A multi-modal multi-target hierarchical distributed comprehensive energy system economic dispatch method

Abstract

The application provides a multi-modal multi-target hierarchical distributed comprehensive energy system economic dispatching method, which combines an analytical multi-modal multi-target method and a hierarchical distributed consistency method to be used for economic dispatching of a comprehensive energy system. First, the analytical multi-modal multi-target method in the method is used to obtain a multi-modal economic dispatching multi-target scheme with diversity. Second, the hierarchical distributed consistency method in the method is used to quickly obtain an accurate economic dispatching decision scheme by using hierarchical operation. The multi-modal multi-target hierarchical distributed comprehensive energy system economic dispatching method can solve the multi-modal problem of large-scale comprehensive energy system economic dispatching, provide a diversified economic dispatching decision scheme, optimize the stability of economic dispatching, and improve the calculation speed and accuracy.

Description

Technical Field

[0001] This invention belongs to the field of power systems and integrated energy, and relates to an analytical multimodal multi-objective hierarchical distributed method, which is applicable to the multimodal multi-objective hierarchical distributed economic dispatch of integrated energy systems. Background Technology

[0002] The continuous expansion of the electricity market and the gradual increase in the scale of integrated energy systems lead to slower calculation speeds and reduced information privacy in the economic dispatch of integrated energy systems. Furthermore, most economic dispatch methods for integrated energy systems are single-objective, with few multi-objective approaches available. Moreover, research on multi-objective economic dispatch problems for integrated energy systems has not considered multimodal issues, resulting in a lack of diversity and comprehensiveness in multi-objective economic dispatch schemes.

[0003] Therefore, a multimodal, multi-objective, hierarchical, distributed economic dispatch method for integrated energy systems is proposed. This method can preserve the multimodal characteristics and, in the multimodal, multi-objective, hierarchical, distributed economic dispatch problem of integrated energy systems, can not only speed up the system's problem-solving speed and ensure the system's privacy and robustness, but also provide multimodal options for the dispatch of integrated energy systems, thus ensuring the stability and security of economic dispatch. Summary of the Invention

[0004] This invention proposes a multimodal, multi-objective, hierarchical, distributed economic dispatch method for integrated energy systems. It combines analytical multimodal, multi-objective methods with hierarchical distributed consensus methods for the economic dispatch of integrated energy systems. This increases the diversity and stability of economic dispatch, improves computational speed and accuracy, and provides decision-makers with a wealth of alternative decision-making schemes. The steps in its application are as follows:

[0005] Step (1): The construction of the economic dispatch model of the integrated energy system includes thermal energy, hydropower, wind energy, solar energy and geothermal energy, and takes cost and carbon emissions as multiple objectives, following the equality constraints and inequality constraints of economic dispatch.

[0006] The objective function for total power generation cost is:

[0007]

[0008]

[0009] in, The total cost of electricity generation; This represents the cost of the i-th generator set at time t; This represents the cost of the j-th thermal power generating unit at time t; This represents the cost of the k-th hydroelectric generator unit at time t; This represents the cost of the l-th wind turbine at time t; This represents the cost of the r-th photovoltaic generator unit at time t; This represents the cost of the s-th geothermal generator unit at time t; The value represents the cost of the u-th clean energy generator unit at time t; N is the total number of generator units; N TG N represents the number of thermal power generating units. HG N represents the number of hydroelectric generator units. WG N represents the number of wind turbine generator units. PG N represents the number of photovoltaic generator units. OG N represents the number of geothermal generator units. CleanG The number of clean energy generator units; T is the statistical time of the objective function; Let be the power output of the i-th generator unit at time t; Let be the power generation of the j-th thermal power generating unit at time t; Let be the power generation of the kth hydroelectric generator unit at time t; Let t be the power generation of the l-th wind turbine at time t; Let r be the power generation of the r-th photovoltaic generator unit at time t; Let be the power generation of the s-th geothermal generator unit at time t; Let a be the power generation of the u-th clean energy generator unit at time t; j b j and c j d represents the cost coefficient of the j-th thermal power generating unit; k e is the cost coefficient for the k-th hydroelectric generator unit; l The cost coefficient for the l-th wind turbine generator; g r Let m be the cost coefficient of the r-th photovoltaic generator unit; s κ is the cost coefficient of the s-th geothermal generator unit; u Let u be the cost coefficient of the uth clean energy generator set; use the cost of clean energy generator sets. Replacement water and electricity costs Wind power cost Cost of light energy and geothermal energy costs

[0010] The objective function for total carbon emissions is:

[0011]

[0012] in, This represents the total carbon emissions; α represents the carbon emissions of the j-th thermal power generating unit at time t;j β j and γ j Let be the carbon emission coefficient of the j-th thermal power generating unit; the power balance equation constraint is:

[0013]

[0014] in, Let be the total load demand at time t;

[0015] The upper and lower limits of power are constrained as follows:

[0016]

[0017] in, This is the lower limit of the power generation of the j-th thermal power generating unit; This represents the upper limit of the power generation capacity of the j-th thermal power generating unit. This represents the lower limit of the power generation of the kth hydroelectric generator unit; This represents the upper limit of the power generation capacity of the kth hydroelectric generator unit; This is the lower limit of the power generation of the l-th wind turbine generator; This is the upper limit of the power generation of the l-th wind turbine unit; This represents the lower limit of the power generation of the r-th photovoltaic generator unit; This represents the upper limit of the power generation of the r-th photovoltaic generator unit; This is the lower limit of the power generation of the sth geothermal generator unit; This is the upper limit of the power generation of the s-th geothermal generator unit;

[0018] The ramp rate constraint for thermal power units is:

[0019]

[0020] in, Let J be the power generation of the j-th thermal power generating unit at time t-1; Let be the downward ramp rate of the j-th thermal power unit; T is the upward ramp rate of the j-th thermal power unit; 60 It lasts for 60 minutes;

[0021] The time constraint for the optoelectronic module is:

[0022]

[0023] Step (2): Initialize the weighting factor η according to the proportion R, let

[0024]

[0025] Where R is the proportion of the weighting factor values; η is the weighting factor, whose value range is 0≤η≤1; n η This indicates the number of segments into which the range of values ​​for the weight factor is divided;

[0026] Step (3): Transform the multi-objective problem into a single-objective problem using a linear weighting method;

[0027]

[0028] Among them, f Total The total objective value is the linearly weighted sum. The target value of the j-th thermal power unit at time t is the line-weighted value.

[0029] Step (4): Set the initial iteration time t = 1;

[0030] Step (5): Input the predicted load value and set the initial iteration step number k = 0;

[0031] Step (6): Divide the integrated energy system into four layers: The first layer is divided into region A, region B and region C; the second layer is divided into region Aa, region Ab, region Ac, region Ba, region Bb, region Bc, region Ca and region Cb; the third layer is divided into region Aa1, region Aa2, region Ab1, region Ab2, region Ac1, region Ac2, region Ba1, region Ba2, region Ba3, region Bb1, region Bb2, region Bc1, region Bc2, region Ca1, region Ca2, region Cb1, region Cb2 and region Cb3; the fourth layer consists of the regions within the third layer. The generator set; Step (7): Select the leader and follower agents of each layer. Region A of the first layer is the leader, and regions B and C are the followers; Regions Aa, Ba and Ca of the second layer are the leaders, and the other regions of the second layer are the followers; Regions Aa1, Ab1, Ac1, Ba1, Bb1, Bc1, Ca1 and Cb1 of the third layer are the leaders, and the other regions of the third layer are the followers; Step (8): Update the consistency variables of the first layer agents by combining the topology diagram of the agents and formulas (15) and (16);

[0032] The adjacency matrix A = (a, E) of the directed graph G = (V, E) of the distributed network topology and the weight relationships between the corresponding agents. ij The network topology that constitutes multiple agents; the Laplace matrix L of the directed graph G = [l ij ]for:

[0033]

[0034] Among them, l ii and lij It is an element in the Laplace matrix L; a ij These are elements in the adjacency matrix A;

[0035] The agent's row random matrix H ij for:

[0036]

[0037] The consistency variable λ is represented as:

[0038] λ=[λ T , λ Clean (12)

[0039]

[0040]

[0041] Where λ represents the consistency variable of the integrated energy system; λ T λ represents the consistency variable of a thermal power generating unit. Clean Represents the consistency variables of clean energy generator sets;

[0042] The update formulas for the consistency variables of followers and leaders are as follows:

[0043]

[0044]

[0045] Where, N IB The number of agents is represented; k represents the k-th iteration; Let represent the consistency variable of the nth agent at time t in the kth iteration; τ represents the consensus variable of the m-th agent at time t in the (k+1)-th iteration; τ represents the power balance adjustment factor of the distributed consensus method; ΔP t (k) represents the power deviation of the k-th iteration at time t; H mn (k) represents the row random matrix of the agent;

[0046] Step (9): Calculate the active power output of each agent in the first layer according to formula (17);

[0047] The active power formulas for each unit are as follows:

[0048]

[0049] in, Let represent the active power output of the i-th agent at time t in the (k+1)-th iteration; Let represent the active power output of the j-th thermal power generating unit at time t in the (k+1)th iteration; Let represent the active power output of the u-th clean energy generator unit at time t in the (k+1)th iteration; It represents the active power limit of the μ-th region in the ω-th layer, which is the sum of the active power limits of the lower-level sub-regions or units divided by this region; Let represent the consistency variable of the i-th agent at time t in the (k+1)th iteration; Let represent the consistency variable of the j-th thermal power generating unit at time t in the (k+1)th iteration; Let represent the consistency variable of the u-th clean energy unit at time t in the (k+1)th iteration; This represents the number of thermal power units in the μ-th region of the ω-th layer; U represents the number of clean energy generator sets in the μ-th region of the ω-th layer; U represents the total number of regions in this layer; W represents the total number of layers.

[0050] Step (10): Correct the first layer of active power output obtained in step (9) according to formula (18);

[0051] The correction formula for active power is:

[0052]

[0053] in, This represents the lower limit of the active power output of the i-th power generation agent; This represents the upper limit of the active power output of the i-th power generation agent;

[0054] Step (11): Calculate the active power deviation value of the first layer according to formula (19);

[0055] Active power deviation ΔP t (k+1) is:

[0056]

[0057] Step (12): Determine whether the multimodal active power deviation value of the first layer obtained in step (11) meets the accuracy requirements.

[0058]

[0059] Where δ represents the maximum allowable power deviation; Indicates the allowable value for multimodal power deviation;

[0060] if Then let the iteration step number k = k + 1, and return to step (8) to continue the iterative calculation of the first layer;

[0061] if Store the active power output of the obtained set of agents, then iterate for another k = k + 1 steps, and return to step (8) to continue the iterative calculation of the first layer;

[0062] if If the multimodal active power deviation value of the first layer meets the requirements, the iteration ends, and the active power output of multiple agents is obtained.

[0063] Step (13): Perform multimodal judgment and decision congestion distance screening on the active power output of the multiple sets of intelligent agent generator sets obtained in step (12) to obtain multimodal active power output that meets the decision-maker's requirements; the multimodal judgment is as follows:

[0064] |f(x1)-f(x2)|<ε (21)

[0065] Here, x1 and x2 are the two solutions for the multimodal problem; ε represents the critical threshold for the multimodal problem.

[0066] Crowded distance in the decision space is used as the filter:

[0067]

[0068]

[0069]

[0070] in, x represents the crowding distance of the j-th decision variable in the i-th solution set; i+1,j x represents the j-th decision variable in the (i+1)-th solution set; i,j Let j represent the j-th decision variable in the i-th solution set; The j-th decision variable represents the maximum objective value; The j-th decision variable represents the minimum objective value; x i-1,j Let j be the decision variable of the (i-1)th solution set; Let ξ represent the crowding distance of the i-th solution group; D represents the number of decision variables; ξ represents the minimum crowding distance allowed by multimodal decision-making.

[0071] Step (14): Input the multiple sets of multi-mode power generation of the first layer obtained in step (13) as load values ​​into the second layer, and set the initial iteration step number k = 0;

[0072] Step (15): Update the consistency variables of the second-layer agent by combining the topology diagram of the agent with formulas (15) and (16);

[0073] Step (16): Calculate the active power output of each intelligent agent unit in the second layer according to formula (17);

[0074] Step (17): Correct the active power output obtained in step (16) according to formula (18);

[0075] Step (18): Calculate the active power deviation value of the second layer according to formula (19);

[0076] Step (19): Determine whether the multimodal active power deviation value obtained in step (18) meets the accuracy requirements;

[0077] if Then let the iteration step number k = k + 1, and return to step (15) to continue the iteration calculation of the second layer;

[0078] if Store the active power output of the obtained set of agents, then iterate for another k = k + 1 steps, and return to step (15) to continue the iterative calculation of the second layer;

[0079] if If the multimodal active power deviation value of the second layer meets the requirements, the iteration ends, and the active power output of multiple intelligent agent generator sets is obtained.

[0080] Step (20): Perform multimodal judgment and decision congestion distance screening on the active power output of the multiple groups of agents in the second layer obtained in step (19) to obtain multimodal active power output that meets the requirements of the decision-maker;

[0081] Step (21): Input the multiple sets of multi-mode power generation of the second layer obtained in step (20) as load values ​​into the third layer, and set the initial iteration step number k = 0;

[0082] Step (22): Update the consistency variables of the third-layer agent by combining the topology diagram of the agent with formulas (15) and (16);

[0083] Step (23): Calculate the active power output of each intelligent agent unit in the third layer according to formula (17);

[0084] Step (24): Correct the active power output obtained in step (23) according to formula (18);

[0085] Step (25): Calculate the active power deviation value of the third layer according to formula (19);

[0086] Step (26): Determine whether the multimodal active power deviation value obtained in step (25) meets the accuracy requirements;

[0087] if Then let the iteration step number k = k + 1, and return to step (22) to continue the iteration calculation of the third layer;

[0088] if Store the active power output of the obtained set of agents, then iterate for another k = k + 1 steps, and return to step (22) to continue the iterative calculation of the third layer;

[0089] if If the multimodal active power deviation value of the third layer meets the requirements, the iteration ends, and the active power output of multiple intelligent agent generator sets is obtained.

[0090] Step (27): Perform multimodal judgment and decision congestion distance screening on the active power output of the multiple groups of agents in the third layer obtained in step (26) to obtain multimodal active power output that meets the requirements of the decision-maker;

[0091] Step (28): Input the multiple sets of multi-mode power generation of the third layer obtained in step (27) as load values ​​into the fourth layer, and set the initial iteration step number k = 0;

[0092] Step (29): Update the consistency variables of each generator set in the fourth layer by combining the topology diagram of the agent with formulas (15) and (16);

[0093] Step (30): Calculate the active power output of each generator set in the fourth layer according to formula (17);

[0094] Step (31): Correct the active power output obtained in step (30) according to formula (18); Step (32): Calculate the active power deviation value of the fourth layer according to formula (19);

[0095] Step (33): Determine whether the multimodal active power deviation value obtained in step (32) meets the accuracy requirements;

[0096] if Then let the iteration step number k = k + 1, and return to step (29) to continue the iteration calculation of the fourth layer;

[0097] if Store the active power output of the obtained generator set, then iterate for another k = k + 1 step, and return to step (29) to continue the iterative calculation of the fourth layer;

[0098] if If the multimodal active power deviation value of the fourth layer meets the requirements, the iteration ends and the active power output of multiple generator sets is obtained.

[0099] Step (34): Perform multimodal judgment and decision congestion distance screening on the active power output of the multiple generator sets in the fourth layer obtained in step (33) to obtain a multimodal active power output decision scheme that meets the requirements of the decision-maker;

[0100] Step (35): Determine whether t < T is satisfied. If it is satisfied, let t = t + 1 and transfer to step (5); if not, proceed to the next step;

[0101] Step (36): Determine whether η < 1 is satisfied. If it is satisfied, store the multi-modal solution obtained when the value is η, let η = η + R, and transfer to step (4); if not, proceed to the next step;

[0102] Step (37): Inverse-transform all the multi-modal solutions obtained with different η values to obtain multi-modal multi-objective values. The multi-modal multi-objective values obtained under all η values form a Pareto front in the objective space.

[0103] The present invention has the following advantages and effects compared with the prior art:

[0104] (1) An analytical framework of a multi-modal multi-objective hierarchical distributed method is built for the economic dispatch problem of the integrated energy system.

[0105] (2) The constructed analytical multi-modal multi-objective hierarchical distributed method, because it adopts hierarchical dispatch and only exchanges the consistency variables of adjacent regions without exchanging the detailed operation information of each unit, speeds up the calculation speed and better guarantees information privacy; moreover, it solves the multi-modal characteristic problem in the integrated energy system dispatch, increasing the diversity and comprehensiveness of the economic dispatch solutions.

[0106] (3) In the economic dispatch plan of the integrated energy system obtained, there are multiple mutually replaceable economic dispatch plans. When facing some sudden accidents or serious situations of wind curtailment, light curtailment, and water curtailment in some regions, decision-makers have more alternative plans, increasing the stability and security of the economic dispatch. Brief Description of the Drawings

[0107] Figure 1 is the hierarchical distributed schematic diagram of the method of the present invention.

[0108] Figure 2 is the analytical multi-modal multi-objective distributed consistency flowchart of the method of the present invention.

[0109] Figure 3 is the overall flowchart of the method of the present invention. Detailed Embodiment

[0110] A multi-modal multi-objective hierarchical distributed economic dispatch method for an integrated energy system proposed by the present invention is described in detail with reference to the accompanying drawings as follows:

[0111] Figure 1This is a schematic diagram of the hierarchical distributed system of the present invention. The integrated energy system is divided into four layers: the first layer is divided into region A, region B, and region C; the second layer is divided into region Aa, region Ab, region Ac, region Ba, region Bb, region Bc, region Ca, and region Cb; the third layer is divided into region Aa1, region Aa2, region Ab1, region Ab2, region Ac1, region Ac2, region Ba1, region Ba2, region Ba3, region Bb1, region Bb2, region Bc1, region Bc2, region Ca1, region Ca2, region Cb1, region Cb2, and region Cb3; the fourth layer consists of the generator sets within each region of the third layer. A leader and followers are selected for each layer.

[0112] Figure 2 This is an analytical flowchart of the multimodal, multi-objective distributed consensus method of the present invention. The specific steps are as follows:

[0113] Step 1: Transform the multi-objective problem into a single-objective problem according to formula (9);

[0114] Step 2: Input the predicted load value and set the initial iteration step number k = 0;

[0115] Step 3: Determine the leader and followers of the agent, and update the consistency variables of the agent by combining the topology diagram of the agent and formulas (15) and (16);

[0116] Step 4: Calculate the active power output of each intelligent agent unit according to formula (17);

[0117] Step 5: Correct the active power output obtained in Step 4 according to formula (18);

[0118] Step 6: Calculate the active power deviation value according to formula (19);

[0119] Step 7: Determine whether the multimodal active power deviation value obtained in Step 6 meets the accuracy requirements. If... Then let the iteration step number k = k + 1, and return to step 3; if Store the active power output of the obtained set of agents, then iterate for another k = k + 1 steps, and return to step 3; if If the multimodal active power deviation value meets the requirements, the iteration ends, and the active power output of multiple intelligent agent generator sets is obtained.

[0120] Step 8: Perform multimodal judgment and decision congestion distance screening on the active power output of the multiple sets of intelligent agent generator sets obtained in Step 7 to obtain a multimodal active power output scheduling scheme that meets the decision-maker's requirements.

[0121] Figure 3 This is an overall flowchart of the method of the present invention. The specific steps are as follows:

[0122] Step 1: Initialize the weighting factors, let η = 1 / b η ;

[0123] Step 2: Set the initial iteration time t = 1; Step 3: Input the predicted load value and set the initial iteration step number k = 0;

[0124] Step 4: Divide the system into four layers. The first layer is divided into region A, region B, and region C; the second layer is divided into region Aa, region Ab, region Ac, region Ba, region Bb, region BC, region Ca, and region Cb; the third layer is divided into region Aa1, region Aa2, region Ab1, region Ab2, region Ac1, region Ac2, region Ba1, region Ba2, region Ba3, region Bb1, region Bb2, region Bc1, region Bc2, region Ca1, region Ca2, region Cb1, region Cb2, and region Cb3; the fourth layer consists of the generator sets within each region of the third layer.

[0125] Step 5: Select the leader and follower agents for each layer. In the first layer, region A is the leader, and regions B and C are the followers. In the second layer, regions Aa, Ba, and Ca are the leaders, and the other regions in the second layer are the followers. In the third layer, regions Aa1, Ab1, Ac1, Ba1, Bb1, Bc1, Ca1, and Cb1 are the leaders, and the other regions in the third layer are the followers.

[0126] Step 6: Utilize Figure 2 The analytical multimodal multi-objective distributed consensus method in the paper optimizes the solution of the first layer of the system.

[0127] Step 7: Input the multiple sets of multi-mode power generation in the three regions of the first layer obtained in Step 6 as load values ​​into the second layer;

[0128] Step 8: Utilize Figure 2 The analytical multimodal multi-objective distributed consensus method in the paper optimizes the solution of the second layer of the system;

[0129] Step 9: Input the multiple sets of multi-mode power generation of the eight regions in the second layer obtained in Step 8 as load values ​​into the third layer;

[0130] Step 10: Utilize Figure 2 The analytical multimodal multi-objective distributed consensus method in the paper optimizes the solution of the third layer of the system.

[0131] Step 11: Respectively take the multi-group multi-modal power generation amounts of the 18 regions in the third layer obtained in Step 10 as load values and input them into the fourth layer. Regard each generator set in the fourth layer as an agent, and select the leader and follower agents;

[0132] Step 12: Use the Figure 2 analytical formula in the multi-modal multi-objective distributed consensus method to perform an optimization solution on the fourth layer of the system;

[0133] Step 13: Judge whether t < T is satisfied. If it is satisfied, let t = t + 1 and transfer to Step 2; if it is not satisfied, execute the next step;

[0134] Step 14: Judge whether η < 1 is satisfied. If it is satisfied, store the multi-modal solution obtained when the value is η, let η = η + R, and transfer to Step 2; if it is not satisfied, execute the next step; Step 15: Obtain the multi-modal multi-objective values through inverse calculation for all the multi-modal solutions obtained with different η values; Step 16: The multi-modal multi-objective values obtained under all η values form a Pareto front in the objective space.

[0135] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. An economic dispatch method for a multimodal, multi-objective, hierarchical, distributed integrated energy system, characterized in that: This paper combines analytical multimodal multi-objective methods with hierarchical distributed consensus methods for the economic dispatch of integrated energy systems. This increases the diversity and stability of economic dispatch, improves computational speed and accuracy, and provides decision-makers with a wealth of alternative decision-making schemes. The steps involved in its application are as follows: Step (1): The construction of the economic dispatch model of the integrated energy system includes thermal energy, hydropower, wind energy, solar energy and geothermal energy, and takes cost and carbon emissions as multiple objectives, following the equality constraints and inequality constraints of economic dispatch. The objective function for total power generation cost is: in, The total cost of electricity generation; This represents the cost of the i-th generator set at time t; This represents the cost of the j-th thermal power generating unit at time t; This represents the cost of the k-th hydroelectric generator unit at time t; This represents the cost of the l-th wind turbine at time t; This represents the cost of the r-th photovoltaic generator unit at time t; This represents the cost of the s-th geothermal generator unit at time t; The value represents the cost of the u-th clean energy generator unit at time t; N is the total number of generator units; N TG N represents the number of thermal power generating units. HG N represents the number of hydroelectric generator units. WG N represents the number of wind turbine generator units. PG N represents the number of photovoltaic generator units. OG N represents the number of geothermal generator units. CleanG The number of clean energy generator units; T is the statistical time of the objective function; Let be the power output of the i-th generator unit at time t; Let be the power generation of the j-th thermal power generating unit at time t; Let be the power generation of the kth hydroelectric generator unit at time t; Let t be the power generation of the l-th wind turbine at time t; Let r be the power generation of the r-th photovoltaic generator unit at time t; Let be the power generation of the s-th geothermal generator unit at time t; Let a be the power generation of the u-th clean energy generator unit at time t; j b j and c j d represents the cost coefficient of the j-th thermal power generating unit; k e is the cost coefficient for the k-th hydroelectric generator unit; l The cost coefficient for the l-th wind turbine generator; g r Let m be the cost coefficient of the r-th photovoltaic generator unit; s Let k be the cost coefficient of the s-th geothermal generator unit; u Let u be the cost coefficient of the uth clean energy generator set; use the cost of clean energy generator sets. Replacement water and electricity costs Wind power cost Cost of light energy and geothermal energy costs The objective function for total carbon emissions is: in, This represents the total carbon emissions; α represents the carbon emissions of the j-th thermal power generating unit at time t; j β j and γ j Let be the carbon emission coefficient of the j-th thermal power generating unit; The equality constraints for power balance are: in, Let be the total load demand at time t; The upper and lower limits of power are constrained as follows: in, This is the lower limit of the power generation of the j-th thermal power generating unit; This represents the upper limit of the power generation capacity of the j-th thermal power generating unit. This represents the lower limit of the power generation of the kth hydroelectric generator unit; This represents the upper limit of the power generation capacity of the kth hydroelectric generator unit; This is the lower limit of the power generation of the l-th wind turbine generator; This is the upper limit of the power generation of the l-th wind turbine unit; This represents the lower limit of the power generation of the r-th photovoltaic generator unit; This represents the upper limit of the power generation of the r-th photovoltaic generator unit; This is the lower limit of the power generation of the sth geothermal generator unit; This is the upper limit of the power generation of the s-th geothermal generator unit; The ramp rate constraint for thermal power units is: in, Let J be the power generation of the j-th thermal power generating unit at time t-1; Let be the downward ramp rate of the j-th thermal power unit; T is the upward ramp rate of the j-th thermal power unit; 60 It lasts for 60 minutes; The time constraint for the optoelectronic module is: Step (2): Initialize the weighting factor η according to the proportion R, let Where R is the proportion of the weighting factor values; η is the weighting factor, whose value range is 0≤η≤1; n η This indicates the number of segments into which the range of values ​​for the weight factor is divided; Step (3): Transform the multi-objective problem into a single-objective problem using a linear weighting method; Among them, f Total The total objective value is the linearly weighted sum. The target value of the j-th thermal power unit at time t is the line-weighted value. Step (4): Set the initial iteration time t = 1; Step (5): Input the predicted load value and set the initial iteration step number k = 0; Step (6): Divide the integrated energy system into four layers: The first layer is divided into region A, region B and region C; the second layer is divided into region Aa, region Ab, region Ac, region Ba, region Bb, region Bc, region Ca and region Cb; the third layer is divided into region Aa1, region Aa2, region Ab1, region Ab2, region Ac1, region Ac2, region Ba1, region Ba2, region Ba3, region Bb1, region Bb2, region Bc1, region Bc2, region Ca1, region Ca2, region Cb1, region Cb2 and region Cb3; the fourth layer consists of the regions within the third layer. The generator set; Step (7): Select the leader and follower agents of each layer. Region A of the first layer is the leader, and regions B and C are the followers; Regions Aa, Ba and Ca of the second layer are the leaders, and the other regions of the second layer are the followers; Regions Aa1, Ab1, Ac1, Ba1, Bb1, Bc1, Ca1 and Cb1 of the third layer are the leaders, and the other regions of the third layer are the followers; Step (8): Update the consistency variables of the first layer agents by combining the topology diagram of the agents and formulas (15) and (16); The adjacency matrix A = (a, E) of the directed graph G = (V, E) of the distributed network topology and the weight relationships between the corresponding agents. ij The network topology that constitutes multiple agents; the Laplace matrix L of the directed graph G = [l ij ]for: Among them, l ii and l ij It is an element in the Laplace matrix L; a ij These are elements in the adjacency matrix A; The agent's row random matrix H ij for: The consistency variable λ is represented as: λ=[λ T ,l Clean ] (12) Where λ represents the consistency variable of the integrated energy system; λ T λ represents the consistency variable of a thermal power generating unit. Clean Represents the consistency variables of clean energy generator sets; The update formulas for the consistency variables of followers and leaders are as follows: Where, N IB The number of agents is represented; k represents the k-th iteration; Let represent the consistency variable of the nth agent at time t in the kth iteration; τ represents the consensus variable of the m-th agent at time t in the (k+1)-th iteration; τ represents the power balance adjustment factor of the distributed consensus method; ΔP t (k) represents the power deviation of the k-th iteration at time t; H mn (k) represents the row random matrix of the agent; Step (9): Calculate the active power output of each agent in the first layer according to formula (17); The active power formulas for each unit are as follows: in, Let represent the active power output of the i-th agent at time t in the (k+1)-th iteration; Let represent the active power output of the j-th thermal power generating unit at time t in the (k+1)th iteration; Let represent the active power output of the u-th clean energy generator unit at time t in the (k+1)th iteration; It represents the active power limit of the μ-th region in the ω-th layer, which is the sum of the active power limits of the lower-level sub-regions or units divided by this region; Let represent the consistency variable of the i-th agent at time t in the (k+1)th iteration; Let represent the consistency variable of the j-th thermal power generating unit at time t in the (k+1)th iteration; Let represent the consistency variable of the u-th clean energy unit at time t in the (k+1)th iteration; This represents the number of thermal power units in the μ-th region of the ω-th layer; U represents the number of clean energy generator sets in the μ-th region of the ω-th layer; U represents the total number of regions in this layer; W represents the total number of layers. Step (10): Correct the first layer of active power output obtained in step (9) according to formula (18); The correction formula for active power is: in, This represents the lower limit of the active power output of the i-th power generation agent; This represents the upper limit of the active power output of the i-th power generation agent; Step (11): Calculate the active power deviation value of the first layer according to formula (19); Active power deviation ΔP t (k+1) is: Step (12): Determine whether the multimodal active power deviation value of the first layer obtained in step (11) meets the accuracy requirements. Where δ represents the maximum allowable power deviation; Indicates the allowable value for multimodal power deviation; if Then let the iteration step number k = k + 1, and return to step (8) to continue the iterative calculation of the first layer; if Store the active power output of the obtained set of agents, then iterate for another k = k + 1 steps, and return to step (8) to continue the iterative calculation of the first layer; if If the multimodal active power deviation value of the first layer meets the requirements, the iteration ends, and the active power output of multiple agents is obtained. Step (13): Perform multimodal judgment and decision congestion distance screening on the active power output of the multiple sets of intelligent agent generator sets obtained in step (12) to obtain multimodal active power output that meets the requirements of the decision-maker; Multimodal determination is as follows: |f(x1)-f(x2)|<ε (21) Here, x1 and x2 are the two solutions for the multimodal problem; ε represents the critical threshold for the multimodal problem. Crowded distance in the decision space is used as the filter: in, x represents the crowding distance of the j-th decision variable in the i-th solution set; i+1,j x represents the j-th decision variable in the (i+1)-th solution set; i,j Let j represent the j-th decision variable in the i-th solution set; The j-th decision variable represents the maximum objective value; The j-th decision variable represents the minimum objective value; x i-1,j Let j be the decision variable of the (i-1)th solution set; Let ξ represent the crowding distance of the i-th solution group; D represents the number of decision variables; ξ represents the minimum crowding distance allowed by multimodal decision-making. Step (14): Input the multiple sets of multi-mode power generation of the first layer obtained in step (13) as load values ​​into the second layer, and set the initial iteration step number k = 0; Step (15): Update the consistency variables of the second-layer agent by combining the topology diagram of the agent with formulas (15) and (16); Step (16): Calculate the active power output of each intelligent agent unit in the second layer according to formula (17); Step (17): Correct the active power output obtained in step (16) according to formula (18); Step (18): Calculate the active power deviation value of the second layer according to formula (19); Step (19): Determine whether the multimodal active power deviation value obtained in step (18) meets the accuracy requirements; if Then let the iteration step number k = k + 1, and return to step (15) to continue the iteration calculation of the second layer; if Store the active power output of the obtained set of agents, then iterate for another k = k + 1 steps, and return to step (15) to continue the iterative calculation of the second layer; if If the multimodal active power deviation value of the second layer meets the requirements, the iteration ends, and the active power output of multiple intelligent agent generator sets is obtained. Step (20): Perform multimodal judgment and decision congestion distance screening on the active power output of the multiple groups of agents in the second layer obtained in step (19) to obtain multimodal active power output that meets the requirements of the decision-maker; Step (21): Input the multiple sets of multi-mode power generation of the second layer obtained in step (20) as load values ​​into the third layer, and set the initial iteration step number k = 0; Step (22): Update the consistency variables of the third-layer agent by combining the topology diagram of the agent with formulas (15) and (16); Step (23): Calculate the active power output of each intelligent agent unit in the third layer according to formula (17); Step (24): Correct the active power output obtained in step (23) according to formula (18); Step (25): Calculate the active power deviation value of the third layer according to formula (19); Step (26): Determine whether the multimodal active power deviation value obtained in step (25) meets the accuracy requirements; if Then let the iteration step number k = k + 1, and return to step (22) to continue the iteration calculation of the third layer; if Store the active power output of the obtained set of agents, then iterate for another k = k + 1 steps, and return to step (22) to continue the iterative calculation of the third layer; if If the multimodal active power deviation value of the third layer meets the requirements, the iteration ends, and the active power output of multiple intelligent agent generator sets is obtained. Step (27): Perform multimodal judgment and decision congestion distance screening on the active power output of the multiple groups of agents in the third layer obtained in step (26) to obtain multimodal active power output that meets the requirements of the decision-maker; Step (28): Input the multiple sets of multi-mode power generation of the third layer obtained in step (27) as load values ​​into the fourth layer, and set the initial iteration step number k = 0; Step (29): Update the consistency variables of each generator set in the fourth layer by combining the topology diagram of the agent and formulas (15) and (16); Step (30): Calculate the active power output of each generator set in the fourth layer according to formula (17); Step (31): Correct the active power output obtained in step (30) according to formula (18); Step (32): Calculate the active power deviation value of the fourth layer according to formula (19); Step (33): Determine whether the multimodal active power deviation value obtained in step (32) meets the accuracy requirements; if Then let the iteration step number k = k + 1, and return to step (29) to continue the iteration calculation of the fourth layer; if Store the active power output of the obtained generator set, then iterate for another k = k + 1 step, and return to step (29) to continue the iterative calculation of the fourth layer; if If the multimodal active power deviation value of the fourth layer meets the requirements, the iteration ends and the active power output of multiple generator sets is obtained. Step (34): Perform multimodal judgment and decision congestion distance screening on the active power output of the multiple generator sets in the fourth layer obtained in step (33) to obtain a multimodal active power output decision scheme that meets the requirements of the decision-maker; Step (35): Determine whether t < T is satisfied. If it is satisfied, set t = t + 1 and transfer to step (5); if not, proceed to the next step; Step (36): Determine whether η < 1 is satisfied. If it is satisfied, store the multi-modal solution obtained when the value is η, set η = η + R, and transfer to step (4); if not, proceed to the next step; Step (37): Back-calculate all the multi-modal solutions obtained with different η values to obtain multi-modal multi-objective values. The multi-modal multi-objective values obtained at all η values form a Pareto front in the objective space.

Images