Method and System for Constructing Wind Balance Model, Wind Balance Method and System

Abstract

The present invention discloses a method and system for constructing a wind balance model, a wind balance method and system. The input of the wind balance model is flow data, and the output is corresponding prediction data. The prediction data includes predicted values corresponding to each adjustment item. The adjustment items include fan voltage, fan power, and several damper angles. The construction method includes the following steps: obtaining sample data; configuring hyperparameter data, where the hyperparameter data includes the value ranges of angle constraint parameters, voltage constraint parameters, and power constraint parameters; constructing an objective function, where the objective function includes an error loss function, an energy-saving constraint strategy, and a regularization term; determining a mapping matrix based on the sample data, the hyperparameter data, and the objective function, and constructing a wind balance model based on the mapping matrix. By designing the angle constraint parameters, voltage constraint parameters, and power constraint parameters, the present invention is more energy-saving on the premise of ensuring measurement accuracy.

Description

Technical Field

[0001] The present invention relates to the field of ventilation control, and particularly to a technology for constructing a wind balance model and a wind balance technology implemented by using the constructed wind balance model. Background Art

[0002] With the continuous development of green buildings in China, the demand for intelligent ventilation systems is increasing. Green buildings refer to high-quality buildings that save resources, protect the environment, reduce pollution, provide healthy, applicable, and efficient living spaces for people, and maximize the harmonious coexistence between humans and nature throughout their life cycle. As the main component of building energy consumption, the research on ventilation systems is of great significance for the development of green buildings. For intelligent ventilation systems, their main components are air valves and fans, and the control of air valves and fans is an important factor in ventilation system energy conservation.

[0003] Wind balance is a technology that distributes appropriate airflows to the required terminals, which can effectively avoid energy waste and improve the energy utilization rate of ventilation systems. However, in actual complex ventilation systems, the distribution of airflows is usually uneven. Insufficient ventilation will lead to poor indoor air quality, reduce the comfort of occupants, and even cause sick building syndrome. On the other hand, excessive ventilation will waste a large amount of energy. Summary of the Invention

[0004] Aiming at the disadvantages of unbalanced adjustment at each terminal and high energy consumption in the existing ventilation systems, the present invention provides a technology for constructing a wind balance model and a wind balance technology implemented by using the constructed wind balance model. The wind balance model constructed by the present invention can predict the fan voltage, power, and the opening degrees of each air valve through machine learning methods, and can effectively reduce the energy consumption of the corresponding ventilation system while meeting the target air volume.

[0005] To solve the above technical problems, the present invention is solved by the following technical solutions:

[0006] A method for constructing a wind balance model, which constructs a wind balance model based on a mapping matrix. The input of the wind balance model is flow data, and the output is corresponding prediction data. The prediction data includes the predicted values corresponding to each adjustment item, and the adjustment items include fan voltage, fan power, and several air valve angles.

[0007] The method for obtaining the mapping matrix includes the following steps:

[0008] Obtain sample data, where the sample data includes flow sample data and working sample data. The flow sample data and the working sample data correspond to each other, and the working sample data includes the true values corresponding to each adjustment item.

[0009] Configure hyperparameter data, where the hyperparameter data includes the value ranges of angle constraint parameters, voltage constraint parameters, and power constraint parameters;

[0010] Construct an objective function, which includes an error loss function, an energy-saving constraint strategy, and a regularization term. Among them, the energy-saving constraint strategy is constructed based on the angle constraint parameter, the voltage constraint parameter, and the power constraint parameter, and the regularization term is constructed based on a mapping matrix;

[0011] Determine the mapping matrix based on the sample data, the hyperparameter data, and the objective function.

[0012] As an implementable manner:

[0013] The hyperparameter data further includes a value condition, which is that the power constraint parameter is greater than the angle constraint parameter and the power constraint parameter is greater than the voltage constraint parameter.

[0014] As an implementable manner:

[0015] The energy-saving constraint strategy L c The calculation formula of is:

[0016]

[0017]

[0018] Where:

[0019] k + 2 is the total number of adjustment terms. The first k adjustment terms are the damper angles, the (k + 1)-th adjustment term is the fan voltage, and the (k + 2)-th adjustment term is the fan power;

[0020] N is the total number of the input sample data;

[0021] Represents the constraint strategy corresponding to the i-th adjustment term in the n-th sample data;

[0022] W i n Represents the predicted value corresponding to the i-th adjustment term in the n-th sample data;

[0023] μ θ Represents the angle constraint parameter, μ U Represents the voltage constraint parameter, μ P Represents the power constraint parameter.

[0024] As an implementable manner:

[0025] The regularization term L r The calculation formula of is:

[0026]

[0027] Wherein:

[0028] a represents the mapping matrix, and λ represents the regularization parameter.

[0029] As an implementable manner:

[0030] Divide the sample data into training data and test data;

[0031] Based on the training data, hyperparameter data, and objective function, determine the hyperparameter group and the mapping matrix corresponding to the hyperparameter group;

[0032] Based on the mapping matrix, construct the corresponding intermediate model;

[0033] Test the intermediate model based on the test data, and use the intermediate model with the best test result as the wind balance model.

[0034] Furthermore:

[0035] Based on the hyperparameter data, determine the hyperparameter group corresponding to the current intermediate model, and the hyperparameter group includes an angle constraint parameter, a voltage constraint parameter, and a power constraint parameter;

[0036] With the goal of minimizing the objective function, use the gradient method to calculate the corresponding mapping matrix based on the training data and the hyperparameter group.

[0037] Furthermore, the intermediate model is a kernel function model, and the kernel function model is:

[0038]

[0039] Wherein:

[0040] 1 ≤ l ≤ N, and 1 ≤ i ≤ k + 2, k + 2 is the total number of adjustment terms, N is the number of sample data input during training, q represents the input flow data, q l represents the l-th group of flow sample data;

[0041] W i represents the predicted value corresponding to the i-th adjustment term, K(,) is the kernel function, and a represents the mapping matrix.

[0042] The present invention also proposes a construction system for a wind balance model, which is used to construct a wind balance model based on a mapping matrix. The input of the wind balance model is flow data, and the output is the corresponding predicted data. The predicted data includes the predicted values corresponding to each adjustment term. The adjustment terms include fan voltage, fan power, and several damper angles;

[0043] Including:

[0044] A sampling module for obtaining sample data, where the sample data includes flow sample data and working sample data, the flow sample data corresponds to the working sample data, and the working sample data includes the true values corresponding to each adjustment item;

[0045] A configuration module for configuring hyperparameter data, where the hyperparameter data includes the value ranges of angle constraint parameters, voltage constraint parameters, and power constraint parameters;

[0046] A calculation module for constructing an objective function, where the objective function includes an error loss function, an energy-saving constraint strategy, and a regularization term. Among them, the energy-saving constraint strategy is constructed based on the angle constraint parameter, the voltage constraint parameter, and the power constraint parameter, and the regularization term is constructed based on a mapping matrix; it is also used to determine the mapping matrix based on the sample data, the hyperparameter data, and the objective function.

[0047] The present invention also proposes a wind balance method, including the following steps:

[0048] Input target flow data into a wind balance model, and the wind balance model outputs corresponding prediction data. The prediction data includes the predicted values corresponding to each adjustment item. The adjustment items include fan voltage, fan power, and several wind valve angles. Among them, the wind balance model is constructed by using any one of the above construction methods;

[0049] Control the target ventilation system to work in balance based on the prediction data.

[0050] The present invention also proposes a wind balance system, including:

[0051] A prediction module for inputting target flow data into a wind balance model, and the wind balance model outputs corresponding prediction data. The prediction data includes the predicted values corresponding to each adjustment item. The adjustment items include fan voltage, fan power, and several wind valve angles. Among them, the wind balance model is constructed by using any one of the above construction methods;

[0052] A control module for controlling the target ventilation system to work in balance based on the prediction data.

[0053] Due to the adoption of the above technical solutions, the present invention has remarkable technical effects:

[0054] Through the design of power constraint, when predicting the fan voltage and the opening degree of the wind valve, the fan power is constrained, so that the energy consumption of the ventilation system can be effectively reduced. Through the design of angle constraint and voltage constraint, the prediction accuracy can be increased. The three constraints work together to be more energy-saving on the premise of ensuring the measurement accuracy. Brief Description of the Drawings

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0056] Figure 1 It is a comparison schematic diagram of the designed flow rate corresponding to the first air outlet and the experimental flow rate in the experiment;

[0057] Figure 2 It is a comparison schematic diagram of the designed flow rate corresponding to the second air outlet and the experimental flow rate in the experiment;

[0058] Figure 3 It is a comparison schematic diagram of the designed flow rate corresponding to the third air outlet and the experimental flow rate in the experiment;

[0059] Figure 4 It is a comparison schematic diagram of the designed flow rate corresponding to the fourth air outlet and the experimental flow rate in the experiment;

[0060] Figure 5 It is a comparison schematic diagram of the designed flow rate corresponding to the fifth air outlet and the experimental flow rate in the experiment;

[0061] Figure 6 It is a comparison schematic diagram of the designed power and the experimental power in the experiment;

[0062] Figure 7 It is a schematic diagram of the module connection of a system for constructing a wind balance model of the present invention. Detailed Embodiments

[0063] The following will further elaborate on the present invention in combination with embodiments. The following embodiments are explanations of the present invention, and the present invention is not limited to the following embodiments.

[0064] Embodiment 1: A method for constructing a wind balance model;

[0065] The input of the wind balance model is flow rate data, and the output is corresponding prediction data;

[0066] The flow rate data includes the designed flow rate corresponding to each air valve;

[0067] The prediction data includes the predicted values corresponding to each adjustment item, and the adjustment items include fan voltage, fan power, and several air valve angles.

[0068] In the existing air balance method, by adjusting the air valve angle and the fan voltage, on the premise of meeting the designed flow rate, the air valve angle and the fan voltage are made to reach the lowest values, that is, it is determined that the best energy-saving effect is achieved; for the purpose of achieving energy conservation; however, in fact, the most direct factor affecting the energy consumption of the ventilation system is the power of the fan. In this embodiment, the fan power is used as the adjustment item, which overcomes the thinking misunderstanding of those skilled in the art that only adjusting the air valve angle and the fan voltage can achieve the best energy-saving effect, and further improves the energy-saving effect.

[0069] The specific construction method includes the following steps:

[0070] S100. Obtain sample data;

[0071] The sample data includes flow sample data and working sample data, and the flow sample data corresponds to the working sample data;

[0072] The flow sample data includes the actual flow rates corresponding to the respective air valves;

[0073] The working sample data includes the actual values corresponding to the respective adjustment items.

[0074] In this embodiment, the method for obtaining sample data is specifically as follows:

[0075] Randomly select k angles and randomly select voltages, where k is the number of air valves of the target ventilation system;

[0076] Set the air valve angle based on the selected angles and set the fan voltage according to the selected voltages;

[0077] After the target ventilation system runs for 1 minute, record the air valve flow rates of the respective air valves and the fan power;

[0078] Construct flow sample data based on the obtained air valve flow rates, and construct working sample data based on the selected air valve angles, air valve voltages, and the detected fan power.

[0079] S200. Configure hyperparameter data;

[0080] The hyperparameter data includes the value ranges of the respective hyperparameters;

[0081] In addition to the regularly used regularization parameters and kernel parameters, the hyperparameters also include one or more constraint parameters. In this embodiment, the hyperparameters also include an angle constraint parameter, a voltage constraint parameter, and a power constraint parameter.

[0082] In this embodiment, the value ranges of the respective hyperparameters are all [0, 1].

[0083] Optionally, the hyperparameter data may further include value conditions;

[0084] In this embodiment, the value-taking condition is that the power constraint parameter is greater than the angle constraint parameter and the power constraint parameter is greater than the voltage constraint parameter, with the power constraint as the dominant factor.

[0085] If only the power constraint parameter is added to reduce the fan power, as the energy-saving effect increases, the error also increases. Within the allowable range of the error, the energy-saving effect is limited. Therefore, this embodiment also adds angle constraint and voltage constraint to increase the prediction accuracy, so as to further improve the energy-saving effect on the premise of ensuring the prediction accuracy.

[0086] S300. Construct the objective function;

[0087] The objective function includes an error loss function, an energy-saving constraint strategy, and a regularization term. The energy-saving constraint strategy is constructed based on the angle constraint parameter, the voltage constraint parameter, and the power constraint parameter, and the regularization term is constructed based on the mapping matrix;

[0088] The mapping matrix is a parameter matrix to be trained.

[0089] In this embodiment, the calculation formula of the objective function L is:

[0090] L = L e + L c + L r ;

[0091] Among them, L e is the error loss function, L c is the energy-saving constraint strategy, and L r is the regularization term.

[0092] The error loss function L c is used to indicate the wind balance error, and its calculation formula is:

[0093]

[0094] Among them:

[0095] k + 2 is the number of adjustment items, k is the number of damper angles in the adjustment items, and N is the number of input sample data, that is, the number of sample data as training data;

[0096] W i n represents the predicted value corresponding to the i-th adjustment item in the n-th sample data, represents the true value corresponding to the i-th adjustment item in the n-th sample data;

[0097] Note that in this embodiment, both the flow data and the flow sample data are k×1 matrices, and the prediction data and the working sample data are (k + 2)×1 matrices.

[0098] Energy-saving constraint strategy L c The calculation formula is as follows:

[0099]

[0100]

[0101] Where:

[0102] k + 2 is the total number of adjustment terms. The first k adjustment terms are the damper angles, the (k + 1)-th adjustment term is the fan voltage, and the (k + 2)-th adjustment term is the fan power;

[0103] represents the constraint strategy corresponding to the i-th adjustment term in the n-th sample data;

[0104] W i n represents the predicted value corresponding to the i-th adjustment term in the n-th sample data;

[0105] μ θ represents the angle constraint parameter, μ U represents the voltage constraint parameter, μ P represents the power constraint parameter.

[0106] Regularization term L r The calculation formula is as follows:

[0107]

[0108] Where:

[0109] a represents the mapping matrix, and λ represents the regularization parameter.

[0110] That is, based on the above error loss function, energy-saving constraint strategy, and regularization term, it can be deduced that the calculation formula of the objective function L is:

[0111]

[0112] Where R is the residual matrix, ||·|| F represents the F-norm, θ is the calculated angle matrix, U is the calculated voltage matrix, P is the calculated power matrix, a is the mapping matrix, where L(a) indicates that a is an unknown parameter to be identified.

[0113] S400. Model construction;

[0114] Based on the sample data, the hyperparameter data, and the objective function, determine the mapping matrix, and construct the air balance model based on the mapping matrix.

[0115] Specifically, it includes the following steps:

[0116] S410. Determine the training data and test data:

[0117] Divide the sample data into training data and test data;

[0118] In this embodiment, the obtained sample data is randomly divided into five groups, and 4 groups of sample data are randomly selected as the training data, and the remaining 1 group of sample data is used as the test data.

[0119] S420. Based on the training data, hyperparameter data, and objective function, determine the hyperparameter group and the corresponding mapping matrix;

[0120] S421. Based on the hyperparameter data, determine the hyperparameter group corresponding to the current intermediate model;

[0121] That is, based on the value range and value conditions configured in the hyperparameter data, generate several different hyperparameter groups, and calculate the mapping matrix based on each hyperparameter group respectively;

[0122] In this embodiment, the hyperparameter group includes an angle constraint parameter, a voltage constraint parameter, and a power constraint parameter;

[0123] Those skilled in the art can set the selection rules of the hyperparameter group in the process of constructing the intermediate model according to actual needs. For example, generate several hyperparameter groups in advance, extract each hyperparameter group in turn, and can also adjust the hyperparameter group used last time according to the preset rules. This specification does not limit it in detail.

[0124] S422. With the goal of minimizing the objective function, use the gradient method to calculate the corresponding mapping matrix based on the training data and the hyperparameter group.

[0125] That is, determine the angle constraint parameter μ θ , the voltage constraint parameter μ U , the power constraint parameter μ P , and the regularization parameter λ, as well as the training data and the objective function. With the goal of minimizing the objective function, the mapping matrix as an unknown parameter can be calculated, and the optimal hyperparameter group and the optimal mapping matrix can be determined to construct the wind balance model;

[0126] In this embodiment, L(a) is a convex function, and its gradient is:

[0127]

[0128] a = {a θ , a U , a P}

[0129] A = [μθ a θ μ U a U μ P a P

[0130] Among them, K is the kernel function, a is the mapping matrix, a θ is the angle mapping matrix, a U is the voltage mapping matrix, a P is the power mapping matrix, and

[0131] Let Obtain the optimal weight matrix under the current hyperparameter group and obtain the mapping matrix;

[0132] Calculate the mapping matrix a = {a θ , a U , a P} by the formula:

[0133]

[0134]

[0135]

[0136] Among them, I represents the identity matrix.

[0137] S430. Based on the mapping matrix, construct the corresponding intermediate model;

[0138] In this embodiment, the constructed intermediate model is a kernel function model, specifically:

[0139]

[0140] Among them:

[0141] 1 ≤ l ≤ N, and 1 ≤ i ≤ k + 2, k + 2 is the total number of adjustment items, N is the number of sample data input during training, that is, the number of sample data used as training data during training, q represents the input flow data, q l represents the l-th group of flow sample data;

[0142] W i represents the predicted value corresponding to the i-th adjustment item, K(,) is the kernel function, and a represents the mapping matrix;

[0143] In this embodiment, the adjustment items corresponding to the first k columns of the prediction data and the mapping matrix are all damper angles, the adjustment items corresponding to the (k + 1)-th column are all fan voltages, and the adjustment items corresponding to the (k + 2)-th column are all fan powers. ​

[0144] In this embodiment, the kernel function adopts a Gaussian kernel function, specifically:

[0145]

[0146] where r is the kernel parameter in the hyperparameters.

[0147] S440. Test the intermediate model based on the test data, and use the intermediate model with the best test result as the air balance model.

[0148] The test method is as follows:

[0149] Input the flow sample data in the test data into the intermediate model to obtain corresponding prediction data;

[0150] Adjust the air valves and fans of the ventilation system based on the prediction data, and then detect the flow rate of each air valve to obtain flow rate test data;

[0151] Calculate the error based on the flow sample data and the flow rate test data to obtain the corresponding predicted mean square error;

[0152] Use the intermediate model with the smallest predicted mean square error as the best intermediate model.

[0153] Experiment:

[0154] Build an air balance experiment platform, which includes 1 fan and 5 air outlets, and each air outlet is equipped with an intelligent air valve, that is, the adjustment items include 5 air valve angles, 1 fan voltage, and 1 fan power;

[0155] Build an air balance model based on the construction method disclosed in this embodiment;

[0156] In this experiment, the kernel parameter r of the built air balance model is 0.92, the regularization parameter λ is 0.02, and the angle constraint parameter μ θ is 0.11, the voltage constraint parameter μ U is 0.26, and the power constraint parameter μ P is 0.56.

[0157] Input the flow sample data in the test data into the air balance model, and the air balance model outputs the predicted values of each adjustment item, and adjust the fan and each air valve in the air balance experiment platform according to the obtained predicted values;

[0158] Measure the air flow rate of each air outlet to obtain the experimental flow rate.

[0159] Prediction accuracy analysis:

[0160] The comparison chart of the designed flow rate and the experimental flow rate corresponding to the 5 air outlets is as Figures 1 to 5 shown, refer toFigures 1 to 5 It can be seen that the designed flow rates corresponding to each experiment are close to the experimental flow rates;

[0161] In this experiment, the prediction accuracy corresponding to each air valve was also calculated based on the designed flow rate and the experimental flow rate. In this experiment, the prediction accuracy was measured by the mean percentage error, and the calculation method is as follows:

[0162] Based on the experimental flow rate q exp and the designed flow rate q ref the corresponding absolute percentage error (APE) was calculated, and the calculation formula is as follows:

[0163]

[0164] Based on the calculated APE, an average calculation was performed to obtain the mean percentage error corresponding to each air valve. In this experiment, the mean percentage errors of the five air valves were 3.61%, 5.34%, 3.35%, 4.01%, and 3.78% respectively, all of which meet the requirements for prediction accuracy (mean percentage error < 10%).

[0165] Energy-saving analysis:

[0166] The true values of the fan power corresponding to each test data were used as the designed power, and the predicted values of the fan power output by the air balance model were used as the experimental power;

[0167] Referring to Figure 6 it can be seen that the experimental powers are all lower than the designed power, and there is a large gap between the designed power and the experimental power corresponding to some experiments;

[0168] In this experiment, the energy-saving effect was also measured by P save and the calculation formula of P save is as follows:

[0169]

[0170] where P represents the experimental power and P ref represents the designed power.

[0171] After calculation, the experimental power was reduced by 23.55% compared with the designed power, and the energy-saving effect is good.

[0172] In summary, when the angle of the air valve remains unchanged, reducing the voltage and power of the fan will cause the air flow rate inside the air valve terminal to decrease. When the voltage and power of the fan remain unchanged, reducing the angle of the air valve will cause the air flow rate inside the air valve terminal to increase. By adding power constraint parameters, angle constraint parameters, and voltage constraint parameters, and taking the power constraint as the dominant factor, the three constraints act together, so as to effectively improve the energy-saving efficiency on the premise of ensuring the prediction accuracy.

[0173] Embodiment 2. A system for constructing a wind balance model is used to construct a wind balance model based on a mapping matrix. The input of the wind balance model is flow data, and the output is corresponding prediction data. The prediction data includes predicted values corresponding to each adjustment item. The adjustment items include fan voltage, fan power, and several damper angles;

[0174] As Figure 7 shown, it includes:

[0175] A sampling module 100 is used to obtain sample data. The sample data includes flow sample data and working sample data. The flow sample data corresponds to the working sample data. The working sample data includes true values corresponding to each adjustment item;

[0176] A configuration module 200 is used to configure hyperparameter data. The hyperparameter data includes the value ranges of angle constraint parameters, voltage constraint parameters, and power constraint parameters;

[0177] A calculation module 300 is used to construct an objective function. The objective function includes an error loss function, an energy-saving constraint strategy, and a regularization term. Among them, the energy-saving constraint strategy is constructed based on the angle constraint parameter, the voltage constraint parameter, and the power constraint parameter, and the regularization term is constructed based on the mapping matrix; it is also used to determine the mapping matrix based on the sample data, the hyperparameter data, and the objective function.

[0178] Further, the calculation module 300 includes:

[0179] A division unit 310 is used to divide the sample data into training data and test data;

[0180] A calculation unit 320 is used to determine a hyperparameter group and a mapping matrix corresponding to the hyperparameter group based on the training data, the hyperparameter data, and the objective function;

[0181] A construction unit 330 is used to construct a corresponding intermediate model based on the mapping matrix;

[0182] An output unit 340 is used to test the intermediate model based on the test data and use the intermediate model with the best test result as the wind balance model.

[0183] Embodiment 3. A wind balance method includes the following steps:

[0184] Input the target flow data into the wind balance model, and the wind balance model outputs corresponding prediction data. The prediction data includes predicted values corresponding to each adjustment item. The adjustment items include fan voltage, fan power, and several damper angles. Among them, the wind balance model is a wind balance model constructed by using the construction method described in Embodiment 1.

[0185] Control the target ventilation system to work in balance based on the predicted data.

[0186] Embodiment 4. A wind balance system includes:

[0187] A prediction module for inputting target flow data into a wind balance model, and the wind balance model outputs corresponding predicted data. The predicted data includes predicted values corresponding to each adjustment item. The adjustment items include fan voltage, fan power, and several damper angles. Among them, the wind balance model is the wind balance model constructed by the construction method described in Embodiment 1;

[0188] A control module for controlling the target ventilation system to work in balance based on the predicted data.

[0189] For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple. For the relevant parts, refer to the partial description of the method embodiment.

[0190] Each embodiment in this specification is described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the embodiments, refer to each other.

[0191] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, devices, or computer program products. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0192] The present invention is described with reference to the flowcharts and / or block diagrams of methods, terminal devices (systems), and computer program products according to the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, and the combination of processes and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing terminal devices to generate a machine, so that the instructions executed by the processors of the computer or other programmable data processing terminal devices generate means for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0193] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including an instruction device that implements the functions specified in one process Figure 1 or more processes and / or blocks Figure 1 or more blocks specified in the function.

[0194] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device, such that a series of operational steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable terminal device provide steps for implementing the functions specified in one process Figure 1 or more processes and / or blocks Figure 1 or more blocks specified in the function.

[0195] It should be noted that:

[0196] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed as including the preferred embodiments as well as all changes and modifications falling within the scope of the present invention.

[0197] In addition, it should be noted that any equivalent or simple changes made according to the structure, features, and principles described in the inventive concept of the present invention are included within the protection scope of the present invention. Those skilled in the technical field of the present invention can make various modifications, supplements, or use similar methods to replace the specific embodiments described, as long as they do not deviate from the structure of the present invention or exceed the scope defined by this claim book, they should all fall within the protection scope of the present invention.

Claims

1. A method for constructing a wind balance model, characterized in that Construct a wind balance model based on a mapping matrix. The input of the wind balance model is flow data, and the output is corresponding prediction data. The prediction data includes predicted values corresponding to each adjustment item. The adjustment items include fan voltage, fan power, and several damper angles; The method for obtaining the mapping matrix includes the following steps: Obtain sample data, where the sample data includes flow sample data and working sample data. The flow sample data corresponds to the working sample data, and the working sample data includes the true values corresponding to each adjustment item; Configure hyperparameter data, where the hyperparameter data includes the value ranges of angle constraint parameters, voltage constraint parameters, and power constraint parameters; Construct an objective function. The objective function includes an error loss function, an energy-saving constraint strategy, and a regularization term. Among them, the energy-saving constraint strategy is constructed based on the angle constraint parameter, the voltage constraint parameter, and the power constraint parameter, and the regularization term is constructed based on the mapping matrix, including: The calculation formula of the objective function L is: where R is the residual matrix, ||·|| F represents the Frobenius norm, θ is the calculated angle matrix, U is the calculated voltage matrix, P is the calculated power matrix, and a is the mapping matrix; L e is the error loss function, L c is the energy-saving constraint strategy, L r is the regularization term; k + 2 is the number of adjustment items, k is the number of damper angles among the adjustment items, and N is the number of input sample data; W i n represents the predicted value corresponding to the i-th adjustment item in the n-th sample data, represents the true value corresponding to the i-th adjustment item in the n-th sample data; Denote the constraint strategy corresponding to the i-th adjustment item in the n-th sample data; μ θ represents the angle constraint parameter, μ U represents the voltage constraint parameter, μ P represents the power constraint parameter; λ represents the regularization parameter; Based on the sample data, the hyperparameter data, and the objective function, determine the mapping matrix.

2. The construction method of a wind balance model according to claim 1, characterized in that: The hyperparameter data further includes a value condition, and the value condition is that the power constraint parameter is greater than the angle constraint parameter, and the power constraint parameter is greater than the voltage constraint parameter.

3. The construction method of a wind balance model according to claim 1, characterized in that: Energy-saving constraint strategy L c The calculation formula is as follows: Wherein: k + 2 is the total number of adjustment items. The first k adjustment items are damper angles, the (k + 1)-th adjustment item is fan voltage, and the (k + 2)-th adjustment item is fan power; N is the total number of input sample data; Indicates the constraint strategy corresponding to the i-th adjustment item in the n-th sample data; W i n represents the predicted value corresponding to the i-th adjustment item in the n-th sample data; μ θ represents the angle constraint parameter, μ U represents the voltage constraint parameter, μ P represents the power constraint parameter.

4. The construction method of a wind balance model according to claim 3, characterized in that: Regularization term L r The calculation formula is as follows: Wherein: a represents the mapping matrix, and λ represents the regularization parameter.

5. The construction method of a wind balance model according to any one of claims 1 to 4, characterized in that: Divide the sample data into training data and test data; Based on the training data, hyperparameter data, and objective function, determine a hyperparameter group and a mapping matrix corresponding to the hyperparameter group; Based on the mapping matrix, construct a corresponding intermediate model; Test the intermediate model based on the test data, and use the intermediate model with the best test result as the wind balance model.

6. The construction method of a wind balance model according to claim 5, characterized in that: Based on the hyperparameter data, determine the hyperparameter group corresponding to the current intermediate model. The hyperparameter group includes angle constraint parameters, voltage constraint parameters, and power constraint parameters; With the minimization of the objective function as the goal, use the gradient method to calculate the corresponding mapping matrix based on the training data and the hyperparameter group.

7. The construction method of a wind balance model according to claim 6, characterized in that The intermediate model is a kernel function model, and the kernel function model is: Wherein: 1 ≤ l ≤ N, and 1 ≤ i ≤ k + 2, where k + 2 is the total number of adjustment terms, N is the number of sample data input during the training process, q represents the input traffic data, q l represents the l-th group of traffic sample data; W i represents the predicted value corresponding to the i-th adjustment item, K(,) is the kernel function, and a represents the mapping matrix.

8. A system for constructing a wind balance model, characterized in that, For constructing a wind balance model based on a mapping matrix, the input of the wind balance model is flow data, and the output is corresponding prediction data. The prediction data includes predicted values corresponding to each adjustment item. The adjustment items include fan voltage, fan power, and several damper angles; Comprising: A sampling module for obtaining sample data. The sample data includes flow sample data and working sample data. The flow sample data corresponds to the working sample data. The working sample data includes true values corresponding to each adjustment item; A configuration module for configuring hyperparameter data. The hyperparameter data includes the value ranges of angle constraint parameters, voltage constraint parameters, and power constraint parameters; A calculation module for constructing an objective function. The objective function includes an error loss function, an energy-saving constraint strategy, and a regularization term. Among them, the energy-saving constraint strategy is constructed based on the angle constraint parameter, the voltage constraint parameter, and the power constraint parameter, and the regularization term is constructed based on the mapping matrix; Comprising: The calculation formula of the objective function L is: where R is the residual matrix, ||·|| F denotes the Frobenius norm, θ is the calculated angle matrix, U is the calculated voltage matrix, P is the calculated power matrix, and a is the mapping matrix; L e is the error loss function, L c is the energy-saving constraint strategy, L r is the regularization term; k + 2 is the number of adjustment items, k is the number of damper angles among the adjustment items, and N is the number of input sample data; W i n represents the predicted value corresponding to the i-th adjustment item in the n-th sample data, represents the true value corresponding to the i-th adjustment item in the n-th sample data; Indicates the constraint strategy corresponding to the i-th adjustment item in the n-th sample data; μ θ represents the angle constraint parameter, μ U represents the voltage constraint parameter, μ P represents the power constraint parameter; λ represents the regularization parameter; It is also used to determine the mapping matrix based on the sample data, the hyperparameter data, and the objective function.

9. A wind balance method, characterized in that Including the following steps: Input the target flow data into the wind balance model, and the wind balance model outputs corresponding prediction data. The prediction data includes predicted values corresponding to each adjustment item. The adjustment items include fan voltage, fan power, and several damper angles. Among them, the wind balance model is constructed by the construction method of any one of claims 1-7; Control the balanced operation of the target ventilation system based on the prediction data.

10. A wind balance system, characterized in that, Comprising: A prediction module for inputting the target flow data into the wind balance model, and the wind balance model outputs corresponding prediction data. The prediction data includes predicted values corresponding to each adjustment item. The adjustment items include fan voltage, fan power, and several damper angles. Among them, the wind balance model is constructed by the construction method of any one of claims 1-7; A control module for controlling the balanced operation of the target ventilation system based on the prediction data.

Images