Air conditioner load cluster control and scheduling optimization method

Abstract

The invention discloses an air conditioner load cluster control and scheduling optimization method, which comprises the following steps: constructing an air conditioner load state queue model according to an equivalent thermal parameter model of a single air conditioner; according to the influence of temperature control on the air conditioner aggregation power, a dynamic response model of the air conditioner load group state queue model is constructed; according to the power characteristics of the dynamic response model and the comfort degree requirements of the users, scheduling optimization of the aggregation air conditioner load participating in the demand response process is completed, starting from improving the control precision of the aggregation power of the air conditioner load, the stabilization of the rebound load and the consistency of the comfort degree of the controlled air conditioner users are comprehensively considered, and the control precision of the aggregation power of the air conditioner load is improved. And a scheduling optimization strategy allowing the air conditioner users to participate in demand response in turns for multiple times is provided.

Description

Technical Field

[0001] The present invention relates to the field of air conditioning load control, and in particular to an air conditioning load cluster control and scheduling optimization method. Background Art

[0002] With the large-scale grid connection of intermittent renewable energy (such as wind power, solar energy) and new grid loads (such as electric vehicles), the contradiction between supply and demand of the power system has become more prominent, and its safe and stable operation faces huge challenges. Under the traditional grid operation mode, it often costs a lot to adjust the output of the supply-side generator sets. Therefore, it has higher application value to guide users to participate in demand response and tap the adjustable potential of the load side to alleviate the power supply pressure of the grid. Among the various flexible loads that can interact with the power grid in both directions, air-conditioning loads have the characteristics of thermal energy storage, flexible scheduling, large user scale, fast response speed, strong controllability, etc., which can provide greater adjustment potential without affecting or less affecting the user experience. In recent years, the rapid development of two-way communication technology and advanced measurement systems of smart grids has provided technical support for monitoring and controlling user-side loads, and is also the equipment basis for implementing demand response operations.

[0003] Implementing reasonable control strategies for massive air-conditioning loads can reduce the peak load pressure of the power system, ensure the safe and stable operation of the system, and have good social and economic benefits. Research on air-conditioning loads participating in demand response mainly includes two aspects: control method and dispatch optimization. The former focuses on the study of refined control methods for aggregated air conditioners or single air conditioners, and is committed to suppressing load fluctuations and load rebounds; the latter focuses on evaluating the response potential based on the aggregation modeling of air-conditioning loads, and selecting appropriate dispatch strategies to maximize social benefits. Direct load control (DLC) is an incentive-based demand response regulation method in the power market, and is also the most important control method for air-conditioning loads to participate in demand response. When implementing DLC ​​projects, users can be remotely regulated through strategies such as switch control and temperature control. Switch control has a fast response speed and great short-term regulation potential, but it is prone to large power fluctuations during its control process. Temperature control can better guarantee user experience and meet regulation needs on a long time scale, so it is used in a large number of studies.

[0004] The existing mature technologies are:

[0005] 1. The state queueing model (SQ model) is used to analyze the reasons for the oscillation of the aggregated power of the air conditioning load, and a method of separating the upper and lower limits of the temperature setting value is proposed to smooth the power fluctuation during the response period;

[0006] 2. The mechanism of load diversity destruction was analyzed, and the load fluctuation was effectively smoothed by independently setting the temperature setting value of each air conditioner in combination with switch control and temperature control. However, the above literature did not consider the differences in air conditioning parameters and user comfort;

[0007] 3. The power drop phenomenon that occurs during collective control of air-conditioning loads is modeled specifically, and a correction coefficient is proposed to reduce the model error. However, the changes in the air-conditioning load control cycle under different temperature setting values ​​are ignored. In the centralized control of large-scale air-conditioning loads, the errors in the aggregation model will greatly affect the scheduling results.

[0008] 4. The dynamic response process of polymerization power in the initial stage of temperature control is analyzed relatively comprehensively, but the changes of polymerization power in the stable stage and temperature recovery stage are not discussed.

[0009] 5. An approximate aggregation model of temperature control load is established based on the first-order equivalent thermal parameter model, the influence of air conditioning parameters, temperature setting value and ambient temperature on the aggregated power is analyzed, and a response potential evaluation method considering the uncertainty of user participation response is proposed;

[0010] 6. The lock time constraint when switching between the on and off states of the air conditioner is introduced to improve the accuracy of the adjustable capacity assessment;

[0011] 7. The control process is divided into two stages: response control and load recovery. The scheduling optimization is carried out with the highest user comfort and the shortest temperature recovery time as the goals respectively. However, the decision model needs to judge whether each air conditioner is controlled or not in turn, which requires a large amount of calculation. In addition, once the air conditioner is controlled during the response period, it cannot be adjusted back, resulting in large differences in the length of time different users participate in the control.

[0012] 8. An optimal scheduling plan is formulated on the day before to guide the selection of the scheduling strategy of the aggregation group on the next day, and preparation time is introduced to reduce the impact of a large load drop in the initial response. However, this scheduling strategy sacrifices the regulation potential during the aggregation power drop stage and does not consider the temperature recovery process.

[0013] Based on the existing research, we can see that:

[0014] (1) Existing studies lack a classified discussion and complete modeling of the dynamic response of aggregate power at each stage of the temperature control process (including power drop stage, power stable stage, and temperature recovery stage);

[0015] (2) The influence of air conditioning parameter differences and temperature setting value changes on the air conditioning start-stop cycle is ignored, resulting in large errors in the established aggregate power model;

[0016] (3) At present, most scheduling strategies stipulate that each air conditioner is only allowed to participate in regulation once during the demand response period, which limits the potential of air conditioner load regulation. At the same time, under this regulation mode, the comfort experience of users varies greatly. Summary of the invention

[0017] In view of the above-mentioned deficiencies in the prior art, the present invention provides an air conditioning load cluster control and scheduling optimization method to solve the above-mentioned problems.

[0018] In order to achieve the above-mentioned purpose of the invention, the technical solution adopted by the present invention is: an air conditioning load cluster control and scheduling optimization method, comprising:

[0019] According to the equivalent thermal parameter model of a single air conditioner, an air conditioner load state queue model is constructed;

[0020] According to the influence of temperature control on the aggregate power of air conditioners, a dynamic response model of the state queue model of air conditioner load group is constructed;

[0021] Based on the power characteristics of the dynamic response model and the user's comfort requirements, the scheduling optimization of the aggregated air-conditioning load participating in the demand response process is completed.

[0022] Further: the air conditioning load state queue model divides the start and stop cycle of the air conditioning into t off +t on state unit;

[0023] Among them, the former off The state unit indicates that the air conditioner compressor is in the off state, and the indoor temperature gradually increases. on A status unit indicates that the air conditioner is in the on-state of cooling;

[0024] After each time interval Δt, the air conditioners in each state unit are uniformly transferred to the next state. The number of air conditioners contained in each state group is N / (t off +t on) ;

[0025] Where N is the total number of air conditioners.

[0026] Furthermore, the dynamic response model of the air-conditioning load group state queue model divides the aggregate power change process of the air-conditioning load after response control into:

[0027] Load reduction process during the response period: including the power drop phase and power stable phase of demand response;

[0028] The power drop stage of the demand response is used to represent the change of the number of air-conditioning load startup state groups and the aggregate power over time from the time t0 when the control command is issued to the time t3 when the first state group of the state queue cools down to the lower limit of the new temperature;

[0029] The state group represents the air conditioner in the state unit;

[0030] The power stable stage is used to indicate that the air conditioning state group is in a new temperature range [T min,2 ,T max,2 ], the number of air-conditioning load startup groups and aggregate power change over time, corresponding to t3 to t recv moment; which includes a number of complete cycles;

[0031] Each complete cycle includes: time t3 when the first state group cools down to the lower limit of the new temperature to time t7 when all vacant / overlapping state groups have switched to the shutdown state;

[0032] Temperature recovery process after the response ends: used to indicate that the temperature of the air conditioning cluster has recovered from a higher temperature range [T min,2 ,T max,2 ] Cool down to a lower temperature range [T min,1 ,T max,1 ] process.

[0033] Further: the power stabilization stage is based on the state queues at both ends reaching T min,2 The time dependency required is modeled, which includes three cases:

[0034] Case 1: The first state group in the state queue reaches T min,2 The time is equal to the time when the last state group in the state queue reaches T min,2 time;

[0035] Case 2: The first state group in the state queue reaches T min,2 The time is greater than the time when the last state group in the state queue reaches T min,2 time;

[0036] Case 3: The first state group in the state queue reaches T min,2 The time is less than the time when the last state group in the state queue reaches T min,2 time;

[0037] Among them, T min,2 Indicates the minimum value of the indoor temperature after heating.

[0038] Further: during the temperature recovery process after the response is completed, for the air conditioner in the off state, the temperature setting value is set to the temperature setting value after the temperature is increased;

[0039] For air conditioners that are turned on, control them separately according to whether there is state overlap:

[0040] If there is no state overlap, the temperature setting value is set to the temperature setting value before the temperature rise;

[0041] If there is state overlap, the overlapping state groups will be separated and called back separately.

[0042] Further: the method for scheduling optimization of the dynamic response model of the load group state queue model includes:

[0043] The air conditioners are clustered into K clusters, and the user comfort is quantified by predicting the average number of votes;

[0044] Construct a scheduling optimization model for one-time user response based on user comfort;

[0045] The scheduling optimization model for one-time user response is optimized to obtain a scheduling optimization model that allows users to respond multiple times.

[0046] Further: the scheduling optimization model of the user's one-time response includes a first stage and a second stage;

[0047] The first stage performs scheduling optimization on the load reduction process during the response period, and the optimization goal is to minimize the deviation between the actual load reduction amount and the demand amount;

[0048] The second stage performs scheduling optimization on the temperature recovery process after the response is completed, and the optimization goal is to keep the aggregated power of the temperature recovery process stable within the rebound load limit.

[0049] Further: the methods for optimizing the scheduling optimization model of the user's one-time response include:

[0050] Allow the air conditioner to end the response early after the temperature is controlled;

[0051] Limit the number of adjustments per aggregation group;

[0052] Limit the number of air conditioners in controlled state at any time to no more than the total number of air conditioners in the aggregation group;

[0053] Rank the priorities of air conditioner users' participation in the response.

[0054] The beneficial effects of the present invention are:

[0055] 1. Completely describe the changes of aggregate power in the power drop stage, power stable stage and temperature recovery stage, especially when there is state overlap / vacancy, and point out the temperature callback time and specific control process to maintain temperature diversity;

[0056] 2. Further grouping according to the initial temperature setting value not only ensures that the aggregate power remains stable after the temperature is adjusted back, but also reduces the load rebound caused by the simultaneous temperature reduction of the air conditioner to a certain extent;

[0057] 3. The scheduling strategy of selecting air conditioners to enter temperature control in batches and time periods during the demand response period can avoid power troughs in the early stage of response. The power fluctuations between aggregation groups in the power drop stage and the power stable stage can be comprehensively utilized to effectively reduce the load reduction deviation;

[0058] 4. Combining the scheduling strategy of multiple rotations of air-conditioning users to participate in demand response not only reduces the impact of temperature control on the comfort of air-conditioning users to a certain extent, but also narrows the comfort differences between users. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Flowchart of the air conditioning load cluster control and scheduling optimization method.

[0060] Figure 2 This is the equivalent thermal parameter model diagram of a single air conditioning load.

[0061] Figure 3 Schematic diagram of the air conditioner's room temperature, switch status, and power changes over time.

[0062] Figure 4 Schematic diagram of air conditioning load state queue model.

[0063] Figure 5 Schematic diagram of the state transfer process of air conditioning load.

[0064] Figure 6 Schematic diagram of the load reduction process during the response period.

[0065] Figure 7 Schematic diagram of three cases considered during the power steady phase.

[0066] Figure 8 This is a diagram showing the simulation verification results of the aggregation power model in the embodiment.

[0067] Fig. 9 This is a comparison chart of air conditioning polymerization power at different temperature setting initial values ​​when the temperature is increased by 1°C in the embodiment.

[0068] Fig.10 This is a comparison diagram of the aggregate power based on whether to set the initial value for grouping or not based on the temperature in the embodiment.

[0069] Fig.11 It is a result diagram of scheduling optimization scheme 4 and scheduling optimization scheme 5 in the embodiment.

[0070] Fig.12It is a scheduling effect diagram of scheduling optimization schemes 3, 4, and 5 under different demand reduction ratios in the embodiment.

[0071] Fig.13 This is a comparison chart of group polymerization power when implementation schemes 3 and 4 are implemented at a reduction ratio of 10% for the 5th polymerization group in the embodiment. DETAILED DESCRIPTION

[0072] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0073] like Figure 1 As shown, in one embodiment of the present invention, a method for controlling and optimizing the clustering of air-conditioning loads is provided, comprising:

[0074] According to the equivalent thermal parameter model of a single air conditioner, an air conditioner load state queue model is constructed;

[0075] According to the influence of temperature control on the aggregate power of air conditioners, a dynamic response model of the state queue model of air conditioner load group is constructed;

[0076] Based on the power characteristics of the dynamic response model and the user's comfort requirements, the scheduling optimization of the aggregated air-conditioning load participating in the demand response process is completed.

[0077] like Figure 2 As shown in the figure, it is the equivalent thermal parameter model diagram of a single air conditioner load. From the perspective of simplifying the calculation, the indoor air temperature T a and indoor solid temperature T m Equal, and then get the first-order ETP model that characterizes the indoor temperature change and the air conditioner switch state:

[0078]

[0079] Where C is the equivalent gas specific heat capacity, R is the equivalent thermal resistance; Q = ηP rated is the cooling / heating capacity of the air conditioner, η is the air conditioner energy efficiency ratio, P rated is the rated power of the air conditioner, s(t) is the on / off state of the air conditioner at time t;

[0080] The s(t) is developed for the cooling working mode of the fixed-frequency air conditioner. The air conditioner compressor starts and stops periodically in a certain environment, and the corresponding indoor temperature changes periodically within a certain range. The basic principle is: when the indoor temperature is higher than the upper temperature limit, the air conditioner compressor is turned on and the indoor temperature gradually decreases; when the indoor temperature is lower than the lower temperature limit, the air conditioner compressor is turned off and the indoor temperature gradually rises. Therefore, s(t) is expressed as:

[0081]

[0082] Where: T max With T min Respectively represent the upper and lower limits of the air conditioner operating temperature, which are determined by the temperature setting value T set and the temperature dead zone δ.

[0083] Combining the above analysis, the discretized calculation formula describing the indoor temperature change of the fixed-frequency air conditioner during periodic operation can be obtained:

[0084]

[0085] Where Δt represents the simulation time interval.

[0086] Further, consider the air conditioner when the indoor temperature is [T min ,T max ], and assuming that the outdoor temperature remains unchanged during the control period, according to equations (3) and (4), the on / off time of a single air-conditioning load in a single start / stop cycle is obtained:

[0087]

[0088] Among them, τ off Indicates the off time of the air conditioning compressor, τ on Indicates the duration of the air conditioning compressor being on.

[0089] like Figure 3 The instantaneous power of a fixed frequency air conditioner depends on the rated power P of the air conditioner. rated The on / off status of the compressor, the on / off status of a single air conditioner, the room temperature, and the instantaneous power change over time, the air conditioner load cluster with consistent or similar parameters, and its aggregated total power P agg (t) is shown in formula (7):

[0090]

[0091] Among them, N on (t) represents the number of air conditioners that are turned on at time t.

[0092] When the number of air conditioners N is large enough and the load of each air conditioner is in a steady state (i.e., the temperature setting value remains unchanged), the air conditioner cluster is considered to have a high load diversity, that is, the indoor temperature of each user is evenly distributed in [T min ,T max ] temperature range, this application introduces a state queue model to describe the aggregate power change process of the air conditioning load, such as Figure 4 As shown, the air conditioning load state queue model divides the start and stop cycle of the air conditioning into t off +t on state units; among them, the first t off The state unit indicates that the air conditioner compressor is in the off state, and the indoor temperature gradually increases. on The state unit indicates that the air conditioner is in the on-state cooling state. After each time interval Δt, the air conditioners in each state unit are uniformly transferred to the next state. The number of air conditioners contained in each state group is N / (t off +t on) ; Where N is the total number of air conditioners.

[0093] exist Figure 4 In, t off is 7, so the first 7 state units represent shutdown, t on The value is 5, so the last 5 status units represent startup, and there are 12 status units in total.

[0094] Therefore, the aggregate power change of the air conditioning load is only determined by the number of state groups in the on state under the current parameter conditions, and has nothing to do with the number of loads, that is,

[0095]

[0096] Among them, S off (t) and S on (t) represents the number of air conditioner status groups that are in the off state and the on state at time t.

[0097] Direct load control (DLC) is an incentive-based demand response control method in the power market, and is also the main way for air conditioning loads to participate in demand response. When implementing DLC ​​projects, users can be remotely controlled through a variety of control strategies such as switch control and temperature control. Switch control is that the relevant departments issue a shutdown command to the air conditioning load in advance to control the switch state of the air conditioning, and then turn on the air conditioning again after the response is completed, so as to achieve the purpose of load reduction or transfer. This control method has a large adjustable capacity and a fast response, but it cannot continuously reduce the load for a long time due to user comfort restrictions. Moreover, a large number of controlled loads are put into operation at the same time after the demand response is completed, resulting in a rapid and severe impact of the rebound load, which is bound to have an adverse impact on the operation of the power grid. The load reduction principle of temperature control is to reduce the start-up duty cycle of the air conditioning during the periodic operation process by changing the air conditioning temperature setting value. From formulas (5) and (6), it can be seen that when the external ambient temperature remains unchanged, the shutdown and start-up time of the air conditioning compressor will change with the increase or decrease of the temperature setting value. According to formula (8), the aggregate power of the air conditioning cluster will also change accordingly. This control method has a small adjustable load capacity but can meet the control requirements on a long time scale.

[0098] Traditional temperature control methods will cause huge power fluctuations in aggregate power because uniformly issuing temperature setpoint change commands to air-conditioning load groups with the same parameters will seriously destroy the continuity and diversity of the state queue.

[0099] Therefore, in order to keep the aggregate power relatively stable, the present application adopts the SP control method, sets the temperature setting values ​​for the air conditioners in the on state and the off state respectively, and allows the state queue to operate between the initial temperature lower limit and the temperature upper limit after the temperature rises, thereby ensuring the continuity and temperature diversity of the state transfer process. The control method is implemented by establishing a dynamic response model of the state queue model of the air conditioner load group;

[0100] Since the start-stop cycle (total number of states) and the on-off duration (state distribution) of the air conditioner when it is running at different temperature setting values ​​are different, before constructing the dynamic response model of the air conditioner load group state queue model under the temperature control mode, the following data before and after the air conditioner load temperature rise needs to be calculated:

[0101]

[0102] Among them, S off,m Indicates that the temperature setting value is T set,m (m=1,2), the number of shutdown and startup state units of the air conditioning load in a start-stop cycle, T set,1 and T set,2 Respectively represent the temperature setting values ​​before and after heating; S on,m The corresponding indoor temperature range [T min,m,T max,m ] and T set,m Satisfies the relationship (3); ΔF 1,2 Indicates that the air conditioner is from T max,1 Heating to T max,2 The number of time intervals required, ΔN 2,1 Indicates that the air conditioner is from T min,2 Cool down to T min,1 The desired number of time intervals.

[0103] In this application, the dynamic response model of the air conditioning load group state queue model divides the aggregate power change process of the air conditioning load after response control into:

[0104] Load reduction process during the response period: including the power drop phase and power stable phase of demand response;

[0105] The power drop stage of the demand response is used to represent the change of the number of air-conditioning load startup state groups and the aggregate power over time from the time t0 when the control command is issued to the time t3 when the first state group of the state queue cools down to the lower limit of the new temperature;

[0106] The state group represents the air conditioner in the state unit;

[0107] The power stable stage is used to indicate that the air conditioning state group is in a new temperature range [T min,2 ,T max,2 ], the number of air-conditioning load startup groups and aggregate power change over time, corresponding to t3 to t recv moment; which includes a number of complete cycles;

[0108] Each complete cycle includes: time t3 when the first state group cools down to the lower limit of the new temperature to time t7 when all vacant / overlapping state groups have switched to the shutdown state;

[0109] Temperature recovery process after the response ends: used to indicate that the temperature of the air conditioning cluster has recovered from a higher temperature range [T min,2 ,T max,2 ] Cool down to a lower temperature range [T min,1 ,T max,1 ] process.

[0110] like Figure 5 As shown in the figure, it is a schematic diagram of the state transition process of the air conditioning load during the load reduction process during the response period. In order to more intuitively show the process of the state transition of the air conditioning load over time during the heating process, a continuous curve is used to represent the continuous state queue, where the blue curve is connected by the state units of the air conditioning state group at the current moment, and is calculated according to Figure 2 The numbering sequence in Figure 5The air conditioner status group numbers from left to right in the continuous status queue should include (S off,1 +1) to (S off,1 +S on,1 ) and the temperature drop from 1 to S off,1 That is to say, when the state is kept continuous, the state queue can be considered to be at the beginning and end of the Sth off,1 , (S off,1 +1) group of air conditioners, the state unit numbering method is the same as the state group, and the state unit numbering when operating in the initial temperature range is 1~(S off,1 +S on,1 ), in the new temperature range, numbered 1′~(S off,1 +S on,1 )′ to distinguish them.

[0111] Figure 5 (a) represents the state distribution of air conditioning load at time t0 when the control command is issued. Figure 5 In (b), use t c Indicates the duration of the air conditioning load participating in temperature control. It can be seen that when the state queue moves to a higher temperature range, the total number of states remains unchanged, the number of power-on state groups decreases, and the number of power-off state groups increases. At this time, the aggregate power is in a decreasing stage, with the following relationship:

[0112]

[0113] t1=t0+min{S on,1 ,ΔF 1,2} (15)

[0114] t2=t0+max{S on,1 ,ΔF 1,2} (16)

[0115] Among them, t1 indicates that the aggregate power reaches the minimum amplitude P min t2 represents the moment when the aggregate power starts to rise, P agg,1 Indicates the aggregate power of air conditioning load before it participates in regulation.

[0116] When S off,1 The group air conditioner cools down to the new lower temperature limit T min,2 , the time is t3, then:

[0117] t3=t0+ΔF 1,2 +S on,2 (17)

[0118] The aggregate power will remain relatively stable for a period of time after t3. Figure 6The load reduction process during the response period is shown, that is, the changes in the number of air-conditioning load on-state groups and aggregate power over time during the period t0 to t3.

[0119] Since the temperature control method will change the on / off time of the air conditioner, to analyze the power stabilization stage, that is, the aggregate power change law when the air conditioner load is cyclically running in the new temperature range after time t3, we first need to consider the state queue at the head and tail ends reaching the minimum value of the indoor temperature after heating, T min,2 the time required;

[0120] That is, S off,1 Group (state queue head) from T max,1 Heating to T max,2 , then switch to the power-on state and cool down to T min,2 The time required is (ΔF 1,2 +S on,2 ), the first (S off,1 +1) group (end of status queue) from T max,1 Cool down to T min,1 , then switch to shutdown state and heat up to T min,2 The time required is (S on,1 +S off,1 +ΔF 1,2 -S off,2 ).

[0121] like Figure 7 As shown, the size relationship between the two is reflected in the state queue diagram, which includes three situations:

[0122] Case 1: The first state group in the state queue reaches T min,2 The time is equal to the time when the last state group in the state queue reaches T min,2 time;

[0123] Case 2: The first state group in the state queue reaches T min,2 The time is greater than the time when the last state group in the state queue reaches T min,2 time;

[0124] Case 3: The first state group in the state queue reaches T min,2 The time is less than the time when the last state group in the state queue reaches T min,2 time;

[0125] Figure 7 (a) indicates that the start-stop cycle of the air conditioner remains unchanged after the temperature rises, that is, the state distribution without state overlap / vacancy phenomenon; Figure 7 (b) indicates that the start-stop cycle of the air conditioner is shortened after the temperature rises, that is, the state distribution with state overlap phenomenon, and the air conditioner state has not yet fully entered the new temperature range at time t3; Figure 7(c) indicates that the start-stop cycle becomes longer, that is, the state distribution with state vacancy phenomenon, at (t0+S on,1 +S off,1 +ΔF 1,2 -S off,2 ) At this moment, all air conditioners have entered the new temperature range.

[0126] Figure 7 The aggregate power of the air conditioning load in the case shown in (a) will remain constant after time t3, while the remaining two cases will cause the aggregate power to fluctuate slightly periodically in subsequent cycles. In order to facilitate subsequent derivation, ΔS is used to represent the change in the start-stop cycle before and after the air conditioning load temperature rises. The expression is:

[0127] ΔS=|S off,1 +S on,1 -(S off,2 +S on,2 )| (18)

[0128] for Figure 7 (b), Sth time from time t3 off,1 The group will switch to the shutdown state, that is, transfer to state unit 1', and at the same time (S on,1 +ΔS) group will also heat up to T min,2 Entering state unit 1′, there will be an overlap of air conditioner state groups (that is, the same state unit contains two groups of air conditioners); at (t3+ΔS) time, all air conditioners enter [T min,2 ,T max,2 ] temperature range, there are ΔS overlapping state units at this time, and when the overlapping state units pass through the new temperature upper and lower limits, the number of power-on state groups will be S on (t) In [S on,2 ,S on,2 +ΔS], resulting in periodic fluctuations in the polymerization power.

[0129] for Figure 7 (c), at (t3-ΔS) all air conditioners enter [T min,2 ,T max,2 ] temperature range, the air conditioning load from time t3 (S off,1 +S on,1 ) state groups will be in a longer period (S off,2 +S on,2 ) state units are transferred cyclically, then there will be ΔS state units that do not contain air conditioning groups, which are called vacant state units; similar to the overlapping situation, when the vacant state units pass through the new temperature upper and lower limits, the number of power-on state groups S on (t) In [S on,2 -ΔS,S on,2 ] range, resulting in periodic fluctuations in the aggregate power.

[0130] Based on the above analysis, we can fully model the change process of the number of power-on groups during the load reduction process during the response period:

[0131] First, some key moments are calculated:

[0132]

[0133] Among them, t4 represents the moment when the first vacant / overlapping state unit reaches the new temperature upper limit. Before and after t4, the vacant / overlapping state units are distributed in the range of 1′ to S off,2 In the state unit, from time t4+1, the vacant / overlapping state unit gradually transfers to the state unit indicating power on; t5 indicates that the last vacant / overlapping state unit reaches the state unit (S off,2 +1)′ (switching from off to on); t6 and t7 respectively represent the arrival of the head end of the vacant / overlapping state unit at the state unit (S off,2 +S on,2 )′ and the moment when the tail end of the queue returns to state unit 1′ again, from time t6+1, the vacant / overlapping state units gradually transfer to the state units indicating shutdown, and by time t7, all vacant / overlapping state groups have been switched to the shutdown state.

[0134] Before time t3, the number of air conditioner groups S in the on state in three cases on (t) has the same expression, both

[0135]

[0136] After time t3, the start-stop cycle of the air-conditioning load group changes. The variable J is used to represent the comparison result of the new and old start-stop cycles. Its expression is as follows:

[0137] J=sign(S off,1 +S on,1 -S off,2 -S on,2 ) (twenty one)

[0138] The Sign() function refers to the sign function in mathematics, where J takes -1, 0 and 1, respectively. Figure 7 Three situations correspond.

[0139] Therefore, the number of air conditioner groups in the on state after time t3 can be expressed as follows:

[0140]

[0141] In the above formula, the number of air conditioner power-on status groups changes cyclically from time t7+1, and the cycle period is S off,2+S on,2 It is worth noting that, when the start-stop cycle becomes longer, the aggregate power starts the first cycle at t3-ΔS and ends at t6; when the start-stop cycle is shortened, the aggregate power starts the first cycle at t3+ΔS and ends at t7; this application uses t3 as the starting point of the aggregate air conditioner's cyclic operation within the new temperature range, and the period from t3 to the start of the temperature adjustment operation is called the power stable stage.

[0142] In order to maintain the continuity of the state queue and make the air conditioning load aggregate power return to the amplitude level before regulation after the temperature setting value is restored, the callback method needs to be improved similar to the above-mentioned temperature rise control process, that is, to select a suitable temperature callback start time, and set different temperature setting values ​​for the air conditioners in the on and off states for regulation. In the state queue, the first and last ends are respectively S off,1 Group and (S off,1 +1) group, during the callback process, it is necessary to ensure that the S off,1 The group first cooled down to T min,2 Below. When S off,1 The group air conditioner returns to the state unit S off,1 At this time, S on (t) = S on,1 , S off (t) = S off,1 , the temperature callback process ends.

[0143] During the temperature recovery process after the response is completed, for the air conditioner in the off state, the temperature setting value is set to T set,2 , to keep them heated up;

[0144] For air conditioners that are turned on, control them separately according to whether there is state overlap:

[0145] If there is no state overlap, that is, S off,1 +S on,1 ≤S off,2 +S on,2 When the temperature is adjusted back to the temperature setting value before the temperature rises, set,1 , so that each air-conditioning state group can be transferred to the original temperature lower limit in an orderly manner;

[0146] If there is state overlap, that is, S off,1 +S on,1 >S off,2 +S on,2 When the temperature is directly controlled, it is not possible to effectively split the merged overlapping state groups into their original states. Figure 7(b) It can be seen that the overlap of the state queue refers to the overlap of the head end and the tail end. If the temperature setting value is uniformly called back, the ΔS state groups at the tail end will follow the corresponding overlapping state groups at the head end to T with one less cycle. min,1 Transfer, resulting in the destruction of the temperature diversity of the air conditioner after the temperature is adjusted back. The method adopted in this application is: separating the overlapping state groups and then adjusting them respectively;

[0147] Specifically: at time t3, the first (S off,1 +1)~(S off,1 +ΔS) group is marked as the keep group, and the rest of the air conditioners are classified as the normal group. After the callback operation starts, the air conditioners in the normal group are called back in a way that there is no state overlap, while the air conditioners in the keep group continue to cycle in the new temperature range for a full cycle, and then call back the temperature in the way of the normal group, thereby achieving the effect of separating the overlapping state groups.

[0148] After determining the above scheme, the temperature recovery process after the response is completed can be modeled. Before modeling, the actual callback start time of the air conditioning load needs to be determined. In order to ensure that the aggregate power of the air conditioning load remains at P after the temperature callback, agg,1 Nearby, each air conditioner needs to complete the same number of complete cycles in the new temperature range before being adjusted back. The earliest adjustment time of the air conditioner load is t recv satisfy:

[0149]

[0150] t recv =t3+l min ×(S on,2 +S off,2 ) (twenty four)

[0151] Among them, t d is the moment when the air conditioner receives the callback signal, then l min Indicates the minimum number of cycles for the air conditioning load to operate within the new temperature range.

[0152] If t recv As the starting time of the air conditioning temperature callback, combined with ΔN calculated by formula (12) 2,1 , then the temperature recovery process after the response ends in the absence of overlapping state can be expressed as:

[0153]

[0154] For the case of overlapping states, ΔN needs to be considered 2,1 The relationship between the size of the overlapping state number ΔS, the key moments and the number of air conditioner start-up groups are changed, and the temperature recovery process after the response is completed is expressed as:

[0155]

[0156] Among them, t8 represents the recv After a while, S remained constant. on (t) The time when the change starts (ΔN 2,1 ≠ΔS), t9 represents the moment when the power rebound reaches the peak value, t 10 Indicates the moment when the aggregate power starts to fall back, and the state queue head (S off,1 The air conditioner of group) returns to state unit S off,1 The moment, t fin Indicates the time when the temperature recovery operation is completed.

[0157] like Figure 8 As shown, in one embodiment of the present application, in order to verify the state vacancy and overlapping phenomenon mentioned in the present application, two groups of aggregation groups with different parameters are selected for simulation analysis. The simulation time interval is 30s. It is assumed that both aggregation groups receive a control instruction to increase the temperature by 1°C in the 15th minute, and a temperature recovery instruction is issued after 30 minutes of response. The control terminal then performs corresponding operations on all air conditioners in each group according to the improved temperature control method.

[0158] Figure 8 (a) and Figure 8 (b) shows the group aggregation power curves of the state overlap phenomenon due to the shortening of the start-stop cycle and the state vacancy phenomenon due to the lengthening of the start-stop cycle. The specific parameters and calculation results of the key moments are shown in Table 1.

[0159] Table 1 Calculation results of key parameters of the state queue model for two air-conditioning aggregation groups

[0160]

[0161]

[0162] Figure 8 The horizontal axis scales marked with black dashed lines correspond to t0, t3 and t recv moment. As you can see, Figure 8 The aggregate power in (a) is not from t recv It did not rebound from moment t, but maintained until moment t8, which is consistent with the analysis of the temperature recovery process with overlapping states in this application. The previous aggregation power model assumed that the aggregation power of all air conditioners after entering the new temperature range remained constant, and did not consider the impact of the start-stop cycle changes on the aggregation power. There were errors in the evaluation of the adjustable potential of the air-conditioning load group. Obviously, the power curve of the model proposed in this application can better fit the actual situation.

[0163] In order to meet the power reduction requirements and minimize power fluctuations during the demand response period, the present application also proposes a scheduling optimization method to optimize the dynamic response model of the load group state queue model, which includes:

[0164] The air conditioners are clustered into K clusters, and the user comfort is quantified by predicting the average number of votes;

[0165] Construct a scheduling optimization model for one-time user response based on user comfort;

[0166] The scheduling optimization model for one-time user response is optimized to obtain a scheduling optimization model that allows users to respond multiple times.

[0167] Assume that a load aggregator can control N air conditioners under its jurisdiction on a certain day, and cluster them into K aggregation groups. The load aggregator evaluates the load control potential based on the outdoor temperature forecast, air conditioning parameters, initial temperature settings of each user, user acceptability information and other data and reports it to the power company. The power company assigns the control task of the day based on the load forecast results and requires the load aggregator to provide a certain control capacity P through demand response. need The load aggregator formulates a dispatch plan for each aggregation group based on this.

[0168] When formulating a scheduling plan, load aggregators need to ensure the user's comfort experience as much as possible. Thermal comfort is the user's subjective evaluation and feeling of the indoor thermal environment. In order to quantify the impact of temperature on human thermal comfort, this application uses the predicted mean vote (PMV) indicator to measure the user's thermal comfort;

[0169] ISO7730 recommends that the PMV value be between ±0.5. Therefore, when all user parameters except the indoor temperature are known, the temperature range acceptable to each user can be inferred based on the calculation formula of the PMV index. For most residential users' air conditioners, the adjustment unit of the temperature setting value is 0.5°C, so this application sets 0.5°C as the minimum unit of temperature increase.

[0170] The temperature increase of the i-th air conditioner The optional set of and the maximum temperature increase of the aggregation group I (I∈1,2,…,K) It can be expressed by the following formula:

[0171]

[0172] Where Z represents a set of integers, represents the maximum temperature increase that the user of the i-th air conditioner can accept, represents the desired temperature setting value of the user of the i-th air conditioner. In this application, it is considered that the initial temperature setting value of user i and equal.

[0173] Based on the classification modeling of the aggregate power change process of air conditioning load after response control, the heat capacity C of the indoor gas of the aggregation group is obtained through the load aggregator. I 、Thermal resistance of indoor gas R I , air conditioning refrigeration efficiency η I and rated power After the key parameters are determined, the amount can be adjusted according to the temperature assigned by the group. Calculate the number of on and off state groups before and after temperature rise and the corresponding and The number of air conditioners involved in the regulation of each group Control start and end time and It is used as the decision variable for scheduling optimization to build a scheduling optimization model for user one-time response.

[0174] The scheduling optimization model for one-time user response includes a first stage and a second stage;

[0175] The first stage performs scheduling optimization on the load reduction process during the response period, and the optimization goal is to minimize the deviation between the actual load reduction amount and the demand amount;

[0176] Specifically, the load aggregator finds the starting time for each aggregation group to participate in regulation. Temperature increase and the number of participants The optimal combination of variables makes the actual load reduction ΔP agg (t) and demand P need The deviation is the smallest, and its objective function is:

[0177] minF1=Var(ΔP agg (t)-P need ) ,t start ≤t≤t end (29)

[0178]

[0179] Among them, t start With t end represent the start and end time of the demand response period, and in the first stage of optimization, each aggregation group defaults to t endAccording to the improved method, the temperature callback operation is implemented. Based on the above response process, the power-on status data of each aggregation group is The relationship between its variables is briefly expressed by functions f1 and f2. Son and Soff in formula (31) can be calculated according to formulas (9) and (10). Calculated according to equations (11) and (12), ts is a variable, tend is a given value, and the variable relationship in (9) to (12) corresponds exactly to f2 in equation (32).

[0180] The constraints of the first stage are:

[0181]

[0182] The second stage performs scheduling optimization on the temperature recovery process after the response is completed, and the optimization goal is to keep the aggregate power of the temperature recovery process stable within the rebound load limit;

[0183] Since the disorderly lowering of the temperature setting value by users after demand response will inevitably lead to a large load rebound, endangering the safety of the power grid, this application adds a delay unit to each group in the temperature recovery stage: each aggregation group maintains the current temperature setting value for several cycles and then adjusts the temperature.

[0184] The aggregate power in the temperature recovery stage keeps the power stable within the rebound load limit. The objective function of the second stage is:

[0185]

[0186] The constraints of the second stage are:

[0187]

[0188] in, Indicates that the first aggregation group has optimized the temperature callback time The number of power-on states at time t after represents the moment when the last group of air conditioners returns to the original temperature state, which can be calculated according to equations (25) and (26). At this time, the aggregate power will be restored to the initial power P agg,1 , β represents the rebound load fluctuation percentage.

[0189] After completing the scheduling optimization model for the user's one-time response, according to previous temperature control research, each air conditioner can only participate in at most one response process, and the temperature is not allowed to be adjusted back during the period from entering the control to the end of the response. This mode can easily lead to different comfort experience differences among users due to different control durations of the air conditioners selected at different times.

[0190] In order to solve the above problems, a scheduling optimization model that allows users to participate in demand response multiple times is proposed, and the following improvements are made based on the scheduling optimization model of user one-time response proposed in this application:

[0191] Allow the air conditioner to end the response early after the temperature is controlled;

[0192] Limit the number of adjustments per aggregation group;

[0193] Limit the number of air conditioners in controlled state at any time to no more than the total number of air conditioners in the aggregation group;

[0194] Rank the priorities of air conditioner users' participation in the response.

[0195] Since this control mode allows the air conditioner to end the response early after the temperature is controlled, the control variable in the first stage should include the temperature callback start time Secondly, considering the user comfort experience and reducing the computing pressure of the load aggregator when formulating the control strategy and the communication pressure when implementing the control, the number of control times of each aggregation group is limited to no more than n ctrl Second-rate.

[0196] Correspondingly, the decision variables such as the control start and end time, temperature increase amount, number of participating air conditioners, and the number of power-on status groups are converted into the following n ctrl dimensional vector, and the objective function F1 still satisfies equation (29);

[0197]

[0198] To ensure that the selected air conditioner at any time can only participate in one response plan, that is, the number of air conditioners in the controlled state at time t does not exceed the total number of air conditioners in the aggregation group, constraint (33) should be replaced by:

[0199]

[0200] in, represents the number of air conditioners drawn by the aggregation group in the jth regulation, and They correspond to the start and end times of the j-th regulation respectively.

[0201] Based on the optimization results of the demand response phase, the second phase only requires the temperature callback time of the last regulation of each aggregation group. The optimization objective function and constraints are consistent with the basic scheduling model and will not be described here.

[0202] Since the load aggregator issues a one-time control instruction to each aggregation group according to the day-ahead dispatch plan (assuming that the demand capacity will not change once it is formulated, and the power company will take other measures to deal with power shortages caused by emergencies), before implementing the control strategy during the day, the load aggregator needs to select the air-conditioning users that each aggregation group will participate in the control:

[0203] First, the distribution of air conditioner temperature states under the parameter conditions of each aggregation group is calculated by equations (9) and (10), and then the air conditioners with similar indoor temperatures are divided into the same state group. The state units correspond one to one.

[0204] During a regulation process, the load aggregator evenly extracts air conditioners participate in the response. Usually, the incentive cost of users in the same aggregation group to participate in demand response is similar. It can be considered that the probability of air conditioner users in each status group being selected is the same.

[0205] In order to ensure that the degree of comfort of users in the same aggregation group is similar, this application uses the comfort factor as a measurement indicator to guide the load aggregator to prioritize the response of air-conditioning users. The comfort factor uses the product of the temperature adjustment change of the air conditioner and the controlled time to characterize the degree of user comfort. The comfort factor C accumulated by each user in the aggregation group I at the current moment is calculated. i The expression of (t) is:

[0206]

[0207] The smaller the result is, the less the comfort of user i is affected before the jth adjustment, and the higher the priority of the air conditioner in selection.

[0208] In one embodiment of the present invention, in order to verify the effectiveness of the aggregate power dynamic response model, the k-means clustering method is first used to preliminarily group the air conditioner clusters with heterogeneous parameters. The clustering features are usually selected from the temperature change parameter RC and the characteristic temperature difference QR of the controlled air conditioner. However, in fact, when the initial temperature setting values ​​of homogeneous air conditioners with similar parameters are different, even if the same temperature is adjusted upward, their aggregate power curves are also quite different, such as Fig. 9 shown.

[0209] Fig. 9Through simulation and control of 500 homogeneous air conditioners using the improved temperature control method, it is found that different curves represent the aggregate power when the user's expected temperature (initial temperature setting value) is 25.5℃, 26℃, and 26.5℃, respectively. Assuming that the demand response period lasts from 30min to 110min, the temperature increase is 1℃. Combined with formulas (8) to (10), the state queue distribution (on state duty cycle) of homogeneous air conditioners at different temperature setting values ​​is different. The higher the initial temperature setting value of the air conditioner, the lower the initial level of aggregate power. In addition, even if the temperature increase is 1℃, the change in the start-stop cycle of the air conditioner increases with the increase in the initial temperature setting value, and an increasingly obvious "state loss" phenomenon occurs, resulting in more severe power fluctuations in the power stability stage.

[0210] Considering the above situation, the 1500 homogeneous air conditioners are regulated in two ways: a. No further grouping, uniformly regulated according to the control method with the initial temperature setting of 26°C; b. Divided into 3 groups according to the initial temperature setting value, and regulated according to 3 initial temperature setting values ​​respectively. The response start and end time of each group are still set to 30min and 110min, and the aggregate power curve is as follows Fig.10 As shown. It is not difficult to see that directly and uniformly regulating the air conditioning clusters with different initial temperature settings will destroy the continuity of the state queue and the temperature diversity, causing the aggregate power of the red curve to fluctuate cyclically after the temperature is adjusted back. If further grouping is performed according to the initial temperature setting value, on the one hand, the aggregate power can be guaranteed to remain stable after the temperature is adjusted back. On the other hand, due to the different start-stop cycles, there is a time difference between the actual temperature callback moments of each group after the end instruction is issued, which to a certain extent reduces the load rebound caused by the simultaneous temperature reduction of the air conditioners.

[0211] Considering that air conditioners with different initial temperature settings have different adjustable ranges when accepting temperature control, for example, the higher the initial temperature setting value, the smaller the temperature increase that users can accept, so the load reduction that each air conditioner in the same aggregation group can provide varies. Combined with formula (28), if users are not differentiated according to the initial temperature setting value, it will be unfavorable to exert the control potential of air conditioning load in the mode where the aggregation group participates in the unified control. In summary, the initial temperature setting value should be an important reference factor when clustering and grouping air conditioning loads.

[0212] In one embodiment of the present invention, it is assumed that there are more than 12,000 air conditioners in a certain area (residential community or office building) that can participate in demand response during the control period. According to the user data collected recently (including the expected temperature setting value, the temperature adjustable range, air conditioning parameters, etc.), they are divided into 20 aggregation groups to accept the load aggregator scheduling. The air conditioners and simulation parameters are shown in Table 2.

[0213] Table 2 Air conditioning indoor parameters and simulation parameters

[0214]

[0215] On a certain day, the power company requires the region to continuously provide a load reduction of a ratio of α during the period from 13:00 to 15:00 to ease the power supply pressure, and requires that the rebound load percentage β does not exceed 2α, and the load is fully restored at no more than 16:00. In order to verify the superiority and effectiveness of the demand response scheduling strategy proposed in this paper, the following scheduling strategies that may be adopted by load aggregators are compared:

[0216] Solution 1: Using the traditional temperature control method, each polymerization group simultaneously raises the temperature setting value at 13:00 and restores the temperature setting value at 15:00.

[0217] Solution 2: Use an improved temperature control method to achieve separate control of the air conditioner on and off. In this solution, each group of air conditioners still increases and restores the temperature at the same time at the start and end times of the control task.

[0218] Solution 3: Based on Solution 2, the control preparation time is introduced to guide each group to enter the temperature control process in time before the response period to reduce the impact of the load drop process on the power grid. This solution requires that all air conditioners participating in the control have entered the new temperature range before 13:00. In this paper, the preparation time is set to 1h.

[0219] Solution 4: A scheduling optimization model for one-time user response, that is, starting from 13:00, with 15 minutes as an optimization time window, some air conditioners in different groups are guided to enter the temperature control process in each time window, but the temperature is not allowed to be adjusted before 15:00. The aggregate power reduction in each time window is provided by two parts:

[0220] (1) Some air conditioners in the current time window begin to enter the response process, resulting in a rapid drop in power;

[0221] (2) The air conditioners that have completed the temperature rise process are in a stable power stage in the current time window, but they can still provide a certain power reduction due to the reduction in the startup duty cycle. Through appropriate optimization combinations, the power complementary characteristics of the controlled air conditioners between the aggregation groups and in each time window are comprehensively utilized to meet the control needs.

[0222] Solution 5: A scheduling optimization model that allows users to respond multiple times. Users can participate in regulating or adjusting the temperature at any time between 13:00 and 15:00 according to the improved temperature control method. They can participate in the response again after the temperature is restored. The temperature increase amount and the number of air conditioners adjusted by each aggregation group each time are independent of each other.

[0223] Fig.11(a) and (b) can more intuitively show the changes in the aggregate power of a certain aggregation group when implementing schemes 4 and 5. The demand response stage is from the 60th minute to the 180th minute. Scheme 4 selects two batches of air conditioners to participate in the response in the first and third time windows. The load reduction of each regulation is superimposed on each other, and the air conditioner load is adjusted back in time after the response period ends; Scheme 5 allows the air conditioners that have participated in the regulation to enter the second regulation after the temperature recovers. The temperature curve of the air conditioner that has participated in two consecutive regulations is shown as the red dotted line, where the temperature increase of the two regulations is 0.5℃ and 1℃ respectively, and the group aggregate power has two drops and two rebounds accordingly.

[0224] When conducting simulation experiments on Schemes 4 and 5, considering that this application adopts temperature control to ensure temperature diversity, it is impossible for a large number of air conditioners to complete the on-off state switching at the same time in a short period of time to achieve the effect of a sharp drop in aggregate power instantly. Therefore, the aggregation group is allowed to enter the temperature adjustment process 10 minutes in advance as a preparation stage for demand response.

[0225] Table 3 shows the load reduction deviation and rebound fluctuation percentage β of the above five scheduling schemes when the power reduction target ratio α is 5%, 10%, and 15% respectively:

[0226] Table 3 Load reduction deviation and rebound load percentage of each scheduling scheme

[0227]

[0228] Since the aggregate power fluctuations of Scheme 1 and Scheme 2 are relatively large, this embodiment will further analyze the scheduling optimization results of Schemes 3, 4, and 5.

[0229] Combining Table 1 and Fig.12 , Schemes 4 and 5 have more advantages in load reduction capacity and rebound suppression effect, and the two schemes allow the air conditioner to participate in the temperature control process in batches at different times during the demand response period, avoiding the power trough phenomenon that occurs in schemes 1 to 3.

[0230] Scheme 3 is an improvement based on Scheme 2. Its essence is to concentrate the power drop stage of each aggregation group when participating in temperature control before the start of the regulation task, ensuring that each group is in the power stable stage after temperature control at 13:00, and the total aggregate power is maintained near the regulation target. Due to the introduction of a 1-hour preparation time, this scheme alleviates the situation where all air conditioners increase the temperature at the same time to a certain extent, but it is obvious that the aggregate power will still drop significantly before 13:00, causing an impact on the operation of the power grid. In addition, this scheme relies on the increase in the temperature set value - the load reduction brought about by the decrease in the startup duty cycle is relatively limited. As the reduction ratio increases from 5% to 15%, the aggregate power during the response period will be difficult to maintain near the target power, and the load fluctuation will become more severe.

[0231] Compared with Scheme 3, Scheme 4 is based on the dynamic response model of the air-conditioning load group state queue model of this application, and comprehensively utilizes the power reduction generated by the power drop and power stabilization stages of each aggregation group.

[0232] Fig.13 The comparison of group aggregation power when the 5th aggregation group implements schemes 3 and 4 at a reduction ratio of 10% is shown. The response start and end time, temperature increase amount and number of participating air conditioners of the two schemes are shown in Table 4. The user average load reduction index ΔP is defined to measure the load reduction capacity that the air conditioners in the group can provide when implementing different schemes. The calculation formula is:

[0233]

[0234] in, It represents the aggregate power of the fifth group of air conditioners when no scheme is implemented.

[0235] Table 4 Scheduling results of the fifth group's air conditioning implementation plan 3 and 4

[0236]

[0237] As shown in Table 4, when the aggregation team implemented Scheme 4, the power that each air conditioner could provide when the temperature was raised by 1°C was 16W higher than that of Scheme 3, indicating that the control method of allowing the air conditioners to respond multiple times and in batches can better exert the regulation potential of the air conditioning load.

[0238] Further considering the differences in thermal comfort among air-conditioning users, this paper proposes Improvement Scheme 5. As shown in Table 3, Schemes 4 and 5 are generally similar in their ability to control load reduction deviation and suppress rebound load. The difference in the effects of the two schemes is mainly reflected in the accumulated comfort factor of each user during the demand response period. Table 5 shows the calculation results of user comfort factors of Implementation Schemes 4 and 5.

[0239] Table 5 Comfort factor index of air-conditioning users when implementing schemes 4 and 5 respectively under different reduction ratios

[0240]

[0241] Depend on Figure 5It can be seen that as the target reduction ratio increases, the degree of comfort of the controlled air-conditioning users is gradually affected. In order to provide enough power reduction, the load aggregator needs to select more air-conditioning to participate in the demand response. According to the mean and variance of the comfort factor, Scheme 5 reduces the sum of the comfort factors of the air-conditioning cluster to a certain extent, and also reduces the comfort differences between users. The reason is that Scheme 5 gives priority to selecting the air-conditioning with the least impact on the current user's comfort to participate in temperature control, and distributes the power reduction pressure more evenly to the air-conditioning load group, avoiding the extreme situation where only some air-conditioning are continuously subjected to long-term high-temperature control to provide load reduction, while the remaining air-conditioning does not participate in the demand response process at all.

[0242] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for cluster control and scheduling optimization of air conditioning loads, characterized in that: include: According to the equivalent thermal parameter model of a single air conditioner, an air conditioner load state queue model is constructed; According to the influence of temperature control on the aggregate power of air conditioners, a dynamic response model of the state queue model of air conditioner load group is constructed; Based on the power characteristics of the dynamic response model and the user's comfort requirements, the scheduling optimization of the aggregated air-conditioning load participating in the demand response process is completed.

2. The air conditioning load cluster control and scheduling optimization method according to claim 1 is characterized in that: The air conditioning load state queue model divides the start and stop cycle of the air conditioner into t off +t on state unit; Among them, the former off The state unit indicates that the air conditioner compressor is in the off state, and the indoor temperature gradually increases. on A status unit indicates that the air conditioner is in the on-state of cooling; After each time interval Δt, the air conditioners in each state unit are uniformly transferred to the next state. The number of air conditioners contained in each state group is N / (t off +t on) ; Where N is the total number of air conditioners.

3. The air conditioning load cluster control and scheduling optimization method according to claim 1, characterized in that: The dynamic response model of the air-conditioning load group state queue model divides the aggregate power change process of the air-conditioning load after response control into: Load reduction process during the response period: including the power drop phase and power stable phase of demand response; The power drop stage of the demand response is used to represent the change of the number of air-conditioning load startup state groups and the aggregate power over time from the time t0 when the control command is issued to the time t3 when the first state group of the state queue cools down to the lower limit of the new temperature; The state group represents the air conditioner in the state unit; The power stable stage is used to indicate that the air conditioning state group is in a new temperature range [T min,2 ,T max,2 ], the number of air-conditioning load startup groups and aggregate power change over time, corresponding to t3 to t recv time; That Includes multiple full loops; Each complete cycle includes: time t3 when the first state group cools down to the lower limit of the new temperature to time t7 when all vacant / overlapping state groups have switched to the shutdown state; Temperature recovery process after the response ends: used to indicate that the temperature of the air conditioning cluster has recovered from a higher temperature range [T min,2 ,T max,2 ] Cool down to a lower temperature range [T min,1 ,T max,1 ] process.

4. The air conditioning load cluster control and scheduling optimization method according to claim 3 is characterized in that: The power stabilization stage is based on the state of the queue, with the first and last ends reaching T min,2 The time dependency required is modeled, which includes three cases: Case 1: The first state group in the state queue reaches T min,2 The time is equal to the time when the last state group in the state queue reaches T min,2 time; Case 2: The first state group in the state queue reaches T min,2 The time is greater than the time when the last state group in the state queue reaches T min,2 time; Case 3: The first state group in the state queue reaches T min,2 The time is less than the time when the last state group in the state queue reaches T min,2 time; Among them, T min,2 Indicates the minimum value of the indoor temperature after heating.

5. The air conditioning load cluster control and scheduling optimization method according to claim 3 is characterized in that: During the temperature recovery process after the response is completed, for the air conditioner in the off state, the temperature setting value is set to the temperature setting value after the temperature is increased; For air conditioners that are turned on, control them separately according to whether there is state overlap: If there is no state overlap, the temperature setting value is set to the temperature setting value before the temperature rise; If there is state overlap, the overlapping state groups will be separated and called back separately.

6. The air conditioning load cluster control and scheduling optimization method according to claim 1, characterized in that: The method for optimizing the dispatching of the dynamic response model of the load group state queue model includes: The air conditioners are clustered into K clusters, and the user comfort is quantified by predicting the average number of votes; Construct a scheduling optimization model for one-time user response based on user comfort; The scheduling optimization model for one-time user response is optimized to obtain a scheduling optimization model that allows users to respond multiple times.

7. The air conditioning load cluster control and scheduling optimization method according to claim 6, characterized in that: The scheduling optimization model for one-time user response includes a first stage and a second stage; The first stage performs scheduling optimization on the load reduction process during the response period, and the optimization goal is to minimize the deviation between the actual load reduction amount and the demand amount; The second stage performs scheduling optimization on the temperature recovery process after the response is completed, and the optimization goal is to keep the aggregated power of the temperature recovery process stable within the rebound load limit.

8. The air conditioning load cluster control and scheduling optimization method according to claim 6, characterized in that: The methods used to optimize the scheduling optimization model for user one-time response include: Allow the air conditioner to end the response early after the temperature is controlled; Limit the number of adjustments per aggregation group; Limit the number of air conditioners in controlled state at any time to no more than the total number of air conditioners in the aggregation group; Rank the priorities of air conditioner users' participation in the response.

Images