Intelligent building group power cooperative operation method based on EPEC framework

Abstract

The invention relates to a smart building group power cooperative operation method based on an EPEC framework. The method comprises the following steps: acquiring equipment data of each building in a smart building group; constructing a boundary updating model, and obtaining boundary data of each building; constructing a double-layer optimization model of a cluster operator, and converting the double-layer optimization model into an MPEC model based on a KKT condition; performing linearization processing on a nonlinear equation set in the MPEC model based on a large M method and a strong dual theorem; and an EPEC solving framework is constructed, and solving of the linearized MPEC model is realized through double-layer iteration. According to the solving framework provided by the invention, the building group cooperative operation solving can be realized under the condition that only boundary information communication is carried out among buildings, the solving framework does not need to transmit actual data information to the cluster by the buildings, and the leakage risk of sensitive data in the buildings can be effectively reduced; according to the method, the solving process is simplified, the solving difficulty is reduced, and a commercial linear solver such as Cplex can be used for solving.

Description

Technical Field

[0001] This invention relates to the field of energy dispatching technology, and in particular to a method for coordinated power operation of smart building complexes based on the EPEC framework. Background Technology

[0002] The utilization of new energy sources such as wind and solar power is gradually attracting attention. Among various new energy carriers, smart buildings are considered to have good potential for absorbing urban distributed new energy sources due to their flexible operation and diverse load types.

[0003] When smart buildings operate independently, peak energy production and consumption periods typically exhibit significant temporal differences, which can negatively impact their economic viability. With ongoing urbanization, multiple smart buildings of different types may exist within the same power distribution area, exhibiting substantial differences in their power source and load characteristics. This naturally creates a need and capability for interconnection and power allocation. Therefore, building complexes comprised of multiple smart buildings can significantly reduce operational costs and enhance power supply reliability through power sharing.

[0004] Currently, there is limited research on collaborative operation methods for building complexes, and existing collaborative operation methods all require the transmission of actual building operation data during communication, which may lead to the leakage of sensitive load data in the building and affect the building's privacy and security. Summary of the Invention

[0005] To address the sensitive data leakage problem in existing building cluster collaborative operation methods, the present invention aims to provide a smart building cluster power collaborative operation method based on the EPEC framework that effectively reduces the risk of sensitive data leakage in buildings, flexibly schedules each smart building in the building cluster, optimizes the power trading volume of each building, fully explores the interaction potential between buildings, and achieves optimal economic efficiency of the building cluster.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for power collaborative operation of smart building clusters based on the EPEC framework, the method comprising the following sequential steps:

[0007] (1) Obtain equipment data of each building in the smart building complex, including data on new energy sources, loads and electric vehicles;

[0008] (2) Based on the equipment data obtained in step (1), construct a boundary update model to obtain the boundary data of each building, wherein the boundary data is the maximum tradable power of the building at each time.

[0009] (3) Based on the boundary data obtained in step (2), construct a two-layer optimization model for the cluster operator. The two-layer optimization model consists of an upper-layer model and a lower-layer model. Based on the KKT conditions, the two-layer optimization model is transformed into an MPEC model.

[0010] (4) Linearize the nonlinear equations in the MPEC model based on the Big M method and the strong duality theorem to obtain the linearized MPEC model.

[0011] (5) Construct the EPEC solution framework, solve the linearized MPEC model through two-layer iteration, and obtain the final power optimization operation results of each building.

[0012] In step (2), the objective function of the boundary update model is to minimize the operating cost of the smart building, which includes the operating and maintenance costs of building equipment. Flexible load dispatch compensation costs Electric vehicle compensation Cost of power interaction between building and power distribution network Costs arising from the power interaction between buildings and other buildings As shown in equations (1) and (2):

[0013]

[0014]

[0015] In the formula: f b λ represents the total operating cost of smart building b; T represents the total operating cycle of the system; ot,es , λ trans , λ cut and λ ev These represent the operating costs of the energy storage device, the cost of offsetting loads, the cost of offsetting loads, and the cost of offsetting electric vehicles, respectively. and These represent the charging and discharging power of the energy storage device in building b at time t; t tr+ and t tr- These represent the upper and lower limits of the allowable operating time for transferable loads, respectively. and Δt represents the power of building b before and after load transfer scheduling at time t; Δt represents the unit step size of the system. and Let represent the initial power and the power after scheduling that building b can reduce its load at time t, respectively. and Let represent the charging and discharging power of electric vehicle i in building b at time t when it is in V2B mode; This represents the charging power of an electric vehicle at time t in its normal operating mode. These represent the grid purchase price and the electricity sales price at time t, respectively. This represents the cluster transaction price at time t; These represent the power that building b purchases and sells to the power distribution network at time t, respectively. and Let represent the power that building b purchases from other buildings and the power that it sells to other buildings at time t, respectively. This indicates the number of electric vehicles in building b;

[0016] The constraints of the boundary update model include building power production constraints, energy storage device output power constraints, electric vehicle state of charge constraints, and flexible load constraints.

[0017] The building's power production and power consumption remain consistent at all times, and the building's power production constraint is as follows:

[0018]

[0019] In the formula: This represents the photovoltaic output power of building b at time t; This represents the wind power output of building b at time t; This represents the power of the fixed load on building b at time t;

[0020] The output power constraint of the energy storage device is:

[0021]

[0022] In the formula: σ represents the remaining electrical charge of the energy storage device in building b at time t; es Indicates the self-loss rate of the energy storage device; η es,c and η es,d These represent the charging and discharging efficiencies of the energy storage device, respectively; W es,max and W es,min These represent the upper and lower limits of the energy storage device's capacity, respectively; P es,max This represents the maximum charging and discharging power of the energy storage device; ⊥ indicates that the two equations on the left and right are positively complementary.

[0023] The state of charge constraint of the electric vehicle is:

[0024]

[0025] In the formula: This represents the state of charge of electric vehicle i when it arrives at building b. W represents the target SOC of electric vehicle i in building b during its last charge. ev,rated Indicates the rated capacity of the electric vehicle battery; Indicates the time when the electric vehicle arrived at the building; Indicates the time when the electric vehicle charging ends; η ev This indicates the charging and discharging efficiency of an electric vehicle.

[0026] The flexible load constraint is:

[0027]

[0028] In step (3), the upper-level model is a profit maximization model for smart building b when it performs power sharing in the building group. The bidding price of smart building b and the amount of power sharing between smart building b and other buildings are used as decision variables. The objective function of the upper-level model is shown in equation (1):

[0029]

[0030] In the formula, and Let represent the power that building b purchases from other buildings and the power that it sells to other buildings at time t, respectively. This represents the bid price of building b at time t; Δt represents the cluster transaction price at time t, Δt represents the system's unit step size, and T represents the system's total operating cycle.

[0031] The upper-level model's constraint is a bid price constraint: the bid price for a smart building must be less than the price at which the building purchases electricity from the distribution network, and greater than the price at which it sells electricity to the distribution network.

[0032]

[0033] In the formula: These represent the grid purchase price and the electricity sales price at time t, respectively. This represents the bid price of building b at time t;

[0034] The lower-level model is a cluster transaction price maximization model for smart building clusters, with the cluster transaction price of smart building clusters as the decision variable. The objective function of the lower-level model is shown in equation (11):

[0035]

[0036] Where: N sb This indicates the number of buildings in a smart building complex;

[0037] The constraints of the lower-level model include power balance constraints for the building complex and boundary information constraints:

[0038] At any given time, the mutual power supply among the buildings in the smart building complex remains in balance. The power balance constraint of the building complex is the dual variable of the power balance constraint:

[0039]

[0040] The power purchase and sales capacity of each smart building are less than the maximum tradable capacity of the building in the boundary information, whereby the boundary information constraint is:

[0041]

[0042] In the formula: and These are the dual variables corresponding to constraints (13) and (14), respectively. This represents the dual variable corresponding to the right half of the constraint. This represents the dual variable corresponding to the left half of the constraint expression; and Let represent the maximum purchasable power and the maximum salable power of building b at time t, respectively, which are the boundary data obtained in step (2).

[0043] In step (3), the transformation of the bi-level optimization model into an MPEC model based on KKT conditions specifically refers to: transforming the lower-level model in the bi-level optimization model into its equivalent KKT conditions, thereby obtaining the equivalent equation set of the lower-level model, namely equations (15), (16), (17), and (18):

[0044]

[0045] Equations (15), (16), (17), and (18) are added as supplementary constraints to the lower-level model, and the upper-level model and the lower-level model with the added supplementary constraints form the MPEC model.

[0046] The objective function of the MPEC model is the same as the objective function of the upper-level model.

[0047] The constraints of the MPEC model include bid price constraints, building group power balance constraints, boundary information constraints, and equations (15), (16), (17), and (18).

[0048] Step (4) specifically refers to: using the Big M method to handle the linearization problem of complementary relaxation conditions in the MPEC model, with the linearization formula shown in equation (11):

[0049]

[0050] In the formula: θ v1 and θ v2 Let u represent two continuous variables in the complementary relaxation condition; v This is a newly introduced binary variable used to represent the state of a continuous variable; m is a very large positive number, and we take m = 10.5 ;

[0051] In the objective function of the MPEC model, the multiplication of two continuous variables forms a nonlinear term. The objective function of the MPEC model can be equivalently replaced by the strong duality theorem and some constraint terms in the KKT conditions:

[0052]

[0053] In the formula, T represents the total operating cycle of the system; N sb This indicates the number of buildings in a smart building complex; This represents the bid price of building b at time t; and Let represent the power that building b purchases from other buildings and the power that it sells to other buildings at time t, respectively. and Let represent the maximum purchasable power and the maximum salable power of building b at time t, respectively, which are the boundary data obtained in step (2);

[0054] Equation (20) and the constraints of the MPEC model together form the linearized MPEC model.

[0055] Step (5) specifically includes the following sequential steps:

[0056] (5a) Initialize iteration information: Initialize the outer loop iteration count K, let K = 1, and set the outer loop convergence condition κ. outer =10 -6 Inner loop convergence condition κ inner =10 -3 ;

[0057] (5b) Update the boundary information of building b: Solve the boundary update model for all buildings. The building energy management system shares the initial bid price information and the updated building's buying and selling roles and maximum tradable power at each time with the cluster operator. When the smart building runs the boundary update model, it needs to preset a set of initial cluster transaction prices as input data for the first run of the boundary update model.

[0058] (5c) Determine the building solution order: Sort all smart buildings, prioritize the analysis of the building with the lowest bid price, and after completing the building sorting, initialize the inner loop iteration count M by setting M=1;

[0059] (5d) Solve the linearized MPEC model: The cluster operator solves the linearized MPEC model of each building in the established building solution order. Whenever the linearized MPEC model of a building is solved, the bid price of that building in the optimization result is used as the input of the linearized MPEC model of the next building, and then the linearized MPEC model of each building is solved in turn.

[0060] (5e) Determine if the inner loop iteration has converged:

[0061] If the inner loop iteration count M = 1, then let M = M + 1 and execute step (5d);

[0062] If the inner loop iteration count M > 1, then check the convergence criterion of the inner loop iteration. If the convergence criterion of the inner loop iteration is met, then execute step (5f). If the convergence criterion of the inner loop iteration is not met, then let M = M + 1 and execute step (5d). The convergence criterion of the inner loop iteration is shown in equation (21):

[0063]

[0064] In the formula: This represents the bid price of building b at time t during the Mth inner loop iteration; ||x|| ∞ The norm is used to measure the maximum absolute value in a vector.

[0065] (5f) Determine if the outer loop iteration has converged:

[0066] If the outer loop iteration count K = 1, then let K = K + 1 and execute step (5d);

[0067] If the inner loop iteration count K > 1, then check the convergence criterion of the outer loop iteration. If the convergence criterion of the outer loop iteration is met, then execute step (5g). If the convergence criterion of the outer loop iteration is not met, then let K = K + 1 and execute step (5b). The convergence criterion of the outer loop iteration is shown in equations (22) and (23):

[0068]

[0069] In the formula: and Let represent the maximum purchasable power and the maximum salable power of building b at time t during the Kth outer loop iteration, respectively;

[0070] (5g) Output optimization results: The cluster operator sends the optimized cluster transaction price and the power mutual assistance value between buildings to the building energy management system to obtain the final power optimization operation results of each building.

[0071] As can be seen from the above technical solution, the beneficial effects of the present invention are as follows: First, by flexibly scheduling each smart building in the building cluster, the present invention optimizes the power trading volume of each building, fully explores the interaction potential between buildings, reduces the operating cost of the building cluster while ensuring the balance of interests of each building, and optimizes the economy of the building cluster; Second, the solution framework proposed in the present invention can realize the solution of collaborative operation of the building cluster when only boundary information communication is required between buildings. This solution framework does not require buildings to transmit actual data information to the cluster, which can effectively reduce the risk of leakage of sensitive data in buildings; Third, the research object of the present invention is a building cluster, which has multiple independent buildings and is a multi-agent optimization problem. Moreover, this multi-agent problem is nonlinear. The present invention simplifies the solution process by linearizing the independent agent problem and solving multiple independent agent problems simultaneously and iteratively, thereby reducing the solution difficulty and enabling it to be solved using commercial linear solvers such as Cplex. Attached Figure Description

[0072] Figure 1 This is a flowchart of the method of the present invention;

[0073] Figure 2 This is a schematic diagram of the internal framework of a building according to Embodiment 1 of the present invention;

[0074] Figure 3 This is a schematic diagram of the initial output of various building equipment in Embodiment 1 of the present invention;

[0075] Figure 4 This is the power curve of the building complex without considering power interaction in Embodiment 1 of the present invention;

[0076] Figure 5 This is the power curve of a building complex considering power interaction in Embodiment 1 of the present invention. Detailed Implementation

[0077] like Figure 1 As shown, a method for coordinated power operation of smart building clusters based on the EPEC framework is presented, which includes the following sequential steps:

[0078] (1) Obtain equipment data of each building in the smart building complex, including data on new energy sources, loads and electric vehicles;

[0079] (2) Based on the equipment data obtained in step (1), construct a boundary update model to obtain the boundary data of each building, wherein the boundary data is the maximum tradable power of the building at each time.

[0080] (3) Based on the boundary data obtained in step (2), construct a two-layer optimization model for the cluster operator. The two-layer optimization model consists of an upper-layer model and a lower-layer model. Based on the KKT conditions, the two-layer optimization model is transformed into an MPEC model.

[0081] (4) Linearize the nonlinear equations in the MPEC model based on the Big M method and the strong duality theorem to obtain the linearized MPEC model.

[0082] (5) Construct the EPEC solution framework, solve the linearized MPEC model through two-layer iteration, and obtain the final power optimization operation results of each building.

[0083] In step (2), the objective function of the boundary update model is to minimize the operating cost of the smart building, which includes the operating and maintenance costs of building equipment. Flexible load dispatch compensation costs Electric vehicle compensation Cost of power interaction between building and power distribution network Costs arising from the power interaction between buildings and other buildings As shown in equations (1) and (2):

[0084]

[0085] In the formula: f b λ represents the total operating cost of smart building b; T represents the total operating cycle of the system; ot,es , λ trans , λ cut and λ ev These represent the operating costs of the energy storage device, the cost of offsetting loads, the cost of offsetting loads, and the cost of offsetting electric vehicles, respectively. and These represent the charging and discharging power of the energy storage device in building b at time t; t tr+ and t tr- These represent the upper and lower limits of the allowable operating time for transferable loads, respectively. and Δt represents the power of building b before and after load transfer scheduling at time t; Δt represents the unit step size of the system. and Let represent the initial power and the power after scheduling that building b can reduce its load at time t, respectively. and Let represent the charging and discharging power of electric vehicle i in building b at time t when it is in V2B mode; This represents the charging power of an electric vehicle at time t in its normal operating mode. These represent the grid purchase price and the electricity sales price at time t, respectively. This represents the cluster transaction price at time t; These represent the power that building b purchases and sells to the power distribution network at time t, respectively. and Let represent the power that building b purchases from other buildings and the power that it sells to other buildings at time t, respectively. This indicates the number of electric vehicles in building b;

[0086] The constraints of the boundary update model include building power production constraints, energy storage device output power constraints, electric vehicle state of charge constraints, and flexible load constraints.

[0087] The building's power production and power consumption remain consistent at all times, and the building's power production constraint is as follows:

[0088]

[0089] In the formula: This represents the photovoltaic output power of building b at time t; This represents the wind power output of building b at time t; This represents the power of the fixed load on building b at time t;

[0090] The output power constraint of the energy storage device is:

[0091]

[0092] In the formula: σ represents the remaining electrical charge of the energy storage device in building b at time t; es Indicates the self-loss rate of the energy storage device; η es,c and η es,d These represent the charging and discharging efficiencies of the energy storage device, respectively; W es,max and W es,min These represent the upper and lower limits of the energy storage device's capacity, respectively; P es,max This represents the maximum charging and discharging power of the energy storage device; ⊥ indicates that the two equations on the left and right are positively complementary.

[0093] The state of charge constraint of the electric vehicle is:

[0094]

[0095] In the formula: This represents the state of charge of electric vehicle i when it arrives at building b. W represents the target SOC of electric vehicle i in building b during its last charge. ev,rated Indicates the rated capacity of the electric vehicle battery; Indicates the time when the electric vehicle arrived at the building; Indicates the time when the electric vehicle charging ends; ηev This indicates the charging and discharging efficiency of an electric vehicle.

[0096] The flexible load constraint is:

[0097]

[0098] In step (3), the upper-level model is a profit maximization model for smart building b when it performs power sharing in the building group. The bidding price of smart building b and the amount of power sharing between smart building b and other buildings are used as decision variables. The objective function of the upper-level model is shown in equation (1):

[0099]

[0100] In the formula, and Let represent the power that building b purchases from other buildings and the power that it sells to other buildings at time t, respectively. This represents the bid price of building b at time t; Δt represents the cluster transaction price at time t, Δt represents the system's unit step size, and T represents the system's total operating cycle.

[0101] The upper-level model's constraint is a bid price constraint: the bid price for a smart building must be less than the price at which the building purchases electricity from the distribution network, and greater than the price at which it sells electricity to the distribution network.

[0102]

[0103] In the formula: These represent the grid purchase price and the electricity sales price at time t, respectively. This represents the bid price of building b at time t;

[0104] The lower-level model is a cluster transaction price maximization model for smart building clusters, with the cluster transaction price of smart building clusters as the decision variable. The objective function of the lower-level model is shown in equation (11):

[0105]

[0106] Where: N sb This indicates the number of buildings in a smart building complex;

[0107] The constraints of the lower-level model include power balance constraints for the building complex and boundary information constraints:

[0108] At any given time, the mutual power supply among the buildings in the smart building complex remains in balance. The power balance constraint of the building complex is the dual variable of the power balance constraint:

[0109]

[0110] The power purchase and sales capacity of each smart building are less than the maximum tradable capacity of the building in the boundary information, whereby the boundary information constraint is:

[0111]

[0112]

[0113] In the formula: and These are the dual variables corresponding to constraints (13) and (14), respectively. The dual variables represent the opposite variables corresponding to the right half of the constraint, whereby the right half of the constraint refers to... The left half of the constraint represents the dual variable corresponding to the left half of the constraint, which respectively refers to... and Let represent the maximum purchasable power and the maximum salable power of building b at time t, respectively, which are the boundary data obtained in step (2).

[0114] In step (3), the transformation of the bi-level optimization model into an MPEC model based on KKT conditions specifically refers to: transforming the lower-level model in the bi-level optimization model into its equivalent KKT conditions, thereby obtaining the equivalent equation set of the lower-level model, namely equations (15), (16), (17), and (18):

[0115]

[0116] Equations (15), (16), (17), and (18) are added as supplementary constraints to the lower-level model, and the upper-level model and the lower-level model with the added supplementary constraints form the MPEC model.

[0117] The objective function of the MPEC model is the same as the objective function of the upper-level model.

[0118] The constraints of the MPEC model include bid price constraints, building group power balance constraints, boundary information constraints, and equations (15), (16), (17), and (18).

[0119] Step (4) specifically refers to: using the Big M method to handle the linearization problem of complementary relaxation conditions in the MPEC model, with the linearization formula shown in equation (11):

[0120]

[0121] In the formula: θ v1 and θ v2 Let u represent two continuous variables in the complementary relaxation condition; vThis is a newly introduced binary variable used to represent the state of a continuous variable; m is a very large positive number, and we take m = 10. 5 ;

[0122] In the objective function of the MPEC model, the multiplication of two continuous variables forms a nonlinear term. The objective function of the MPEC model can be equivalently replaced by the strong duality theorem and some constraint terms in the KKT conditions:

[0123]

[0124] In the formula, T represents the total operating cycle of the system; N sb This indicates the number of buildings in a smart building complex; This represents the bid price of building b at time t; and Let represent the power that building b purchases from other buildings and the power that it sells to other buildings at time t, respectively. and Let represent the maximum purchasable power and the maximum salable power of building b at time t, respectively, which are the boundary data obtained in step (2);

[0125] Equation (20) and the constraints of the MPEC model together form the linearized MPEC model.

[0126] Step (5) specifically includes the following sequential steps:

[0127] (5a) Initialize iteration information: Initialize the outer loop iteration count K, let K = 1, and set the outer loop convergence condition κ. outer =10 -6 Inner loop convergence condition κ inner =10 -3 ;

[0128] (5b) Update the boundary information of building b: Solve the boundary update model for all buildings. The building energy management system shares the initial bid price information and the updated building's buying and selling roles and maximum tradable power at each time with the cluster operator. When the smart building runs the boundary update model, it needs to preset a set of initial cluster transaction prices as input data for the first run of the boundary update model.

[0129] (5c) Determine the building solution order: Sort all smart buildings, prioritize the analysis of the building with the lowest bid price, and after completing the building sorting, initialize the inner loop iteration count M by setting M=1;

[0130] (5d) Solve the linearized MPEC model: The cluster operator solves the linearized MPEC model of each building in the established building solution order. Whenever the linearized MPEC model of a building is solved, the bid price of that building in the optimization result is used as the input of the linearized MPEC model of the next building, and then the linearized MPEC model of each building is solved in turn.

[0131] (5e) Determine if the inner loop iteration has converged:

[0132] If the inner loop iteration count M = 1, then let M = M + 1 and execute step (5d);

[0133] If the inner loop iteration count M > 1, then check the convergence criterion of the inner loop iteration. If the convergence criterion of the inner loop iteration is met, then execute step (5f). If the convergence criterion of the inner loop iteration is not met, then let M = M + 1 and execute step (5d). The convergence criterion of the inner loop iteration is shown in equation (21):

[0134]

[0135] In the formula: This represents the bid price of building b at time t during the Mth inner loop iteration; ||x|| ∞ The norm is used to measure the maximum absolute value in a vector.

[0136] (5f) Determine if the outer loop iteration has converged:

[0137] If the outer loop iteration count K = 1, then let K = K + 1 and execute step (5d);

[0138] If the inner loop iteration count K > 1, then check the convergence criterion of the outer loop iteration. If the convergence criterion of the outer loop iteration is met, then execute step (5g). If the convergence criterion of the outer loop iteration is not met, then let K = K + 1 and execute step (5b). The convergence criterion of the outer loop iteration is shown in equations (22) and (23):

[0139]

[0140]

[0141] In the formula: and Let represent the maximum purchasable power and the maximum salable power of building b at time t during the Kth outer loop iteration, respectively;

[0142] (5g) Output optimization results: The cluster operator sends the optimized cluster transaction price and the power mutual assistance value between buildings to the building energy management system to obtain the final power optimization operation results of each building.

[0143] Example 1

[0144] This embodiment takes a building complex containing four smart buildings as the research object, and the internal structure of the buildings is as follows: Figure 2 As shown in the figure. Buildings 1 and 2 are residential buildings, building 3 is a commercial building, and building 4 is an industrial building. The time step Δt = 30 minutes, and the total optimized operation time is 24 hours, with a total of T = 48 time periods per day. The initial output of each piece of equipment in the buildings is shown in the figure. Figure 3 As shown in Table 1, the flexible load parameters are as shown in Table 2, and the electricity price adopts the time-of-use pricing. It is assumed that all electric vehicles are of the same model and use the relevant configuration of BYD Qin PLUS Champion Edition, with a battery capacity of 42kWh, a power consumption of 0.116kWh / km, a charge / discharge efficiency of 0.9, and a battery SOC upper and lower limits of [0.3, 0.9].

[0145] Table 1 Flexible Load Parameters

[0146]

[0147] Table 2 Time-of-use Electricity Prices

[0148]

[0149] The implementation plan is divided into two scenarios: Scenario 1, which does not consider power interaction between buildings, and Scenario 2, which considers power difference interaction between buildings. The result diagram for Scenario 1 is shown below. Figure 4 As shown, the result of scenario 2 is as follows. Figure 5 As shown.

[0150] Both Building 1 and Building 2 are residential buildings, constructed by Figure 4 As shown in (a) and (b), the power optimization results are quite similar. Most of the flexible load is transferred to the peak output period of new energy sources, and electric vehicles and energy storage devices discharge during peak electricity price periods and charge during off-peak electricity price periods. The peak load for both buildings is between 17:00 and 24:00, during which time the buildings have a large demand for electricity.

[0151] Building 3 is a commercial building, by Figure 4 As shown in (c), during the shopping mall's operating hours, the building is always under peak load. From 10:00 to 14:00, the renewable energy output is relatively high, sufficient to meet the building's load demand. From 14:00 to 20:00, the renewable energy output gradually decreases, and the building's electricity purchase demand is high. During the shopping mall's closing hours, the building load remains lower than the renewable energy output, and the building's electricity sales demand is high at this time.

[0152] Building 4 is an industrial building, constructed by... Figure 4As shown in (d), during the period from 0:00 to 8:00, buildings are in peak electricity consumption and have a large demand for electricity. This is because industrial buildings usually adopt a staggered production policy, and the electricity price is lower during this period. During the period from 8:00 to 24:00, industrial buildings cease production activities, and the load remains lower than the output of renewable energy sources. During this period, the demand for electricity sales from buildings is large.

[0153] The economic costs of each building are shown in Table 3. Without considering building electrical energy interaction, the total operating cost of the smart building complex is 5113.6 yuan.

[0154] Table 3 Optimization Results for Scenario 1

[0155]

[0156] The power optimization results for each building in Scenario 2 are as follows: Figure 5 As shown. Comparison Figure 4 (a) and Figure 5 As shown in (a), after considering the energy exchange between buildings, buildings reduced their dispatching efforts for electric vehicles, selling some of the power used to charge electric vehicles between 0:00 and 8:00 to other buildings, and purchasing a small amount of power during the nighttime peak load period. (Comparison) Figure 5 As shown in (a) and (b), since both Building 1 and Building 2 are residential buildings, their operational results are similar. At the cluster transaction level, Building 2 sells less power and buys more power than Building 1. This is because Building 2's initial bid price is higher, causing other buildings to prefer selling power to Building 2 rather than buying power from it. At the electric vehicle level, Building 2 does not reduce its dispatching efforts for electric vehicles. This is because after deducting the electricity sold to other buildings, the building still has redundant electricity to charge dispatched electric vehicles, eliminating the need for additional electricity purchases.

[0157] contrast Figure 5 As shown in (c) and (d), during the period from 0:00 to 8:00, Building 4 has a higher demand for electricity. Regardless of whether Building 4's bid price meets expectations, other buildings will choose to sell power to Building 4. However, since Building 1's initial bid price is lower, Building 4 is more inclined to buy electricity from Building 1. Therefore, Building 3 will only be able to purchase electricity when Building 1's sales volume reaches its limit.

[0158] Table 4 shows the economic costs and carbon emissions of each building in Scenario 2. After considering building-to-electricity interaction, the total operating cost of the smart building complex decreased from 5113.6 yuan to 4439.3 yuan. This demonstrates the effectiveness of the proposed solution framework, which can achieve optimized operation of building complexes in scenarios where only boundary information communication is allowed.

[0159] Table 4 Optimization Results for Scenario 2

[0160]

[0161] In summary, this invention optimizes the power trading volume of each smart building in a building cluster by flexibly scheduling them, fully tapping the interaction potential between buildings, and reducing the operating cost of the building cluster while ensuring a balance of interests among the buildings, thus achieving optimal economic efficiency. The proposed solution framework enables collaborative operation of the building cluster with only boundary information communication between buildings. This framework eliminates the need for buildings to transmit actual data to the cluster, effectively reducing the risk of leakage of sensitive data. The research object of this invention is a building cluster with multiple independent buildings, representing a multi-agent optimization problem. This multi-agent problem is nonlinear. This invention simplifies the solution process by linearizing the independent agent problems and solving multiple independent agent problems simultaneously and iteratively, reducing the difficulty of the solution. It can be solved using commercial linear solvers such as Cplex.

Claims

1. A method for coordinated power operation of smart building clusters based on the EPEC framework, characterized in that: The method includes the following steps in sequence: (1) Obtain equipment data of each building in the smart building complex, including data on new energy sources, loads and electric vehicles; (2) Based on the equipment data obtained in step (1), construct a boundary update model to obtain the boundary data of each building, wherein the boundary data is the maximum tradable power of the building at each time. (3) Based on the boundary data obtained in step (2), construct a two-layer optimization model for the cluster operator. The two-layer optimization model consists of an upper-layer model and a lower-layer model. Based on the KKT conditions, the two-layer optimization model is transformed into an MPEC model. (4) Linearize the nonlinear equations in the MPEC model based on the Big M method and the strong duality theorem to obtain the linearized MPEC model. (5) Construct the EPEC solution framework, solve the linearized MPEC model through two-layer iteration, and obtain the final power optimization operation results of each building.

2. The power collaborative operation method for smart building clusters based on the EPEC framework according to claim 1, characterized in that: In step (2), the objective function of the boundary update model is to minimize the operating cost of the smart building, which includes the operating and maintenance costs of building equipment. Flexible load dispatch compensation costs Electric vehicle compensation Cost of power interaction between building and power distribution network Costs arising from the power interaction between buildings and other buildings As shown in equations (1) and (2): In the formula: f b λ represents the total operating cost of smart building b; T represents the total operating cycle of the system; ot,es , λ trans , λ cut and λ ev These represent the operating costs of the energy storage device, the cost of offsetting loads, the cost of offsetting loads, and the cost of offsetting electric vehicles, respectively. and These represent the charging and discharging power of the energy storage device in building b at time t; t tr+ and t tr- These represent the upper and lower limits of the allowable operating time for transferable loads, respectively. and These represent the power of building b before and after load transfer scheduling at time t; Δt represents the unit step size of the system; and Let represent the initial power and the power after scheduling that building b can reduce its load at time t, respectively. and Let represent the charging and discharging power of electric vehicle i in building b at time t when it is in V2B mode; This represents the charging power of an electric vehicle at time t in its normal operating mode. These represent the grid purchase and sale prices of electricity at time t, respectively. This represents the cluster transaction price at time t; These represent the power that building b purchases and sells to the power distribution network at time t, respectively. and These represent the power purchased by building b from other buildings and the power supplied to other buildings at time t, respectively. The power output of electricity sold; This indicates the number of electric vehicles in building b; The constraints of the boundary update model include building power production constraints, energy storage device output power constraints, electric vehicle state of charge constraints, and flexible load constraints. The building's power production and power consumption remain consistent at all times, and the building's power production constraint is as follows: In the formula: This represents the photovoltaic output power of building b at time t; This represents the wind power output of building b at time t; This represents the power of the fixed load on building b at time t; The output power constraint of the energy storage device is: In the formula: σ represents the remaining electrical charge of the energy storage device in building b at time t; es Indicates the self-loss rate of the energy storage device; η es,c and η es,d These represent the charging and discharging efficiencies of the energy storage device, respectively; W es,max and W es,min These represent the upper and lower limits of the energy storage device's capacity, respectively; P es,max ⊥ indicates the maximum charging and discharging power of the energy storage device; ⊥ indicates that the two equations on the left and right are positively complementary. The state of charge constraint of the electric vehicle is: In the formula: This represents the state of charge of electric vehicle i when it arrives at building b. W represents the target SOC of electric vehicle i in building b during its last charge. ev,rated Indicates the rated capacity of the electric vehicle battery; Indicates the time when the electric vehicle arrived at the building; Indicates the time when the electric vehicle charging ends; η ev This indicates the charging and discharging efficiency of an electric vehicle. The flexible load constraint is:

3. The method for coordinated power operation of smart building clusters based on the EPEC framework according to claim 1, characterized in that: In step (3), the upper-level model is a profit maximization model for smart building b when it performs power sharing in the building group. The bidding price of smart building b and the amount of power sharing between smart building b and other buildings are used as decision variables. The objective function of the upper-level model is shown in equation (1): In the formula, and Let represent the power that building b purchases from other buildings and the power that it sells to other buildings at time t, respectively. This represents the bid price of building b at time t; Δt represents the cluster transaction price at time t, Δt represents the system's unit step size, and T represents the system's total operating cycle. The upper-level model's constraint is a bid price constraint: the bid price for a smart building must be less than the price at which the building purchases electricity from the distribution network, and greater than the price at which it sells electricity to the distribution network. In the formula: These represent the grid purchase price and the electricity sales price at time t, respectively. This represents the bid price of building b at time t; The lower-level model is a cluster transaction price maximization model for smart building clusters, with the cluster transaction price of smart building clusters as the decision variable. The objective function of the lower-level model is shown in equation (11): Where: N sb This indicates the number of buildings in a smart building complex; The constraints of the lower-level model include power balance constraints for the building complex and boundary information constraints: At any given time, the mutual power supply among the buildings in the smart building complex remains in balance. The power balance constraint of the building complex is the dual variable of the power balance constraint: The power purchase and sales capacity of each smart building are less than the maximum tradable capacity of the building in the boundary information, whereby the boundary information constraint is: In the formula: and These are the dual variables corresponding to constraints (13) and (14), respectively. This represents the dual variable corresponding to the right half of the constraint. This represents the dual variable corresponding to the left half of the constraint expression; and Let represent the maximum purchasable power and the maximum salable power of building b at time t, respectively, which are the boundary data obtained in step (2).

4. The method for coordinated power operation of smart building clusters based on the EPEC framework according to claim 1, characterized in that: In step (3), the transformation of the bi-level optimization model into an MPEC model based on KKT conditions specifically refers to: transforming the lower-level model in the bi-level optimization model into its equivalent KKT conditions, thereby obtaining the equivalent equation set of the lower-level model, namely equations (15), (16), (17), and (18): Equations (15), (16), (17), and (18) are added as supplementary constraints to the lower-level model, and the upper-level model and the lower-level model with the added supplementary constraints form the MPEC model. The objective function of the MPEC model is the same as the objective function of the upper-level model. The constraints of the MPEC model include bid price constraints, building group power balance constraints, boundary information constraints, and equations (15), (16), (17), and (18).

5. The method for coordinated power operation of smart building clusters based on the EPEC framework according to claim 1, characterized in that: Step (4) specifically refers to: using the Big M method to handle the linearization problem of complementary relaxation conditions in the MPEC model, with the linearization formula shown in equation (11): In the formula: and Let u represent two continuous variables in the complementary relaxation condition; v This is a newly introduced binary variable used to represent the state of a continuous variable; m is a very large positive number, and we take m = 10. 5 ; In the objective function of the MPEC model, the multiplication of two continuous variables forms a nonlinear term. The objective function of the MPEC model can be equivalently replaced by the strong duality theorem and some constraint terms in the KKT conditions: In the formula, T represents the total operating cycle of the system; N sb This indicates the number of buildings in a smart building complex; This represents the bid price of building b at time t; and Let represent the power that building b purchases from other buildings and the power that it sells to other buildings at time t, respectively. and Let represent the maximum purchasable power and the maximum salable power of building b at time t, respectively, which are the boundary data obtained in step (2); Equation (20) and the constraints of the MPEC model together form the linearized MPEC model.

6. The method for coordinated power operation of smart building clusters based on the EPEC framework according to claim 1, characterized in that: Step (5) specifically includes the following sequential steps: (5a) Initialize iteration information: Initialize the outer loop iteration count K, let K = 1, and set the outer loop convergence condition κ. outer =10 -6 Inner loop convergence condition κ inner =10 -3 ; (5b) Update the boundary information of building b: Solve the boundary update model for all buildings. The building energy management system shares the initial bid price information and the updated building's buying and selling roles and maximum tradable power at each time with the cluster operator. When the smart building runs the boundary update model, it needs to preset a set of initial cluster transaction prices as input data for the first run of the boundary update model. (5c) Determine the building solution order: Sort all smart buildings, prioritize the analysis of the building with the lowest bid price, and after completing the building sorting, initialize the inner loop iteration count M by setting M=1; (5d) Solve the linearized MPEC model: The cluster operator solves the linearized MPEC model of each building in the established building solution order. Whenever the linearized MPEC model of a building is solved, the bid price of that building in the optimization result is used as the input of the linearized MPEC model of the next building, and then the linearized MPEC model of each building is solved in turn. (5e) Determine if the inner loop iteration has converged: If the inner loop iteration count M = 1, then let M = M + 1 and execute step (5d); If the inner loop iteration count M > 1, then check the convergence criterion of the inner loop iteration. If the convergence criterion of the inner loop iteration is met, then execute step (5f). If the convergence criterion of the inner loop iteration is not met, then let M = M + 1 and execute step (5d). The convergence criterion of the inner loop iteration is shown in equation (21): In the formula: This represents the bid price of building b at time t during the Mth inner loop iteration; ||x|| ∞ The norm is used to measure the maximum absolute value in a vector. (5f) Determine if the outer loop iteration has converged: If the outer loop iteration count K = 1, then let K = K + 1 and execute step (5d); If the inner loop iteration count K > 1, then check the convergence criterion of the outer loop iteration. If the convergence criterion of the outer loop iteration is met, then execute step (5g). If the convergence criterion of the outer loop iteration is not met, then let K = K + 1 and execute step (5b). The convergence criterion of the outer loop iteration is shown in equations (22) and (23): In the formula: and Let represent the maximum purchasable power and the maximum salable power of building b at time t during the Kth outer loop iteration, respectively; (5g) Output optimization results: The cluster operator sends the optimized cluster transaction price and the power mutual assistance value between buildings to the building energy management system to obtain the final power optimization operation results of each building.

Images