A collaborative control method for virtual energy storage in air-conditioning clusters based on multi-agent system

Abstract

This invention discloses a multi-agent system-based virtual energy storage collaborative control method for air conditioning clusters. By establishing N single-zone HVAC physical load models, this method monitors indoor CO2 concentrations in real time and uses a consensus algorithm to distribute and adjust damper positions, changing the ratio of mixed air to return air. Furthermore, the consensus algorithm also distributes and adjusts the air supply rate to ensure consistent ambient temperatures in the service areas of each HVAC cluster, keeping indoor CO2 concentrations within a comfortable range. Furthermore, the power of each HVAC load is collaboratively adjusted to ensure consistent load rates across all HVAC units, making the overall power of the HVAC cluster flexible and adjustable. This method achieves energy savings while maintaining good indoor air quality and a user-satisfied comfortable temperature, while also achieving operational efficiency in HVAC cluster control.

Description

Technical Field

[0001] The present invention relates to the technical field of air-conditioning control, and in particular to a method for collaborative control of virtual energy storage of an air-conditioning cluster based on a multi-agent system. Background Art

[0002] With the accelerating pace of modern urbanization in my country, the construction industry has also reached a certain stage of development, resulting in a dramatic increase in the number and scale of buildings. Driven by this development, an increasing number of buildings are equipped with heating, ventilation, and air conditioning (HVAC) systems, including commercial offices, residential homes, universities, and industrial plants. HVAC systems can modify the temperature environment within buildings to provide users with the desired comfort. Currently, most HVAC systems utilize multi-zone, single-dimensional load models that flexibly adjust indoor temperature. Little attention is paid to single-zone HVAC cluster control, which collaboratively considers temperature and indoor air quality (IAQ) load models. This makes it difficult to fully meet user needs. Furthermore, the energy consumed by HVAC system operation is enormous. HVAC accounts for 35% to 40% of my country's building energy consumption. Therefore, fully tapping the energy supply potential and coordinated control of HVAC energy use in buildings to effectively alleviate the pressure on energy supply caused by rapidly growing energy demand in buildings is a critical energy issue that needs to be addressed urgently.

[0003] In HVAC cluster systems, due to the insulation of building walls and other enclosing structures, the heat exchange process between indoor and outdoor spaces is slow, resulting in a certain amount of heat storage capacity, very similar to traditional energy storage devices such as electrochemical, battery, and electrical energy storage systems. If the HVAC cluster system is equivalent to a virtual energy storage device, the dynamic change in HVAC power output by adjusting indoor temperature and CO2 concentration can be viewed as the charging and discharging process of a traditional energy storage device, achieving heat release, storage, and fresh air storage, which helps improve energy utilization. Furthermore, for users, it can maximize their comfort in the building.

[0004] Among the existing technologies for HVAC cluster control today, most are the following three methods: rule-based methods, learning-based methods, and model-based methods. Rule-based methods use some simple rules (such as ON / OFF switch control) to simply control the indoor temperature. Learning-based methods directly use interactive information with the building's thermal environment to learn control strategies and can respond to multiple trained cases in real time, but the algorithm may lack the necessary generalization capabilities. Model-based methods need to consider modeling the HVAC thermal dynamics model affected by environmental factors. Model predictive control methods can reduce the energy cost of HVAC systems while ensuring user thermal comfort. These methods consider the future state and disturbances of the system when making decisions, but they rely on a clear model of the building's thermal dynamics.

[0005] The shortcomings of these existing technologies are that most methods only achieve flexible adjustment of indoor temperature in a single dimension, failing to consider multiple coupling factors. They only consider users' thermal comfort, but not their air quality requirements or the energy losses caused by inconsistent indoor temperatures across multiple zones. Furthermore, they fail to address user fairness issues caused by inconsistent virtual energy storage control contributions. Summary of the Invention

[0006] In order to solve at least one technical problem existing in the above background technology, the present invention provides an air-conditioning cluster virtual energy storage collaborative control method based on a multi-agent system.

[0007] To achieve the above object, the technical solution of the present invention is:

[0008] A method for collaborative control of virtual energy storage in an air conditioning cluster based on a multi-agent system, the method comprising:

[0009] Establish a physical load model for N single-zone HVAC systems to monitor indoor CO2 concentration in real time and use a consensus algorithm to adjust the damper position to change the ratio of mixed air to return air; N is a positive integer.

[0010] The air supply rate is distributedly adjusted through a consistency algorithm to ensure a consistent ambient temperature in the service area of ​​an HVAC cluster, thereby maintaining the indoor CO2 concentration within a comfortable range; the HVAC cluster includes N single-zone HVAC systems.

[0011] Furthermore, the multi-agent system-based air conditioning cluster virtual energy storage collaborative control method further includes:

[0012] The load power of each HVAC system is coordinated and adjusted to make the load rate of each HVAC system consistent, so that the overall power of the HVAC cluster can be flexibly adjusted.

[0013] Furthermore, the HVAC system includes an air handling unit (AHU) and a variable air volume (VAV) box.

[0014] Furthermore, the variable air volume box AHU includes a damper, a cooling pipe and a supply fan SAF. The damper is used to mix external fresh air with the air returned from the area, the cooling pipe is used to cool the mixed air, and the supply fan SAF is used to transport the mixed air to the variable air volume box VAV of the equivalent area.

[0015] Furthermore, the physical load model of the single-zone HVAC system is established in the following manner:

[0016] 1) The physical load model of the single-zone HVAC system includes a thermodynamic model and a CO2 dynamic concentration IAQ model, which are described by the following formulas (1) and (2), respectively:

[0017]

[0018]

[0019] Where T in formula (1) i，t is the indoor temperature of zone i in period t, as a state variable, m i，t is the air supply rate of zone i in time period t, as input, T o，t is the outdoor air temperature, q is the thermal disturbance, T s is the air temperature of the fan, C is the heat capacity, R is the thermal resistance, C a is the specific heat capacity of air; O in formula (2) i，t is the indoor CO2 concentration in zone i at time t, as a state variable, and O mix，t is the CO2 concentration of the mixed air at time t, where O mix With (1-β r )O out，t +β r O i，t Indicates that β r is the damper position in the AHU, v is the spatial volume of the single-zone HVAC service area, Z i is the number of people in the single-region spatial volume of zone i, σ is the per capita carbon dioxide generation rate, and ρ is the air density;

[0020] 2) Power generated by the cooling coil in the air handling unit AHU P c , which is expressed by the following equation (3):

[0021]

[0022] The mixed air temperature T m =β r T i+91-β r )T o，t , η c The efficiency coefficient of the cooling pipe, Cop is the performance coefficient of the cooler,

[0023] 3) The power generated by SAF is expressed by the following equation (4):

[0024] P s =μm i，t 3 (4)

[0025] Wherein, μ is the power consumption coefficient of the fan;

[0026] 4) The power of the HVAC service area load in a single zone is:

[0027] P d，i =P c +P s (5).

[0028] Furthermore, the constraints of the physical load model of the single-zone HVAC system include:

[0029] a) Select the appropriate comfort temperature for indoor occupants, in HVAC i The temperature range of the service area is described as:

[0030]

[0031] in and represent the minimum and maximum indoor comfort temperatures respectively;

[0032] b) For each HVAC service area, adjust the VAV box to change the air supply rate to obtain HVAC i Constraint value of gas supply rate in the service area:

[0033]

[0034] in and represent the minimum and maximum air supply rates in zone i respectively;

[0035] c) Since high CO2 concentration levels in an area are harmful to the health and productivity of occupants, the level should be controlled below a threshold value, which is expressed by the following formula (8):

[0036]

[0037] represents the CO2 concentration threshold of region i;

[0038] d) The opening of the damper in AHU is limited. Set β i，t Dynamically adjust within the following ranges:

[0039]

[0040] in Indicates the minimum opening of the damper position. Indicates the maximum opening of the damper position; when β i，t = 1, the mixed air contains only the air returned from the area, and β i，t =0, the mixed air contains only external fresh air.

[0041] Furthermore, the distributed adjustment of the air supply rate through the consistency algorithm to ensure a consistent ambient temperature in the service area of ​​the HVAC cluster and to ensure that the indoor CO2 concentration is maintained within a comfortable range includes:

[0042] 1) There must be at least one communication link between any two HVAC systems i and j to establish a strongly connected distributed communication network. The strongly connected distributed communication network is represented by the Laplace matrix L, which is defined as L = DA, where A is the adjacency matrix. The adjacency matrix A is an n*n matrix. The elements of each row of A are summed up to get N numbers, and then they are placed on the diagonal to form an n*n diagonal matrix, denoted as the degree matrix D; the off-diagonal elements a ij Represents the number of connections from node i to node j, and defines the L matrix:

[0043] L ii =-a ij ,i≠j (11)

[0044]

[0045] 2) In a distributed environment, P cannot be directly obtained d，i The distributed average consistency algorithm (13) is used to realize average power sharing. The average power process is completed in each HVAC to obtain the average power of the service area load of the single-area HVAC system in a distributed environment.

[0046]

[0047] 3) Determine the Laplace matrix L according to the number of HVAC system service areas. Let each HVAC service area system be a node on the distributed strongly connected communication network, and let the Laplace matrix L be multiplied by the indoor temperature T of different HVAC systems. j and single-zone HVAC service area load power P d，j And sum, in addition, P refand total The difference between them is used to generate a control signal, and the difference control signal is weighted to determine the air supply rate of the HVAC system. In this way, the distributed consistency control algorithm formula (14) is obtained, which is expressed as:

[0048]

[0049] 4) After discretizing formula (14) through the proportional integral link, the expression is:

[0050]

[0051] Where τ is the time slot length, in seconds;

[0052] 6) Determine the target factor weight coefficient K i , find the appropriate weight coefficient by training data of HVAC system cluster control; In formula (14), not all HVAC have known P ref Only a few HVAC are assigned P ref The node corresponding to this HVAC is called the leader node, and the nodes corresponding to the other HVAC are called follower nodes. The K3 of this node is equal to 0.

[0053] Furthermore, the multi-agent system-based air conditioning cluster virtual energy storage collaborative control method further includes: collecting CO2 concentration data in the equivalent area. i，t , β i，t Make dynamic adjustments.

[0054] Furthermore, the CO2 concentration data in the acquisition equivalent area is i，t , β i，t Dynamic adjustments include:

[0055] 1) Let the highest CO2 user comfort concentration in the equivalent area be The lowest CO2 household comfort concentration is The CO2 concentration in this range is a comfortable value. On the premise of satisfying thermal comfort, the CO2 concentration is maintained in this range. It can be seen from the indoor temperature adjustment step that this distributed communication network is strongly connected. The specific value of the Laplace L matrix is ​​obtained in the same way as in step 1) of adjusting the indoor temperature in claim 6, and is defined using formulas (11) and (12).

[0056] 2) Distributed acquisition of CO2 concentration in the HVAC service area, multiplying the Laplace matrix L by the indoor CO2 concentration of different HVAC and the regional HVAC service area load power P d，j And sum, in addition, P ref and total The difference control signal is obtained by subtracting the difference between the two, and the difference control signal is multiplied by the weight coefficient K i Later, it was determined that the HVAC system The expression:

[0057]

[0058] 3) After discretizing formula (14) through the proportional integral link, the expression is:

[0059]

[0060] Where τ is the time slot length, in seconds;

[0061] 4) Change the weight coefficient K of the consensus algorithm (14) and (16) i The proportion of K1 can affect the degree of consistency of the indoor temperature, CO2 concentration and load rate of the HVAC cluster system. When K1 is increased / decreased, the indoor temperature T in the HVAC cluster system can be changed. i，t When K4 is increased / decreased, the degree of consistency of the indoor CO2 concentration in the HVAC cluster system can be changed; when K2, K3, K5 and K6 are increased / decreased, the degree of consistency of the load rate of the HVAC cluster system can be increased / decreased; the temperature area and load rate η of each HVAC service area are p Adjust to consistency to achieve the effect of virtual energy storage while maximally satisfying IAQ.

[0062] Compared with the prior art, the present invention has the following beneficial effects:

[0063] A multi-agent system-based virtual energy storage collaborative control method for air conditioning clusters uses N single-zone HVAC physical load models to monitor indoor CO2 concentrations in real time and uses a consensus algorithm to distribute and adjust damper positions, changing the ratio of mixed air to return air. Furthermore, the consensus algorithm also distributes and adjusts the air supply rate to ensure consistent ambient temperatures in the service areas of each HVAC cluster, keeping indoor CO2 concentrations within a comfortable range. Furthermore, the power of each HVAC load is collaboratively adjusted to ensure consistent load rates across all HVAC units, making the overall power of the HVAC cluster flexible and adjustable. This approach achieves energy savings while maintaining good indoor air quality and a comfortable temperature for users, while also achieving economical operation of HVAC cluster control. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 This is a schematic diagram of the physical components of a single-zone HVAC system;

[0065] Figure 2This is a block diagram of HVAC cluster virtual energy storage collaborative control based on a multi-agent system;

[0066] Figure 3 A schematic diagram of the physical and information systems of the HVAC cluster in the verification example;

[0067] Figure 4 The temperature, CO2 concentration and power diagram of the HVAC service area under the consistency algorithm control;

[0068] Figure 5 Graphs of HVAC service area temperature, CO2 concentration, and power at different power ratings;

[0069] Figure 6 are HVAC service area temperature, carbon dioxide concentration, and power with different weight coefficients. DETAILED DESCRIPTION

[0070] Example:

[0071] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0072] The method for collaborative control of virtual energy storage in an air conditioning cluster based on a multi-agent system provided in this embodiment specifically includes the following steps:

[0073] First, it is necessary to establish a physical load model of a single-zone HVAC system in HVAC cluster control, and regard this single-zone load model as the service area in HVAC cluster control. In public buildings, indoor temperature is an important indicator for measuring user comfort. In addition to indoor temperature, occupants are also concerned about indoor air quality (IAQ), which is usually represented by the CO2 concentration level. A single-zone HVAC system usually consists of an air handling unit (AHU) and a variable air volume box (VAV). The AHU mainly consists of dampers, cooling pipes and supply air fans (SAF). The dampers can mix the fresh air outside with the air returned from the area, the cooling pipes cool the mixed air, and the SAF can transport the mixed air to the VAV box of the equivalent area. By changing the SAF power and damper position to change the air supply rate, adjusting the cooling pipe power to affect the mixed air temperature, and thus achieving control of the HVAC indoor temperature and indoor CO2, such as Figure 1 Currently, most building models are data-driven models and high-precision physical models, which make it difficult to ensure the accuracy of simulated temperature predictions outside of training data. Therefore, this application adopts the RC analog network physical model modeling method for modeling, and considers the power regulation constraints of the actual HVAC load cluster. The specific steps are as follows:

[0074] 1) The physical load model of the single-zone HVAC system includes a thermodynamic model and a CO2 dynamic concentration (IAQ) model, which are described by the following formulas (1) and (2), respectively:

[0075]

[0076]

[0077] Where T in formula (1) i，t is the indoor temperature of zone i in period t, as a state variable, m i，t is the air supply rate of zone i in time period t, as input, T o，t is the outdoor air temperature, q is the thermal disturbance, T s is the air temperature of the fan, C is the heat capacity, R is the thermal resistance, C a is the specific heat capacity of air. i，t is the indoor CO2 concentration in zone i at time t, as a state variable, and O mix，t is the CO2 concentration of the mixed air at time t, where O mix With (1-β r )O out，t +β r O i，t Indicates that β r is the damper position in the AHU, v is the spatial volume of the single-zone HVAC service area, Z i is the number of people in the single-region spatial volume of zone i, σ is the per capita carbon dioxide generation rate, and ρ is the air density.

[0078] 2) Power generated by the cooling coil in the air handling unit AHU P c , which can be expressed by the following equation (3):

[0079]

[0080] The mixed air temperature T m =β r T i +(1-β r )T o，t , η c The efficiency coefficient of the cooling pipe, Cop is the performance coefficient of the cooler,

[0081] 3) The power generated by SAF is expressed by the following equation (4):

[0082] P s =μm i，t 3 (4)

[0083] Among them, μ is the power consumption coefficient of the fan.

[0084] 4) The power of the HVAC service area load in a single zone is:

[0085] P d，i =P c +P s (5)

[0086] The constraints are as follows:

[0087] a) Select the appropriate comfort temperature for indoor occupants, in HVAC i The temperature range of the service area is described as:

[0088] in and represent the minimum and maximum indoor comfort temperatures respectively.

[0089] b) For each HVAC service area, adjust the VAV box to change the air supply rate, and you can get the HVAC i Constraint value of gas supply rate in the service area:

[0090]

[0091] in and Represent the minimum and maximum air supply rates in zone i respectively.

[0092] c) Since high CO2 concentration levels in an area are harmful to the health and productivity of occupants, the level should be controlled below a threshold value, which is expressed by the following formula (8):

[0093]

[0094] Indicates the CO2 concentration threshold of area i.

[0095] d) The opening of the damper in AHU is limited. Set β i，t Dynamically adjust within the following ranges:

[0096]

[0097] in Indicates the minimum opening of the damper position. Indicates the maximum opening of the damper position. i，t = 1, the mixed air contains only the air returned from the area, and β i，t =0, the mixed air contains only external fresh air.

[0098] Secondly, ensure that the service area temperature and CO2 concentration are maintained within a reasonable range, the temperature range of each HVAC service area is consistent, and the load rate of each HVAC load is η p The overall load of the HVAC cluster is flexible and adjustable. p It is the ratio of the actual power of each HVAC to the rated (maximum) power of the HVAC.

[0099] Given the rated power of the HVAC cluster P ref Under the premise that the consensus algorithm can eventually converge, it is necessary to have a strong communication network. That is, every node can reach every other node through the communication line. In order to meet the requirements of strong connectivity, there can be no isolated nodes in the system. There must be at least one communication line between any node and other nodes. Using the consensus distributed algorithm to adjust the indoor temperature T i，t The specific steps are as follows:

[0100] 1) There must be at least one communication link between any two HVAC systems i and j to establish a strongly connected distributed communication network ( Figure 2 The strongly connected distributed communication network signal in ② can be represented by the Laplace matrix L, which is defined as L = DA, where A is the adjacency matrix. The adjacency matrix A is an n*n matrix. Add up the elements of each row of A to get N numbers, and then put them on the diagonal (all other places are zero) to form an n*n diagonal matrix, which is recorded as the degree matrix D. The non-diagonal elements a ij Represents the number of connections from node i to node j. For a simple finite connectivity matrix, the adjacency matrix is ​​a (0,1) matrix with all diagonal elements being 0 and the other elements being 1. The L matrix is ​​defined as follows:

[0101] L ii =-a ij ,i≠j (11)

[0102]

[0103] 2) In a distributed environment, P cannot be directly obtained d，i The distributed average consistency algorithm (13) is used to realize average power sharing. The average power process is completed in each HVAC to obtain the average power of the HVAC service area load in a distributed environment.

[0104]

[0105] 3) Determine the Laplace matrix L according to the number of HVAC service areas. Let each HVAC service area system be a node on the distributed strongly connected communication network, and let the Laplace matrix L be multiplied by the indoor temperature T of different HVAC j and single-zone HVAC service area load power P d，j And sum, in addition, P ref and total The difference between them is used to generate a control signal, and this difference control signal is weighted④ to determine the air supply rate of the HVAC system. In this way, the distributed consistency control algorithm formula (14) is obtained, which is expressed as:

[0106]

[0107] 4) After discretizing formula (14) through the proportional integral link ⑤, the expression is:

[0108]

[0109] Where τ is the time slot length, in seconds.

[0110] 5) Determine the target factor weight coefficient K i ,like Figure 2 ④, find the appropriate weight coefficient by training data of HVAC system cluster control. In formula (14), not all HVAC have known P ref Only a few HVAC are assigned P ref The node corresponding to this HVAC is called the leader node, and the nodes corresponding to the other HVAC are called follower nodes. The K3 of this node is equal to 0.

[0111] In addition, from formula (2), we can see that O i，t Air door position β i，t and gas supply rate m i，t The effect of the consistency algorithm (13) is used to adjust the air supply rate m i，t It is impossible to ensure that the temperature and CO2 concentration in the service area are maintained within a reasonable range, so it is also necessary to collect CO2 concentration data in the equivalent area. i，t , β i，t Make dynamic adjustments, such as Figure 2 In order to meet the user's environmental comfort needs and reduce unnecessary energy loss, a local damper β based on CO2 concentration detection is used. i，t The core of the dynamic adjustment method is also to use the distributed consensus algorithm (13) and (14). The steps are similar to adjusting the indoor temperature T i，t Similar, as follows:

[0112] 1) Let the highest CO2 user comfort concentration in the equivalent area be The lowest CO2 household comfort concentration is The CO2 concentration in this range is the comfortable value. Maintaining the CO2 concentration within this range ensures thermal comfort. As can be seen from the indoor temperature adjustment steps, this distributed communication network is strongly connected. Obtaining the specific value of the Laplace L matrix is ​​the same as in step 1) of the indoor temperature adjustment step, defined using equations (11) and (12).

[0113] 2) Distributed acquisition of CO2 concentration in the HVAC service area, multiplying the Laplace matrix L by the indoor CO2 concentration of different HVAC and the regional HVAC service area load power P d，j And sum, in addition, P ref and total The difference control signal is obtained by subtracting the difference between the two, and the difference control signal is multiplied by the weight coefficient K i Later, it was determined that the HVAC system The expression:

[0114]

[0115] 3) After discretizing formula (14) through the proportional integral link ⑥, the expression is:

[0116]

[0117] Where τ is the time slot length, in seconds.

[0118] 4) Change the weight coefficient K of the consensus algorithm (14) and (16) i The proportion of K1 can affect the degree of consistency of the indoor temperature, CO2 concentration and load rate of the HVAC cluster system. When K1 is increased / decreased, the indoor temperature T in the HVAC cluster system can be changed. i，t When K4 is increased or decreased, the degree of consistency of the indoor CO2 concentration in the HVAC cluster system can be changed; when K2, K3, K5 and K6 are increased or decreased, the degree of consistency of the load rate of the HVAC cluster system can be increased or decreased. Due to the conflict of consistency goals, it is impossible to make each target value consistent. Therefore, under the premise of ensuring that the CO2 concentration is maintained in a reasonable range, the temperature zone and load rate η of each HVAC service area are adjusted. p Adjust to consistency to achieve the effect of virtual energy storage while maximally satisfying IAQ.

[0119] The following is a further verification of this method using an application scenario example:

[0120] A small commercial building complex in Shenzhen is selected. The season is summer. The complex has five identical small commercial buildings, collectively referred to as the HVAC service area. The space volume of each building is about 240m3 and is equipped with one HVAC. Thus, a distributed HVAC cluster system with strong connectivity communication network is established. Figure 3 The main system parameters of each small commercial building are shown in Table 1.

[0121] Table 1 HVAC-commercial building system parameters

[0122]

[0123]

[0124] Table 2 Initial parameters of HVAC cluster system

[0125]

[0126] Matlab2020b software was used to simulate this HVAC cluster system. First, the parameters in Table 1 and Table 2 were used to establish the thermodynamic model and IAQ model of this HVAC-commercial building system to further observe the dynamic changes in indoor temperature and indoor CO2 concentration in the HVAC service area. According to China's "Indoor Air Quality Standard" GB / T 18883-2020, the indoor CO2 concentration shall not exceed 1000ppm. Under normal circumstances, the CO2 concentration in nature or outdoors is about 400ppm, the CO2 concentration in an unmanned indoor environment is 500-600ppm (no fresh air entering), and a good indoor comfortable environment is 400-800ppm. People will feel dull and mentally fatigued in an environment exceeding 800ppm. Taking this as a standard, as a reference value for the CO2 concentration comfort in the HVAC service area, the highest CO2 user comfort concentration Set to 800ppm, the lowest CO2 user comfort concentration Set to 400ppm. According to China Meteorological News, when the ambient temperature is between 18℃ and 25℃ and the relative humidity is between 40% and 70%, the human body feels most comfortable. Therefore, the user thermal comfort range is set to 18℃~25℃.

[0127] When the HVAC cluster system rated power P ref =1900W, the weight coefficients are K1 = 0.38, K2 = 0.012, K3 = 0.001, K4 = 2×10 -8 K5=3×10 -6 K6=1×10 -7 When the indoor temperature, indoor CO2 concentration and power of the HVAC cluster system are obtained, Figure 4As shown in the figure, it can be seen that before 20000s, the gas supply rate m is continuously adjusted under the control of consistency algorithms (14) and (17). i，t and air door position β i，t , let the indoor temperature T i，t and power P d，i The difference signal keeps tracking Reference value, tends to be consistent after reaching 20000s. The indoor temperature after consistency is T i，t The highest user comfort temperature is around 21℃, which is within the user's comfort range. The difference between the two can be used as a source of virtual energy storage. The power of each HVAC cluster converges to 380W, that is, the load rate of each HVAC cluster is η p Consistent performance allows for flexible and unified allocation of the HVAC cluster system's overall load, combining real-time electricity prices, user thermal comfort, and demand-side response, ensuring consistent charging and discharging of virtual energy storage. Furthermore, indoor CO2 concentrations within the 400-800 ppm range provide a comfortable indoor environment and meet IAQ requirements.

[0128] When the HVAC cluster system rated power is P ref =1900W is changed to 2500W, and other parameters remain unchanged. The P in the consistency algorithm (14) and (17) ref and total The difference between the control signal and the control effect is enhanced, the overall power of the HVAC cluster system is increased, and the indoor cooling strength is enhanced. Figure 5 It can be seen that the power of a single HVAC unit P d，i Continuously track average power Reference value, from formula (3) and (4), we can know that the air supply rate m is increased i，t , so that the indoor temperature T i，t The temperature dropped from 21℃ to about 20℃, and the indoor CO2 concentration decreased overall. i，t and CO2 concentration are both within the user comfort range, and the load rate η p This shows that the HVAC cluster system can adjust the rated power P according to actual needs without violating the thermal comfort and IAQ. ref , making the HVAC load cluster power flexible and adjustable, thereby improving economic value.

[0129] Adjusting the weight coefficients in the consistency algorithms (14) and (17) can change the coordination speed of the indoor temperature and load rate of the HVAC cluster system. When K1 = 0.38 becomes K1 = 0.95, K2 = 0.012 becomes K2 = 0.048, K3 = 0.001 becomes K3 = 0.002, and K6 = 1 × 10 -7 becomes K6=2×10-7 K5=3×10 -6 becomes K5=9×10 -6 , K4 value remains unchanged, observe the changes in HVAC service area temperature, carbon dioxide concentration and power, such as Figure 6 Compared to Figure 4 , indoor temperature T i，t and load factor η p The coordination speed of CO2 concentration is increased from 25,000s to 15,000s, which shortens the time by two-fifths and increases the response speed of HVAC cluster system. Changing the weight coefficient K4 can also change the coordination speed of CO2 concentration, but since CO2 concentration is closely related to indoor temperature T i，t and load factor η p , and the target values ​​for the three cannot all converge. In practice, maintaining the CO2 concentration within a comfortable range is sufficient, without compromising the effectiveness of virtual energy storage. On the other hand, when adjusting the weight coefficients, it is important to consider that excessive focus on improving convergence speed may lead to overshoot. A balance should be struck between convergence time and overshoot.

[0130] In summary, the present invention has the following technical advantages compared with the prior art:

[0131] (1) Theoretical innovation: The present invention applies consistent distributed regulation and control technology to the coordinated control of indoor temperature, CO2 concentration and load rate of the HVAC cluster of an intelligent building. It also uses target factor weight coefficient adjustment, which is beneficial for the HVAC cluster system to obtain faster response speed and accelerate the virtual energy storage process.

[0132] (2) Social and economic benefits: The method of distributed air-conditioning cluster virtual energy storage collaborative control system based on multi-agent system is a scheduling strategy to balance the energy of the power system. While maintaining stable system operation, it also improves the economy and environmental protection of the microgrid system. In the building microgrid, a virtual energy storage system model is established using the thermal storage characteristics of the temperature control load. Under the premise that the indoor temperature does not affect the user's comfort and ensures a certain air quality, the temperature zone and load rate η of each HVAC service area are adjusted. p Adjusting them to a consistent level and maximizing the goal of virtual energy storage will help power supply companies manage energy, formulate better measures based on their laws, and bring good social and economic benefits.

[0133] The above embodiments are intended only to illustrate the technical concepts and features of the present invention. Their purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. They are not intended to limit the scope of protection of the present invention. Any equivalent changes or modifications made based on the essence of the present invention are intended to be covered by the scope of protection of the present invention.

Claims

1. A method for collaborative control of virtual energy storage in an air conditioning cluster based on a multi-agent system, characterized in that: The method comprises: Establish a physical load model for N single-zone HVAC systems to monitor indoor CO2 concentration in real time and use a consensus algorithm to adjust the damper position to change the ratio of mixed air to return air; N is a positive integer. A consistent algorithm is used to adjust the air supply rate in a distributed manner to ensure consistent ambient temperature across the HVAC cluster's service area and maintain indoor CO2 concentration within a comfortable range. This includes: 1) There must be at least one communication link between any two HVAC systems i and j to establish a strongly connected distributed communication network. The strongly connected distributed communication network is represented by the Laplace matrix L, which is defined as L = DA, where A is the adjacency matrix. The adjacency matrix A is an n*n matrix. The elements of each row of A are summed up to get N numbers, and then they are placed on the diagonal to form an n*n diagonal matrix, denoted as the degree matrix D; the off-diagonal elements a ij Represents the number of connections from node i to node j, and defines the L matrix: L ii =-a ij ,i≠j (11) 2) In a distributed environment, P cannot be directly obtained d,i The distributed average consistency algorithm (13) is used to realize average power sharing. The average power process is completed in each HVAC to obtain the average power of the service area load of the single-area HVAC system in a distributed environment. 3) Determine the Laplace matrix L according to the number of HVAC system service areas. Let each HVAC service area system be a node on the distributed strongly connected communication network, and let the Laplace matrix L be multiplied by the indoor temperature T of different HVAC systems. j and single-zone HVAC service area load power P d,j And sum, in addition, P ref and total The difference between them is used to generate a control signal, and the difference control signal is weighted to determine the air supply rate of the HVAC system. In this way, the distributed consistency control algorithm formula (14) is obtained, which is expressed as: 4) After discretizing formula (14) through the proportional integral link, the expression is: Where τ is the time slot length, in seconds; 5) Determine the target factor weight coefficient K i , find the appropriate weight coefficient by training data of HVAC system cluster control; In formula (14), not all HVAC have known P ref Only a few HVAC are assigned P ref The node corresponding to this HVAC is called the leader node, and the nodes corresponding to the other HVAC are called follower nodes. The K3 of this node is equal to 0; The HVAC cluster includes N single-zone HVAC systems.

2. The air conditioning cluster virtual energy storage collaborative control method based on a multi-agent system according to claim 1, characterized in that: The method further comprises: The load power of each HVAC system is coordinated and adjusted to make the load rate of each HVAC system consistent, so that the overall power of the HVAC cluster can be flexibly adjusted.

3. The air conditioning cluster virtual energy storage collaborative control method based on a multi-agent system according to claim 1 or 2, characterized in that: The HVAC system includes an air handling unit (AHU) and a variable air volume (VAV) box.

4. The air conditioning cluster virtual energy storage collaborative control method based on a multi-agent system according to claim 3, characterized in that: The variable air volume box VAV includes a damper, a cooling pipe and a supply fan SAF; the damper is used to mix the external fresh air with the air returned from the area, the cooling pipe is used to cool the mixed air, and the supply fan SAF is used to transport the mixed air to the variable air volume box VAV of the equivalent area.

5. The air conditioning cluster virtual energy storage collaborative control method based on a multi-agent system according to claim 4, characterized in that: The physical load model of the single-zone HVAC system is established as follows: 1) The physical load model of the single-zone HVAC system includes a thermodynamic model and a CO2 dynamic concentration IAQ model, which are described by the following formulas (1) and (2), respectively: Where T in formula (1) i,t is the indoor temperature of zone i in period t, as a state variable, m i,t is the air supply rate of zone i in time period t, as input, T o,t is the outdoor air temperature, q is the thermal disturbance, T s is the air temperature of the fan, C is the heat capacity, R is the thermal resistance, C a is the specific heat capacity of air; O in formula (2) i,t is the indoor CO2 concentration in zone i at time t, as a state variable, and O mix,t is the CO2 concentration of the mixed air at time t, where O mix,t With (1-β r )O out,t +β r O i,t Indicates that β r is the damper position in the AHU, v is the spatial volume of the single-zone HVAC service area, Z i is the number of people in the single-region spatial volume of zone i, σ is the per capita carbon dioxide generation rate, and ρ is the air density; 2) Power generated by the cooling coil in the air handling unit AHU P c , which is expressed by the following equation (3): The mixed air temperature T m =β r T i,t +(1-β r )T o,t , η c The efficiency coefficient of the cooling pipe, Cop is the performance coefficient of the cooler, 3) The power generated by SAF is expressed by the following equation (4): P s =μm i,t 3 (4) Wherein, μ is the power consumption coefficient of the fan; 4) The power of the HVAC service area load in a single zone is:

6. The air conditioning cluster virtual energy storage collaborative control method based on a multi-agent system according to claim 5, characterized in that: The constraints of the physical load model for the single-zone HVAC system include: a) Select the appropriate comfort temperature for indoor occupants, in HVAC i The temperature range of the service area is described as: in and represent the minimum and maximum indoor comfort temperatures respectively; b) For each HVAC service area, adjust the VAV box to change the air supply rate to obtain HVAC i Constraint value of gas supply rate in the service area: in and represent the minimum and maximum air supply rates in zone i respectively; c) Since high CO2 concentration levels in an area are harmful to the health and productivity of occupants, the level should be controlled below a threshold value, which is expressed by the following formula (8): represents the CO2 concentration threshold of region i; d) The opening of the damper in AHU is limited. Set β i,t Dynamically adjust within the following ranges: in Indicates the minimum opening of the damper position. Indicates the maximum opening of the damper position; when β i,t = 1, the mixed air contains only the air returned from the area, and β i,t =0, the mixed air contains only external fresh air.

7. The air conditioning cluster virtual energy storage collaborative control method based on a multi-agent system according to claim 6, characterized in that: Also includes: Collect CO2 concentration data in the equivalent area i,t , β i,t Make dynamic adjustments.

8. The air conditioning cluster virtual energy storage collaborative control method based on a multi-agent system according to claim 7, characterized in that: The CO2 concentration data in the acquisition equivalent area is i,t , β i,t Dynamic adjustments include: 1) Let the highest CO2 user comfort concentration in the equivalent area be The lowest CO2 household comfort concentration is The CO2 concentration in this range is a comfortable value. Under the premise of meeting thermal comfort, the CO2 concentration is maintained in this range. From the steps of indoor temperature adjustment, it can be seen that this distributed communication network is strongly connected. The specific value of the Laplace L matrix is ​​obtained and defined using formulas (11) and (12). 2) Distributed acquisition of CO2 concentration in the HVAC service area, multiplying the Laplace matrix L by the indoor CO2 concentration of different HVAC and the regional HVAC service area load power P d,j And sum, in addition, P ref and total The difference control signal is obtained by subtracting the difference between the two, and the difference control signal is multiplied by the weight coefficient K i Later, it was determined that the HVAC system The expression: 3) After discretizing formula (16) through the proportional integral link, the expression is: Where τ is the time slot length, in seconds; 4) Change the weight coefficient K of the consensus algorithm (14) and (16) i The proportion of K1 can affect the degree of consistency of the indoor temperature, CO2 concentration and load rate of the HVAC cluster system. When K1 is increased / decreased, the indoor temperature T in the HVAC cluster system can be changed. i,t When K4 is increased / decreased, the degree of consistency of the indoor CO2 concentration in the HVAC cluster system can be changed; when K2, K3, K5 and K6 are increased / decreased, the degree of consistency of the load rate of the HVAC cluster system can be increased / decreased; the temperature area and load rate η of each HVAC service area are p Adjust to consistency to achieve the effect of virtual energy storage while maximally satisfying IAQ.

Images