents of each time period have finished calculating. If yes, jump to step (13); otherwise, jump to step (10); (13) Generate decision variable y; (14) Calculate the objective function of the upper-level model, the economic objective and the environmental objective; (15) Select the configuration scheme by using the minimum probability maximum regret value criterion; (16) Determine whether the termination condition is met. If yes, jump to step (17); otherwise, jump to step (5); (17) Terminate the iteration and output the Pareto solution set of the multi-objective robust optimization configuration of the integrated energy system; (18) End.

[0165] Compared with the prior art, the beneficial effects of the present invention are:

[0166] (1) This invention proposes a multi-objective robust two-layer joint optimization configuration method for integrated energy systems, which can provide an optimized configuration scheme for integrated energy systems, improve energy utilization efficiency, reduce total cost during the planning period, and reduce pollutant emissions.

[0167] (2) This method takes minimizing the total cost of optimization configuration and minimizing the amount of pollutant gas emissions within the planning period as multiple objectives, and optimizes the Pareto solution set of the configuration scheme to ensure the diversity and comprehensiveness of the solution, providing decision-makers with more alternative schemes and meeting their multi-faceted needs.

[0168] (3) This method takes into account the uncertainty of electricity, heat, cooling and gas loads of industry, commerce and residents, establishes a multi-load uncertainty model, introduces a fluctuation coefficient to control the fluctuation range of load, coordinates the optimality and robustness of the configuration scheme, and establishes a multi-objective robust two-layer joint optimization configuration model for integrated energy system, which can effectively cope with the risks caused by multi-load uncertainty.

[0169] (4) A multi-objective robust optimization model for integrated energy systems based on multiple agents is proposed. For the lower-level model, a time period sub-agent partitioning method based on multi-variable load characteristics is proposed to decompose the typical daily optimization time period into multiple time period sub-agents, reduce the variable dimension, and thus improve the solution efficiency. Detailed Implementation

[0170] The present invention will be further described in detail below with reference to specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0171] This invention discloses a multi-objective robust optimization configuration method for integrated energy systems, which is implemented through the following technical solution.

[0172] 3. Establish a multivariate load uncertainty model:

[0173] The uncertainty model of multiple loads in an integrated energy system is as follows:

[0174]

[0175]

[0176] In the formula, J represents the number of load objects applicable to the integrated energy system, K represents the load type, and S represents the number of typical daily scenarios per year, including 3 or 4 typical daily scenarios. For a typical daily scenario s, the actual load value of the j-th load user and the k-th load type at the i-th time; L s,j,k,i The baseline predicted load value for the j-th load user and the k-th load type at the i-th time in a typical daily scenario s; ω represents the maximum load fluctuation value at time i for the j-th load user and the k-th load type in a typical daily scenario; s,j,k,i Let ω be the load fluctuation coefficient of the j-th load user and the k-th load type at the i-th time in a typical daily scenario s. s,j,k,i ∈[-1,1];

[0177] 4. Multi-objective robust two-layer joint optimization configuration model for integrated energy systems:

[0178] The upper-level model is the main model, which is the equipment selection and capacity optimization model. The lower-level model is the sub-model, which is the operation scheduling optimization model. The equipment configuration scheme determined by the upper-level model is used as the boundary condition.

[0179] The upper-level model optimizes the configuration strategies and planned capacities of various equipment and passes them to the lower-level model. The lower-level model optimizes the scheduling and operation of the integrated energy system and returns the optimal operation plan and total operating cost to the upper-level model. Through the optimization iteration between the upper and lower levels, the optimal configuration strategy of the integrated energy system is obtained.

[0180] 2.1 Upper-level model

[0181] 2.1.1 Objective Function

[0182] The objective function of the upper-level model is to minimize the total cost and the amount of pollutant emissions within the planning period, as shown in the following formula:

[0183] minF(x,y)=[f c (x,y),f e (x,y)] T

[0184] Where F(x,y) is the multi-objective programming optimization objective function of the integrated energy system; f c (x,y) represents the objective function for minimizing the total cost of optimal allocation of the integrated energy system; f e (x,y) represents the objective function for minimizing pollutant emissions from the integrated energy system; x represents the equipment configuration decision variable; and y represents the system operation optimization decision variable.

[0185] (1) Economic objectives

[0186] The total cost of joint optimization configuration of the integrated energy system is as follows:

[0187]

[0188] Where T is the planning period, i.e., the total planning period; r is the discount rate. Let t be the total investment cost in year t; Let t be the total operating cost in year t.

[0189] Total investment cost in year t:

[0190]

[0191] In the formula, i is the energy supply production equipment number; Ω1 is the set of energy supply production equipment; The unit capacity investment cost for the i-th energy supply production equipment; Ω2 represents the installed capacity of the i-th energy supply production equipment in year t; j is the energy storage equipment number; Ω2 is the set of energy storage equipment. Let $ be the unit capacity investment cost of the j-th energy storage device. Let the installed capacity of the j-th energy storage device in year t be denoted as .

[0192] (2) Environmental protection goals

[0193] The pollutant emissions from the integrated energy system are configured in a coordinated and optimized manner as follows:

[0194]

[0195]

[0196]

[0197]

[0198]

[0199] In the formula, E t,CO E represents the CO emissions in year t. t,CO2 E represents the CO2 emissions in year t. t,SO2 E represents the SO2 emissions in year t. t,NOx P represents the NOx emissions in year t. e,t,s,h Ω3 represents the electricity purchase amount during the h-hour period of a typical day in year t; Ω3 represents the set of energy supply production equipment consuming natural gas; G g,i,t,s,h α represents the amount of natural gas purchased by the energy supply production equipment consuming natural gas during the h-th time period of a typical day in year t; e,CO α e,CO2 α e,SO2 and α e,NOx These are the emission coefficients for CO, CO2, SO2, and NOx generated from the consumption of electrical energy, respectively; α g,i,CO α g,i,CO2 α g,i,SO2 and α g,i,NOx Let G be the emission coefficients of CO, CO2, SO2, and NOx generated by the i-th energy supply production equipment consuming natural gas; gL,o,t,s,h For the typical daily scenario h-hour period in year t, other natural gas load power; α g,o,CO、 α g,o,CO2、 α g,o,SO2 and α g,o,NOx These are the emission coefficients for CO, CO2, SO2, and NOx generated from natural gas consumption other than in energy supply production equipment that consumes natural gas; 2.1.2 Constraints

[0200] (1) Upper limit constraint of equipment planning capacity

[0201] The upper limit constraint on the planned capacity of equipment includes energy production equipment and energy storage equipment. Due to limitations such as geographical location, land area, space, geographical environment and natural environment, each type of equipment has a maximum planned capacity, and the actual planned capacity must not exceed the upper limit.

[0202]

[0203]

[0204] In the formula, These are the planned capacity limits for energy production equipment and energy storage equipment, respectively.

[0205] (2) Equipment commissioning planning constraints

[0206] Ensure that the external energy supply capacity, the total capacity of energy supply equipment, and the total discharge power of energy storage equipment are greater than or equal to the peak load during the h-hour period of a typical day in year t.

[0207]

[0208]

[0209]

[0210]

[0211] In the formula, These refer to the maximum external power supply and gas supply capacity, respectively. These are the rated capacities of the electrical, thermal, cold, and natural gas production equipment in year t, respectively, with the distributed power source considering the maximum output based on the reliable capacity. These represent the maximum energy release power of the energy storage, thermal storage, cold storage, and gas storage devices in year t, respectively. These represent the peak values ​​of the actual electricity load, heat load, cooling load, and natural gas load in year t, respectively, including the energy supply production equipment that consumes the corresponding energy; H g η is the calorific value of natural gas. P2G To improve the energy conversion efficiency of the electro-gas conversion equipment, The upper limit of the input electrical power of the electro-gas conversion device;

[0212] 2.2 Lower-level model

[0213] 2.2.1 Objective Function

[0214] The objective function of the lower-level model is to minimize the total annual operating cost, as shown in the following formula:

[0215]

[0216] In the formula, The equipment operation and maintenance cost in year t; Let t be the energy purchase cost of the system in year t.

[0217]

[0218] In the formula, The annual operating and maintenance cost per unit capacity of the i-th energy supply production equipment; Let $ be the annual operating and maintenance cost per unit capacity of the $j$-th energy storage device.

[0219] The energy purchase cost of an integrated energy system includes the cost of purchasing electricity and the cost of purchasing natural gas, as shown in the following formula:

[0220]

[0221] In the formula, N s H represents the number of days in a typical daily scene s throughout the year; H represents the total number of time periods in a day; C represents the number of days in a typical daily scene s throughout the year. e,t,s,h C g,t,s,h The electricity purchase price and natural gas purchase price per unit area for a typical day in year t, during the h-hour period; P e,t,s,h G g,t,s,h These represent the electricity and natural gas purchases during the h-hour period of a typical day in year t;

[0222] 2.1.2 Constraints

[0223] (1) Power balance constraint

[0224]

[0225] In the formula, Ω4 represents the set of energy supply production equipment that generates electrical energy; Ω5 represents the set of energy storage devices; Ω6 represents the set of energy supply production equipment that consumes electrical energy; P e,i,t,s,h P represents the electrical power generated by the i-th energy supply production equipment during time period h in a typical day of year t; e,j,ES-dis,t,s,h P represents the discharge power of the j-th energy storage device during time period h in a typical daily scenario of year t; e,k,t,s,h The electrical power consumed by the k-th energy supply device during the h-hour period of a typical day in year t; For other actual electrical load power during the h-hour period of a typical day in year t; P e,j,ES-ch,t,s,h The charging power of the j-th energy storage device during time period h in a typical day scenario of year t;

[0226] (2) Thermal power balance constraint

[0227]

[0228] In the formula, Ω7 represents the set of energy supply production equipment that generates heat; Ω8 represents the set of thermal storage equipment; Ω9 represents the set of energy supply production equipment that consumes heat; Q q,i,t,s,h Q represents the heat output of the i-th energy supply production equipment that generates heat during time period h in a typical day of year t; q,j,HS-dis,t,s,h Q represents the heat release power of the j-th thermal storage device during the h-hour period of a typical day in year t; q,k,t,s,h The heat load of the energy supply production equipment consuming heat energy during the h-th time period of a typical day in year t is given. Q represents the actual heat load power during the h-hour period of a typical day in year t;q,j,HS-ch,t,s,h The charging power of the j-th thermal storage device during the h-hour period of a typical day in year t;

[0229] (3) Cooling power balance constraint

[0230]

[0231] In the formula, Ω 10 A collection of energy supply and production equipment for generating cold energy; Ω 11 A collection of cold storage equipment; C c,i,t,s,h C represents the cooling power generated by the i-th energy supply production equipment during time period h in a typical day of year t; c,j,CS-dis,t,s,h The cooling power of the j-th cold storage device during the h-hour period of a typical day in year t; C represents the actual cooling load power during the h-hour period of a typical day in year t; c,j,CS-ch,t,s,h The charging power of the j-th cold storage device during the h-hour period of a typical day in year t;

[0232] (4) Natural gas power balance constraints

[0233]

[0234] In the formula, Ω3 represents the collection of energy supply and production equipment that consumes natural gas; Ω 12 A collection of gas storage devices; G g,t,s,h G represents the natural gas power during the h-hour period of a typical day in year t; g,P2G,t,s,h G represents the natural gas power generated by the electro-gas conversion equipment during the h-hour period of a typical day in year t; g,i,GS-dis,t,s,h G represents the gas release power of the j-th gas storage device during the h-hour period of a typical day in year t; g,i,t,s,h The natural gas power of the energy supply production equipment consuming natural gas during the time period h in a typical day of year t; For other actual natural gas load power during the h-hour period of a typical day in year t; G g,i,GS-ch,t,s,h The gas charging power of the j-th gas storage device during the h-hour period of a typical day in year t;

[0235] (5) Operational constraints of energy production and conversion equipment

[0236]

[0237] In the formula, P represents the lower and upper limits of the operating power of the i-th energy production and conversion equipment in year t, respectively. i,t,s,h The operating power of the energy production and conversion equipment s during a typical daily scenario h in year t;

[0238] (6) Operational constraints of energy storage devices

[0239]

[0240]

[0241]

[0242]

[0243] P k,t,s,h,ESS-dis P k,t,s,h,ESS-ch =0

[0244] E k,t,s,0,ESS =E k,t,s,H,ESS

[0245] In the formula, k represents the energy storage type; P k,ESS-ch,t,s,h The charging power during the Δh period; P k,ESS-dis,t,s,h τ represents the energy released during the time interval Δh; k,ES η is the self-discharge rate of the energy storage device. k,ESch The charging efficiency during the Δh time period; η k,ESdis E represents the energy release efficiency during the Δh time period. k,t,s,h,ESS The energy storage status of the energy storage device; This refers to the rated capacity of the energy storage device. These are the minimum and maximum values ​​of the allowable ratio of stored energy to stored capacity, respectively; P k,t,s,h,ESS-ch P k,t,s,h,ESS-dis These refer to the charging and discharging power of the energy storage device, respectively. These are the maximum energy storage capacity and energy release capacity of the energy storage device, respectively; E k,t,s,0,ESS E k,t,s,H,ESS These represent the energy stored by the energy storage device at the beginning and end of the day, respectively.

[0246] (7) Power exchange constraints with the superior network

[0247] Due to network architecture limitations, the maximum power delivered from the upper-level network to the integrated energy system is restricted. Therefore, the power exchange between the integrated energy system and the upper-level network cannot exceed the upper limit.

[0248]

[0249]

[0250] In the formula, These are the minimum and maximum power purchase capacities, respectively. These are the minimum and maximum gas purchasing power, respectively;

[0251] 3. Solution Method

[0252] The upper-level model passes the configuration scheme to the lower-level model, which uses the configuration scheme as boundary conditions, considers multivariate load uncertainties, optimizes to obtain the optimal operating strategy, and returns the optimal operating strategy and total operating cost to the upper-level model. From the perspective of both the upper and lower-level models, intelligent optimization algorithms are needed for solving the problem. For the multivariate load uncertainty model of the lower-level model, a multi-scenario technique and a minimum-maximum regret criterion are used for joint solution, ultimately obtaining the Pareto solution set for the multi-objective robust configuration scheme of the integrated energy system.

[0253] For the lower-level model, firstly, a multi-scenario technique is used to describe the uncertainty of multi-variable loads; secondly, the particle swarm optimization algorithm is used to obtain the optimal solution for each scenario; finally, the minimum-maximum regret criterion is used to obtain the Pareto solution set of the final robust configuration scheme of the integrated energy system.

[0254] 3.1 Multi-scenario technical description of multi-load uncertainty

[0255] To accurately represent the uncertainty of multivariate loads, a sufficient number of multivariate load fluctuation scenarios are generated using the Latin hypercube sampling method. However, too many samples will slow down the solution speed. To improve computational efficiency, a technique must be used to reduce the sample size while maintaining the maximum sample fitting accuracy. A backward scenario reduction technique is used to reduce the sample size. Let the number of sampled samples be Z, and the target number of samples be z. Taking the j-th load user and k-th load type in a typical daily scenario s as an example, the specific process is explained below.

[0256] (1) Latin hypercube sampling

[0257] 1) Set the sampling size to Z;

[0258] 2) The fluctuation coefficient ω s,j,k,i The interval [-1, 1] is divided into Z subintervals on average;

[0259] 3) Generate a Latin hypercube matrix of dimension Z×H, where H is the total number of time periods in a day. Each column of the matrix is ​​a random sequence of integers from 1 to Z, and each column corresponds to an uncertain decision variable. Each row corresponds to a sample. At this point, all Z samples have been obtained.

[0260] (2) Scene reduction

[0261] After generating Z samples, a backward scene reduction technique is used to reduce the multivariate load samples. The Z samples are represented by λ. r (i = 1, 2, ..., Z) represents that λ r The probability of occurrence is denoted by Pr(r), and the specific process is as follows:

[0262] 1) Let Ψ be the initial set of scenes, and ψ be the set of scenes removed from set Ψ. Initially, set ψ is empty. Calculate the distance between any two scenes;

[0263] D r,u =d(λ r ,λ u r,u=1,...,Z

[0264]

[0265] 2) For each scenario λ r Determine the scene λ that has the smallest distance to it. q A multi-dimensional load maximum allowable fluctuation constraint variable Δω is introduced. As the number of scenarios is reduced, Δω gradually increases to ensure that some scenarios that are far away are not deleted, thus ensuring scenario diversity.

[0266] D r,q =min{D r,u :r,u∈Ψ,r≠u}

[0267]

[0268] 3) Calculate PD r,q =Pr(r)×D r,q r∈Ψ, and obtain PD d,q =min{PD r,q d :r∈Ψ}.

[0269] 4) Delete scene Ψ from the initial scene set Ψ d And update each set and its corresponding scenario λ. q The probability of;

[0270] Ψ=Ψ-Ψ d

[0271] ψ=ψ+Ψ d

[0272] Pr(q) = Pr(q) + Pr(d)

[0273] 5) Determine if the target number of scenarios z has been reached. If so, return to step 2); otherwise, proceed to step 6. If no scenario in this round of reduction meets the maximum allowable fluctuation constraint of multi-dimensional load, update Δω to the smallest deviation among the existing scenarios.

[0274] 6) Scene reduction operation completed:

[0275] In all typical daily scenarios, various load types of industrial, commercial, and residential load users are obtained using the above method to acquire scenarios with a target sample size of z. The obtained scenarios are combined one by one to obtain the multi-variable load uncertainty scenario set Π of the integrated energy system.

[0276] 3.2 Minimum and Maximum Regret Criterion for Selecting the Final Execution Strategy

[0277] The minimum maximum regret criterion is an effective method for obtaining the final decision from multiple scenarios. Considering the probability of each multivariate load scenario, the minimum maximum regret criterion is calculated as follows;

[0278]

[0279]

[0280] In the formula, g (x,y),r To determine the objective function value of strategy y in scenario r under configuration scheme x; g p,r Y represents the minimum objective function of the running strategy in scenario r under configuration scheme x; Y is the set of running strategies under the scenario uncertainty set under configuration scheme x.

[0281] 3.3 Multi-objective robust optimization model for integrated energy systems based on multiple agents

[0282] Using multi-scenario technology to handle the uncertainty of diverse loads will inevitably lead to a significant increase in computational load. Therefore, it is necessary to reduce the dimensionality of the optimization problem and reduce computational complexity. For the lower-level model, all time periods of a typical day are set as the general agent. Based on the characteristics of diverse loads, a time period sub-agent partitioning method is proposed to divide the model into multiple time period sub-agents according to the time period, thereby reducing the dimensionality of variables.

[0283] 3.3.1 Time-based Sub-agent Partitioning Method Based on Multi-variable Load Characteristics

[0284] When dividing time-segment sub-agents, the optimization time-segment cannot be divided across time-segments. All time-segments in each sub-agent must be consecutive. Set the number of time-segment sub-agents to M. The first and last moments of each time-segment sub-agent are the boundary moments. The method for dividing time-segment sub-agents is as follows.

[0285] (1) Calculate load fluctuation. Calculate the load fluctuation within the sub-agent of the k-th load quality of the j-th load user in the m-th time period. The calculation formula is as follows:

[0286]

[0287]

[0288] In the formula: N represents the load fluctuation of the j-th load user, the k-th load type, and the m-th time period; m This represents the number of moments in the load curve within the m-th time period. The load value at time i within the m-th time period for the j-th load user and the k-th load type;

[0289] (2) Normalization: Due to the large differences in load values ​​between different load applications and different load types, directly calculating the load fluctuation can easily lead to significant errors. Therefore, it is necessary to normalize the load fluctuation. The calculation formula is as follows:

[0290]

[0291] In the formula, This is the normalized load fluctuation. The maximum load value of the j-th load user, the k-th load type, within the m-th time period;

[0292] (3) Calculate the total load fluctuation: Calculate the sum of the load fluctuation of J load users, K load types, and M time period sub-agents. Determine the time period sub-agent division strategy with the goal of minimizing the load fluctuation. Therefore, optimizing the time period sub-agent division can be described as the following optimization problem.

[0293]

[0294] (4) Constraints: The sum of the number of moments in all sub-agents is the total optimization period H; the optimization period cannot be divided across time periods, and all time periods in each sub-agent must be continuous; the load fluctuation between any two moments in a sub-agent does not exceed the maximum threshold.

[0295]

[0296]

[0297]

[0298] In the formula, U m Let be the set of times within the m-th time period, and ΔL be the maximum threshold for load fluctuation;

[0299] The optimization problem of the time-period sub-agent partitioning method based on multi-variable load characteristics described above is a nonlinear discrete optimization problem, which is solved using the particle swarm optimization algorithm. The solution process is as follows:

[0300] 1) Parameter initialization: Input the load values ​​of electricity, heat, cooling and gas for each time period of the representative daily industrial load, commercial load and residential load, as well as parameters such as the number of iterations of the intelligent optimization algorithm;

[0301] 2) Set up a time-segment sub-agent M;

[0302] 3) Optimize variable initialization, specifically the number of time periods within each sub-agent;

[0303] 4) Calculate the load fluctuation of the j-th load user, the k-th load type, within the m-th time period (m-th time period sub-agent);

[0304] 5) Normalization process;

[0305] 6) Calculate the total load fluctuation;

[0306] 7) Determine whether the termination condition is met. If it is met, terminate the iteration and jump to step (8); if it is not met, the optimization variable is moved according to the particle swarm optimization algorithm to obtain the next optimization variable and jump to step (4).

[0307] 8) Output the time-segment sub-agent partitioning results;

[0308] 3.3.2 Lower-level model multi-agent computation model

[0309] A time-segment general agent and time-segment sub-agent model was established. The calculation process was decomposed into general agent calculation and time-segment sub-agent calculation. The two parts of the calculation are related by the boundary time of the sub-agent. The decision variables at each boundary time can be obtained through the general agent calculation. These are also the boundary conditions in the time-segment sub-agent calculation. After each time-segment sub-agent calculation is completed, the obtained running strategy is passed to the general agent in sequence as the initial condition for the next iteration calculation until the calculation ends.

[0310] Since the calculation process is decomposed into general agent calculation and time-segment sub-agent calculation, the lower-level model needs to be revised. The mathematical models of these two parts of the calculation are analyzed below.

[0311] (1) General Agent Calculation Mathematical Model

[0312] The purpose of the general agent calculation is to obtain the decision variables at each boundary time. Its objective function is the operation and maintenance cost and energy purchase cost corresponding to the decision variables at each boundary time, as shown in the following formula:

[0313]

[0314]

[0315]

[0316] In the formula, C G2 Calculate the objective function for the general agent; C G2,m The objective function computed by the sub-agent in the m-th time period; t1 mt2 represents the minimum boundary time of the m-th time-segment sub-agent; m Let be the maximum boundary time of the m-th time period sub-agent; if the m-th time period sub-agent has only one moment, then the sub-agent is a boundary sub-agent;

[0317] The constraints for the general agent's calculations are the same as those for the lower-level model mentioned above;

[0318] (2) Time-based sub-agent computational mathematical model

[0319] The purpose of time-segment sub-agent computation is to obtain the decision variables for each time-segment sub-agent. Its objective function is the operation and maintenance cost and energy purchase cost corresponding to the decision variables at other times except the boundary time. The objective function is shown in the following formula:

[0320]

[0321]

[0322]

[0323] In the formula, C G3 Calculate the objective function for the time-phase sub-agent; C G3,m The sum of the objective functions of the subordinate time steps within the m-th time-period sub-agent; t3 m For the m-th time period sub-agent, the decision variables and objective function at the boundary time are not calculated repeatedly, but directly inherit the calculation results of the general agent, while the decision variables at the boundary time are transformed as known quantities into boundary conditions or initial conditions for the calculation of the time period sub-agent.

[0324] The constraints for time-phase sub-agent computation are the same as those for the lower-level model mentioned above;

[0325] 3.4 Solution Process of Multi-Objective Robust Optimization Model for Integrated Energy System Based on Multi-Agent

[0326] The solution steps for the multi-objective robust optimization model of a multi-agent integrated energy system are as follows:

[0327] (1) Parameter initialization, including equipment parameters, load data, energy price policy, emission coefficient, planning cycle, number of iterations, number of particles, and optimization variables;

[0328] (2) The sampling scene was obtained by using Latin hypercube sampling and backward scene reduction techniques;

[0329] (3) Solve the time-segment sub-proxies to obtain the partitioning results;

[0330] (4) Optimization begins, iteration count n = 1;

[0331] (5) Determine if the number of iterations is greater than the maximum value. If it is greater than the maximum value, jump to step (15); otherwise, jump to step (6).

[0332] (6) Calculate the objective function, economic objective, and environmental objective of the upper-level model;

[0333] (7) The optimization algorithm is used to generate the decision variable x;

[0334] (8) Calculate the lower-level objective function and schedule the total running cost of the optimization subproblems;

[0335] (9) The general agent optimizes the boundary moment operation strategy;

[0336] (10) The sub-agent optimization process yields the internal time-based operation strategy of the sub-agent;

[0337] (11) Sub-agents at each time period upload the optimization results to the main agent, and the main agent performs combined solutions;

[0338] (12) Determine whether all time period sub-agents have been calculated. If yes, proceed to step (13); otherwise, proceed to step (10).

[0339] (13) Generate decision variable y;

[0340] (14) Calculate the objective function, economic objective, and environmental objective of the upper-level model;

[0341] (15) Select the configuration scheme using the minimum probability maximum regret value criterion;

[0342] (16) Determine whether the termination condition is met. If it is met, jump to step (17); otherwise, jump to step (5).

[0343] (17) Terminate the iteration and output the Pareto solution set of the multi-objective robust optimization configuration of the integrated energy system;

[0344] (18) End.

[0345] The above description is only a preferred embodiment of the present invention. It should be noted that, for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A multi-objective robust optimization configuration method for an integrated energy system, characterized in that, Follow these steps: I. Establishing a multivariate load uncertainty model: The uncertainty model of multiple loads in an integrated energy system is as follows: In the formula, the number of load users within the integrated energy system is The load variety is , The number of typical day scenarios per year, including 3 or 4 typical day scenarios. Typical daytime scene No. The load user is the first The first load variety The actual load value at each moment; Typical daytime scene No. The load user is the first The first load variety The baseline forecast load value at each time point; Typical daytime scene No. The load user is the first The first load variety The maximum load fluctuation value at any given time; Typical daytime scene No. The load user is the first The first load variety The load fluctuation coefficient at any given time. ; II. Multi-objective robust two-layer joint optimization configuration model for integrated energy systems: The upper-level model is the main model, which is the equipment selection and capacity optimization model. The lower-level model is the sub-model, which is the operation scheduling optimization model. The equipment configuration scheme determined by the upper-level model is used as the boundary condition. The upper-level model optimizes the configuration strategies and planned capacities of various equipment and passes them to the lower-level model. The lower-level model optimizes the scheduling and operation of the integrated energy system and returns the optimal operation plan and total operating cost to the upper-level model. Through the optimization iteration between the upper and lower levels, the optimal configuration strategy of the integrated energy system is obtained. 2.1 Upper-level model 2.1.1 Objective Function The objective function of the upper-level model is to minimize the total cost and the amount of pollutant emissions within the planning period, as shown in the following formula: in Optimize the objective function for multi-objective programming of integrated energy systems; The objective function is to optimize the overall energy system to minimize the total cost. The objective function is to minimize the pollutant emissions of the integrated energy system. Configure decision variables for the equipment; Optimize decision variables for system operation; (1) Economic objectives The total cost of joint optimization configuration of the integrated energy system is as follows: in The planning period is the total number of years of the plan. The discount rate for funds; For the first Total annual investment cost; For the first Total annual operating costs; No. Total annual investment cost: In the formula, Number the energy supply production equipment; A collection of production equipment for energy supply; For the first Unit capacity investment cost of energy supply production equipment; For the first The energy supply production equipment Annual construction capacity; Number the energy storage device; A collection of energy storage devices; For the first Unit capacity investment cost of an energy storage device; For the first The first energy storage device Annual construction capacity; (2) Environmental protection goals The pollutant emissions from the integrated energy system are configured in a coordinated and optimized manner as follows: In the formula, For the first Annual CO emissions; For the first Annual CO2 emissions; For the first Annual SO2 emissions; For the first Annual NOx emissions; For the first Year Typical daytime scene Electricity purchases during a given period; A collection of production equipment for energy supply that consumes natural gas; For the first Year Typical daytime scene Time period The amount of natural gas purchased by the energy supply production equipment that consumes natural gas; , , and These are the emission coefficients for CO, CO2, SO2, and NOx, which are pollutants generated from the consumption of electrical energy. , , and The first The emission coefficients of pollutants CO, CO2, SO2 and NOx generated by a natural gas-consuming energy supply production facility; For the first Year Typical daytime scene Other natural gas load power during the period; , , and These are the emission coefficients for CO, CO2, SO2, and NOx generated from natural gas consumption other than in energy supply production equipment that consumes natural gas; 2.1.2 Constraints (1) Upper limit constraint of equipment planning capacity The upper limit constraint on the planned capacity of equipment includes energy production equipment and energy storage equipment. Due to limitations such as geographical location, land area, space, geographical environment and natural environment, each type of equipment has a maximum planned capacity, and the actual planned capacity must not exceed the upper limit. In the formula, , These are the planned capacity limits for energy production equipment and energy storage equipment, respectively. (2) Equipment commissioning planning constraints Ensure the Year Typical daytime scene The external energy supply capacity, the total capacity of energy supply equipment, and the total discharge power of energy storage equipment during the period are greater than or equal to the peak load. In the formula, , These refer to the maximum external power supply and gas supply capacity, respectively. , , , The first The rated capacity of annual power, heat, cold energy and natural gas production equipment, of which the maximum output of distributed power sources is considered based on the reliable capacity; , , , These are respectively energy storage, thermal storage, cold storage, and gas storage equipment in the first... Maximum annual energy release power; , , , The first The annual peak values ​​of electricity load, heat load, cooling load and natural gas actual load, including energy supply production equipment that consumes the corresponding energy; The calorific value of natural gas. To improve the energy conversion efficiency of the electro-gas conversion equipment, The upper limit of the input electrical power of the electro-gas conversion device; 2.2 Lower-level model 2.2.1 Objective Function The objective function of the lower-level model is to minimize the total annual operating cost, as shown in the following formula: In the formula, For the first Annual equipment operation and maintenance costs; For the first Annual system energy purchase cost; In the formula, For the first Annual operating and maintenance cost per unit capacity of energy supply production equipment; For the first Annual operating and maintenance cost per unit capacity of an energy storage device; The energy purchase cost of an integrated energy system includes the cost of purchasing electricity and the cost of purchasing natural gas, as shown in the following formula: In the formula, Typical daily scene of the year The number of days it lasts; The total number of time periods in a day. , The first Year Typical daytime scene Electricity purchase price and natural gas purchase price for different time periods; , The first Year Typical daytime scene Electricity and natural gas purchases during the period; 2.1.2 Constraints (1) Electric power balance constraint In the formula, A collection of energy supply production equipment for generating electricity; A collection of energy storage devices; A collection of energy supply production equipment that consumes electricity; For the first Year Typical daytime scene Time period The electrical power generated by an energy supply production device that produces electricity; For the first Year Typical daytime scene Time period Discharge power of individual energy storage devices; For the first Year Typical daytime scene Time period The electrical power consumed by an energy supply device that consumes electrical energy; For the first Year Typical daytime scene Other actual electrical load power during the period; For the first Year Typical daytime scene Time period The charging power of each energy storage device; (2) Thermal power balance constraint In the formula, A collection of energy supply production equipment for generating heat; A collection of thermal storage equipment; A collection of production equipment that supplies energy that consumes heat. For the first Year Typical daytime scene Time period The heat power generated by the energy supply production equipment that produces heat; For the first Year Typical daytime scene Time period The heat release capacity of each thermal storage device; For the first Year Typical daytime scene Time period The heat load of energy supply production equipment that consumes heat energy; For the first Year Typical daytime scene The actual heat load power during the time period; For the first Year Typical daytime scene Time period The charging power of each thermal storage device; (3) Cold power balance constraint In the formula, A collection of energy supply production equipment for generating cold energy; A collection of cold storage equipment; For the first Year Typical daytime scene Time period The cooling power generated by an energy supply production device that produces cold energy; For the first Year Typical daytime scene Time period The cooling capacity of each cold storage device; For the first Year Typical daytime scene The actual cooling load power during the period; For the first Year Typical daytime scene Time period The cooling capacity of each cold storage device; (4) Natural gas power balance constraints In the formula, A collection of production equipment for energy supply that consumes natural gas; A collection of gas storage devices; For the first Year Typical daytime scene Natural gas power output during the period; For the first Year Typical daytime scene The natural gas power generated by the power-to-gas conversion equipment during certain periods; For the first Year Typical daytime scene Time period Gas discharge capacity of each gas storage device; For the first Year Typical daytime scene Time period The natural gas power supply for the production equipment that consumes natural gas; For the first Year Typical daytime scene Other actual natural gas load power during the period; No. Year Typical daytime scene Time period The gas filling power of each gas storage device; (5) Operational constraints of energy production and conversion equipment In the formula, , The first Year The lower and upper limits of the operating power of each energy production and conversion device. For the first Year Energy production and conversion equipment Typical daytime scene Operating power during a given time period; (6) Operational constraints of energy storage equipment In the formula, It is an energy storage type; for Charging power during a given period; for Power released during a given time period; The self-discharge rate of the energy storage device; for Time-of-use charging efficiency; for Energy release efficiency over time period; The energy storage status of the energy storage device; This refers to the rated capacity of the energy storage device. , These are the minimum and maximum values ​​of the ratio of allowable stored energy to stored energy capacity, respectively. , These refer to the charging and discharging power of the energy storage device, respectively. , These are the maximum energy storage capacity and energy release capacity of the energy storage device, respectively. , These represent the energy stored by the energy storage device at the beginning and end of the day, respectively. (7) Power exchange constraints with the superior network Due to network architecture limitations, the maximum power delivered from the upper-level network to the integrated energy system is restricted. Therefore, the power exchange between the integrated energy system and the upper-level network cannot exceed the upper limit. In the formula, , These are the minimum and maximum power purchase capacities, respectively. , These are the minimum and maximum gas purchasing power, respectively; III. Solution Method The upper-level model passes the configuration scheme to the lower-level model, which uses the configuration scheme as boundary conditions, considers the uncertainty of multiple loads, optimizes to obtain the optimal operating strategy, and returns the optimal operating strategy and total operating cost to the upper-level model. From the perspective of the upper-level and lower-level models, intelligent optimization algorithms are required to solve the problem. For the multi-load uncertainty model of the lower-level model, a multi-scenario technology and the minimum-maximum regret criterion are used to solve the problem together, and finally, the Pareto solution set of the multi-objective robust configuration scheme of the integrated energy system is obtained. For the lower-level model, firstly, a multi-scenario technique is used to describe the uncertainty of multi-variable loads; secondly, the particle swarm optimization algorithm is used to obtain the optimal solution for each scenario; finally, the minimum-maximum regret criterion is used to obtain the Pareto solution set of the final robust configuration scheme of the integrated energy system. 3.1 Multi-scenario technical description of multi-load uncertainty To accurately represent the uncertainty of multivariate loads, a Latin hypercube sampling method is used to generate a sufficient number of multivariate load fluctuation scenarios. However, too many samples will slow down the solution speed. To improve computational efficiency, a technique must be used to reduce the number of samples while maintaining the maximum sample fitting accuracy. A backward scenario reduction technique is used to reduce the number of samples. Let the number of samples be... The target sample size is ; Typical daytime scene No. The load user is the first Taking a single load variety as an example, the specific process is explained as follows; (1) Latin hypercube sampling 1) Set the sampling size to ; 2) The volatility coefficient interval Average score Sub-intervals; 3) Generate dimension is The Latin hypercube matrix, where The total number of time periods in a day, with each column of the matrix consisting of integers from 1 to... The matrix is ​​composed of a random sequence of numbers, where each column corresponds to an uncertain decision variable, and each row corresponds to a sample; at this point, All samples were obtained; (2) Scene reduction produce After one sample, the backward scenario reduction technique is used to reduce the multivariate load samples; One sample express, The probability of occurrence is used The specific process is as follows: 1) Set as The initial set of scenes, For from set The set of deleted scenes, the original set Empty; calculate the distance between any two scenes; In the formula, For the first Typical daytime scene Time period The load user is the first The first load variety The sample's volatility coefficient; For the first Typical daytime scene Time period The load user is the first The first load variety The sample's volatility coefficient; 2) For each scenario Determine the scene with the smallest distance. Introduce multi-variable maximum allowable load fluctuation constraints As the number of scenes is reduced, Gradually increase the size to ensure that some scenes that are far away are not deleted, thus maintaining scene diversity; ， 3) Calculation and be satisfied of ; 4) From the initial scene set Delete scene And update each set and its corresponding scenario. The probability of; 5) Determine if the target number of scenarios has been reached. If the current round of load reduction does not satisfy the maximum allowable fluctuation constraint for multi-variable loads, then update... It represents the smallest deviation in the existing scenario; 6) Scene reduction operation completed: In all typical daily scenarios, the target sample size for various load types of industrial, commercial, and residential users was obtained using the method described above. The scenarios are combined one by one to obtain a set of multiple load uncertainties for the integrated energy system. ; 3.2 Minimum and Maximum Regret Criterion for Selecting the Final Execution Strategy The minimum maximum regret criterion is an effective method for obtaining the final decision from multiple scenarios. Considering the probability of each multivariate load scenario, the minimum maximum regret criterion is calculated as follows; In the formula, In the configuration scheme Run strategy In the scene The objective function value in; In the configuration scheme The following execution strategy is used in the scenario Find the minimum value of the objective function in the given information; Configuration scheme The set of operational strategies under the uncertainty set of the following scenario; 3.3 Multi-objective robust optimization model for integrated energy systems based on multiple agents Using multi-scenario technology to handle the uncertainty of diverse loads will inevitably lead to a significant increase in computational load. Therefore, it is necessary to reduce the dimensionality of the optimization problem and reduce computational complexity. For the lower-level model, all time periods of a typical day are set as the general agent. Based on the characteristics of diverse loads, a time period sub-agent partitioning method is proposed to divide the model into multiple time period sub-agents according to the time period, thereby reducing the dimensionality of variables. 3.3.1 Time-based Sub-agent Partitioning Method Based on Multi-variable Load Characteristics When dividing time-segment sub-agents, optimization time periods cannot span across time periods; all time periods within each sub-agent must be consecutive. Set the time-segment sub-agent as... There are 1 sub-agent, and the first and last moments of each time period are the boundary moments. The method for dividing time period sub-agents is as follows: (1) Calculate the load fluctuation: Calculate the first The load user is the first The first load variety The load fluctuation within a sub-agent during a given time period is calculated using the following formula: In the formula: For the first The load user is the first The first load variety Load fluctuation within a time period; For the first The number of moments in the load curve within a time period; For the first The load user is the first The first load variety Within the time period, the first Load values ​​at each moment; (2) Normalization: Since the load values ​​of different load applications and different load types vary greatly, directly calculating the load fluctuation is prone to large errors. Therefore, it is necessary to normalize the load fluctuation. The calculation formula is as follows: In the formula, This is the normalized load fluctuation. For the first The load user is the first The first load variety The maximum load within a time period; (3) Calculate the total load fluctuation: Calculate Individual load users, Individual load varieties, The sum of load fluctuations of each time period sub-agent is used to determine the time period sub-agent partitioning strategy with the goal of minimizing the load fluctuation. Therefore, optimizing the time period sub-agent partitioning can be described as the following optimization problem. (4) Constraint: The sum of the number of time periods in all sub-agents is equal to the total number of time periods to be optimized. The optimization period cannot be divided into agents across time periods; all time periods in each sub-agent must be continuous; the load fluctuation between any two times within a time period sub-agent shall not exceed the maximum threshold. In the formula, For the first A set of moments within a time period The maximum threshold for load fluctuation; The optimization problem of the time-period sub-agent partitioning method based on multi-variable load characteristics described above is a nonlinear discrete optimization problem, which is solved using the particle swarm optimization algorithm. The solution process is as follows: 1) Parameter initialization: Input the load values ​​of electricity, heat, cooling and gas for each time period of each representative day's industrial load, commercial load and residential load, as well as parameters such as the number of iterations of the intelligent optimization algorithm; 2) Set up time-segment sub-agents M ; 3) Optimize variable initialization, specifically the number of time periods within each sub-agent; 4) Calculate the first The load user is the first The first load variety Within the time period (the first time period) The load fluctuation of each time period sub-agent; 5) Normalization process; 6) Calculate the total load fluctuation; 7) Determine whether the termination condition is met. If it is met, terminate the iteration and jump to step (8); if it is not met, the optimization variable is moved according to the particle swarm optimization algorithm to obtain the next optimization variable and jump to step (4). 8) Output the time-segment sub-agent partitioning results; 3.3.2 Lower-level multi-agent computation model A time-segment general agent and time-segment sub-agent model was established. The calculation process was decomposed into general agent calculation and time-segment sub-agent calculation. The two parts of the calculation are related by the boundary time of the sub-agent. The decision variables at each boundary time can be obtained through the general agent calculation. These are also the boundary conditions in the time-segment sub-agent calculation. After each time-segment sub-agent calculation is completed, the obtained running strategy is passed to the general agent in sequence as the initial condition for the next iteration calculation until the calculation ends. Since the calculation process is decomposed into general agent calculation and time-segment sub-agent calculation, the lower-level model needs to be revised. The mathematical models of these two parts of the calculation are analyzed below. (1) General Agent Calculation Mathematical Model The purpose of the general agent calculation is to obtain the decision variables at each boundary time. Its objective function is the operation and maintenance cost and energy purchase cost corresponding to the decision variables at each boundary time, as shown in the following formula: In the formula, Calculate the objective function for the general agent; For the first The objective function computed by the sub-agent in each time period; For the first The minimum boundary time of each time-segment sub-agent; For the first The maximum boundary time of the sub-agent in the nth time period; if the nth time period... If a sub-agent has only one time period, then the sub-agent is a boundary sub-agent; The constraints for the general agent's calculations are the same as those for the lower-level model mentioned above; (2) Mathematical model for time-segment sub-agent computation The purpose of time-segment sub-agent computation is to obtain the decision variables for each time-segment sub-agent. Its objective function is the operation and maintenance cost and energy purchase cost corresponding to the decision variables at other times except the boundary time. The objective function is shown in the following formula: In the formula, Calculate the objective function for the time-segment sub-agent; For the first The sum of the objective functions of the subordinate time periods within each time period sub-agent; For the first The subordinate time of each time-period sub-agent; the decision variables and objective functions at the boundary time are not calculated repeatedly, but directly inherit the calculation results of the general agent, while the decision variables at the boundary time are transformed as known quantities into boundary conditions or initial conditions for the calculation of the time-period sub-agent; The constraints for time-phase sub-agent computation are the same as those for the lower-level model mentioned above; 3.4 Solution Process of Multi-Objective Robust Optimization Model for Integrated Energy System Based on Multi-Agent The solution steps for the multi-objective robust optimization model of a multi-agent integrated energy system are as follows: (1) Parameter initialization, including equipment parameters, load data, energy price policy, emission coefficient, planning cycle, number of iterations, number of particles, and optimization variables; (2) The sampling scene was obtained by using Latin hypercube sampling and backward scene reduction techniques; (3) Solve the time-segment sub-proxies to obtain the partitioning results; (4) Optimization begins, iteration count n=1; (5) Determine if the number of iterations is greater than the maximum value. If it is greater than the maximum value, jump to step (15); otherwise, jump to step (6). (6) Calculate the objective function of the upper-level model, including the economic objective and the environmental objective; (7) Optimization algorithm to generate decision variables x ; (8) Calculate the lower-level objective function and schedule the total running cost of the optimization subproblems; (9) The general agent optimizes the boundary time operation strategy; (10) The sub-agent optimization method is used to obtain the internal time-based operation strategy of the sub-agent; (11) Sub-agents at each time period upload the optimization results to the main agent, and the main agent performs combined solutions; (12) Determine whether all time period sub-agents have been calculated. If yes, proceed to step (13); otherwise, proceed to step (10). (13) Generate decision variables y ; (14) Calculate the objective function of the upper-level model, including the economic objective and the environmental objective; (15) Select the configuration scheme using the minimum probability maximum regret criterion; (16) Determine whether the termination condition is met. If it is met, jump to step (17); otherwise, jump to step (5). (17) Terminate the iteration and output the Pareto solution set of the multi-objective robust optimization configuration of the integrated energy system; (18) End.

Images