A power distribution network double-layer dispatching method based on model prediction control under temperature control load participation

Abstract

The application discloses a kind of temperature control load participation based on model predictive control's distribution network double-layer dispatching method, and the technical scheme steps of application include: first, design the objective function and constraint condition of temperature control load distribution network dispatching, form day-ahead optimal dispatching plan based on particle swarm;Based on the heat balance equation of building, the quantitative mathematical relationship between indoor temperature and refrigeration power and external environment is constructed, and the heat dynamic characteristic model considering is established;Finally, according to the day-ahead temperature control load dispatching plan and building heat dynamic characteristic model, based on model predictive control, single air conditioner control strategy is established, system stability and regulation economy are considered in a larger range, and the economy of distribution network day-ahead optimal dispatching is further improved.

Description

Technical Field

[0001] This invention relates to the problem of distribution network optimization scheduling and temperature-controlled load control strategies under demand response, and in particular to a two-layer scheduling method for distribution networks based on model predictive control with the participation of temperature-controlled loads. Technical Background

[0002] The stability of power angle and frequency is essential for the safe and reliable operation of a power system, and it is closely related to the power balance between generation and consumption. Traditionally, power systems have adopted a load-tracking approach on the generation side to achieve power balance and stability, while the load side is considered a passive physical terminal and not fully integrated into the power system control system. When traditional methods of allocating generator output are insufficient to maintain system stability or require significant costs, the current load shedding / curtailment measures incur substantial socio-economic costs. Furthermore, with the continuous increase in power load, the centralized integration of numerous intermittent power sources into the power system, and the increasing proportion of large-capacity supercritical units in the system, the ability of the generation side to flexibly dispatch power output is gradually weakening.

[0003] The development of the social economy and the improvement of residents' living standards have led to a continuous increase in the proportion of flexible loads with rapid response dispatching capabilities, such as central air conditioning, water heaters, and some large industrial users, in the total electricity load. Demand response, as a means for loads to participate in power system regulation, has been proposed. The development of source-grid-load interactive operation has attracted widespread attention to utilizing existing demand-side resources to supplement traditional generation dispatching for power system regulation. This paper takes temperature-controlled loads, represented by air conditioning, as the main research object, and studies the participation of temperature-controlled loads in day-ahead optimal dispatching of the distribution network from two aspects: model construction and control strategies. Summary of the Invention

[0004] The purpose of this invention is to propose a day-ahead optimal dispatch control strategy for distribution networks involving temperature-controlled loads. This invention designs day-ahead optimal dispatch control involving wind, solar, and air conditioning, and establishes an air conditioning thermal dynamic characteristic model based on the building heat balance equation. It proposes a single-unit air conditioning control strategy based on Model Predictive Control (MPC), which balances system stability and control economy over a wider range, further improving the economy of day-ahead optimal dispatch in distribution networks.

[0005] This invention adopts the following technical solution: A two-layer dispatching method for distribution networks based on model predictive control with the participation of temperature-controlled loads is established, comprising the following steps:

[0006] (1) Design the objective function and constraints for the dispatching of distribution networks with temperature-controlled loads, and form a day-ahead optimization dispatching plan based on particle swarm optimization.

[0007] (2) Based on the building’s heat balance equation, construct a quantitative mathematical relationship between indoor temperature, cooling power and external environmental conditions, and establish a model that considers thermal dynamic characteristics.

[0008] (3) Based on the day-to-day temperature control load scheduling plan and the building thermal dynamic characteristics model, establish a single air conditioning control strategy based on model predictive control.

[0009] Specifically, in step (1), the participation of temperature-controlled load aggregators in the day-ahead dispatch model of the distribution network is described as follows:

[0010] The objective function is to minimize the overall daily operating cost of the scheduling system. The objective function is as follows:

[0011]

[0012] In the formula: C is the total cost within the scheduling period; C DG (t) represents the cost of a traditional generator unit during time period t; C ES (t) represents the operation and maintenance cost of the energy storage battery during time period t; C c (t) represents the demand response cost of temperature control load during time period t; C WP (t) represents the penalty cost for wind and solar power curtailment during time period t; N is the total number of time periods within a scheduling cycle;

[0013] The mathematical models for each scheduling cost are as follows:

[0014] (1) Cost of traditional units

[0015]

[0016] In the formula: α, β, and γ are the traditional unit dispatch cost coefficients; P DG (t) represents the output of the traditional generating unit during time period t;

[0017] (2) Energy storage battery cost

[0018]

[0019] In the formula: f ES The unit capacity installation cost of the battery; Q ES The capital recovery factor for the battery; a ES For the capacity factor of the battery; Y ES This refers to the annual operating hours of the battery; m ES P represents the operating and management cost coefficient for the battery. ES (t) represents the charging and discharging power of the battery during time period t;

[0020] (3) Temperature control load demand response cost

[0021]

[0022] In the formula: μ0(t) is the initial electricity price of the temperature-controlled load during time period t; This represents the electricity volume that did not participate in demand response during time period t. The electricity generated by the temperature-controlled load after participating in demand response during time period t; μ c (t) represents the electricity price after the temperature-controlled load participates in demand response during time period t;

[0023] (4) Punishment for abandoning wind and light

[0024]

[0025] In the formula: ε is the penalty cost per unit of abandoned air volume; and These represent the expected output power and energy consumption of the wind turbine generator on the day before time period t; δ is the penalty fee per unit of curtailed solar power. and P v (t) represents the day-ahead expected output power and power consumption of the photovoltaic generator unit during time period t.

[0026] The constraints are as follows:

[0027] (1) Power balance constraint

[0028]

[0029] In the formula: This refers to the discharge power of the energy storage battery. Power for charging energy storage batteries; P L (t) is the base load; P c (t) represents the demand response power of the temperature-controlled load;

[0030] (2) Constraints of traditional units

[0031]

[0032] In the formula: These are the upper and lower limits of the output power of traditional generator sets; The ramp rate for power increase and decrease of traditional generating units; Δt is the time difference for dispatching;

[0033] (3) Energy storage battery charge and discharge constraints

[0034]

[0035] In the formula: This represents the maximum charge and discharge power of the energy storage battery.

[0036]

[0037] Equation (9) is the constraint condition for the remaining capacity of the energy storage battery, where E ES (t), E ES (t-1) represents the remaining battery capacity at time t and time t-1, respectively. The upper and lower limits of the remaining capacity are given by η, and the charge / discharge efficiency is given by η.

[0038] With the goal of minimizing scheduling costs and considering constraints, a particle swarm optimization algorithm is used to complete day-ahead optimal scheduling, resulting in the output plans of traditional units, wind power, photovoltaic, and energy storage batteries, as well as the reduction of temperature-controlled loads.

[0039] Specifically, in step (2), based on the building's heat balance equation, a quantitative mathematical relationship is constructed between indoor temperature, cooling power, and external environmental conditions. A model considering thermal dynamic characteristics is established, and the temperature control load model is as follows:

[0040] Ignoring air leakage through gaps in the room, we assume the air conditioning vent is the only ventilation opening in the room. Thermal convection is the process by which cold / heat energy is transferred from one place to another in a gas, including forced convection of indoor air towards the inner surface of the building envelope.

[0041]

[0042] In the formula: Q in Forced convection of indoor air to the interior surface of the building; Q out Forced convection of outdoor air on the building exterior; T in For indoor temperature in smart buildings; T im T om T represents the internal and external surface temperatures of the material. out Outdoor temperature; A is the surface area of ​​the building envelope; k in k out These are the surface heat transfer coefficients for the inner and outer surfaces, respectively.

[0043] Because of the temperature gradient between the inner and outer surfaces of the building envelope, heat will transfer from the high-temperature side to the low-temperature side, which is the heat conduction process of the building envelope, including walls, windows, and roof. The thermal dynamic prediction model established in this invention will consider the layered structure of the building envelope. The heat conduction process of each layer of material can be obtained by Fourier's law as shown in equation (11):

[0044]

[0045] In the formula: Q k D represents the heat conducted per unit time by each layer of material; D represents the thickness of the building envelope; k k denoted as the thermal conductivity of the material.

[0046] During the heating / cooling process of building envelope materials and air, heat absorption and release are usually involved. Therefore, building envelope materials and indoor air have a certain heat storage capacity, and their heat storage process can be represented by equation (12):

[0047]

[0048] In the formula: T om Related to continuous time τ; Q m The heat storage capacity of the building envelope materials and indoor air per unit time; c is the specific heat capacity of the material; m m For the quality of materials.

[0049] The indoor air absorbs cold / heat from the air conditioning system. The indoor air absorbs cold / heat and transfers it to the inner surfaces of the walls, windows, roof, and floor through forced convection. The heat is then conducted and stored / released layer by layer along the building envelope until it reaches the outer surface of the building envelope. Finally, the heat is dissipated to the outside environment through natural convection between the building's outer surface and the outdoor air.

[0050] Specifically, in step (3), based on the daytime temperature control load scheduling plan and the building's thermal dynamic characteristics model, a single-unit air conditioning control strategy is established based on model predictive control. The model control strategy is described as follows:

[0051] According to the thermodynamic model of air conditioning, the continuous state-space equation of air conditioning can be expressed as:

[0052]

[0053] In the formula: T = [T in T im T om ] T Let T be the state vector. in Indoor temperature, T im T om These represent the indoor and outdoor envelope temperatures, respectively; u = [T out ,ψ,u] T Let be the input vector, where ψ is the solar irradiance, u is the air conditioner start-up status, u = 0 indicates that the air conditioner is not started, and u = 1 indicates that the air conditioner is started; matrix A represents the dynamic behavior of the system, and matrix B represents the influence of the input elements (outdoor temperature, solar irradiance, air conditioner start-up status) on the system.

[0054] With T s For the sampling period, model (13) is transformed into a discrete state-space expression:

[0055]

[0056] In the formula: T(k)=[T in(k), T im (k), T om (k)] T U(k) = [T out (k), ψ(k), u(k)] T y(k) is the output vector.

[0057] The goal of the air conditioning MPC control strategy is to maintain the indoor temperature close to the set temperature while reducing air conditioning energy consumption. Therefore, the MPC objective function consists of two parts: minimizing air conditioning power consumption and a user discomfort index, measured by the difference between the actual room temperature and the set temperature. The objective function expression is as follows:

[0058]

[0059] In the formula: Δk is the air conditioner operating time step; M is the number of predicted time periods; u(k) indicates whether the air conditioner is started at time k, u(k) = 0 indicates that the air conditioner is not started, and u(k) = 1 indicates that the air conditioner is started; P c (k) represents the air conditioning power; in this invention, Δk is set to 15 min, and M is set to 4, meaning the total predicted duration is 1 hour; T set The preset temperature for the air conditioner; λ is the weighting coefficient.

[0060] Let u(k)=[u1,u2,…,u M ] T To predict the air conditioning control vector within the time domain M, let the acceptable indoor temperature range for the user be . The steps of the MPC control strategy are as follows:

[0061] Step 1: Measure the temperature T(k) at time t using a temperature sensor;

[0062] Step 2: Optimize the air conditioner control vector u(k) by using the start-stop status of the air conditioner at different times as the decision variable. The MPC optimization model is shown in equations (16) and (17):

[0063] u(k)=arg min J (16)

[0064]

[0065] Step 3: Control the start / stop state of the air conditioner by using step 1 u1 in the control vector u(k);

[0066] Step 4: Let k = k + Δk, proceed to the next time period optimization, and return to step 1.

[0067] The technical solution provided by this invention has the following beneficial effects:

[0068] This invention designs the objective function and constraints for day-ahead temperature-controlled load distribution network scheduling, generates an optimal scheduling plan based on particle swarm optimization, and establishes a thermal dynamic characteristic model for air conditioners. Finally, based on the day-ahead temperature-controlled load scheduling plan and the building thermal dynamic characteristic model, a model predictive control strategy is proposed to establish a control strategy for individual air conditioners. This approach balances system stability and control economy, further improving the economic efficiency of day-ahead optimal scheduling in the distribution network. Attached Figure Description

[0069] The present invention will be further described below with reference to the accompanying drawings:

[0070] Figure 1 This is a flowchart of the present invention;

[0071] Figure 2 For wind power, solar power, and load forecast curves;

[0072] Figure 3 A diagram showing the results of optimized dispatching of the power distribution network;

[0073] Figure 4 This is a diagram showing the indoor temperature and the on / off status of the air conditioner. Detailed Implementation Plan

[0074] To better understand the purpose, technical solution, and technical effects of this invention, the invention will be further explained and described below in conjunction with the accompanying drawings.

[0075] This invention proposes a two-layer dispatching method for distribution networks based on model predictive control with the participation of temperature-controlled loads. Figure 1 The flowchart of this invention includes the following detailed steps.

[0076] Step 1: Design the objective function and constraints for the dispatching of distribution networks with temperature-controlled loads, and generate a day-ahead optimization dispatching plan based on particle swarm optimization.

[0077] With the large-scale grid connection of new energy sources, in order to fully utilize wind power and photovoltaic power and reduce grid operating costs, based on the day-ahead forecasts of wind power, photovoltaic power and load, the objective function is to minimize the overall daily operating cost of dispatching, as shown in the following formula:

[0078]

[0079] In the formula: C is the total cost within the scheduling period; C DG (t) represents the cost of a traditional generator unit during time period t; C ES (t) represents the operation and maintenance cost of the energy storage battery during time period t; C c (t) represents the demand response cost of temperature control load during time period t; C WP (t) represents the penalty cost for wind and solar power curtailment during time period t; N represents the total number of time periods within a scheduling cycle.

[0080] The mathematical models for each scheduling cost are as follows:

[0081] (1) Cost of traditional units

[0082]

[0083] In the formula: α, β, and γ are the traditional unit dispatch cost coefficients; P DG (t) represents the output of the traditional unit during time period t.

[0084] (2) Energy storage battery cost

[0085]

[0086] In the formula: f ES The unit capacity installation cost of the battery; Q ES The capital recovery factor for the battery; a ES For the capacity factor of the battery; Y ES This refers to the annual operating hours of the battery; m ES P represents the operating and management cost coefficient for the battery. ES (t) represents the charging and discharging power of the battery during time period t.

[0087] (3) Temperature control load demand response cost

[0088]

[0089] In the formula: μ0(t) is the initial electricity price of the temperature-controlled load during time period t; This represents the electricity volume that did not participate in demand response during time period t. The electricity generated by the temperature-controlled load after participating in demand response during time period t; μ c (t) represents the electricity price after the temperature-controlled load participates in demand response during time period t.

[0090] (4) Punishment for abandoning wind and light

[0091]

[0092] In the formula: ε is the penalty cost per unit of abandoned air volume; and P w (t) represents the expected output power and absorption capacity of the wind turbine generator on the day before time period t; δ is the penalty fee per unit of curtailed solar power. and P v (t) represents the day-ahead expected output power and power consumption of the photovoltaic generator unit during time period t.

[0093] The constraints are:

[0094] (1) Power balance constraint

[0095]

[0096] In the formula: This refers to the discharge power of the energy storage battery. Power for charging energy storage batteries; P L (t) is the base load; P c (t) represents the demand response power of the temperature-controlled load;

[0097] (2) Constraints of traditional units

[0098]

[0099] In the formula: These are the upper and lower limits of the output power of traditional generator sets; The ramp rate for power increase and decrease of traditional generating units; Δt is the time difference for dispatching;

[0100] (3) Energy storage battery charge and discharge constraints

[0101]

[0102] In the formula: This represents the maximum charge and discharge power of the energy storage battery.

[0103]

[0104] Equation (9) is the constraint condition for the remaining capacity of the energy storage battery, where E ES (t), E ES (t-1) represents the remaining battery capacity at time t and time t-1, respectively. The upper and lower limits of the remaining capacity are given by η, and the charge / discharge efficiency is given by η.

[0105] With the goal of minimizing scheduling costs and considering constraints, a particle swarm optimization algorithm is used to complete day-ahead optimal scheduling, resulting in the output plans of traditional units, wind power, photovoltaic, and energy storage batteries, as well as the reduction of temperature-controlled loads.

[0106] Step 2: Based on the building's heat balance equation, construct a quantitative mathematical relationship between indoor temperature, cooling power, and external environmental conditions, and establish a model that considers thermal dynamic characteristics.

[0107] Ignoring air leakage through gaps in the room, and assuming the air conditioning vent is the only ventilation opening in the room, thermal convection is the process of transferring cold / heat energy from one place to another in a gas, including forced convection of indoor air towards the inner surface of the building envelope.

[0108]

[0109] In the formula: Q in Forced convection of indoor air to the interior surface of the building; Qout Forced convection of outdoor air on the building exterior; T in For indoor temperature in smart buildings; T im T om T represents the internal and external surface temperatures of the material. out Outdoor temperature; A is the surface area of ​​the building envelope; k in k out These are the surface heat transfer coefficients for the inner and outer surfaces, respectively.

[0110] Because of the temperature gradient between the inner and outer surfaces of the building envelope, heat will transfer from the high-temperature side to the low-temperature side, which is the heat conduction process of the building envelope, including walls, windows, and roof. The thermal dynamic prediction model established in this invention will consider the layered structure of the building envelope. The heat conduction process of each layer of material can be obtained by Fourier's law as shown in equation (11):

[0111]

[0112] In the formula: Q k D represents the heat conducted per unit time by each layer of material; D represents the thickness of the building envelope; k k denoted as the thermal conductivity of the material.

[0113] During the heating / cooling process of building envelope materials and air, heat is usually absorbed and released. Therefore, building envelope materials and indoor air have a certain heat storage capacity, and their heat storage process can be represented by equation (12).

[0114]

[0115] In the formula: T om Related to continuous time τ; Q m The heat storage capacity of the building envelope materials and indoor air per unit time; c is the specific heat capacity of the material; m m For the quality of materials.

[0116] The indoor air absorbs cold / heat from the air conditioning system. The indoor air absorbs cold / heat and transfers it to the inner surfaces of the walls, windows, roof, and floor through forced convection. The heat is then conducted and stored / released layer by layer along the building envelope until it reaches the outer surface of the building envelope. Finally, the heat is dissipated to the outside environment through natural convection between the building's outer surface and the outdoor air.

[0117] Step 3: Based on the day-ahead temperature control load scheduling plan and the building thermal dynamic characteristics model, establish a single-unit air conditioning control strategy based on model predictive control.

[0118] According to the thermodynamic model of air conditioning, the continuous state-space equation of air conditioning can be expressed as:

[0119]

[0120] In the formula: T = [T in T im T om ] T Let T be the state vector. in Indoor temperature, T im T om These represent the indoor and outdoor envelope temperatures, respectively; U = [T out ,ψ,u] T Let be the input vector, where ψ is the solar irradiance, u is the air conditioner start-up status, u = 0 indicates that the air conditioner is not started, and u = 1 indicates that the air conditioner is started; matrix A represents the dynamic behavior of the system, and matrix B represents the influence of the input elements (outdoor temperature, solar irradiance, air conditioner start-up status) on the system.

[0121] With T s For the sampling period, model (13) is transformed into a discrete state-space expression:

[0122]

[0123] In the formula: T(k)=[T in (k), T im (k), T om (k)] T U(k) = [T out (k), ψ(k), u(k)] T y(k) is the output vector.

[0124] The goal of the air conditioning MPC control strategy is to maintain the indoor temperature close to the set temperature while reducing air conditioning energy consumption. Therefore, the MPC objective function consists of two parts: minimizing air conditioning power consumption and a user discomfort index, measured by the difference between the actual room temperature and the set temperature. The objective function expression is as follows:

[0125]

[0126] In the formula: Δk is the air conditioner operating time step; M is the number of predicted time periods; u(k) indicates whether the air conditioner is started at time k, u(k) = 0 indicates that the air conditioner is not started, and u(k) = 1 indicates that the air conditioner is started; P c (k) represents the air conditioning power; in this invention, Δk is set to 15 min, and M is set to 4, meaning the total predicted duration is 1 hour; T set The preset temperature for the air conditioner; λ is the weighting coefficient.

[0127] Let u(k)=[u1,u2,…,u M ] TTo predict the air conditioning control vector within the time domain M, let the acceptable indoor temperature range for the user be . The steps of the MPC control strategy are as follows:

[0128] Step 1: Measure the temperature T(k) at time t using a temperature sensor;

[0129] Step 2: Optimize the air conditioner control vector u(k) by using the start-stop status of the air conditioner at different times as the decision variable. The MPC optimization model is shown in equations (16) and (17):

[0130] u(k)=arg min J (16)

[0131]

[0132] Step 3: Control the start / stop state of the air conditioner by using step 1 u1 in the control vector u(k);

[0133] Step 4: Let k = k + Δk, proceed to the next time period optimization, and return to step 1.

[0134] To further understand this invention and verify the accuracy and effectiveness of the model predictive control in establishing a single-unit air conditioner control strategy, a simulation example was conducted. The parameters were set as follows: In the distribution network optimization scheduling model, the scheduling duration was 24 hours, and the time interval was 1 hour; the installed capacity of wind power and photovoltaic power was 3MW each, the output range of traditional units was 0-4MW, the maximum charging and discharging power of energy storage was 1.2MW, and the capacity was 4.5MWh. The air conditioner used had a power of 1.5kW and a cooling capacity of 5.1kW. The user-set temperature was 25℃, ΔT was 1℃, and the initial values ​​of the indoor temperature, indoor building envelope temperature, and outdoor building envelope temperature were 25.3℃, 22.7℃, and 33.3℃, ​​respectively. The time period length Δt was 15 minutes, the number of prediction periods was 4 (i.e., calculating the start and stop of the air conditioner in the next hour), and the weighting coefficient λ was 2. Figure 2 Wind power, solar power, and load forecast curves are provided.

[0135] To illustrate the load output regulation process of distributed power sources and the main grid after the energy storage system and temperature-controlled load participate in the distribution network dispatch operation, the simulation results are as follows: Figure 3As shown, due to the participation of energy storage and temperature-controlled loads, the renewable energy absorption capacity may be significantly improved, the power purchase ratio of the distribution network from the main grid and the output ratio of gas turbines will be significantly reduced, and the load will tend to flatten. The role of energy storage and temperature-controlled loads is mainly manifested in the following ways: During periods of surplus renewable energy output, energy storage devices use excess power to charge, which shortens the periods of wind and solar curtailment (wind and solar curtailment still exists from 1:00 to 5:00 in the morning and from 10:00 to 16:00), and the amount of renewable energy power reduction is significantly reduced to 2.82 MWh; During peak load periods, energy storage devices discharge to the distribution network, and the air conditioning load responds to dispatch instructions to reduce the power of the air conditioning load, thereby reducing the peak load of the distribution network and reducing the peak-valley difference of the load.

[0136] To illustrate the temperature-controlled load demand response regulation strategy Figure 4 The diagram shows the indoor temperature and air conditioner start / stop status. Taking 6 PM to 10 PM as an example, the control variables and indoor temperature for the MPC-based air conditioner control in 16 time periods over a 4-hour period are shown below. Figure 4 ,Depend on Figure 3 It can be seen that, since the comfort index coefficient is 2, which accounts for a large proportion, the indoor temperature fluctuates slightly around the set value of 25 degrees Celsius, but the air conditioner starts and stops relatively frequently.

Claims

1. A two-layer dispatching method for distribution networks based on model predictive control with the participation of temperature-controlled loads, characterized in that, Including the following steps: (1) Design the objective function and constraints for the dispatching of distribution networks with temperature-controlled loads, and form a day-ahead optimized dispatching plan based on the particle swarm optimization algorithm; (2) Based on the building’s heat balance equation, construct a quantitative mathematical relationship between indoor temperature, cooling power and external environmental conditions, and establish a model that considers thermal dynamic characteristics. (3) Based on the day-ahead temperature control load scheduling plan and the building thermal dynamics model, design a single-unit air conditioning control strategy based on model predictive control; The thermal dynamic characteristic model considered in step (2) is as follows: In the formula: Q in Forced convection of indoor air to the interior surface of the building; Q out Forced convection of outdoor air on the building exterior; T in For indoor temperature in smart buildings; T im T om T represents the internal and external surface temperatures of the material. out Outdoor temperature; A is the surface area of ​​the building envelope; k in k out These are the surface heat transfer coefficients of the inner and outer surfaces, respectively. The objective function expression for the air conditioning model predictive control in step (3) is as follows: In the formula: Δk is the air conditioner operating time step; M is the number of predicted time periods; u(k) indicates whether the air conditioner is started at time k, u(k) = 0 indicates that the air conditioner is not started, and u(k) = 1 indicates that the air conditioner is started; P c (k) represents the air conditioner power; T set The preset temperature for the air conditioner; λ is the weighting coefficient; T(k) = [T in (k), T im (k), T om (k)] T Let T be the state vector. in Indoor temperature, T im T om These refer to the indoor and outdoor envelope temperatures, respectively. Let u(k)=[u1,u2,...,u M ] T To predict the air conditioning control vector within the time domain M, let the acceptable indoor temperature range for the user be . The steps of the MPC control strategy are as follows: Step 1: Measure the temperature T(k) at time t using a temperature sensor; Step 2: Optimize the air conditioner control vector u(k) by using the start-stop status of the air conditioner at different times as the decision variable. The MPC optimization model is shown in equations (3) and (4): u(k)=arg min J (3) In the formula: U(k)=[T out (k), ψ(k), u(k)] T Let be the input vector, where ψ(k) is the solar irradiance; Matrix A represents the dynamic behavior of the system, matrix B represents the impact of input elements including external temperature, solar irradiance, and air conditioning start-up status on the system, and T s Where τ is the sampling period, and τ is the continuous time; C = [1 0 0] T ; Step 3: Control the start / stop state of the air conditioner by using step 1 u1 in the control vector u(k); Step 4: Let k = k + Δk, proceed to the next time period optimization, and return to step 1.

2. The two-layer dispatching method for distribution networks based on model predictive control with the participation of temperature-controlled loads as described in claim 1, characterized in that, In step (1), the load aggregator participates in the day-ahead dispatch model of the distribution network: The objective function is to minimize the overall daily operating cost of the scheduling system. The objective function is as follows: In the formula: C is the total cost within the scheduling period; C DG (t) represents the cost of a traditional generator unit during time period t; C ES (t) represents the operation and maintenance cost of the energy storage battery during time period t; C c (t) represents the demand response cost of temperature control load during time period t; C WP (t) represents the penalty cost for wind and solar power curtailment during time period t; N is the total number of time periods within a scheduling cycle; The mathematical models for each scheduling cost are as follows: (1) Cost of traditional units In the formula: α, β, and γ are the traditional unit dispatch cost coefficients; P DG (t) represents the output of the traditional generating unit during time period t; (2) Energy storage battery cost In the formula: f ES The unit capacity installation cost of the battery; Q ES The capital recovery factor for the battery; a ES For the capacity factor of the battery; Y ES This refers to the annual operating hours of the battery; m ES P represents the operating and management cost coefficient for the battery. ES (t) represents the charging and discharging power of the battery during time period t; (3) Temperature control load demand response cost In the formula: μ0(t) is the initial electricity price of the temperature-controlled load during time period t; This represents the electricity volume that did not participate in demand response during time period t. The electricity generated by the temperature-controlled load after participating in demand response during time period t; μ c (t) represents the electricity price after the temperature-controlled load participates in demand response during time period t; (4) Punishment for abandoning wind and light In the formula: ε is the penalty cost per unit of abandoned air volume; and P w (t) represents the expected output power and power consumption of the wind turbine generator on the day before time period t; δ represents the penalty fee per unit of wasted light. and P v (t) represents the day-ahead expected output power and power consumption of the photovoltaic power generation unit during time period t, respectively; The main constraints of the power distribution network are as follows: in, This refers to the discharge power of the energy storage battery. Power for charging energy storage batteries; P L (t) is the base load; P c (t) represents the demand response power of the temperature-controlled load; These are the upper and lower limits of the output power of traditional generator sets; The ramp rate for power increase and decrease of traditional generating units; Δt is the time difference for dispatching; E represents the maximum charge and discharge power of the energy storage battery. ES (t), E ES (t-1) represents the remaining battery capacity at time t and time t-1, respectively. η represents the upper and lower limits of the remaining capacity, and η represents the charging and discharging efficiency; Equations (10)-(13) are the comprehensive constraints that day-ahead scheduling should satisfy; With the goal of minimizing scheduling costs and considering constraints, the particle swarm optimization algorithm is used to complete day-ahead optimal scheduling, resulting in the output plans of traditional units, wind power, photovoltaic, and energy storage batteries, as well as the reduction of temperature-controlled loads.

3. The two-layer dispatching method for distribution networks based on model predictive control with the participation of temperature-controlled loads as described in claim 1, characterized in that, The heat conduction process of each layer of material can be obtained from Fourier's law as shown in equation (14): In the formula: Q k D represents the heat conducted per unit time by each layer of material; D represents the thickness of the building envelope; k k The thermal conductivity of the material; During the heating / cooling process of building envelope materials and air, heat absorption and release are usually involved. Therefore, building envelope materials and indoor air have a certain heat storage capacity, and their heat storage process can be represented by equation (15): In the formula: T om Related to continuous time τ; Q m The heat storage capacity of the building envelope materials and indoor air per unit time; c is the specific heat capacity of the material; m m For the quality of materials; The indoor air absorbs cold / heat from the air conditioning system. The indoor air absorbs cold / heat and transfers it to the inner surfaces of the walls, windows, roof, and floor through forced convection. The heat is then conducted and stored / released layer by layer along the building envelope until it reaches the outer surface of the building envelope. Finally, the heat is dissipated to the outside environment through natural convection between the building's outer surface and the outdoor air.

4. The two-layer dispatching method for distribution networks based on model predictive control with the participation of temperature-controlled loads as described in claim 1, characterized in that, The individual air conditioner control strategy in step (3) is as follows: According to the thermodynamic model of air conditioning, the continuous state-space equation of air conditioning can be expressed as: In the formula: T = [T in T im T om ] T Let T be the state vector. in Indoor temperature, T im T om These represent the indoor and outdoor envelope temperatures, respectively; U = [T out ,ψ,u] T Let be the input vector, where ψ is the solar irradiance and u is the air conditioner's start / stop status, where u = 0 indicates the air conditioner is not started and u = 1 indicates the air conditioner is started; Matrix A represents the dynamic behavior of the system, and matrix B represents the impact of input elements, including external temperature, solar irradiance, and air conditioning start-up status, on the system. With T s For the sampling period, model (16) is transformed into a discrete state-space expression: In the formula: T(k)=[T in (k), T im (k), T om (k)] T U(k) = [T out (k), ψ(k), u(k)] T y(k) is the output vector. C = [1 0 0] T .

Images