A quantification method for the incentive-response characteristics of central air-conditioning load resources based on BP neural network

Abstract

The present invention discloses a method for quantifying the excitation-response characteristics of central air-conditioning load resources based on BP neural network, which belongs to the field of power load regulation and control, and solves the problems of large deviation and low accuracy when the existing quantitative evaluation method is used to evaluate the excitation-response characteristics of central air-conditioning load. The method steps are as follows: first, the central air-conditioning operation information is collected, then, a central air-conditioning startup and shutdown operation constraint model is constructed, secondly, a central air-conditioning start-stop loss model based on BP neural network is constructed, and finally, a central air-conditioning excitation-response characteristic model based on BP neural network is constructed. The present invention more accurately quantifies and evaluates the excitation-response characteristics of the central air-conditioning, thereby improving the effectiveness of central air-conditioning regulation and control, and at the same time improves the ability to mine central air-conditioning load resources, and finally improves the overall regulation ability of the new power system.

Description

Technical Field

[0001] The invention belongs to the field of electric load regulation and control, and in particular relates to a method for quantifying the excitation-response characteristics of central air-conditioning load resources based on a BP neural network. Background Art

[0002] It is of great significance to accurately and quantitatively evaluate the incentive-response characteristics of central air-conditioning loads. Through accurate and quantitative evaluation, the response behavior of central air-conditioning loads can be predicted more accurately, so as to formulate more effective load regulation strategies and avoid over-regulation or under-regulation. This helps the power grid to conduct more accurate demand response management during high-load periods, reduce the load pressure of the power grid, and improve the stability and reliability of the power grid. Accurate evaluation helps to discover the maximum regulation potential of the central air-conditioning system, optimize its operation strategy, and achieve more efficient energy use. The potential of the central air-conditioning system in demand response can be better utilized to participate in the auxiliary services such as peak load regulation and frequency regulation of the power grid, thereby improving resource utilization efficiency. Therefore, mastering the incentive-response characteristics of central air-conditioning loads is a prerequisite for effectively dispatching central air-conditioning to participate in the operation of new power systems. However, the existing methods still have the following shortcomings: 1) The factors considered are not comprehensive enough. Many studies have not fully considered the parameters such as the structure and thermal insulation of the building, which have a significant impact on the energy consumption and load response of the air conditioner; the change of the outside temperature directly affects the operation mode and energy consumption of the air conditioner, and some studies have not fully considered this factor when modeling; the start and stop frequency of the air conditioner will affect the life and energy efficiency of the equipment, but when modeling, the minimum start and stop time limit of the air conditioner is often ignored; the density and activity level of people in the building will affect the load demand of the air conditioner, but this factor is often ignored in existing studies; in the process of load excitation response, the impact of load loss on the system is not fully considered, which may lead to deviations in the response characteristics. 2) The response characteristics are not accurate enough. Since the influence of the above factors has not been fully included, the existing response characteristic model may not accurately reflect the actual situation, and the prediction results of the response characteristics may be biased. 3) The participation constraint and operation constraint are not fully considered. The participation constraint refers to the degree to which the air conditioning system can participate in the load excitation response process, which is restricted by many factors, such as the user's comfort requirements and the load capacity of the equipment. The existing methods have not fully considered these restrictions. The operation constraints include the physical operation limitations and maintenance requirements of the equipment. At the same time, these operational constraints were not fully considered during modeling, which may lead to operability problems in actual applications of the model. Summary of the invention

[0003] Based on the above problems, the present invention proposes a quantification method for the excitation-response characteristics of central air-conditioning load resources based on BP neural network, which solves the problem of large deviation and low accuracy when the existing quantitative evaluation method is used to evaluate the excitation-response characteristics of central air-conditioning load.

[0004] The technical solution adopted by the present invention is as follows: a method for quantifying the excitation-response characteristics of central air-conditioning load resources based on a BP neural network, the steps are as follows:

[0005] S1. Collect central air-conditioning operation information: including the rated cooling capacity of the chiller, chilled water temperature parameters, indoor comfort temperature range, air-conditioning operation status, total passenger flow inside the building, cooling area, effective area coefficient inside the building, passenger flow density per unit area, indoor temperature, refrigeration unit shutdown time, startup time, and incentive price;

[0006] S2. Construct a central air-conditioning startup and shutdown operation constraint model: including a startup operation constraint model and a shutdown operation constraint model. The startup operation constraint model aims to ensure that the indoor temperature does not drop to the lower limit of the comfort temperature after startup and meets the minimum shutdown time requirement; the shutdown operation constraint model aims to ensure that the indoor temperature does not rise to the upper limit of the comfort temperature after shutdown and meets the minimum startup time requirement.

[0007] S3. Constructing a central air-conditioning start-stop loss model based on BP neural network: training the collected central air-conditioning operation information based on BP neural network, constructing a central air-conditioning start-stop loss model based on BP neural network, and adjusting the coefficients in the model to obtain an accurate central air-conditioning start-stop loss model;

[0008] S4. Construct a central air conditioning incentive-response characteristic model based on BP neural network: Based on the accurate central air conditioning start-stop loss model obtained, construct a central air conditioning start-stop participation model and combine it with the central air conditioning startup and shutdown operation constraint model, and finally quantify and obtain accurate central air conditioning load incentive-response characteristics.

[0009] Furthermore, in step S1, formula (1) is used to collect the central air-conditioning operation information.

[0010]

[0011] Where X(t) is the central air conditioning information collected at time t; Q p is the rated cooling capacity of the chiller; β c AT is the chilled water temperature parameter; min is the minimum indoor temperature; AT max is the maximum indoor temperature; c (t) is the air conditioning operation state; N r (t) is the total passenger flow inside the building at time t; S c is the cooling area; is the effective area coefficient inside the building; η p (t) is the flow density of people per unit area at time t; cis the threshold of the density of people flow per unit area; T(t) is the indoor temperature at time t; is the most recent refrigeration unit shutdown time before time t; is the most recent start-up time of the refrigeration unit before time t; f ex (t) is the incentive price at time t.

[0012] Furthermore, in step S2, the central air-conditioning startup operation constraint model constructed is as shown in formula (2), δ c (t) = 0 means the central air conditioner is in shutdown state. Indicates that the central air conditioner shutdown time needs to be greater than 5 minutes, T on (t+Δt)>AT min Indicates that the indoor temperature must not be lower than the lower limit of the comfort range after the central air conditioner is in operation, η p (t)<λ c Indicates that the flow density per unit area shall not exceed the threshold.

[0013]

[0014] In the formula, δ c (t) is the operating state of the central air conditioner at time t, with a value of 0 indicating that it is in the shutdown state, and a value of 1 indicating that it is in the startup state; is the most recent refrigeration unit shutdown time before time t; T on (t+Δt) is the indoor temperature after the machine is turned on at time t for Δt; AT min is the lower limit of indoor comfortable temperature; a 1～3 is the building parameter; Q p is the rated cooling capacity of the chiller; T(t) is the indoor temperature at time t; η p (t) is the flow density of people per unit area at time t; c N is the threshold of pedestrian flow density per unit area; r (t) is the total passenger flow inside the building at time t; S c is the cooling area; is the effective area coefficient inside the building;

[0015] The constructed central air conditioning shutdown operation constraint model is shown in formula (3), where δ c (t) = 1 means the central air conditioner is on. Indicates that the central air conditioner needs to be turned on for more than 3 minutes, T off (t+Δt)<AT max Indicates that the indoor temperature shall not be higher than the upper limit of the comfort range after the central air conditioner participates in the grid operation, η p (t)<λ c Indicates that the flow density per unit area shall not exceed the threshold.

[0016]

[0017] In the formula, δ c (t) is the operating state of the central air conditioner at time t, with a value of 0 indicating that it is in the shutdown state, and a value of 1 indicating that it is in the startup state; is the most recent start-up time of the refrigeration unit before time t; T off (t+Δt) is the indoor temperature after the shutdown time Δt at time t; AT max is the upper limit of indoor comfortable temperature; a 1～4 is the building parameter; β c is the chilled water temperature parameter; T(t) is the indoor temperature at time t; η p (t) is the flow density of people per unit area at time t; c N is the threshold of pedestrian flow density per unit area; r (t) is the total passenger flow inside the building at time t; S c is the cooling area, is the effective area coefficient inside the building.

[0018] Furthermore, in step S3, a central air-conditioning start-stop loss model based on the BP neural network is constructed, including a start-up loss model and a stop-down loss model. Formula (4) is the central air-conditioning start-up loss model, formula (5) is the central air-conditioning stop-down loss model, and formula (6) is the load loss calculation model.

[0019]

[0020] In the formula, is the startup loss of the central air conditioner at time t; is the shutdown loss of the central air conditioner at time t; is the indoor temperature deviation lower limit coefficient; is the remaining startup time coefficient; is the passenger flow impact coefficient; is the load loss factor; is the indoor temperature deviation upper limit coefficient; Remaining downtime factor; T on (t+Δt) is the indoor temperature after the machine is turned on at time t for Δt; T off (t+Δt) is the indoor temperature after the power is turned off at time t for Δt; T set is the set indoor temperature; AT min AT is the lower limit of indoor comfortable temperature; max The upper limit of indoor comfortable temperature; The time until the next natural power-on; is the time until the next natural shutdown; τ onis the power-on cycle of the central air conditioner; τ off is the shutdown period of the central air conditioner; p (t) is the flow density of people per unit area at time t; X s (t) is the loss load of the central air conditioner at time t; X c (t) is the cooling load rate at time t; Q p is the rated cooling capacity of the chiller at time t;

[0021] The BP neural network is used to calculate the indoor temperature deviation lower limit coefficient in the startup loss model and the shutdown loss model. Remaining startup time coefficient Passenger flow impact coefficient Load loss factor Indoor temperature deviation upper limit coefficient Remaining downtime factor Adjustment is performed to obtain an accurate central air conditioning start-stop loss model.

[0022] Furthermore, a starting loss coefficient setting model based on BP neural network is constructed. By training and learning the central air-conditioning operation data samples, the features related to the starting loss are extracted, and then the starting loss coefficient of the central air-conditioning is accurately set. It includes 4 types of inputs and 4 outputs:

[0023] The four types of input are: central air conditioning operation data X(t) at time t, the time to the next natural start-up The incentive price f at time t ex (t), indoor comfort temperature lower limit AT min ;

[0024] The 4 outputs are: Indoor temperature deviation lower limit coefficient Remaining startup time coefficient Passenger flow impact coefficient Load loss factor

[0025] The training and learning specifically include the following methods: collecting central air-conditioning operation data as training samples, training the collected samples, conducting network testing on the trained startup loss coefficient setting model, and then offline training and testing the neural network model to obtain the BP neural network training function, connection weights, and thresholds for startup loss coefficient setting, extract sample features, and use the tansig function in the hidden layer of the neural network. The output of the hidden layer is shown in formula (7).

[0026]

[0027] In the formula, X'(t) represents the input of the training sample of the entire neural network model, X i'(t) represents the set of training samples that need to be introduced in each layer, i is the number of layers of neural network training, and 5 layers of training are required for the samples; h j The output of the hidden layer of the neural network is adjusted for the startup loss coefficient; W ij b. Adjust the connection weights from the neural network input layer unit to the hidden layer unit for the startup loss coefficient; b. j is the threshold value of the hidden layer unit of the neural network for starting the loss coefficient adjustment; j ranges from 1 to n; n is the number of hidden layer units of the neural network for starting the loss coefficient adjustment;

[0028] The output layer of the neural network uses the purelin function. The output of the neural network is shown in formula (9).

[0029]

[0030] In the formula, k m The output of the neural network is adjusted for the startup loss coefficient; w jm The connection weights from the hidden layer units to the output layer units of the neural network are adjusted to start the loss coefficient, and the range of m is 1 to 3;

[0031] Get accurate indoor temperature deviation lower limit coefficient Remaining startup time coefficient Passenger flow impact coefficient Load loss factor Then, substitute it into formula (4) to obtain an accurate startup loss model of the central air conditioner; based on the same principle, a shutdown loss coefficient setting model based on BP neural network is constructed to obtain an accurate shutdown loss model of the central air conditioner.

[0032] Furthermore, in step S4, a central air conditioner start-stop participation model is first constructed. Formula (10) is the central air conditioner start-up participation model, and formula (11) is the central air conditioner stop participation model.

[0033]

[0034] In the formula, f ex (t) is the incentive electricity price at time t; is the startup loss of central air conditioner c at time t; is the shutdown loss of central air conditioner c at time t;

[0035] Then, according to the central air-conditioning start-stop participation model, the central air-conditioning start-stop incentive-response characteristic curve can be obtained. When the central air-conditioning satisfies formula (2) and formula (10), according to the incentive electricity price f ex (t) Participate in the upward power response; When the central air conditioner satisfies formula (3) and formula (11), that is, according to the incentive electricity price fex (t) Participate in down-regulating power response.

[0036] Another object of the present invention is to achieve a central air-conditioning load resource excitation-response characteristic quantification system based on a BP neural network, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements a central air-conditioning load resource excitation-response characteristic quantification method based on a BP neural network as described above.

[0037] Another object of the present invention is to be achieved through a computer-readable storage medium, on which a program for implementing information transmission is stored, and when the program is executed by a processor, the steps of a method for quantifying the excitation-response characteristics of central air-conditioning load resources based on a BP neural network are implemented as described above.

[0038] Advantages and beneficial effects of the present invention: The present invention comprehensively considers the influence of factors such as building parameters, external temperature, minimum start-stop operation time of air conditioner, passenger flow density, load loss, etc., and more accurately quantifies and evaluates the incentive response characteristics of central air conditioners, thereby improving the effectiveness of central air conditioner regulation, while improving the ability to tap central air conditioner load resources, and ultimately improving the overall regulation capability of the new power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a flow chart of the quantification method of the present invention;

[0040] Figure 2 It is a startup loss coefficient setting diagram based on BP neural network;

[0041] Figure 3 It is the central air conditioning load excitation-response characteristic curve. DETAILED DESCRIPTION

[0042] The present invention will be further described below with reference to the accompanying drawings as examples:

[0043] Example 1

[0044] like Figure 1 As shown, a method for quantifying the incentive-response characteristics of central air-conditioning load resources based on BP neural network is as follows:

[0045] S1. Collect central air-conditioning operation information: including the rated cooling capacity of the chiller, chilled water temperature parameters, indoor comfort temperature range, air-conditioning operation status, total passenger flow inside the building, cooling area, effective area coefficient inside the building, passenger flow density per unit area, indoor temperature, chiller shutdown time, startup time, and incentive price. Formula (1) is used to collect central air-conditioning operation information.

[0046]

[0047] Where X(t) is the central air conditioning information collected at time t; Q p is the rated cooling capacity of the chiller; β c AT is the chilled water temperature parameter; min is the minimum indoor temperature; AT max is the maximum indoor temperature; c (t) is the air conditioning operation state; N r (t) is the total passenger flow inside the building at time t; S c is the cooling area; is the effective area coefficient inside the building; η p (t) is the flow density of people per unit area at time t; c is the threshold of the density of people flow per unit area; T(t) is the indoor temperature at time t; is the most recent refrigeration unit shutdown time before time t; is the most recent start-up time of the refrigeration unit before time t; f ex (t) is the incentive price at time t.

[0048] S2. Construct a central air-conditioning startup and shutdown operation constraint model: The premise for the central air-conditioning load to participate in the operation of the new power system is to meet its own operation constraints, that is, to ensure that the indoor temperature is within a certain range, and the operation time must not be less than the minimum startup time, and the shutdown time must not be less than the minimum shutdown time, so as to minimize the impact on the life of the central air-conditioning. The constraint model constructed in this embodiment includes a startup operation constraint model and a shutdown operation constraint model, wherein the startup operation constraint model is intended to ensure that the indoor temperature does not drop to the lower limit of the comfort temperature after startup, and meets the minimum shutdown time requirement; the shutdown operation constraint model is intended to ensure that the indoor temperature does not rise to the upper limit of the comfort temperature after shutdown, and meets the minimum startup time requirement.

[0049] The central air-conditioning startup operation constraint model constructed is shown in formula (2), δ c (t) = 0 means the central air conditioner is in shutdown state. Indicates that the central air conditioner shutdown time needs to be greater than 5 minutes, T on (t+Δt)>AT min Indicates that the indoor temperature must not be lower than the lower limit of the comfort range after the central air conditioner is in operation, η p (t)<λ c Indicates that the flow density per unit area shall not exceed the threshold.

[0050]

[0051] In the formula, δ c (t) is the operating state of the central air conditioner at time t, with a value of 0 indicating that it is in the shutdown state, and a value of 1 indicating that it is in the startup state; is the most recent refrigeration unit shutdown time before time t; T on (t+Δt) is the indoor temperature after the machine is turned on at time t for Δt; AT min is the lower limit of indoor comfortable temperature; a 1～3 is the building parameter; Q p is the rated cooling capacity of the chiller; T(t) is the indoor temperature at time t; η p (t) is the flow density of people per unit area at time t; c N is the threshold of pedestrian flow density per unit area; r (t) is the total passenger flow inside the building at time t; S c is the cooling area; is the effective area coefficient inside the building;

[0052] The constructed central air conditioning shutdown operation constraint model is shown in formula (3), where δ c (t) = 1 means the central air conditioner is on. Indicates that the central air conditioner needs to be turned on for more than 3 minutes, T off (t+Δt)<AT max Indicates that the indoor temperature shall not be higher than the upper limit of the comfort range after the central air conditioner participates in the grid operation, η p (t)<λ c Indicates that the flow density per unit area shall not exceed the threshold.

[0053]

[0054] In the formula, δ c (t) is the operating state of the central air conditioner at time t, with a value of 0 indicating that it is in the shutdown state, and a value of 1 indicating that it is in the startup state; is the most recent start-up time of the refrigeration unit before time t; T off (t+Δt) is the indoor temperature after the shutdown time Δt at time t; AT max is the upper limit of indoor comfortable temperature; a 1～4 is the building parameter; β c is the chilled water temperature parameter; T(t) is the indoor temperature at time t; η p (t) is the flow density of people per unit area at time t; c N is the threshold of pedestrian flow density per unit area; r (t) is the total passenger flow inside the building at time t; S c is the cooling area, is the effective area coefficient inside the building.

[0055] S3. Construct a central air-conditioning start-stop loss model based on a BP neural network: The central air-conditioning will generate start-up loss and shutdown loss during the operation of the power grid. Only when the incentive price is greater than the start-up / shutdown loss, the central air-conditioning will respond to the power grid operation. However, the start-up loss and shutdown loss of the central air-conditioning are affected by many factors and are difficult to simulate accurately. Therefore, this embodiment trains the collected central air-conditioning operation information based on the BP neural network, constructs a central air-conditioning start-stop loss model based on the BP neural network, and adjusts the coefficients in the model to obtain an accurate central air-conditioning start-stop loss model;

[0056] Among them, a central air-conditioning start-stop loss model based on BP neural network is constructed, including a start-up loss model and a stop-loss model. Formula (4) is the central air-conditioning start-up loss model, formula (5) is the central air-conditioning stop-loss model, and formula (6) is the load loss calculation model.

[0057]

[0058] In the formula, is the startup loss of the central air conditioner at time t; is the shutdown loss of the central air conditioner at time t; is the indoor temperature deviation lower limit coefficient; is the remaining startup time coefficient; is the passenger flow impact coefficient; is the load loss factor; is the indoor temperature deviation upper limit coefficient; Remaining downtime factor; T on (t+Δt) is the indoor temperature after the machine is turned on at time t for Δt; T off (t+Δt) is the indoor temperature after the power is turned off at time t for Δt; T set is the set indoor temperature; AT min AT is the lower limit of indoor comfortable temperature; max The upper limit of indoor comfortable temperature; The time until the next natural power-on; is the time until the next natural shutdown; τ on is the power-on cycle of the central air conditioner; τ off is the shutdown period of the central air conditioner; p (t) is the flow density of people per unit area at time t; X s (t) is the loss load of the central air conditioner at time t; X c (t) is the cooling load rate at time t; Q p is the rated cooling capacity of the chiller at time t;

[0059] The BP neural network is used to calculate the indoor temperature deviation lower limit coefficient in the startup loss model and the shutdown loss model. Remaining startup time coefficient Passenger flow impact coefficient Load loss factor Indoor temperature deviation upper limit coefficient Remaining downtime factor Adjust the settings to obtain an accurate central air conditioning start-stop loss model;

[0060] like Figure 2 As shown in the figure, a starting loss coefficient setting model based on BP neural network is constructed. By training and learning the central air-conditioning operation data samples, the features related to the starting loss are extracted, and then the starting loss coefficient of the central air-conditioning is accurately set. It includes 4 types of inputs and 4 outputs:

[0061] The four types of input are: central air conditioning operation data X(t) at time t, the time to the next natural start-up The incentive price f at time t ex (t), indoor comfort temperature lower limit AT min ;

[0062] The 4 outputs are: Indoor temperature deviation lower limit coefficient Remaining startup time coefficient Passenger flow impact coefficient Load loss factor

[0063] The training and learning specifically include the following methods: collecting central air-conditioning operation data as training samples, training the collected samples, conducting network testing on the trained startup loss coefficient setting model, and then offline training and testing the neural network model to obtain the BP neural network training function, connection weights, and thresholds for startup loss coefficient setting, extract sample features, and use the tansig function in the hidden layer of the neural network. The output of the hidden layer is shown in formula (7).

[0064]

[0065] In the formula, X'(t) represents the input of the training sample of the entire neural network model, X i '(t) represents the set of training samples that need to be introduced in each layer, i is the number of layers of neural network training, and 5 layers of training are required for the samples; h j The output of the hidden layer of the neural network is adjusted for the startup loss coefficient; W ij b. Adjust the connection weights from the neural network input layer unit to the hidden layer unit for the startup loss coefficient; b. jis the threshold value of the hidden layer unit of the neural network for starting the loss coefficient adjustment; j ranges from 1 to n; n is the number of hidden layer units of the neural network for starting the loss coefficient adjustment;

[0066] The output layer of the neural network uses the purelin function. The output of the neural network is shown in formula (9).

[0067]

[0068] In the formula, k m The output of the neural network is adjusted for the startup loss coefficient; w jm The connection weights from the hidden layer units to the output layer units of the neural network are adjusted to start the loss coefficient, and the range of m is 1 to 3;

[0069] Get accurate indoor temperature deviation lower limit coefficient Remaining startup time coefficient Passenger flow impact coefficient Load loss factor Then, substitute into formula (4) to obtain an accurate central air-conditioning startup loss model; based on the same principle, a shutdown loss coefficient setting model based on BP neural network is constructed to obtain an accurate central air-conditioning shutdown loss model.

[0070] S4. Constructing a central air conditioning incentive-response characteristic model based on BP neural network: Based on the accurate central air conditioning start-stop loss model, construct a central air conditioning start-stop participation model and combine it with the central air conditioning startup and shutdown operation constraint model, and finally quantify and obtain accurate central air conditioning load incentive-response characteristics; wherein, firstly construct a central air conditioning start-stop participation model, formula (10) is the central air conditioning startup participation model, formula (11) is the central air conditioning shutdown participation model,

[0071]

[0072] In the formula, f ex (t) is the incentive electricity price at time t; is the startup loss of central air conditioner c at time t; is the shutdown loss of central air conditioner c at time t;

[0073] Then, according to the central air-conditioning start-stop participation model, the central air-conditioning start-stop excitation-response characteristic curve can be obtained, such as Figure 3 As shown in the figure, P up (t) and P dn (t) are the power of the central air conditioner when it is adjusted up or down. When the central air conditioner satisfies formula (2) and formula (10), that is, according to the incentive electricity price f ex(t) Participate in the upward power response; When the central air conditioner satisfies formula (3) and formula (11), that is, according to the incentive electricity price f ex (t) Participate in down-regulating power response.

[0074] Since mastering the central air-conditioning load incentive-response characteristics is a prerequisite for effectively dispatching the central air-conditioning to participate in the operation of the new power system, the present invention can provide more accurate central air-conditioning incentive-response characteristics, thereby improving the effectiveness of central air-conditioning regulation and the ability to tap central air-conditioning load resources.

Claims

1. A method for quantifying the stimulus-response characteristics of central air-conditioning load resources based on BP neural network, characterized in that: Here are the steps: S1. Collect central air-conditioning operation information: including the rated cooling capacity of the chiller, chilled water temperature parameters, indoor comfort temperature range, air-conditioning operation status, total passenger flow inside the building, cooling area, effective area coefficient inside the building, passenger flow density per unit area, indoor temperature, refrigeration unit shutdown time, startup time, and incentive price; wherein, formula (1) is used to collect central air-conditioning operation information. Where X(t) is the central air conditioning information collected at time t; Q p is the rated cooling capacity of the chiller; c AT is the chilled water temperature parameter; min is the minimum indoor temperature; AT max is the maximum indoor temperature; c (t) is the air conditioning operation state; N r (t) is the total passenger flow inside the building at time t; S c is the area of ​​the cooling zone; is the effective area coefficient inside the building; η p (t) is the flow density of people per unit area at time t; c is the threshold of the density of people flow per unit area; T(t) is the indoor temperature at time t; is the most recent refrigeration unit shutdown time before time t; is the most recent start-up time of the refrigeration unit before time t; f ex (t) is the incentive price at time t; S2. Constructing the central air-conditioning startup and shutdown operation constraint model: including the startup operation constraint model and the shutdown operation constraint model. The startup operation constraint model aims to ensure that the indoor temperature does not drop to the lower limit of the comfort temperature after startup and meets the minimum shutdown time requirement; the shutdown operation constraint model aims to ensure that the indoor temperature does not rise to the upper limit of the comfort temperature after shutdown and meets the minimum startup time requirement. The constructed central air-conditioning startup operation constraint model is shown in formula (2), δ c (t) = 0 means the central air conditioner is in shutdown state. Indicates that the central air conditioner should be shut down for more than 5 minutes. on (t+Δt)>AT min Indicates that the indoor temperature must not be lower than the lower limit of the comfort range after the central air conditioner is in operation, η p (t)<λ c Indicates that the flow density per unit area shall not exceed the threshold. In the formula, δ c (t) is the operating state of the central air conditioner at time t, with a value of 0 indicating that it is in the shutdown state, and a value of 1 indicating that it is in the startup state; is the most recent refrigeration unit shutdown time before time t; T on (t+Δt) is the indoor temperature after the machine is turned on at time t for Δt; AT min is the lower limit of indoor comfortable temperature; a 1～3 is the building parameter; Q p is the rated cooling capacity of the chiller; T(t) is the indoor temperature at time t; η p (t) is the flow density of people per unit area at time t; c N is the threshold of pedestrian flow density per unit area; r (t) is the total passenger flow inside the building at time t; S c is the area of ​​the cooling zone; is the effective area coefficient inside the building; Among them, the central air conditioning shutdown operation constraint model constructed is shown in formula (3), δ c (t) = 1 means the central air conditioner is on. Indicates that the central air conditioner needs to be turned on for more than 3 minutes, T off (t+Δt)<AT max Indicates that the indoor temperature shall not be higher than the upper limit of the comfort range after the central air conditioner participates in the grid operation, η p (t)<λ c Indicates that the flow density per unit area shall not exceed the threshold. In the formula, δ c (t) is the operating state of the central air conditioner at time t, with a value of 0 indicating that it is in the shutdown state, and a value of 1 indicating that it is in the startup state; is the most recent start-up time of the refrigeration unit before time t; T off (t+Δt) is the indoor temperature after the shutdown time Δt at time t; AT max is the upper limit of indoor comfortable temperature; a 1～4 is the building parameter; β c is the chilled water temperature parameter; T(t) is the indoor temperature at time t; η p (t) is the flow density of people per unit area at time t; c N is the threshold of pedestrian flow density per unit area; r (t) is the total passenger flow inside the building at time t; S c is the cooling area, is the effective area coefficient inside the building; S3. Constructing a central air-conditioning start-stop loss model based on BP neural network: Based on the BP neural network, the collected central air-conditioning operation information is trained to construct a central air-conditioning start-stop loss model based on the BP neural network, and the coefficients in the model are adjusted to obtain an accurate central air-conditioning start-stop loss model; wherein, the central air-conditioning start-stop loss model based on the BP neural network is constructed, including a start-up loss model and a shutdown loss model, formula (4) is the central air-conditioning start-up loss model, formula (5) is the central air-conditioning shutdown loss model, and formula (6) is the load loss calculation model. In the formula, is the startup loss of the central air conditioner at time t; is the shutdown loss of the central air conditioner at time t; is the indoor temperature deviation lower limit coefficient; is the remaining startup time coefficient; is the passenger flow impact coefficient; is the load loss factor; is the indoor temperature deviation upper limit coefficient; Remaining downtime factor; T on (t+Δt) is the indoor temperature after the machine is turned on at time t for Δt; T off (t+Δt) is the indoor temperature after the power is turned off at time t for Δt; T set is the set indoor temperature; AT min AT is the lower limit of indoor comfortable temperature; max The upper limit of indoor comfortable temperature; The time until the next natural power-on; is the time until the next natural shutdown; τ on is the power-on cycle of the central air conditioner; τ off is the shutdown period of the central air conditioner; p (t) is the flow density of people per unit area at time t; X s (t) is the loss load of the central air conditioner at time t; X c (t) is the cooling load rate at time t; Q p is the rated cooling capacity of the chiller at time t; The BP neural network is used to calculate the indoor temperature deviation lower limit coefficient in the startup loss model and the shutdown loss model. Remaining startup time coefficient Passenger flow impact coefficient Load loss factor Indoor temperature deviation upper limit coefficient Remaining downtime factor Adjust the settings to obtain an accurate central air conditioning start-stop loss model; S4. Constructing a central air conditioning incentive-response characteristic model based on BP neural network: Based on the accurate central air conditioning start-stop loss model, construct a central air conditioning start-stop participation model and combine it with the central air conditioning startup and shutdown operation constraint model, and finally quantify and obtain accurate central air conditioning load incentive-response characteristics. First, construct a central air conditioning start-stop participation model. Formula (10) is the central air conditioning startup participation model, and formula (11) is the central air conditioning shutdown participation model. In the formula, f ex (t) is the incentive electricity price at time t; is the startup loss of central air conditioner c at time t; is the shutdown loss of central air conditioner c at time t; Then, according to the central air-conditioning start-stop participation model, the central air-conditioning start-stop incentive-response characteristic curve can be obtained. When the central air-conditioning satisfies formula (2) and formula (10), according to the incentive electricity price f ex (t) Participate in the upward power response; When the central air conditioner satisfies formula (3) and formula (11), that is, according to the incentive electricity price f ex (t) participate in down-regulating power response; A starting loss coefficient setting model based on BP neural network is constructed. By training and learning the central air-conditioning operation data samples, the features related to the starting loss are extracted, and then the starting loss coefficient of the central air-conditioning is accurately set. It includes 4 types of inputs and 4 outputs: The four types of input are: central air conditioning operation data X(t) at time t, the time to the next natural start-up The incentive price f at time t ex (t), indoor comfort temperature lower limit AT min ; The 4 outputs are: Indoor temperature deviation lower limit coefficient Remaining startup time coefficient Passenger flow impact coefficient Load loss factor The training and learning specifically include the following methods: collecting central air-conditioning operation data as training samples, training the collected samples, conducting network testing on the trained startup loss coefficient setting model, and then offline training and testing the neural network model to obtain the BP neural network training function, connection weights, and thresholds for startup loss coefficient setting, extract sample features, and use the tansig function in the hidden layer of the neural network. The output of the hidden layer is shown in formula (7). In the formula, X'(t) represents the input of the training sample of the entire neural network model, X' i (t) represents the set of training samples that need to be introduced in each layer, i is the number of layers of neural network training, and 5 layers of training are required for the samples; h j The output of the hidden layer of the neural network is adjusted for the startup loss coefficient; W ij b. Adjust the connection weights from the neural network input layer unit to the hidden layer unit for the startup loss coefficient; b. j is the threshold value of the hidden layer unit of the neural network for starting the loss coefficient adjustment; j ranges from 1 to n; n is the number of hidden layer units of the neural network for starting the loss coefficient adjustment; The output layer of the neural network uses the purelin function. The output of the neural network is shown in formula (9). In the formula, k m The output of the neural network is adjusted for the startup loss coefficient; w jm The connection weights from the hidden layer units to the output layer units of the neural network are adjusted to start the loss coefficient, and the range of m is 1 to 3; Get accurate indoor temperature deviation lower limit coefficient Remaining startup time coefficient Passenger flow impact coefficient Load loss factor Then, substitute into formula (4) to obtain an accurate central air-conditioning startup loss model; based on the same principle, a shutdown loss coefficient setting model based on BP neural network is constructed to obtain an accurate central air-conditioning shutdown loss model.

2. A central air-conditioning load resource incentive-response characteristic quantification system based on BP neural network, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, a method for quantifying the excitation-response characteristics of central air-conditioning load resources based on a BP neural network as described in claim 1 is implemented.

3. A computer-readable storage medium, on which is stored an implementation program for information transmission, which, when executed by a processor, implements the steps of a method for quantifying the excitation-response characteristics of central air-conditioning load resources based on a BP neural network as described in claim 1.

Images