Day-ahead optimal scheduling method for air conditioning load clustering considering multiple uncertainties

Abstract

The present application relates to the technical field of integrated air conditioning load power distribution network optimization, and discloses a kind of air conditioning load cluster day-ahead optimization scheduling method considering multiple uncertainties.According to air conditioning summer refrigeration operation scene, establish air conditioning load operation model, determine indoor temperature range constraint, construct air conditioning load operation feasible region, then use geometric form to represent air conditioning load scheduling feasible region, and reorganize as a group of compact form constraint representation convex polyhedron;According to convex polyhedron, obtain air conditioning load cluster operation feasible region, introduce distributed robust chance constraint to simulate air conditioning load parameter uncertainty, obtain aggregated air conditioning load cluster approximate operation feasible region;With the minimum system operation cost as objective function, build air conditioning load cluster day-ahead optimization scheduling model, and solve it.The present application uses distributed robust chance constraint to build integrated air conditioning load power distribution network day-ahead optimization scheduling model under multiple uncertainties, which guarantees economic operation, safety and reliability of power grid.

Description

Technical Field

[0001] This invention relates to the field of power distribution network optimization technology for integrated air conditioning loads, specifically to a day-ahead optimization scheduling method for air conditioning load clusters that considers multiple uncertainties. Background Technology

[0002] With the increase in urban building area and population density, the demand for air conditioning, as a significant component of building energy consumption, continues to rise. Clustered management can effectively regulate load, achieve a balance between power supply and demand, and improve grid stability. Furthermore, with the development of technologies such as the Internet of Things and big data, clustered air conditioning loads can achieve precise control and optimized interaction.

[0003] Under the premise of meeting the overall operational constraints of the air conditioning cluster (such as indoor temperature limits, power regulation capabilities, energy balance, etc.), the permissible set of power adjustment amounts of all individual air conditioners within the cluster is called the approximate operational feasible region of the air conditioning cluster. It is the basis for the air conditioning cluster to participate in demand response or optimized scheduling such as peak shaving and frequency regulation of the power system, and is used to quantify the maximum adjustable capacity of the cluster within a specific time period (such as maximum load reduction and load increase power), while ensuring the comfort of air conditioning users and the safety of equipment.

[0004] Currently, research on the approximate operational feasible region of air conditioning load clusters has three methods: box polyhedron, chino polyhedron, and isomorphic polyhedron. Existing research has demonstrated that the isomorphic polyhedron-based method has better aggregation effects than the box polyhedron and chino polyhedron methods. However, these studies still have certain limitations: the modeling of air conditioning loads is mostly deterministic aggregation modeling, with less consideration for the impact of dynamic changes in outdoor temperature on the aggregation effect; research on the participation of air conditioning load clusters in distribution network optimization scheduling is mostly based on deterministic assumptions, failing to address the uncertainties of renewable energy output such as wind and solar power. In fact, in energy and power systems, both wind and solar power generation (supply side) and outdoor temperature (demand side) are difficult to predict accurately, and their random fluctuations can overlap, creating more complex and extreme risks. Random fluctuations in photovoltaic and wind power can easily lead to significant deviations between scheduling plans and real-time operation. Failure to fully consider the multiple uncertainties of wind and solar power may impair the economic operation and reliability of the distribution network, triggering power balance problems. Summary of the Invention

[0005] To address the aforementioned problems, the present invention aims to provide a day-ahead optimal scheduling method for air conditioning load clusters that considers multiple uncertainties. This method defines the equivalent operating cost of air conditioning loads by decomposing the aggregated feasible region and considers the uncertainties in wind and solar power output. A day-ahead optimal scheduling model for integrated air conditioning loads under multiple uncertainties is constructed using distributed blue bar chance constraints, ensuring the economic operation and reliability of the power grid. The technical solution is as follows:

[0006] A day-ahead optimization scheduling method for air conditioning load clusters considering multiple uncertainties includes the following steps:

[0007] Step 1: For the summer cooling operation scenario of air conditioning, establish an air conditioning load operation model, determine the indoor temperature range constraint, construct the air conditioning load operation feasible region, and then express it in geometric form as the scheduling feasible region of air conditioning load, and reorganize it into a set of convex polyhedra characterized by compact form constraints;

[0008] Step 2: Obtain the feasible region of the air conditioning load cluster based on the convex polyhedron. Form the maximum internal approximation feasible region of a single air conditioning load by scaling and transferring the basic isomorphic polyhedron. Then optimize the scaling and transfer coefficients of the isomorphic polyhedron. Introduce the Bruker chance constraint to simulate the uncertainty of the air conditioning load parameters and obtain the approximate feasible region of the aggregated air conditioning load cluster.

[0009] Step 3: The air conditioning load aggregator submits the approximate operational feasible region of the aggregated air conditioning load cluster to the distribution network operator. The distribution network operator constructs a day-ahead optimization scheduling model for the air conditioning load cluster with the objective function of minimizing system operating costs, and then obtains the scheduling plan for individual air conditioners.

[0010] The beneficial effects of this invention are:

[0011] This invention focuses on the day-ahead optimal scheduling of air conditioning load clusters considering multiple uncertainties. Based on an isomorphic polyhedral structure, a sub-Bruker chance constraint is introduced to model the air conditioning load cluster, accurately characterizing the impact of ambient temperature changes on the aggregated feasible region. Based on the aggregation model decomposition algorithm, an equivalent operating cost function for the air conditioning load is defined, and the aggregated feasible region and equivalent operating cost function are embedded into the day-ahead optimal scheduling model of the distribution network. This verifies that with the increase of multiple uncertainties, the system's operating cost increases accordingly. The system operating cost is significantly lower when the uncertainty confidence level of wind power, photovoltaic, and air conditioning loads is 0.95 compared to 0.99, achieving a trade-off between system robustness and operational economy. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of air conditioning load cluster scheduling.

[0013] Figure 2 A schematic diagram illustrating the solution process for the day-ahead optimization scheduling model of air conditioning load clusters.

[0014] Figure 3 This is a branch flow power model diagram.

[0015] Figure 4 This is a schematic diagram of the IEEE 33-node distribution network.

[0016] Figure 5The graph shows the aggregated results under different prediction error percentages ζ and violation probability constraints ε.

[0017] Figure 6 This diagram illustrates the system operating costs under different conditions. Detailed Implementation

[0018] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0019] This invention focuses on the day-ahead optimal scheduling of air conditioning load clusters considering multiple uncertainties. First, to accurately characterize the impact of ambient temperature changes on the aggregate feasible region, a distributed robust chance constrained (DRCC) is introduced to model the air conditioning load cluster based on an isomorphic polyhedral structure. Based on the aggregation model decomposition algorithm, an equivalent operating cost function for the air conditioning load is defined, and the aggregate feasible region and equivalent operating cost function are embedded into the day-ahead optimal scheduling model of the distribution network. Simultaneously, considering the uncertainties in wind and solar power output, the impact of different confidence levels and uncertainty parameters on system operating costs is analyzed, verifying the effectiveness of the proposed strategy.

[0020] 1. Air conditioning load operation model:

[0021] For the summer cooling operation scenario of air conditioning, an air conditioning load operation model is established, the indoor temperature range constraint is determined, the air conditioning load operation feasible region is constructed, and then the scheduling feasible region of the air conditioning load is expressed in geometric form, and reorganized into a set of convex polyhedra characterized by compact form constraints.

[0022] For the summer cooling operation scenario of air conditioners, the air conditioner load operation model is as follows:

[0023] (1);

[0024] in, and These represent intermediate variables respectively; and These represent thermal resistance and heat capacity, respectively. and They represent Real-time indoor and outdoor ambient temperatures; and They are respectively Indoor heat load at any time and the first The cooling capacity provided by each air conditioning load.

[0025] Discretizing formula (1) yields a linear expression, as shown below:

[0026] (2);

[0027] In the formula, , , Both represent intermediate variables; Let be the output power of the b-th air conditioning load at time t. The cooling energy efficiency ratio of an air conditioner; Indicates a time interval.

[0028] (3);

[0029] In the formula, For the user's set temperature, The acceptable temperature adjustment value for the user; after combining the above formulas (2) and (3), the feasible region for air conditioning load operation is the scheduling feasible region of the b-th air conditioning load. It can be expressed as follows:

[0030] (4);

[0031] The feasible region for scheduling the b-th air conditioning load can be obtained by expressing formula (4) in geometric form. Using active power as a variable, the feasible region is reorganized into a convex polyhedron characterized by a set of compact constraints:

[0032] (5);

[0033] In the formula, T=24; The temperature of the matrix is ​​represented by a row of T and a column of T. and It is a coefficient matrix.

[0034] 2. Air Conditioning Feasibility Domain Aggregation Method Considering Parameter Uncertainty:

[0035] The feasible region for the operation of the air conditioning load cluster is obtained based on the convex polyhedron. The maximum internal approximation feasible region of a single air conditioning load is formed by scaling and transferring the basic isomorphic polyhedron. The scaling and transfer coefficients of the isomorphic polyhedron are then optimized and solved. The uncertainty of the air conditioning load parameters is simulated by introducing the Bruker chance constraint, and the aggregated approximate feasible region is obtained.

[0036] The feasible domain for air conditioning load cluster operation is represented as follows:

[0037] (6);

[0038] In the formula, For the Minkowski summation operation; The scheduling feasibility domain is the precise aggregation of air conditioning load clusters; This refers to the collection of air conditioning loads. This represents the aggregated power of the air conditioning cluster.

[0039] The Minkowski summation of the aforementioned polygons is a nondeterministic polynomial (NP-hard) problem, and thus unsolvable. Since isomorphic polyhedra offer better processing performance than chino polyhedra, this invention uses isomorphic polyhedra to approximate the feasible region of the air conditioning load. The maximum inner approximation feasible region for a single air conditioning load is formed by scaling and transforming the basic isomorphic polyhedra, as shown in the following equation:

[0040] (7);

[0041] (8);

[0042] (9);

[0043] In the formula, and These are the transfer factor and the scaling factor, respectively. It is a basic isomorphic polyhedron. and These are the coefficient matrices of the basic isomorphic polyhedra. This represents the total number of air conditioning loads. To obtain the maximum feasible region for a single air conditioning load, the scaling and transfer coefficients of the isomorphic polyhedron are optimized, as shown in the following equation:

[0044] (10);

[0045] (11);

[0046] Assume ξ is the uncertainty parameter caused by temperature prediction error, and is a probability distribution function composed of mean and covariance, which has the matrix of formula (5). After considering the outdoor ambient temperature prediction error, formula (11) can be expressed as:

[0047] (12);

[0048] In the formula, This falls under the probability of violation. Formula (12) introduces the uncertainty of DRCC simulation air conditioning load parameters. An auxiliary variable is introduced. , sum matrix Based on Farkas' theorem, formulas (10) and (12) can be transformed into a linear programming problem solvable by formula (13):

[0049] (13);

[0050] In the formula, 4T represents the dimension of matrix N; and Representing matrices respectively sum coefficient matrix The value of the s-th row ,in and for The mean and standard deviation caused by prediction errors; the scaling factor and transfer factor of the air conditioning feasible region can be obtained by calling the commercial solver GUROBI; the aggregated transfer coefficients. and scaling factor As shown below:

[0051] (14);

[0052] The approximate operational feasible region of the aggregated air conditioning load cluster can be represented as:

[0053] (15).

[0054] 3. Day-ahead optimization scheduling strategy for air conditioning load considering multiple uncertainties:

[0055] Based on the previously obtained air conditioning load aggregation results, the distribution network operator needs to formulate an economic dispatch model and issue dispatch trajectories to controllable equipment. The day-ahead optimization dispatch model framework for air conditioning load clustering proposed in this invention is as follows: Figure 1 As shown in the diagram, firstly, the air conditioning load aggregator uses an aggregation method based on isomorphic polyhedra to obtain an approximate feasible operating region for the air conditioning load cluster, and submits this approximate feasible operating region to the distribution network operator. Secondly, the distribution network operator meets the power supply and demand balance through energy storage, photovoltaics, wind power, grid power purchase, and demand-side response. Finally, the distribution network operator obtains the operating power of the air conditioning load cluster through economic operation optimization, and thus obtains the power of a single air conditioner.

[0056] During the decomposition process, air conditioning load aggregators need to decompose the aggregated scheduling trajectory to individual air conditioning loads. Since the air conditioning loads within the same aggregator originate from the same basic isomorphic polyhedron, they can be directly decomposed based on the air conditioning load parameters, as shown in the following formula:

[0057] (16);

[0058] The equivalent operating cost function of air conditioning load is the dispatch compensation cost given by the distribution network operator to the air conditioning load aggregator. Based on the decomposition rules, the equivalent operating cost function of the air conditioning load aggregator is... for:

[0059] (17);

[0060] In the formula, For air conditioning load clusters Aggregate power at any given time; and The operating cost coefficients for air conditioning load aggregators are as follows:

[0061] (18);

[0062] (19);

[0063] In the formula, Compensation cost per unit.

[0064] Power distribution operators are constructing a day-ahead optimization scheduling model for air conditioning load clusters, with the objective function being to minimize system operating costs, including energy storage operating costs. Costs of wind and solar power curtailment Distribution network electricity purchase cost Network loss cost and equivalent operating costs of air conditioning load .

[0065] The objective function of the day-ahead optimization scheduling model for air conditioning load clusters is shown in the following equation:

[0066] (20);

[0067] (twenty one);

[0068] (twenty two);

[0069] (twenty three);

[0070] (twenty four);

[0071] (25);

[0072] In the formula, and These represent the actual charging and discharging power of the energy storage power station at time t; for The power that the distribution network purchases from the main grid at all times; , , , and These are the unit energy storage charging and discharging loss cost, the unit operation and maintenance cost of wind power, the unit operation and maintenance cost of photovoltaic power, the unit electricity purchase cost and the unit network loss cost of the distribution network operator from the upper-level grid; For the collection of distribution network branches, This indicates the total number of air conditioning load aggregators.

[0073] The method and process for solving the model are as follows: Figure 2 As shown. First, basic parameters for each air conditioning load are collected, and then calculated. and The matrix is ​​then calculated. The maximum approximate feasible region of each air conditioning load in Equation (13) is then calculated. The feasible region of the air conditioning load cluster operation (see Equation (15)) and the equivalent operating cost function (see Equation (17)) are then calculated. The distribution network operator's operation model is then substituted into the model (see Equation (20)) to obtain the scheduling plan of the air conditioning load aggregator. Finally, the air conditioning degree plan obtained after aggregation is decomposed to individual air conditioners (see Equation (16)).

[0074] The constraints of the day-ahead optimization scheduling model for air conditioning load clusters include:

[0075] (1) Operational constraints of wind power and photovoltaic power generation;

[0076] The scheduling plan for wind and solar power should be limited by the maximum available wind and solar power forecast. Considering the uncertainty of wind and solar power, the uncertainty of wind and solar power in this invention is expressed by DRCC.

[0077] (26);

[0078] Transform formula (26) into a linear constraint:

[0079] (27);

[0080] In the formula, and These represent the uncertainty parameters caused by temperature prediction errors in photovoltaic power generation and wind power generation, respectively. and It follows a specific probability distribution; and These are the average values ​​of the wind and solar forecasts; and These represent the probabilities of violating the rules regarding scenic views; and These represent the standard deviations of wind and solar forecasts.

[0081] (2) Branch flow constraints;

[0082] This invention takes a radial distribution network as the research object, and selects the state of one branch at time t to establish a branch power flow node model, such as... Figure 3 As shown, where, and These are the node numbers; and They are nodes and The voltage; branch road The current; and They are nodes and nodes The active power injected; and They are nodes and reactive power injection; and Branch roads The active and reactive power; and Branch roads The active and reactive power; k:j→k is the set of child nodes with node j as the parent node. and These are resistance and reactance, respectively.

[0083] The model is transformed into a linear programming problem using Second Order Cone Programming (SOCP). The optimal top current constraints after the transformation are as follows:

[0084] (28);

[0085] (29);

[0086] (30);

[0087] (31);

[0088] (32);

[0089] (33);

[0090] (34);

[0091] In the formula, and The nodes within time period t are respectively and nodes The square of the voltage; and Branch roads The active and reactive power; and Branch roads Reactance and resistance; branch road The current; For nodes The active power injected; For nodes reactive power injection; and Branch roads The active and reactive power; branch road The square of the current; k:j→k is the number of nodes The set of child nodes of the parent node; for Nodes within the time period The square of the voltage; For the first The power aggregation of air conditioning loads by individual aggregators.

[0092] (3) Operational constraints of energy storage power stations;

[0093] The operational constraints of energy storage power stations include charging / discharging power constraints, energy balance constraints, and tie-line power constraints. Specifically:

[0094] 1) Energy storage charging / discharging power constraints:

[0095] (35);

[0096] In the formula, and These represent the maximum charging and discharging power of the energy storage power station, respectively.

[0097] 2) Energy balance constraints for energy storage:

[0098] (36);

[0099] (37);

[0100] In the formula, and These represent the energy storage power values ​​at times t and t-1, respectively. and The state variables representing the charging and discharging of the energy storage power station are 0-1 variables. A value of 1 indicates a charging state. A value of 1 indicates a discharge state; and These are the charging and discharging efficiencies of the energy storage power station, respectively. This refers to the self-consumption rate of the energy storage power station. and These represent the maximum and minimum operating power of the energy storage power station, respectively. and These represent the remaining electricity at the initial and final moments of the energy storage power station, respectively. and represent The state variables of the energy storage power station during charging and discharging at any given time.

[0101] 3) Tie line power constraints:

[0102] (38);

[0103] In the formula, Let t be the power purchased by the distribution network from the main grid; This is the upper limit for the amount of electricity that can be purchased from the main grid.

[0104] The proposed optimized scheduling model was simulated and tested in IEEE 33-node, as shown below. Figure 4 As shown. Figure 4 In this system, nodes 18 and 33 are configured with wind power, nodes 12 and 30 are configured with photovoltaic power, node 16 is connected to energy storage, and nodes 12, 20 and 31 are connected to air conditioning load aggregator 1, air conditioning load aggregator 2 and air conditioning load aggregator 3, respectively. Assuming that aggregator 1, aggregator 2 and aggregator 3 contain 200, 250 and 300 air conditioning loads, respectively, the prediction error percentage ζ is defined as the ratio of the outdoor prediction standard deviation to the outdoor temperature prediction average, and ε is the allowed violation probability of the chance constraint.

[0105] (1) The impact of uncertain environmental temperature parameters on the feasible region of polymerization:

[0106] Taking load aggregator 1 as an example, the aggregation results of the average proportional factor of the air conditioning load under different uncertain outdoor temperatures are as follows: Figure 4 As shown. Average scaling factor is used. This indicates the scaling factor of the aggregated feasible region relative to the prototype, where , This represents the total scaling factor; in the approximation method within the aggregation process, a larger average scaling factor indicates higher aggregation accuracy. The scaling factor, as shown in the formula, is influenced by the violation probability and prediction error parameters, which affect the aggregation result. Figure 5 It can be seen that as the percentage of prediction error ζ increases or the probability of violation decreases, the average scaling factor will decrease, and the aggregation of feasible regions will become more conservative.

[0107] (2) The impact of forecast uncertainty on system operating costs:

[0108] In the optimized scheduling model, the uncertainties of photovoltaic, wind power, and outdoor temperature prediction errors are modeled using DRCC (Dynamic Radiation Control Center), and the probability of violating the opportunity constraints of all resources is set to ε = 0.05. Assuming the uncertainty parameter ζ for wind, solar, and air conditioning loads ranges from 0.01 to 0.05, the impact of different parameters on system operating costs is as follows: Figure 5 As shown.

[0109] Figure 6 The results show that when studying the uncertainty of a single resource, the system operating cost increases monotonically with the increase of the percentage of prediction error of the uncertain parameter; when considering multiple uncertainties, the system operating cost is higher than that considering only a single uncertainty. This is because as the prediction error increases, distribution operators become more conservative in their flexibility regarding the operation of wind and solar power generation and air conditioning load, thus increasing the system operating cost.

Claims

1. A day-ahead optimization scheduling method for air conditioning load clusters considering multiple uncertainties, characterized in that, Includes the following steps: Step 1: For the summer cooling operation scenario of air conditioning, establish an air conditioning load operation model, determine the indoor temperature range constraint, construct the air conditioning load operation feasible region, and then express it in geometric form as the scheduling feasible region of air conditioning load, and reorganize it into a set of convex polyhedra characterized by compact form constraints; Step 2: Obtain the feasible region of the air conditioning load cluster based on the convex polyhedron. Form the maximum internal approximation feasible region of a single air conditioning load by scaling and transferring the basic isomorphic polyhedron. Then optimize the scaling and transfer coefficients of the isomorphic polyhedron. Introduce the Bruker chance constraint to simulate the uncertainty of the air conditioning load parameters and obtain the approximate feasible region of the aggregated air conditioning load cluster. Step 3: The air conditioning load aggregator submits the approximate operational feasible region of the aggregated air conditioning load cluster to the distribution network operator. The distribution network operator constructs a day-ahead optimization scheduling model for the air conditioning load cluster with the objective function of minimizing system operating costs, and solves the scheduling plan for the cluster air conditioners, thereby obtaining the scheduling plan for individual air conditioners. Step 3 specifically includes: Step 3.1: The air conditioning load aggregator allocates the aggregated air conditioning load power to individual air conditioning loads according to parameters, as shown in the following formula: （16）； In the formula, For the power matrix, and These are the transfer factor and the scaling factor, respectively. and These are the transfer coefficients and scaling factors after aggregation; The aggregated power of the air conditioning cluster; This refers to the collection of air conditioning loads. Based on the decomposition rules, the equivalent operating cost function of the air conditioning load aggregator. for: （17）； In the formula, Let be the aggregate power of the air conditioning load cluster at time t; and The operating cost coefficients for air conditioning load aggregators are as follows: （18）； （19）； In the formula, To compensate for unit costs; Step 3.2: The power distribution operator constructs a day-ahead optimization scheduling model for air conditioning load clusters, with the objective function being to minimize the system operating cost, including the energy storage operating cost. Costs of wind and solar power curtailment Distribution network electricity purchase cost Network loss cost and equivalent operating costs of air conditioning load ; The objective function F of the day-ahead optimization scheduling model for air conditioning load clusters is shown in the following equation: （20）； （21）； （22）； （23）； （24）； （25）； In the formula, and These represent the actual charging and discharging power of the energy storage power station at time t; for The power that the distribution network purchases from the main grid at all times; , , , and These are the unit energy storage charging and discharging loss cost, the unit operation and maintenance cost of wind power, the unit operation and maintenance cost of photovoltaic power, the unit electricity purchase cost and the unit network loss cost of the distribution network operator from the upper-level grid; For the collection of distribution network branches, This indicates the total number of air conditioning load aggregators; and The outputs are solar power and wind power, respectively. branch road The square of the current; branch road The resistance; For the first The equivalent operating cost of an air conditioning load aggregator at time t.

2. The day-ahead optimization scheduling method for air conditioning load clusters considering multiple uncertainties according to claim 1, characterized in that, Step 1 specifically includes: Step 1.1: For the summer cooling operation scenario of air conditioning, the air conditioning load operation model is as follows: （1）； in, and These represent intermediate variables respectively; and These represent thermal resistance and heat capacity, respectively. and They represent Real-time indoor and outdoor ambient temperatures; and They are respectively Indoor heat load at any time and the first The cooling capacity provided by each air conditioning load; for The first derivative; Discretizing formula (1) yields a linear expression, as shown below: （2）； In the formula, , and Both represent intermediate variables; For the first Output power of an air conditioning load at time t The cooling energy efficiency ratio of an air conditioner; Indicates a time interval; Step 1.2: Determine the indoor temperature range constraints as follows: （3）； In the formula, For the user's set temperature, The temperature adjustment value is acceptable to the user; Step 1.3: Combine the above formulas (2) and (3) to form the feasible region for the operation of a single air conditioning load, that is, the scheduling feasible region for the b-th air conditioning load. It is expressed as follows: （4）； In the formula, and These are intermediate variables, representing the upper and lower limits of indoor temperature, respectively. and It is divided into the upper and lower limits of the air conditioner's output power; Step 1.4: Express formula (4) in geometric form to obtain the scheduling feasible region of the air conditioning load. With active power as the variable, the scheduling feasible region of the b-th air conditioning load is... As shown in equation (5), it is essentially a set of convex polyhedra characterized by linear inequalities; （5）； In the formula, T is the power matrix. the number of rows, Representative power matrix The dimension is T rows and one column; M b and N b It is a coefficient matrix.

3. The day-ahead optimization scheduling method for air conditioning load clusters considering multiple uncertainties according to claim 2, characterized in that, Step 2 specifically includes: Step 2.1: Represent the feasible region for the operation of the air conditioning load cluster as follows: （6）； in, For the Minkowski summation operation; The scheduling feasibility domain is the precise aggregation of air conditioning load clusters; This refers to the collection of air conditioning loads. The aggregated power of the air conditioning cluster; Step 2.2: By scaling and transferring the basic isomorphic polyhedron, an approximate feasible region for a single air conditioning load is formed, as shown in the following equation: （7）； （8）； （9）； In the formula, and These are the transfer factor and the scaling factor, respectively. It is a basic isomorphic polyhedron. and These are the coefficient matrices of the basic isomorphic polyhedra. P0 represents the total number of air conditioning loads; P0 represents the air conditioning power that satisfies the feasible domain of the basic isomorphic polyhedron. Step 2.3: Optimize the scaling and transfer coefficients of the isomorphic polyhedron, as shown in the following equation: （10）； （11）； In the formula, This represents the approximate feasible region for a single air conditioning load. It is an uncertainty parameter caused by temperature prediction error, and is a probability distribution function composed of mean and covariance, with a coefficient matrix given by formula (5). In the middle; after considering the outdoor ambient temperature prediction error, formula (11) is expressed as: （12）； in, This violates probability; The parameter representing the uncertainty caused by temperature prediction error. It follows a specific probability distribution; Step 2.4: Introduce the partial blue bar chance constraint into formula (12) to simulate the uncertainty of air conditioning load parameters; introduce auxiliary variables. , sum matrix ; Based on Farkas' theorem, formulas (10) and (12) are transformed into linear programming problems solvable by formula (13): （13）； In the formula, 4T represents the dimension of matrix N; and These represent the introduced matrices respectively. sum coefficient matrix The value of the s-th row ,in and for Mean and standard deviation caused by prediction error; aggregation transfer coefficient and scaling factor As shown below: （14）； The approximate feasible operating region of the aggregated air conditioning cluster Represented as: （15）。 4. The day-ahead optimization scheduling method for air conditioning load clusters considering multiple uncertainties according to claim 3, characterized in that, In step 3, the constraints of the day-ahead optimization scheduling model for air conditioning load clusters include: (1) Operational constraints of wind power and photovoltaic power generation; （26）； In the formula, and These represent the uncertainty parameters caused by temperature prediction errors in photovoltaic power generation and wind power generation, respectively. and It follows a specific probability distribution; and These are the average values ​​of the wind and solar forecasts; and These represent the probabilities of violating the rules regarding scenic views; Transformed into linear constraints as follows: （27）； In the formula, and These are the standard deviations of the wind and solar forecasts; (2) Branch flow constraints; The branch power flow model is transformed into a linear programming problem using second-order cone programming. The optimal power flow constraints after the transformation are as follows: （28）； （29）； （30）； （31）； （32）； （33）； （34）； In the formula, and They are respectively Nodes within the time period and nodes The square of the voltage; and Branch roads The active and reactive power; and Branch roads Reactance and resistance; branch road The current; For nodes The active power injected; For nodes reactive power injection; and Branch roads The active and reactive power; branch road The square of the current; k:j→k is the number of nodes The set of child nodes of the parent node; for Nodes within the time period The square of the voltage; For the first The aggregated power of an air conditioning load aggregator cluster; Represents the norm; (3) Operational constraints of energy storage power stations; The operational constraints of energy storage power stations include energy storage charging / discharging power constraints, energy balance constraints, and tie-line power constraints; specifically as follows: 1) Energy storage charging / discharging power constraints: （35）； In the formula, and These are the maximum charging and discharging power of the energy storage power station, respectively. 2) Energy balance constraints for energy storage: （36）； （37）； In the formula, and These represent the energy storage power values ​​at times t and t-1, respectively. and These represent the state variables for charging and discharging in an energy storage power station, and are 0-1 variables. A value of 1 indicates a charging state. A value of 1 indicates a discharge state; and These are the charging and discharging efficiencies of the energy storage power station, respectively. This refers to the self-consumption rate of the energy storage power station. and These represent the maximum and minimum operating power of the energy storage power station, respectively. and These represent the remaining electricity at the initial and final moments of the energy storage power station, respectively. and Represent The state variables of the energy storage power station during charging and discharging at any given moment; 3) Tie line power constraints: （38）； In the formula, for The power that the distribution network purchases from the main grid at all times; This is the upper limit for the amount of electricity that can be purchased from the main grid.

Images