Model predictive control intelligent air conditioner load aggregation regulation system and method

Abstract

The application provides a model predictive control intelligent air conditioner load aggregation regulation system and method. A central controller of the application constructs a first-order equivalent heat parameter model of a single air conditioner load considering random disturbance, and solves characteristic parameters of each air conditioner by using a least square method. A certain temperature range is evenly divided into multiple temperature intervals, an on state set, an off state set, state variables of an air conditioner set at a current time, and a state space model are constructed and set, and each matrix in the state space model is solved. State variables at multiple future times are predicted, and optimal control information at the multiple future times is solved. A regulation error at the current time is calculated as a feedback signal and an input of an MPC controller, and a state control vector of multiple intelligent air conditioners at a next time is obtained by rolling optimization of the MPC controller. The central controller controls states of the multiple air conditioners at the next time according to the state control vector of the multiple intelligent air conditioners at the next time. The application realizes overall scheduling of an air conditioner group.

Description

Technical Field

[0001] This invention belongs to the field of demand response, and particularly relates to an intelligent air conditioning load aggregation and control system and method based on model predictive control. Background Technology

[0002] Demand response is currently a hot research topic for power system researchers. It can alter users' electricity consumption habits through signals such as electricity prices or incentives, regulate demand-side load, increase generation stability, maintain power balance to a certain extent, and also regulate peak loads, reducing pressure on the power system and saving grid operating costs. The air conditioning load market is large, diverse, and highly flexible in its regulation, making it an important resource for participating in demand response.

[0003] However, how to uniformly schedule large-scale air conditioning loads is a key concern for both research and industry. In the context of my country's market, regulating individual air conditioning loads is meaningless and cannot meet demand response requirements. Therefore, it is necessary to aggregate air conditioning loads. Currently, the main method for air conditioning load aggregation is the Monte Carlo method. The Monte Carlo method is simple in principle and can establish a relatively accurate aggregation model for air conditioning loads with similar parameter characteristics. However, when the heterogeneity of load parameters is significant, the accuracy of the Monte Carlo method decreases accordingly. Therefore, there is an urgent need to propose an air conditioning load aggregation method that considers parameter heterogeneity. Summary of the Invention

[0004] To address the above technical problems, this invention provides a model predictive control-based intelligent air conditioning load aggregation and regulation system and method.

[0005] The technical solution of this invention is a model predictive control intelligent air conditioning load aggregation and control system, comprising:

[0006] Central controller, multiple smart air conditioners;

[0007] The central controller is wirelessly connected to multiple smart air conditioners in sequence;

[0008] The central controller constructs a first-order equivalent thermal parameter model of a single air conditioning load considering random disturbances, and uses the least squares method to solve for the characteristic parameters of each air conditioner. A certain temperature range is uniformly divided into multiple temperature intervals, and a state space model is constructed, including sets of on / off states, sets of off states, and the current set of air conditioners. Each matrix in the state space model is solved. The state variables at multiple future moments are predicted, and the optimal control information for these future moments is determined. The adjustment error at the current moment is calculated as a feedback signal and used as input to the MPC controller. Combined with the MPC controller, rolling optimization is performed to obtain the state control vectors of the multiple intelligent air conditioners at the next moment. Based on the state control vectors of the multiple intelligent air conditioners at the next moment, the central controller controls the state of the multiple air conditioners at the next moment.

[0009] The technical solution of this invention is a model predictive control-based intelligent air conditioning load aggregation and regulation method, as detailed below:

[0010] Step 1: Construct a first-order equivalent thermal parameter model of a single air conditioning load considering random disturbances, and use the least squares method to solve for the characteristic parameters of each air conditioner;

[0011] Step 2: Obtain the current on / off status and indoor temperature of multiple air conditioners. Divide a certain temperature range evenly into multiple temperature intervals. Assign each air conditioner to its corresponding temperature interval based on its current indoor temperature. Within the same temperature interval, construct and set an on / off state set. Place the air conditioners in the on / off state into the on / off set and the air conditioners in the off state into the off set. Normalize the number of air conditioners in each set to obtain the ratio of the number of air conditioners in each set to the total number of air conditioners at the current moment. Construct the state variables and state space model of the air conditioner sets at the current moment and solve for each matrix in the state space model.

[0012] Step 3: Based on the state variables at the current moment, predict the state variables at multiple future moments using the state space model, and solve for the optimal control information at multiple future moments based on the state variables at multiple future moments;

[0013] Step 4: Calculate the adjustment error at the current moment as a feedback signal, use the adjustment error at the current moment as the input of the MPC controller, and combine the MPC controller to perform rolling optimization to obtain the state control vector of multiple smart air conditioners at the next moment.

[0014] Step 5: Based on the status control vector of multiple smart air conditioners at the next moment, the central controller randomly selects multiple smart air conditioners as the on state and wirelessly transmits the corresponding on control signals to the corresponding air conditioners. The remaining smart air conditioners are in the off state, and the corresponding off control signals are wirelessly transmitted to the corresponding air conditioners.

[0015] Preferably, the first-order equivalent thermal parameter model for a single air conditioning load considering random disturbances described in step 1 is specifically defined as follows:

[0016] T in,l,k+1 =a l *T in,l,k +(1-a l )*(T out,l -λ l *T g,l )+w l,k

[0017] Where, λ l R is a dimensionless quantity for the l-th air conditioner, which is 1 when the air conditioner is on and 0 when it is off; l C l For the equivalent thermal resistance and equivalent heat capacity of the air conditioner, l = 1, 2, 3…, N tcls N is the number for the air conditioning load. tcls This represents the total number of air conditioners. Δt is the time step, T. g,l For the temperature gain of the l-th air conditioner, T in,l,k Let T be the indoor temperature at time k of the l-th air conditioner. out,l,k Let be the outdoor temperature of the l-th air conditioner at time k;

[0018] Step 1 describes using the least squares method to solve for the characteristic parameters of each air conditioner, as follows:

[0019] Let be a characteristic parameter of the l-th air conditioner. This parameter is obtained using the least squares method based on the temperature change data of the air conditioner over a period of time. The specific solution is as follows:

[0020]

[0021] Preferably, step 2 involves uniformly dividing a certain temperature range into multiple temperature intervals, as detailed below:

[0022] Temperature ranges are distinguished by numbers m = 1, 2, 3, ..., M; the temperature range [T] is further divided into... min ,T max If a temperature is uniformly divided into M temperature intervals, then the m-th temperature interval can be represented as [T m ,T m+1 ];

[0023] The set of all air conditioners that are currently off and the set of all air conditioners that are currently on are defined as follows:

[0024] close k ={Statebin close,k,1 Statebin close,k,2Statebin close,k,3 ,…,Statebin close,k,M}

[0025] open k ={Statebin open,k,1 Statebin open,k,2 Statebin open,k,3 ,…,Statebin open,k,M}

[0026] Among them, close k Statebin is the set of all air conditioners that are turned off at time k. close,k,m The set of all air conditioners in the off state at time k, temperature range m; open k Statebin is the set of all air conditioners turned on at time k. open,k,m Let k represent the set of all air conditioners in the m-th temperature range at time k, where k represents the current time.

[0027] Since there are two sets for each temperature interval, there are a total of 2M sets for M temperature intervals.

[0028] Step 2 involves normalizing the number of air conditioners in each air conditioner set, as detailed below:

[0029] Define Load k,n n = 1, 2, 3, ..., N represents the set of the nth air conditioning group at the kth time, where N = 2M, Load k,n Satisfy the following relations:

[0030]

[0031] The number of air conditioners in each air conditioner set is normalized to the form shown below:

[0032]

[0033] Where, x k,n This represents the ratio of the number of air conditioners in the nth air conditioner set at time k to the total number of air conditioners, where N represents the total number of air conditioner sets, M represents the total number of temperature ranges, and k represents the current time.

[0034] Step 2 involves constructing the state variables of the current air conditioner set, as detailed below:

[0035] x k,n Defined as the state of an air conditioner set, for 2M air conditioner sets, there are a total of 2M state variables, defined as follows:

[0036] X k =[xk,1 ,x k,2 ,x k,3 ,…,x k,N ]

[0037] Among them, X k Let k be the state variable of the air conditioning set at time k, where k represents the current time.

[0038] For state variable X k+1 The factors determining the change are as follows: the state variable X at time k. k The control signal U at time k k And the random disturbance W during the control process at time k. k ;

[0039] The state-space model described in step 2 is defined as follows:

[0040] X k+1 =AX k +BU k +W k

[0041] y k =CX k

[0042] U k =[u k,1 ,u k,2 ,u k,3 ,…,u k,N ]

[0043] Among them, U k Let u represent the control variable at time k. k,n To apply the load to the controller at time k k,n , y k W is the sum of the power of fixed-frequency air conditioners across all temperature ranges at time k. k Let A be the random disturbance at time k, matrix A be the state transition matrix, matrix B be the control input matrix, matrix C be the output matrix, and k represent the current time.

[0044] Step 2 involves solving for each matrix in the state-space model, as detailed below:

[0045] Matrix A is an N×N matrix. The elements of A are P. i,j , i,j∈N. Where, P i,j Let be the transition probability between two state boxes, where the subscript indicates the transition from the air conditioner in the j-th state box to the air conditioner in the i-th state box. Its specific form is as follows:

[0046]

[0047] Among them, T j ,T j+1 T represents the upper and lower bounds of the temperature range in which the j-th state box is located, respectively. i ,T i+1 Let a1 and a2 represent the upper and lower bounds of the temperature range of the i-th state box, respectively. i ,T i+1 The two parameters that determine this are M, where M is the number of temperature ranges;

[0048] Where p(a) is the probability density function of a uniform distribution, in the following form:

[0049]

[0050] Where a min ,a max It can be obtained by least squares fitting as described in step 1.

[0051] Write B in the following form:

[0052]

[0053] The form of C is as follows:

[0054] C = P rate *N tcls *[0...0|1...1] 1×N

[0055] Among them, P rate For the rated power of the air conditioner, N tcls Let N be the total number of air conditioning loads, and N be the total number of air conditioning units.

[0056] As a preferred embodiment, step 3, which involves predicting the state variables at multiple future time points, is specifically as follows:

[0057] Given the state variables at time k, the state-space model can predict the state variables at time k+1. Through an iterative process, N can be predicted. p The state variables at each time point are as follows:

[0058]

[0059] Where matrix A is the state transition matrix, matrix B is the control input matrix, and N... p To predict the step size, X k Let U be the state variable of the air conditioning set at time k. k+q Let q represent the control variable at time k+q, where k represents the current time, k+1 represents the next time, and q∈[0,N]. p ];

[0060] Step 3, which involves solving for the optimal control information for multiple future time periods based on the state variables at multiple future time periods, is as follows:

[0061] According to N p The state variables at each time step can be used to calculate N. p The predicted output at time step 1 is shown below:

[0062]

[0063] Among them, y k+q Let q represent the sum of the power of fixed-frequency air conditioners across all temperature ranges k+q, where q∈[0,N]. p ];

[0064] make:

[0065]

[0066]

[0067]

[0068] Where matrix A is the state transition matrix, matrix B is the control input matrix, matrix C is the output matrix, and N... p To predict the step size, For N p The control signal at time t, Φ is the state variable transition matrix, and F is the control signal transition matrix. For N p Control signals at each moment;

[0069] Therefore, we can conclude that:

[0070]

[0071] Among them, X k Let Y be the state variable of the air conditioning set at time k, and let N be the state variable of the set. p The predicted output at each time step, where Φ is the state variable transition matrix and F is the control signal transition matrix;

[0072] Define the error performance metrics for power point tracking as follows:

[0073]

[0074] Among them, R s Let Y be the input reference signal, and Y be N. p The predicted output at each time step. For N p The control signal at each moment, G is the weight matrix of the control action, which can limit the frequent switching of the system, and can also give different weights to the control sequence according to the simulation expectation, that is, change the value of the G matrix.

[0075] Based on the above objective function and constraints, the optimal N can be solved using the quadratic programming solver in MATLAB. p Control signals at each moment;

[0076] Preferably, step 4, calculating the adjustment error at the current moment, is as follows:

[0077] e k =r sk -y k +w k

[0078] Among them, e k Let r be the adjustment error at time k. sk Let y be the target power at time k. k w is the sum of the power of fixed-frequency air conditioners across all temperature ranges at time k. k Let be the prediction error at time k;

[0079] Step 4 involves combining the MPC controller with rolling optimization to obtain the control action value for the next time step, as detailed below:

[0080] The rolling optimization method in the MPC controller involves continuously adjusting U(k) based on the predicted output power at the next moment using the state-space model to minimize the difference between the output power and the target power. Specifically:

[0081]

[0082] Among them, R s Let Y be the input reference signal, and Y be N. p The predicted output at each time step. For N p The control signal at each time point, where G is the weight matrix of the control action;

[0083] The state control vector for multiple smart air conditioners at the next moment, as described in step 4, includes:

[0084] The state control vector for multiple smart air conditioners at the next moment includes the number of smart air conditioners in the on state and the number of smart air conditioners in the off state at the next moment.

[0085] The aforementioned model predictive control method for regulating air conditioning load groups predicts future power output by establishing a aggregate state-space model of the air conditioning load group. Furthermore, in the presence of random disturbances, it employs the concept of rolling optimization in the model prediction control, using the random disturbances as input to the controller at future time steps. The MPC controller then solves for the control quantity at the next time step and performs adjustments. As time progresses, rolling optimization is performed, and the entire negative feedback control system continuously adjusts the load group output to approach the target power, thus achieving overall scheduling of the air conditioning group. Attached Figure Description

[0086] Figure 1 : Flowchart of the method according to an embodiment of the present invention;

[0087] Figure 2 : A diagram of the first-order equivalent thermal parameter model of air conditioning load in an embodiment of the present invention;

[0088] Figure 3 : A one-dimensional state box model diagram of an embodiment of the present invention;

[0089] Figure 4 : Flowchart of the control system according to an embodiment of the present invention;

[0090] Figure 5 : Target power variation diagram of an embodiment of the present invention;

[0091] Figure 6 The diagram showing the power variation of the air conditioning group in Experiment 1 of this embodiment of the invention;

[0092] Figure 7 The diagram showing the power variation of the air conditioning group in Experiment 2 of this embodiment of the invention; Detailed Implementation

[0093] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0094] In specific implementation, the method proposed in the technical solution of this invention can be automatically executed by those skilled in the art using computer software technology. System devices for implementing the method, such as computer-readable storage media storing the corresponding computer program of the technical solution of this invention and computer equipment including the computer program running the corresponding computer program, should also be within the protection scope of this invention.

[0095] The technical solution of the system in this embodiment of the invention is a model predictive control intelligent air conditioning load aggregation and control system, comprising:

[0096] Central controller, multiple smart air conditioners;

[0097] The central controller is wirelessly connected to multiple smart air conditioners in sequence;

[0098] The central controller is model Midea KJR-37B / bp2;

[0099] The model of the smart air conditioner is Midea KFR-35GW / WDAA21AU1;

[0100] In a specific scenario of an embodiment of the present invention, the control capability of one thousand air conditioners is selected; and a control experiment is set up regarding the number of state boxes to determine the impact of different numbers of state boxes on the tracking error.

[0101] The following is combined with Figure 1-7 The technical solution of the method in the embodiments of the present invention is introduced, namely, a model predictive control method for intelligent air conditioning load aggregation and regulation, as detailed below:

[0102] like Figure 1 The above is a flowchart of a method according to an embodiment of the present invention.

[0103] Step 1: Construct a first-order equivalent thermal parameter model of a single air conditioning load considering random disturbances, and use the least squares method to solve for the characteristic parameters of each air conditioner;

[0104] The first-order equivalent thermal parameter model of a single air conditioning load considering random disturbances described in step 1 is as follows: Figure 2 As shown, the specific definitions are as follows:

[0105] T in,l,k+1 =a l *T in,l,k +(1-a l )*(T out,l -λ l *T g,l )+w l,k

[0106] Where, λ l R is a dimensionless quantity for the l-th air conditioner, which is 1 when the air conditioner is on and 0 when it is off; l C l For the equivalent thermal resistance and equivalent heat capacity of the air conditioner, l = 1, 2, 3…, N tcls N is the number for the air conditioning load. tcls =1000 represents the total number of air conditioners; Δt=15 represents the time step, T g,l For the temperature gain of the l-th air conditioner, T in,l,k Let T be the indoor temperature at time k of the l-th air conditioner. out,l,k =38 represents the outdoor temperature of the l-th air conditioner at time k;

[0107] Step 1 describes using the least squares method to solve for the characteristic parameters of each air conditioner, as follows:

[0108] Let be a characteristic parameter of the l-th air conditioner. This parameter is obtained using the least squares method based on the temperature change data of the air conditioner over a period of time. The specific solution is as follows:

[0109]

[0110] Step 2: Obtain the current on / off status and indoor temperature of multiple air conditioners. Divide a certain temperature range evenly into multiple temperature intervals. Assign each air conditioner to its corresponding temperature interval based on its current indoor temperature. Within the same temperature interval, construct and set an on / off state set. Place the air conditioners in the on / off state into the on / off set and the air conditioners in the off state into the off set. Normalize the number of air conditioners in each set to obtain the ratio of the number of air conditioners in each set to the total number of air conditioners at the current moment. Construct the state variables and state space model of the air conditioner sets at the current moment and solve for each matrix in the state space model.

[0111] Step 2 involves uniformly dividing a certain temperature range into multiple temperature intervals, as detailed below:

[0112] Temperature ranges are distinguished by numbers m = 1, 2, 3, ..., M; the temperature range [T] is further divided into... min ,T max If a temperature is uniformly divided into M temperature intervals, then the m-th temperature interval can be represented as [T m ,T m+1 ];

[0113] The set of all air conditioners currently off, and the set of all air conditioners currently on, as shown below. Figure 3 As shown, they are defined as follows:

[0114] close k ={Statebin close,k,1 Statebin close,k,2 Statebin close,k,3 ,…,Statebin close,k,M}

[0115] open k ={Statebin open,k,1 Statebin open,k,2 Statebin open,k,3 ,…,Statebin open,k,M}

[0116] Among them, close k Statebin is the set of all air conditioners that are turned off at time k. close,k,m The set of all air conditioners in the off state at time k, temperature range m; open k Statebin is the set of all air conditioners turned on at time k. open,k,m Let k represent the set of all air conditioners in the m-th temperature range at time k, where k represents the current time.

[0117] Since there are two sets for each temperature interval, there are a total of 2M sets for M temperature intervals.

[0118] Step 2 involves normalizing the number of air conditioners in each air conditioner set, as detailed below:

[0119] Define Load k,n n = 1, 2, 3, ..., N represents the set of the nth air conditioning group at the kth time, where N = 2M, Load k,n Satisfy the following relations:

[0120]

[0121] The number of air conditioners in each air conditioner set is normalized to the form shown below:

[0122]

[0123] Where, x k,n This represents the ratio of the number of air conditioners in the nth air conditioner set at time k to the total number of air conditioners, where N represents the total number of air conditioner sets, M represents the total number of temperature ranges, and k represents the current time.

[0124] Step 2 involves constructing the state variables of the current air conditioner set, as detailed below:

[0125] x k,n Defined as the state of an air conditioner set, for 2M air conditioner sets, there are a total of 2M state variables, defined as follows:

[0126] X k =[x k,1 ,x k,2 ,x k,3 ,…,x k,N ]

[0127] Among them, X k Let k be the state variable of the air conditioning set at time k, where k represents the current time.

[0128] For state variable X k+1 The factors determining the change are as follows: the state variable X at time k. kThe control signal U at time k k And the random disturbance W during the control process at time k. k ;

[0129] The state-space model described in step 2 is defined as follows:

[0130] X k+1 =AX k +BU k +W k

[0131] y k =CX k

[0132] U k =[u k,1 ,u k,2 ,u k,3 ,…,u k,N ]

[0133] Among them, U k Let u represent the control variable at time k. k,n To apply the load to the controller at time k k,n , y k W is the sum of the power of fixed-frequency air conditioners across all temperature ranges at time k. k Let A be the random disturbance at time k, matrix A be the state transition matrix, matrix B be the control input matrix, matrix C be the output matrix, and k represent the current time.

[0134] Step 2 involves solving for each matrix in the state-space model, as detailed below:

[0135] Matrix A is an N×N matrix. The elements of A are P. i,j , i,j∈N. Where, P i,j Let be the transition probability between two state boxes, where the subscript indicates the transition from the air conditioner in the j-th state box to the air conditioner in the i-th state box. Its specific form is as follows:

[0136]

[0137] Among them, T j ,T j+1 T represents the upper and lower bounds of the temperature range in which the j-th state box is located, respectively. i ,T i+1 Let a1 and a2 represent the upper and lower bounds of the temperature range of the i-th state box, respectively. i ,T i+1 The two parameters that determine this are M, where M is the number of temperature ranges;

[0138] Where p(a) is the probability density function of a uniform distribution, in the following form:

[0139]

[0140] Where a min ,a max It can be obtained by least squares fitting as described in step 1.

[0141] Write B in the following form:

[0142]

[0143] The form of C is as follows:

[0144] C = P rate *N tcls *[0...0|1...1] 1×N

[0145] Among them, P rate For the rated power of the air conditioner, N tcls Let N be the total number of air conditioning loads, and N be the total number of air conditioning units.

[0146] Step 3: Based on the state variables at the current moment, predict the state variables at multiple future moments using the state-space model, and then solve for the optimal control information at these multiple future moments based on the state variables at these multiple future moments.

[0147] Step 3, which involves predicting the state variables at multiple future time points, is detailed below:

[0148] Given the state variables at time k, the state-space model can predict the state variables at time k+1. Through an iterative process, N can be predicted. p The state variables at each time point are as follows:

[0149]

[0150] Where matrix A is the state transition matrix, matrix B is the control input matrix, and N... p To predict the step size, X k Let U be the state variable of the air conditioning set at time k. k+q Let q represent the control variable at time k+q, where k represents the current time, k+1 represents the next time, and q∈[0,N]. p ];

[0151] Step 3, which involves solving for the optimal control information for multiple future time periods based on the state variables at multiple future time periods, is as follows:

[0152] According to N p The state variables at each time step can be used to calculate N. pThe predicted output at time step 1 is shown below:

[0153]

[0154] Among them, y k+q Let q represent the sum of the power of fixed-frequency air conditioners across all temperature ranges k+q, where q∈[0,N]. p ];

[0155] make:

[0156]

[0157]

[0158]

[0159] Where matrix A is the state transition matrix, matrix B is the control input matrix, matrix C is the output matrix, and N... p To predict the step size, For N p The control signal at time t, Φ is the state variable transition matrix, and F is the control signal transition matrix. For N p Control signals at each moment;

[0160] Therefore, we can conclude that:

[0161]

[0162] Among them, X k Let Y be the state variable of the air conditioning set at time k, and let N be the state variable of the set. p The predicted output at each time step, where Φ is the state variable transition matrix and F is the control signal transition matrix;

[0163] Define the error performance metrics for power point tracking as follows:

[0164]

[0165] Among them, R s Let Y be the input reference signal, and Y be N. p The predicted output at each time step. For N p The control signal at each moment, G is the weight matrix of the control action, which can limit the frequent switching of the system, and can also give different weights to the control sequence according to the simulation expectation, that is, change the value of the G matrix.

[0166] Based on the above objective function and constraints, the optimal N can be solved using the quadratic programming solver in MATLAB. p Control signals at each moment;

[0167] Step 4: Calculate the adjustment error at the current moment as a feedback signal, use the adjustment error at the current moment as the input of the MPC controller, and combine the MPC controller to perform rolling optimization to obtain the state control vector of multiple smart air conditioners at the next moment.

[0168] Step 4, which involves calculating the adjustment error at the current moment, is as follows:

[0169] e k =r sk -y k +w k

[0170] Among them, e k Let r be the adjustment error at time k. sk Let y be the target power at time k. k w is the sum of the power of fixed-frequency air conditioners across all temperature ranges at time k. k Let be the prediction error at time k;

[0171] Step 4 involves combining the MPC controller with rolling optimization to obtain the control action value for the next time step, as detailed below:

[0172] The rolling optimization method in the MPC controller involves continuously adjusting U(k) based on the predicted output power at the next moment using the state-space model to minimize the difference between the output power and the target power. Specifically:

[0173]

[0174] Among them, R s Let Y be the input reference signal, and Y be N. p The predicted output at each time step. For N p The control signal at each time point, where G is the weight matrix of the control action;

[0175] The state control vector for multiple smart air conditioners at the next moment, as described in step 4, includes:

[0176] The state control vector for multiple smart air conditioners at the next moment includes the number of smart air conditioners in the on state and the number of smart air conditioners in the off state at the next moment.

[0177] Step 5: Based on the status control vector of multiple smart air conditioners at the next moment, the central controller randomly selects multiple smart air conditioners as the on state and wirelessly transmits the corresponding on control signals to the corresponding air conditioners. The remaining smart air conditioners are in the off state, and the corresponding off control signals are wirelessly transmitted to the corresponding air conditioners.

[0178] In one detailed embodiment, we built a simulation platform containing one thousand air conditioners, such as... Figure 4 As shown, with a random target power as... Figure 5 The signal shown is a reference signal.

[0179] The following two control experiments were conducted:

[0180] Experiment 1: Set the number of state boxes to 20 and evenly distribute the temperatures of the state boxes.

[0181] Experiment 2: Set the number of state boxes to 40 and evenly distribute the temperatures of the state boxes.

[0182] For the above experiments, the scheduling method described in this invention was adopted. From the experimental results... Figure 6 and Figure 7 It can be seen that the experimental scheme can schedule the air conditioning group within a certain time step and can greatly reduce the impact of random interference on the entire control process, enabling it to track the target power.

[0183] Furthermore, the comparison between the two experiments shows that increasing the number of state boxes can reduce tracking error, but it also increases communication lines, thereby raising control costs. Users can choose the number of state boxes according to their own control needs.

[0184] Compared with other schemes, the above-mentioned model predictive control scheme for air conditioning load group scheduling only requires the equivalent capacitance and conductance parameters of the air conditioning loads to model them, and the aggregate model is not affected by the distribution of air conditioning units. During the control process, this scheme achieves target power tracking of the air conditioning group considering random disturbances in the model prediction.

[0185] It should be understood that any parts not described in detail in this specification belong to the prior art.

[0186] Although this document uses terms such as central controller, mobile user terminal, and network access point frequently, the possibility of using other terms is not excluded. These terms are used merely for the convenience of describing the essence of the invention, and interpreting them as any additional limitation would contradict the spirit of the invention.

[0187] It should be understood that the above description of the preferred embodiments is quite detailed, but it should not be considered as a limitation on the scope of protection of this invention. Those skilled in the art, under the guidance of this invention, can make substitutions or modifications without departing from the scope of protection of the claims of this invention, and all such substitutions or modifications fall within the scope of protection of this invention. The scope of protection of this invention should be determined by the appended claims.

Claims

1. A model predictive control-based intelligent air conditioning load aggregation and control system, characterized in that, include: Central controller, multiple smart air conditioners; The central controller is wirelessly connected to multiple smart air conditioners in sequence; The central controller constructs a first-order equivalent thermal parameter model of a single air conditioning load considering random disturbances, and uses the least squares method to solve for the characteristic parameters of each air conditioner. A certain temperature range is uniformly divided into multiple temperature intervals, and a state space model is constructed, including sets of on / off states, sets of off states, and the current set of air conditioners. Each matrix in the state space model is solved. The state variables at multiple future moments are predicted, and the optimal control information for these future moments is determined. The adjustment error at the current moment is calculated as a feedback signal and used as input to the MPC controller. Combined with the MPC controller, rolling optimization is performed to obtain the state control vectors of the multiple intelligent air conditioners at the next moment. Based on the state control vectors of the multiple intelligent air conditioners at the next moment, the central controller controls the state of the multiple air conditioners at the next moment.

2. A model predictive control method for intelligent air conditioning load aggregation control applied to the model predictive control intelligent air conditioning load aggregation control system of claim 1, characterized in that: Includes the following steps: Step 1: Construct a first-order equivalent thermal parameter model of a single air conditioning load considering random disturbances, and use the least squares method to solve for the characteristic parameters of each air conditioner; Step 2: Obtain the current on / off status and indoor temperature of multiple air conditioners. Divide a certain temperature range evenly into multiple temperature intervals. Assign each air conditioner to its corresponding temperature interval based on its current indoor temperature. Within the same temperature interval, construct and set an on / off state set. Place the air conditioners in the on / off state into the on / off set and the air conditioners in the off state into the off set. Normalize the number of air conditioners in each set to obtain the ratio of the number of air conditioners in each set to the total number of air conditioners at the current moment. Construct the state variables and state space model of the air conditioner sets at the current moment and solve for each matrix in the state space model. Step 3: Based on the state variables at the current moment, predict the state variables at multiple future moments using the state space model, and solve for the optimal control information at multiple future moments based on the state variables at multiple future moments; Step 4: Calculate the adjustment error at the current moment as a feedback signal, use the adjustment error at the current moment as the input of the MPC controller, and combine the MPC controller to perform rolling optimization to obtain the state control vector of multiple smart air conditioners at the next moment. Step 5: Based on the status control vector of multiple smart air conditioners at the next moment, the central controller randomly selects multiple smart air conditioners to be in the "on" state and wirelessly transmits the corresponding "on" control signals to the corresponding air conditioners. The remaining smart air conditioners are in the "off" state, and the corresponding "off" control signals are wirelessly transmitted to the corresponding air conditioners.

3. The intelligent air conditioning load aggregation and regulation method based on model predictive control according to claim 2, characterized in that: The first-order equivalent thermal parameter model for a single air conditioning load considering random disturbances, as described in step 1, is specifically defined as follows: in, For the first The dimensionless quantity of an air conditioner is 1 when the air conditioner is on and 0 when it is off. The equivalent thermal resistance and equivalent heat capacity of the air conditioner. The numbering of the air conditioning load. This represents the total number of air conditioners; For time step, For the first Temperature gain of the air conditioner For the first Air conditioner k Indoor temperature at any given time For the first Air conditioner k The outdoor temperature at any given time; Step 1 describes using the least squares method to solve for the characteristic parameters of each air conditioner, as follows: For the first The characteristic parameter of the air conditioner is obtained by using the temperature change data of the air conditioner over a period of time and the least squares method. The specific solution is as follows: 。 4. The intelligent air conditioning load aggregation and regulation method based on model predictive control according to claim 3, characterized in that: Step 2 involves uniformly dividing a certain temperature range into multiple temperature intervals, as detailed below: Use number To distinguish temperature ranges; to divide temperature ranges Evenly divided into The first temperature range, then the second... Each temperature range is represented as ; The set of all air conditioners that are currently off and the set of all air conditioners that are currently on are defined as follows: in, For the first The collection of all air conditioners turned off at any given moment. For the first At the [time]th moment The set of all air conditioners in a closed state within a temperature range; For the first The collection of all air conditioners turned on at any given moment. For the first At the [time]th moment A collection of all air conditioners in operation within a given temperature range. Indicates the current time; Since a temperature range has two sets, then for There are a total of temperature ranges A set; Step 2 involves normalizing the number of air conditioners in each air conditioner set, as detailed below: definition Indicates the first The moment of the first A collection of air conditioning groups, among which , Satisfy the following relations: The number of air conditioners in each air conditioner set is normalized to the form shown below: in, Indicates the first At the [time]th moment The ratio of the number of air conditioners in a single air conditioning unit to the total number of air conditioners. N This indicates the total number of air conditioners in the unit. M Indicates the total number of temperature ranges. Indicates the current moment.

5. The intelligent air conditioning load aggregation and regulation method based on model predictive control according to claim 4, characterized in that: Step 2 involves constructing the state variables of the current air conditioner set, as detailed below: Will Defined as the state of an air conditioning unit, for There are 1 air conditioner collection, totaling 1 There are several state variables, defined as follows: in, For the first The state variables of the air conditioning set at any given time. Indicates the current time; For state variables The factors determining the changes are as follows: Time-state variables , No. Time control signal and the Random disturbances in the time-control process ; The state-space model described in step 2 is defined as follows: in, Indicates the first Control variables at each moment, In the first The controller applies at each moment. , control signals, For the first The sum of the power of a fixed-frequency air conditioner across all temperature ranges at any given moment. For the first Random disturbances at time t, matrix The state transition matrix is ​​a matrix that... To control the input matrix, the matrix For the output matrix, Indicates the current moment.

6. The intelligent air conditioning load aggregation and regulation method based on model predictive control according to claim 5, characterized in that: Step 2 involves solving for each matrix in the state-space model, as detailed below: Matrix A is a Matrix; The middle element is , ;in, Let be the transition probability between two state boxes, and let the subscript indicate the nth state box. The air conditioner in the status box is directing the first... The specific forms of air conditioner switching in each status box are as follows: in, They represent the first The upper and lower bounds of the temperature range in which each state chamber is located. They represent the first The upper and lower bounds of the temperature range in which each state chamber is located; They represent respectively by The two parameters that determine this are: The number of temperature ranges; in The probability density function is a uniform distribution, and its form is as follows: in It can be obtained by least squares fitting as described in step 1; Will Write it in the following form: The format is as follows: in, This refers to the rated power of the air conditioner. This represents the total number of air conditioning loads. This represents the total number of air conditioner units.

7. The intelligent air conditioning load aggregation control method based on model predictive control according to claim 6, characterized in that: Step 3, which involves predicting the state variables at multiple future time points, is detailed below: Based on the state-space model, in the known... When predicting the state variables at time 1, the first time is... The state variable at any given time is predicted through an iterative process. The state variables at each time point are as follows: Among them, matrix The state transition matrix is ​​a matrix that... To control the input matrix, To predict the step size, For the first The state variables of the air conditioning set at any given time. Indicates the first Control variables at each moment, Indicates the current moment. +1 indicates the next time step, q∈[0, ].

8. The intelligent air conditioning load aggregation control method based on model predictive control according to claim 7, characterized in that: Step 3, which involves solving for the optimal control information for multiple future time periods based on the state variables at multiple future time periods, is as follows: according to The state variables at each time point can be obtained. The predicted output at time step 1 is shown below: in, Indicates the first The sum of the power of fixed-frequency air conditioners across all temperature ranges, q∈[0, ]; make: Among them, matrix The state transition matrix is ​​a matrix that... To control the input matrix, the matrix For the output matrix, To predict the step size, for Control signals at each moment, For the state variable transition matrix, To control the signal transition matrix, for Control signals at each moment; Therefore, we can conclude that: in, For the first The state variables of the air conditioning set at any given time. for The predicted output at each time step. For the state variable transition matrix, For control signal transition matrix; Define the error performance metrics for power point tracking as follows: in, The input reference signal, for The predicted output at each time step. for Control signals at each moment, The weight matrix serves as the control effect, used to limit frequent switching of the system. Simultaneously, based on simulation expectations, different weights are assigned to the control sequence, i.e., the weights are changed. The value of the matrix; Based on the objective function and constraints, the optimal solution is obtained using the quadratic programming solver in MATLAB. Control signals at each moment.

9. The intelligent air conditioning load aggregation control method based on model predictive control according to claim 8, characterized in that: Step 4, which involves calculating the adjustment error at the current moment, is as follows: in, For the first Timing adjustment error, For the first The target power at each moment For the first The sum of the power of a fixed-frequency air conditioner across all temperature ranges at any given moment. For the first The prediction error at each time point; Step 4 involves combining the MPC controller with rolling optimization to obtain the control action value for the next time step, as detailed below: The rolling optimization method in the MPC controller involves continuously adjusting the output power based on the prediction of the next time step's output power using the state-space model. To minimize the difference between the output power and the target power, specifically: in, The input reference signal, for The predicted output at each time step. for Control signals at each moment, The weight matrix is ​​for control purposes; The state control vector for multiple smart air conditioners at the next moment, as described in step 4, includes: The state control vector for multiple smart air conditioners at the next moment includes the number of smart air conditioners in the on state and the number of smart air conditioners in the off state at the next moment.

Images