Distributed cooperative power aggregation tracking control method and system for massive air conditioners

Abstract

The invention discloses a consistency control method and system for air conditioner group grouping participating demand response to complete a power adjustment target, and the method comprises the steps: building all air conditioners into a multi-agent network through an air conditioner equipment Internet of Things terminal based on a graph theory, and obtaining a network topology connected graph, an adjacent matrix, a Laplacian matrix and a row random matrix; the air conditioner group is divided into a plurality of subgroups according to the models or the positions of the air conditioners, the intelligent agent with the highest algebraic connectivity in each subgroup is set as a master, and the other intelligent agents are set as followers; a second-order equivalent thermal parameter model is adopted to establish the corresponding relation between the air conditioner load and the indoor air temperature and the corresponding relation between the air conditioner load and the solid temperature, meanwhile, the change rate of the indoor air temperature in the regulation and control process is considered, and the indoor air temperature and the indoor air temperature change rate serve as consistency variables; iterating the indoor air temperature and the air temperature change rate of each intelligent agent in the system at the next moment by adopting a second-order distributed collaborative iteration protocol; and the expected air conditioner power at the current moment is deduced and calculated through the discretized second-order equivalent thermal parameter model, the corresponding working frequency is obtained through calculation according to the expected air conditioner power, and frequency adjustment is conducted through the air conditioner Internet of Things terminal. The indoor air temperature and the air temperature change rate of each air conditioner user are consistent, air conditioner load resources are uniformly developed and utilized, the influence of demand response on the comfort level of the air conditioner users is reduced, and the air conditioner load resources are more efficiently utilized.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid control, and in particular to a method and system for flexible load participation of an air-conditioning group at a power user side in demand response and air-conditioning equipment control, which realizes control of the air-conditioning group through a distributed consistency algorithm. Background Art

[0002] With the continuous advancement of urbanization, the power load in the core areas of cities continues to rise. Air conditioning load accounts for more than 50% of the peak load in major Chinese cities. In summer or winter, the short-term load peak of some regional power grids is often caused by the simultaneous use of massive air conditioning loads. This not only exacerbates the phenomenon of power shortage in cities, but also causes the load characteristics of the power system to deteriorate, which has an adverse impact on the normal operation of the power grid.

[0003] As a flexible resource that is numerous, adjustable and has a certain heat storage capacity, air-conditioning load has considerable regulation potential in participating in demand response. According to the received power adjustment instructions, the air-conditioning operation status is timely regulated by the Internet of Things regulation terminal, which can not only achieve the peak-shaving target, but also realize the reasonable development of the regulation potential of each air-conditioning equipment in the air-conditioning group. In recent years, most studies have participated in demand response and grid peak regulation in the form of aggregation of air-conditioning loads, and adjusted the power of the air-conditioning group through direct load control and other methods. Due to the long-term lack of popularity of user-side air-conditioning Internet of Things terminals, few people pay attention to the operation control process of each air-conditioning in the air-conditioning group during the demand response process, and rarely consider the impact of participating in demand response on each air-conditioning user. Therefore, how to efficiently control the air-conditioning load resources and achieve the required target power is a problem that the present invention needs to solve. Summary of the invention

[0004] The present invention aims to overcome the above-mentioned shortcomings of the prior art and provides a consistency control method and system for air conditioner groups to participate in demand response to achieve power adjustment targets, so as to solve the problem that the existing consistency control method takes the marginal cost of demand response as the target and ignores the difference in the rate of change of air conditioner temperature during the regulation process. The present invention uses indoor temperature and temperature change rate as consistency variables, links the scattered air conditioner load resources, and has the same impact on the comfort of each air conditioner user while achieving the power adjustment target, thereby more efficiently developing air conditioner load resources.

[0005] In order to achieve the above object, the present invention adopts the following technical solution:

[0006] The first aspect of the invention relates to a method for distributed collaborative power aggregation tracking control of massive air conditioners, comprising the following steps:

[0007] S1: Collect information about each air-conditioning device in the air-conditioning group and set the upper and lower limits of the adjustable temperature range T a,maxand T a,min , power range upper and lower limits P max and P min , measure the initial temperature T of each air-conditioning user in the air-conditioning group through the IoT terminal temperature sensor and other devices a (0), and calculate the initial indoor temperature change rate, and obtain the relevant thermodynamic parameters of the building where each air-conditioning user is located through parameter identification and other means; the load aggregation control platform of the air-conditioning group receives the target power issued by the dispatching agency, and compares it with the load aggregation control platform power at the start of regulation to obtain the expected adjustment power target P of the air-conditioning group. target , set the actual amount of power adjustment P of the air conditioning group real and the expected adjusted power target value P target The allowable error value of is ψ;

[0008] S2: Based on the principle of graph theory, the air conditioners are grouped into a multi-agent network through the air conditioner IoT terminal, and the network topology connected graph G is obtained, and the network adjacency matrix A, Laplace matrix L and row random matrix R are obtained. The air conditioner group N is divided according to the model or location of the air conditioner. i Divided into d subgroups, N i =N 1i ∪N 2i ∪…∪N di The agent with the highest algebraic connectivity in each subgroup is set as the leader, and the other agents are set as followers. The leader can obtain the power adjustment information of the air conditioning group issued by the load aggregation control platform. For details of the topology, see the attached Figure 1 ;

[0009] S3: Use the second-order ETP model to establish the corresponding relationship between air conditioning load and indoor temperature and solid temperature, and consider the rate of change of indoor temperature during the regulation process, and use indoor temperature and indoor temperature change rate as consistency variables. Construct a second-order integrator model, consider the influence of air conditioning load model and system information interaction, and use the distributed collaborative iteration protocol to iterate the indoor temperature and temperature change rate of each intelligent agent in the system at the next moment. The specific steps are as follows:

[0010] S3-1: The relationship between air conditioning load power and temperature is established based on the second-order ETP model. The specific formula is as follows:

[0011]

[0012] In the formula, C a and C m are the heat capacities of the gas and solid in the room respectively; R a and R m are the thermal resistances between indoor gas and outdoor, and between indoor solid and indoor gas, respectively; T a (t), Tm (t) and T o (t) are the indoor air temperature, indoor solid temperature and outdoor air temperature respectively; Q(t) is the cooling capacity of the air conditioner.

[0013] S3-2: From S3-1, we can get the corresponding relationship between indoor temperature, solid temperature and air conditioning power. In order to develop the control potential of each air conditioner in the group to the same extent and make the comfort of each air conditioner consistent, set the indoor temperature and the indoor temperature change rate as consistency variables. Derivate expression (1) and transform it into the indoor temperature T a The expression of (t) is shown in formula (2):

[0014]

[0015] S3-3: Remember T a (t) is x i (t), v i (t), the distributed iteration protocol described in S3 is shown in equations (3)-(4):

[0016]

[0017] In the formula, u i (t) is the control input; Δt is the control time interval.

[0018]

[0019] Where P target P is the power adjustment index of the air conditioning group; i is the power of the i-th air conditioner in the network; N di is the dth subgroup; α, β, ω are the system coupling strengths; γ is the competition intensity between subgroups; r ij are the elements in the row random matrix R.

[0020] S4: Determine the expected temperature T of each air-conditioning user at the next moment a Is (t+1) within the allowable temperature range [T a,min , T a,max ], if yes, execute S6, if no, execute S5;

[0021] S5: If the air conditioner user expects the temperature T at the next moment a (t+1) exceeds the upper limit T a,max , then T a (t+1) is set to T a,max , if T a (t+1) is less than the allowable lower limit T a,min , then T a(t+1) is set to T a,min ;

[0022] S6: Use the discretized second-order ETP model to derive and calculate the expected air conditioning power P at the current moment i (t), the specific steps are as follows:

[0023] S6-1: Discretize the second-order ETP model (2) and express it in the form of a state space equation, as shown in (5):

[0024]

[0025] S6-2: S3-S5 describe that the expected indoor temperature and temperature change rate of each air-conditioning user at the next moment can be obtained by calculating according to the distributed collaborative iterative protocol. The expected power P of the air-conditioning user at the current moment can be obtained by the discretized second-order ETP model calculation formula (5): i (t) and the indoor solid temperature T at the next moment m (t+1), as shown in formula (6):

[0026]

[0027] S7: Determine the expected power P of the air conditioner user at the current moment i (t) Whether it is within its adjustable power range [P min , P max ], if yes, execute S8, if no, execute S9;

[0028] S8: If P i (t) exceeds the allowable upper limit P max , then P i (k) is set to P max , if P i (k) is less than the allowable lower limit P min , then P i (t) is set to P min ;

[0029] S9: Calculate the expected power P of each air conditioner i (t), each air-conditioning user adjusts its working state according to the corresponding frequency. The calculation formula is as follows:

[0030]

[0031] In the formula, f i (t) is the operating frequency of the air conditioner; a, b and c are the coefficients of the relationship between the cooling capacity of the air conditioner and the operating frequency.

[0032] S9: The load aggregation control platform sums the power of the air conditioning groups under its jurisdiction to obtain the real-time power at the current moment, calculates the difference between the real-time power of the load aggregation control platform and the power at the start of regulation, and obtains the actual power adjustment amount P at the current moment. real , and the air conditioning group power adjustment index P target Compare and get the difference ΔP(t);

[0033] S10: Determine whether the agent has reached the power adjustment limit. If so, adjust the power according to the limit. If not, execute S11;

[0034] S11: Determine whether |ΔP(t)| is less than the allowable error value ψ. If so, stop the iteration and regard the air-conditioning group as having completed the power adjustment target and realizing power tracking. If not, the power tracking target has not been completed as of the current iteration step, and the time window moves forward one iteration step, that is, t=t+1, and return to S4.

[0035] The second aspect of the present invention relates to a massive air-conditioning distributed collaborative power aggregation tracking control system, which includes a load aggregation control platform, an air-conditioning group and an air-conditioning equipment monomer from the top layer to the bottom layer. Each air-conditioning equipment and its corresponding Internet of Things control terminal are regarded as an independent intelligent agent. These intelligent agents communicate and connect with each other, thereby building a multi-agent network; according to the geographical location of the intelligent agent or the type of equipment associated with it, the entire network is further divided into multiple subgroups, each subgroup includes a leading intelligent agent and multiple follower intelligent agents, each intelligent agent in the subgroup is interconnected with its adjacent intelligent agents, and the leading intelligent agent of each subgroup is interconnected with the leading intelligent agent of the adjacent subgroup; the load aggregation control platform is connected to the power grid dispatching mechanism and the leading intelligent agent of each subgroup in the intelligent agent network;

[0036] The load aggregation control platform receives the target power issued by the grid dispatching agency, monitors the real-time power on the interconnection line between it and the grid, calculates the power tracking deviation, and sends it to the leading intelligent agent of each subgroup in the network; the leading intelligent agent of each subgroup in the network has the ability to obtain the power adjustment deviation of the air-conditioning group from the load aggregation control platform, and is responsible for exchanging information with the leading intelligent agents of other subgroups. Through a series of calculations, it sets and plans the next air-conditioning operating frequency. The follower intelligent agent is mainly responsible for transmitting information with adjacent intelligent agents to coordinate the operating status of the entire subgroup.

[0037] The working principle of the present invention is: in view of the problem that the existing air conditioning load consistency control method takes the marginal cost of demand response as the target and ignores the difference in the air conditioning temperature change rate during the regulation process, a consistency control method and system for air conditioning groups to participate in demand response to achieve power adjustment targets is proposed. The system includes a load aggregation control platform, an air conditioning group and an air conditioning device monomer from top to bottom. The workflow includes:

[0038] 1. Each agent is connected to form a multi-agent network. The load aggregation control platform divides the multi-agent network into multiple subgroups, each of which contains a leading agent and multiple follower agents;

[0039] 2. The load aggregation control platform obtains the power adjustment target and sends it to the leading agent of each subgroup. The agents in each subgroup exchange indoor temperature, temperature change rate and power information with the agents connected to it. The information exchange between subgroups is completed by the leading agent.

[0040] 3. Each agent uses a distributed consensus algorithm to calculate the expected temperature and its rate of change for the next step based on its own temperature and temperature change rate as well as similar information received from neighboring agents;

[0041] 4. Each agent calculates the power and sets the air conditioning operating frequency based on the second-order ETP model according to the expected temperature in the next step.

[0042] The advantages of the present invention are as follows: compared with the existing consistency control strategy that takes the marginal cost of air-conditioning load participating in demand response as the target, the present invention takes the indoor temperature of air-conditioning users and the indoor temperature change rate as consistency targets, links scattered air-conditioning load resources, and takes into account the comfort of air-conditioning users while completing the power adjustment index, and has strong practicality; the present invention adopts a second-order consistency algorithm to calculate the indoor temperature and temperature change rate of air-conditioning users, and calculates the power of air-conditioning users based on a second-order ETP model, with low calculation and communication costs, high accuracy and strong reliability; the present invention can make the indoor temperature and temperature change rate of each air-conditioning user consistent, evenly develop and utilize air-conditioning load resources, reduce the impact of demand response on the comfort of air-conditioning users, and utilize air-conditioning load resources more efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is a structural diagram of the air conditioning group demand response of the present invention.

[0044] Figure 2 This is a flow chart of the distributed collaborative iterative algorithm for solving the air conditioning group of the present invention.

[0045] Figure 3 This is the agent network and grouping topology diagram of the present invention.

[0046] Figure 4 It is the indoor temperature curve of each air-conditioning user in the iterative process of the present invention.

[0047] Figure 5 It is the indoor temperature change rate curve of each air-conditioning user during the iteration process of the present invention.

[0048] Figure 6 It is the power curve of each air-conditioning user in the iterative process of the present invention.

[0049] Figure 7 It is the power adjustment curve of the air conditioning group in the iterative process of the present invention. DETAILED DESCRIPTION

[0050] The present invention will be further described below in conjunction with the accompanying drawings:

[0051] Example 1

[0052] In order to enable those skilled in the art to better understand the present invention, a method for distributed collaborative power aggregation tracking of massive air conditioners in this embodiment is described in detail. Figure 1-Figure 7 , including the following components:

[0053] 1. Simulation parameter settings

[0054] This simulation program is implemented in the MATLAB environment, and the power adjustment index of the air conditioning group is a given value to control each air conditioning device under the air conditioning group. Figure 2 As shown in the figure, the multi-agent topology and grouping of the air conditioning group are as follows: Figure 3 As shown in the figure, during the iteration process, the temperature, temperature change rate, power of each air-conditioning user and the power adjustment of the air-conditioning group are as follows: Figure 4 to Figure 7 Other simulation parameters are shown in Table 1.

[0055] Table 1 Air conditioning equipment simulation parameters

[0056]

[0057] 2. Analysis of simulation results under two different grouping topologies

[0058] The temperature curves of air-conditioning users under the two types of intelligent network grouping are as follows: Figure 4 As shown; the air conditioning user temperature change rate curve is as follows Figure 5 As shown; the air conditioning user power curve is as follows Figure 6 As shown; the power adjustment curve of the air conditioning group is as follows Figure 7 According to the simulation results, the following conclusions can be drawn:

[0059] 1) The proposed distributed collaborative power aggregation tracking method for air conditioner groups based on the second-order ETP model can better complete the power adjustment index of the air conditioner group and achieve the goal of demand response power tracking. The air conditioner users in each subgroup can each achieve consistent temperature and temperature change rate, and the indoor temperature difference of air conditioners in different subgroups is very small when iterative convergence is reached, and there is almost no difference between subgroups. Through group control, a large number of dispersed air conditioner load resources are developed more efficiently.

[0060] 2) The proposed distributed collaborative power aggregation tracking method for air-conditioning groups takes into account the temperature and temperature change rate of air-conditioning users during the demand response process, reduces the comfort difference brought to the users under the jurisdiction of the air-conditioning group by the participation of the air-conditioning group in demand response, avoids the drastic fluctuations in power of some air-conditioning users during the control process, and can distribute the power adjustment tasks to each air-conditioning user more fairly.

[0061] The present invention provides a consistency control method for air conditioner groups to participate in demand response and complete power adjustment targets, including: based on graph theory, each air conditioner is grouped into a multi-agent network through an air conditioner equipment Internet of Things terminal to obtain a network topology connectivity graph, an adjacency matrix, a Laplace matrix and a row random matrix; the air conditioner group is divided into multiple subgroups according to the model or location of the air conditioner, and the agent with the highest algebraic connectivity in each subgroup is set as the leader, and the remaining agents are set as followers; a second-order equivalent thermal parameter model is used to establish a corresponding relationship between the air conditioning load and the indoor air temperature and the solid temperature, and the rate of change of the indoor air temperature during the regulation process is considered, and the indoor temperature and the rate of change of the indoor air temperature are used as consistency variables; a second-order distributed collaborative iteration protocol is used to iterate the indoor temperature and the rate of change of the temperature of each agent in the system at the next moment; a discretized second-order equivalent thermal parameter model is used to derive and calculate the expected air conditioning power at the current moment, and the corresponding working frequency is obtained according to the expected air conditioning power calculation, and the air conditioner Internet of Things terminal performs frequency adjustment.

[0062] Example 2

[0063] This embodiment relates to a massive air conditioner distributed collaborative power aggregation tracking control system, such as Figure 1 , from the top layer to the bottom layer, it includes load aggregation control platform, air conditioner group and air conditioner equipment monomer. Each air conditioner and its corresponding IoT control terminal are regarded as an independent intelligent agent. These intelligent agents communicate with each other to build a multi-agent network, obtain the network topology connection graph G, obtain the network adjacency matrix A, Laplace matrix L and row random matrix R, and divide the air conditioner group N into groups according to the model or location of the air conditioner. i Divided into d subgroups, N i =N 1i ∪N 2i ∪…∪N di The agent with the highest algebraic connectivity in each subgroup is set as the leader, and the other agents are set as followers. The leader can obtain the power adjustment information of the air-conditioning group issued by the load aggregation control platform; each agent in the subgroup is connected to its adjacent agents, and the leading agent of each subgroup is connected to the leading agent of the adjacent subgroup; the load aggregation control platform is connected to the grid dispatching organization and the leading agent of each subgroup in the agent network.

[0064] The load aggregation control platform receives the target power issued by the grid dispatching agency, monitors the real-time power on the interconnection line between it and the grid, calculates the power tracking deviation, and sends it to the leading intelligent agent of each subgroup in the network; the leading intelligent agent of each subgroup in the network has the ability to obtain the power adjustment deviation of the air-conditioning group from the load aggregation control platform, and is responsible for exchanging information with the leading intelligent agents of other subgroups; the follower intelligent agent is mainly responsible for transmitting information with adjacent intelligent agents to coordinate the operating status of the entire subgroup.

[0065] The leading agent of each subgroup includes its own indoor temperature and temperature change rate at the current moment in the calculation process, integrates similar information transmitted from adjacent follower agents and other leading agents, and the air conditioning group power adjustment information provided by the load aggregation control platform, and uses a distributed collaborative iterative protocol to calculate and solve the expected value of the indoor temperature and temperature change rate at the next moment; the follower agent of each subgroup uses a distributed collaborative iterative protocol to calculate the expected value of the indoor temperature and temperature change rate at the next moment based on the indoor temperature and corresponding temperature change rate monitored by itself at the current moment, as well as similar information received from adjacent agents. Each agent uses a discretized second-order ETP model to derive and calculate the expected power and operating frequency at the current moment, and its IoT terminal performs the corresponding frequency adjustment action.

Claims

1. A method for distributed collaborative power aggregation tracking control of massive air conditioners, comprising the following steps: S1. Investigate the basic information such as the adjustable power range and energy consumption ratio of each air-conditioning device in the air-conditioning group, obtain the thermodynamic parameters of the room where each air-conditioning device is located through parameter identification and other means, and store the information in the Internet of Things control terminal of the corresponding intelligent body; S2, the load aggregation control platform receives the demand response target power issued by the grid dispatching agency, and monitors the real-time power on the interconnection line between it and the grid, and calculates the deviation between the target power and the real-time power of the load aggregation control platform, and uses it as the power adjustment index P of the air conditioning group. target Send it to the leading agent of each subgroup in the network; S3. The indoor temperature of the space where each air-conditioning device is located at the start of control is obtained by measuring through the Internet of Things control terminal, and the temperature change rate is calculated; S4. Based on the second-order ETP model, the corresponding relationship between the air-conditioning power load and the indoor temperature and solid temperature is established. At the same time, the rate of change of the indoor temperature during the regulation process is considered, and the indoor temperature and the rate of change of the indoor temperature are used as consistency variables; specifically, for the follower agent in the network, based on the indoor temperature and the corresponding temperature change rate monitored by itself at the current moment, and the same information received from the adjacent agent, a distributed collaborative iteration protocol is used for calculation to obtain the expected value of the indoor temperature and the temperature change rate at the next moment; for the leading agent of each subgroup in the network, during the calculation process, not only the indoor temperature and the temperature change rate at the current moment are included, but also the same information transmitted from the adjacent follower agent and other leading agents, as well as the air-conditioning group power adjustment information provided by the load aggregation control platform are integrated, and the expected value of the indoor temperature and the temperature change rate at the next moment is calculated and solved by the distributed collaborative iteration protocol; S5: Determine the expected indoor temperature T of each agent at the next moment calculated in step S4 a Is (t+1) within the allowed range [T a,min , T a,max ], if otherwise, first set T a (t+1) is set as the boundary value, if yes, execute S6 directly; S6, based on the expected indoor temperature T of the air conditioning agent at the next moment a (t+1), the second-order ETP model is used to derive and calculate the expected air conditioning power P at the current moment i (t), determine whether the expected air conditioning power is within the adjustable range of the intelligent agent, if not, first set P i (t) is set to the boundary value, if yes, execute S7 directly; S7. Calculate the expected power P of each air conditioning equipment i (t) and transmits the corresponding working frequency to the IoT terminal, and each intelligent agent adjusts its working state accordingly according to the frequency information; S8, the load aggregation control platform measures the real-time power on the interconnection line with the power grid at the current moment, calculates the difference between the real-time power of the load aggregation control platform and the power at the start of regulation, and obtains the actual power adjustment value P at the current moment. real , and the air conditioning group power adjustment index P target Compare and get the difference ΔP(t); S9, determine whether the agent reaches the power adjustment limit, if so, adjust the power according to the limit and stop iteration, if not, execute step S10; S10. Determine whether |ΔP(t)| is less than the allowable error value ψ. If so, the air-conditioning group is deemed to have completed the power adjustment target and achieved power tracking, and the iteration is stopped. If not, the power tracking target has not been completed as of the current iteration step, and the iteration is moved forward by one iteration step, i.e., t=t+1, and returns to step S4.

2. A method for distributed collaborative power aggregation tracking control of massive air conditioners according to claim 2, characterized in that: The air conditioning load model in step S4 is: In the formula, C a and C m are the heat capacities of the gas and solid in the room respectively; R a and R m are the thermal resistances between indoor gas and outdoor, and between indoor solid and indoor gas, respectively; T a (t), T m (t) and T o (t) are the indoor air temperature, indoor solid temperature and outdoor air temperature at time t respectively; Q(t) is the cooling capacity of the air conditioner at time t.

3. A method for distributed collaborative power aggregation tracking control of massive air conditioners according to claim 2, characterized in that: Indoor temperature T in step S4 a (t) can be expressed as: The control quantity corresponding to the distributed collaborative iteration protocol in step S4 is: The calculation formula for the predicted value of the indoor temperature and its change rate in step S4 is: Where P target P is the power adjustment index of the air conditioning group; i is the power of the i-th air conditioner in the network; N di is the dth subgroup; α, β, ω are the system coupling strengths; ε is the global traction correction coefficient; γ is the competition intensity between subgroups; r ij is an element in the row random matrix R, and a connected graph G is used to describe the multi-agent network structure of the air-conditioning group. According to the graph G, the corresponding degree matrix D and Laplace matrix L can be obtained, and the row random matrix R is constructed according to the Laplace matrix L.

4. A method for distributed collaborative power aggregation tracking control of massive air conditioners according to claim 1, characterized in that: The discretization calculation method of the second-order ETP model in step S6 is: The method for calculating the expected air conditioning power at the current moment in step S6 is as follows: Where, η is the air conditioning energy efficiency ratio; R a (i) R m (i) C a (i) and C m (i) are the air thermal resistance, solid thermal resistance, air heat capacity and solid heat capacity of the room where the i-th air conditioner is located.

5. The method for distributed collaborative power aggregation tracking control of massive air conditioners according to claim 1 is characterized in that: The air conditioner operating frequency calculation method in step S7 is: In the formula, f i (t) is the operating frequency of the air conditioner; a, b and c are the coefficients of the relationship between cooling capacity and operating frequency.

6. A system for implementing the method for distributed collaborative power aggregation tracking control of massive air conditioners according to claim 1, characterized in that: From the top layer to the bottom layer, it includes the load aggregation control platform, air conditioning group and air conditioning equipment monomer. Each air conditioning equipment and its corresponding Internet of Things control terminal are regarded as an independent intelligent agent. These intelligent agents communicate with each other to build a multi-agent network. According to the geographical location of the intelligent agent or the type of equipment it is associated with, the entire network is divided into multiple subgroups. Each subgroup contains a leading intelligent agent and multiple follower intelligent agents. Each intelligent agent in the subgroup is connected to its adjacent intelligent agents, and the leading intelligent agent of each subgroup is connected to the leading intelligent agent of the adjacent subgroup. The load aggregation control platform is connected to the power grid dispatching organization and the leading intelligent agent of each subgroup in the intelligent agent network. The load aggregation control platform receives the target power issued by the grid dispatching agency, monitors the real-time power on the interconnection line between it and the grid, calculates the power tracking deviation, and sends it to the leading intelligent agent of each subgroup in the network; the leading intelligent agent of each subgroup in the network has the ability to obtain the power adjustment deviation of the air-conditioning group from the load aggregation control platform, and is responsible for exchanging information with the leading intelligent agents of other subgroups. Through a series of calculations, it sets and plans the next air-conditioning operating frequency. The follower intelligent agent is mainly responsible for transmitting information with adjacent intelligent agents to coordinate the operating status of the entire subgroup.

Images