Precise air conditioner energy-saving optimization method based on improved PSO

Abstract

The invention discloses a precision air conditioner energy-saving optimization method based on improved PSO. The precision air conditioner energy-saving optimization method comprises the steps that collected precision air conditioner historical operation data are preprocessed; performing energy consumption characteristic engineering: performing correlation calculation on the influence factors of the operation energy consumption of the precise air conditioner by adopting a linear analysis method and a nonlinear analysis method respectively; establishing an energy consumption prediction model according to a correlation calculation result, and performing hyper-parameter optimization; determining a target function, a decision variable and a constraint condition, and establishing an energy-saving optimization model; and the energy-saving optimization model is solved based on an improved PSO algorithm for dynamically adjusting the inertia weight and the compression factor, and the optimal operation parameter combination of the precise air conditioner when the precise air conditioner operates with the minimum energy consumption under the constraint condition is obtained. The method is suitable for energy consumption prediction and energy-saving optimization of the precise air conditioner in the multi-working-condition change scene, the operation parameters of the precise air conditioner can be rapidly and accurately adjusted through the solving result of the precise air conditioner energy-saving optimization model, and the time needed for maintaining the environment temperature is shortened.

Description

Technical Field

[0001] The present invention belongs to the technical field of building energy conservation, and particularly relates to a precise air-conditioning energy-saving optimization method based on improved PSO. Background Art

[0002] The central air-conditioning system of existing data centers accounts for 40% of the total energy consumption. Its energy-saving optimization is a key technology to improve the energy efficiency of data centers and reduce PUE. The energy consumption of the air-conditioning terminal equipment in data centers accounts for more than 30% of the total air-conditioning energy consumption, and is also an important part of reducing the PUE of data centers. To meet the requirements of the temperature and humidity of the computer room environment specified in the "Data Center Design Specification" and ensure the operation safety and reliability of the IT equipment in the computer room, precision air conditioners often operate in a multi-unit and high-wind-speed manner. At present, the average racking rate of IDC cabinets in China is about 58%. Compared with the heat released by IT equipment, there is a large redundancy in the cold supply. At the same time, the operation characteristics of the air-conditioning terminal equipment in the data center computer room are complex. The traditional energy consumption prediction modeling method has limited prediction ability, insufficient consideration of constraints in the energy-saving optimization strategy, and low accuracy of the optimization result. Summary of the Invention

[0003] The main purpose of the present invention is to overcome the disadvantages and deficiencies of the prior art, and provide a precise air-conditioning energy-saving optimization method based on improved PSO. From the perspective of optimization control, research is carried out on the energy consumption prediction modeling and operation parameter optimization of precision air conditioners, the influencing factors of precision air-conditioning energy consumption are comprehensively analyzed, and an energy consumption prediction model is established; with the minimum energy consumption as the objective function and the computer room environment temperature as the constraint, an energy-saving optimization model is established, and IPSO is used to solve the optimal value of the operation parameters, providing new ideas for the related research on the energy-saving optimization of the air conditioners in the data center computer room.

[0004] To achieve the above object, the present invention adopts the following technical solutions: One aspect of the present invention provides a precise air-conditioning energy-saving optimization method based on improved PSO, including the following steps: Preprocess the historical operation data of the air-conditioning terminal equipment collected, and the preprocessing includes outlier removal, missing value imputation, and data normalization; Conduct energy consumption feature engineering of the air-conditioning terminal equipment, including performing correlation calculations on the influencing factors of the operation energy consumption of the air-conditioning terminal equipment by using linear analysis methods and nonlinear analysis methods respectively; Establish an energy consumption prediction model for the air-conditioning terminal equipment according to the correlation calculation results of the energy consumption feature engineering, and optimize the hyperparameters of the energy consumption prediction model; Determine the objective function, decision variables, and constraint conditions, and establish an energy-saving optimization model for the air-conditioning terminal equipment; The improved PSO algorithm based on dynamically adjusting the inertia weight and compression factor is used to solve the energy-saving optimization model, and the optimal operation parameter combination of the air-conditioning terminal equipment when operating with the minimum energy consumption under the constraint conditions is obtained.

[0005] As a preferred embodiment, the linear analysis method uses the Pearson correlation coefficient, specifically: ; In the formula: cov ( X , Y ) is the covariance of the random variables X and Y ; , are respectively the standard deviations of X and Y ; The value range of

[0006] As a preferred embodiment, the non-linear analysis method uses the LightGBM feature importance, specifically: Calculate the split gain of the feature g 0 in each tree, and take the average of the results of all trees to obtain the relative importance, as shown in the following formula: ; In the formula: N’ is the number of trees; T i’ is the i’ th tree; is the relative importance result; The importance of the feature g 0 on each tree, as shown in the following formula: ; In the formula: is the degree of loss reduction after splitting; L is the number of leaf nodes; L -1 is the number of non-leaf nodes; v t is t the node-related variable.

[0007] As a preferred embodiment, the energy consumption prediction model of the air-conditioning terminal equipment is established according to the calculation results of the correlation of the energy consumption feature engineering, specifically: Divide the preprocessed data set into a training set, a validation set, and a test set according to a set ratio; Take the n parameters that are not less than their respective thresholds in the calculation results of the linear analysis method and the non-linear analysis method in the energy consumption feature engineering as energy consumption features, expressed as: { x1, x 2,…, x i ,… x n}; Establish an energy consumption prediction model for the air-conditioning terminal equipment as follows: P = f ( x 1, x 2,…, x i ,… x n ); In the formula, P is the operating energy consumption of the air-conditioning terminal equipment, x i and i ∈{1,2,…, n} is the i th energy consumption feature; f is the functional relationship between the operating energy consumption of the air-conditioning terminal equipment and the energy consumption features.

[0008] As a preferred implementation manner, perform hyperparameter optimization on the energy consumption prediction model, specifically: Use Bayesian optimization combined with cross-validation to optimize the hyperparameters of the energy consumption prediction model, including the kernel function type Kernel , the penalty parameter C , the insensitive loss coefficient epsilon , and the kernel function width gamma . The specific optimization steps are as follows: Construct an objective function, use the model determination coefficient R 2 as the target value, and use the 4 hyperparameters of the energy consumption prediction model Kernel , C , epsilon , gamma as the input of the function, and the output target is R 2 to be the maximum; Define the hyperparameter optimization range. The hyperparameter optimization ranges of the 4 hyperparameters of the energy consumption prediction model are as follows: ; In the formula, rbf represents the radial basis kernel function, linear represents the linear kernel function, ploy represents the polynomial kernel function, sigmoid represents the hyperbolic tangent kernel function.

[0009] As a preferred embodiment, determining the objective function, decision variables, and constraint conditions, and establishing an energy-saving optimization model for the air-conditioning terminal equipment specifically includes: Taking the minimum operating energy consumption of the air-conditioning terminal equipment as the objective, the objective function is as follows: min P = f’ ( y 1, y 2,…, y j ,… y m ); In the formula: P is the operating energy consumption of the air-conditioning terminal equipment; y i and j ∈{1,2,…, m} is the j th decision variable; f’ is the functional relationship between the operating energy consumption of the air-conditioning terminal equipment and the decision variables; The constraint conditions include geometric constraints that limit the value range of the decision variables themselves and associated constraints affected by the decision variables.

[0010] As a preferred embodiment, the improved PSO algorithm based on dynamically adjusting the inertia weight and compression factor is specifically as follows: ; ; In the formula: w is the inertia weight, w initial is the initial inertia weight, w final is the final inertia weight, iter is the number of iterations, maxiter is the maximum number of iterations, c is the compression factor, c initial is the initial compression factor, c final is the final compression factor.

[0011] As a preferred embodiment, solving the energy-saving optimization model specifically includes: Inputting the modeling data during the optimization time period into the energy-saving optimization model; Determining the objective function and various constraint conditions; Set the initial parameters of the improved PSO algorithm. The number of particles is 50, the cognitive factor and social factor are 1.5, the maximum and minimum weights are 0.9 and 0.4, the maximum and minimum compression factors are 1.0 and 0.1, and the maximum number of iterations is 200. Update the particle positions and velocities, calculate the fitness values, and stop the iteration after the fitness values have changed by less than 0.01 for 20 consecutive times or the number of iterations reaches 200 times.

[0012] Another aspect of the present invention also provides a precision air - conditioning energy - saving optimization system based on the improved PSO, which is applied to the above - mentioned precision air - conditioning energy - saving optimization method based on the improved PSO, and includes a data pre - processing module, an energy - consumption feature engineering module, an energy - consumption prediction module, and an energy - saving optimization module. The data pre - processing module is used to pre - process the historical operation data of the air - conditioning terminal equipment collected. The pre - processing includes outlier removal, missing value imputation, and data normalization. The energy - consumption feature engineering module is used to calculate the correlations of the influencing factors of the operating energy consumption of the air - conditioning terminal equipment by using linear analysis methods and non - linear analysis methods respectively. The energy - consumption prediction module is used to establish an energy - consumption prediction model for the air - conditioning terminal equipment according to the correlation calculation results of the energy - consumption feature engineering module, and optimize the hyper - parameters of the energy - consumption prediction model. The energy - saving optimization module is used to determine the objective function, decision variables, and constraint conditions, and establish an energy - saving optimization model for the air - conditioning terminal equipment. Solve the energy - saving optimization model based on the improved PSO algorithm with dynamically adjusted inertia weight and compression factor to obtain the optimal operation parameter combination when the air - conditioning terminal equipment operates with the minimum energy consumption under the constraint conditions.

[0013] Another aspect of the present invention also provides a storage medium storing a program, which, when executed by a processor, implements the above - mentioned precision air - conditioning energy - saving optimization method based on the improved PSO.

[0014] Compared with the prior art, the present invention has the following advantages and beneficial effects: (1) The present invention is applicable to the energy - consumption prediction and energy - saving optimization of precision air conditioners in multi - operating - condition change scenarios. Through the solution results of the precision air - conditioning energy - saving optimization model, the operation parameters of the precision air conditioner can be adjusted quickly and accurately, reducing the time required to maintain the environmental temperature.

[0015] (2) The present invention can perform intelligent energy fine management according to the real - time changes in the computer room thermal environment. Under the condition of ensuring the safe operation of IT equipment, the operation parameters of the precision air conditioner can be accurately adjusted to minimize the air - conditioning energy consumption.

[0016] (3) The present invention considers multiple constraints in energy-saving optimization, belonging to a multi-constraint objective optimization method. Existing technologies often only constrain a certain type of temperature and do not fully consider the requirements of the computer room environment temperature. Description of the Drawings

[0017] Figure 1 is a flowchart of the precision air conditioner energy-saving optimization method based on improved PSO according to an embodiment of the present invention; Figure 2 is a schematic structural diagram of the precision air conditioner energy consumption prediction model according to an embodiment of the present invention; Figure 3 is a schematic diagram of the hyperparameter optimization process based on BOA according to an embodiment of the present invention; Figure 4 is a schematic diagram of the optimization model solving process based on IPSO according to an embodiment of the present invention. Detailed Embodiments

[0018] In order to enable those skilled in the art to better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts shall fall within the protection scope of the present application.

[0019] Embodiment: As Figure 1 shown, this embodiment provides a precision air conditioner energy-saving optimization method based on improved PSO, including the following steps: S1. Obtain various historical data stored in the energy consumption supervision system, and preprocess the historical operation data of the obtained air-conditioning terminal equipment (also known as air handling unit, precision air conditioner, hereinafter referred to as precision air conditioner).

[0020] In one or more preferred embodiments, the preprocessing includes outlier removal, missing value imputation, and data normalization.

[0021] S2. Perform energy consumption feature engineering on the precision air conditioner.

[0022] For the precision air conditioners in the data center computer room, there are many factors affecting their operating energy consumption, such as the precision air conditioner wind speed percentage, various temperatures in the thermal environment, etc. To analyze the influence degree of each influencing factor on the energy consumption of the precision air conditioner and determine the input features of the energy consumption prediction model, in this embodiment, the linear analysis method and the non-linear analysis method are respectively used to calculate the correlation of each influencing factor on the operating energy consumption of each precision air conditioner. According to the calculation results, the n parameters that are all not less than their respective thresholds are used as energy consumption features, expressed as: {x 1, x 2,…, x i ,… x n}。Combining the characteristics of linear analysis methods and non - linear analysis methods to determine the model input features is beneficial to improving the accuracy and interpretability of the energy consumption prediction model.

[0023] In one or more preferred embodiments, the linear analysis method uses the Pearson correlation coefficient. The Pearson coefficient is used to measure the degree of linear correlation between two variables, and the formula is expressed as: (1); In the formula: cov ([[]] X , Y ) is a random variable X and Y covariance; , are respectively X and Y standard deviation; The value range of is [-1, 1].

[0024] In one or more preferred embodiments, the non - linear analysis method uses one of LightGBM feature importance, GBDT feature importance, and grey relational degree analysis.

[0025] Furthermore, the LightGBM feature importance is used to evaluate the contribution degree of feature variables to the prediction target, capture the non - linear relationship between features and the target, and obtain the relative importance by calculating the split gain of feature g 0 in each tree and taking the average of the results of all trees, as shown in Equation (2); The importance calculation formula of feature g 0 on each tree is shown in Equation (3).

[0026] (2); (3); In Equations (2) and (3): N’ is the number of trees; T i’ is the i’ th tree; is the relative importance result; is the degree of loss reduction after splitting; L is the number of leaf nodes; L -1 is the number of non - leaf nodes; v t is t node - related variable.

[0027] S3. Establish an energy consumption prediction model for the precision air conditioner based on the calculation results of the energy consumption feature engineering, and optimize the hyperparameters of the energy consumption prediction model.

[0028] S31. The specific steps to construct the energy consumption prediction model for the precision air conditioner are as follows: 1) Divide the processed data set into a training set, a validation set, and a test set according to a set ratio (in this embodiment, set to 3:1:1).

[0029] 2) Use the parameters that are not less than their respective thresholds in the calculation results of both the linear analysis method and the non - linear analysis method in the energy consumption feature engineering as energy consumption features, expressed as: { n 1, x 1, x 2, …, x i , … x n}.

[0030] 3) Establish an energy consumption prediction model for the precision air conditioner as follows: P = f ( x 1, x 2, …, x i , … x n ); In the formula, P is the operating energy consumption of the precision air conditioner, x i and i ∈ {1, 2, …, n} is the i th energy consumption feature; f is the functional relationship between the operating energy consumption of the precision air conditioner and the energy consumption features.

[0031] In one or more preferred embodiments, as Figure 2 shown, construct an energy consumption prediction model based on support vector regression (SVR), and select the parameters with a Pearson correlation coefficient greater than or equal to 0.2 with the operating energy consumption of the precision air conditioner and a LightGBM feature importance greater than or equal to 0.05 (including the wind speed ratio V , the cold point temperature C , the hot point temperature H and the return air temperature R ) as the energy consumption features and input them into the energy consumption prediction model, as shown in formula (4). The advantage of doing this is that considering the multi - condition changes of the air conditioner and the complex energy consumption changes, combining linear analysis and non - linear analysis is beneficial to improving the accuracy of energy consumption prediction.

[0032] P = f ( V , C , H , R ) (4).

[0033] S32. As Figure 3 shown, Bayesian optimization combined with cross - validation is used to optimize the hyperparameters of the energy consumption prediction model, including the kernel function type ( Kernel ), penalty parameter ( C ), insensitive loss coefficient ( epsilon ), and kernel function width ( gamma ). The Bayesian optimizer is constructed through the "scikit - optimize" library in Python. The specific optimization steps are as follows: 1) Construct the objective function, taking the coefficient of determination R 2 of the model as the target value, and taking the 4 hyperparameters of the SVR algorithm Kernel , C , epsilon , gamma as the input of the function, with the output target being R 2 maximum; 2) Define the hyperparameter optimization range. The optimization ranges of the 4 hyperparameters of SVR are as shown in Equation (5); (5); In the formula, rbf represents the radial basis kernel function, linear represents the linear kernel function, ploy represents the polynomial kernel function, sigmoid represents the hyperbolic tangent kernel function.

[0034] 3) Construct the Bayesian optimizer. The "scikit - optimize" library allows optimizing computationally expensive Gaussian functions and guiding the search through Gaussian processes to find the best parameter configuration.

[0035] S4. Determine the objective function, decision variables, and constraint conditions, and establish an energy - saving optimization model for precision air conditioners. The specific steps are as follows: 1) Determine the objective function and decision variables of the energy - saving optimization model. The objective function is the core of the optimization problem and is used to quantitatively evaluate the quality of the solution. It is usually expressed as a function of the decision variables. The decision variables are the undetermined parameters in the optimization problem, and the values of these variables will directly affect the result of the optimization problem. In this invention, the research on the energy - saving optimization of precision air conditioners aims to minimize the energy consumption of precision air conditioners, as shown in Equation (6).

[0036] min P = f’ ( y 1, y 2,…, y j ,… y m ) (6); In the formula: P is the operating energy consumption of the precision air conditioner; y i and j ∈{1,2,…, m} is the j th decision variable; f’ is the functional relationship between the operating energy consumption of the precision air conditioner and the decision variable.

[0037] 2) Determine the constraint conditions of the energy-saving optimization model. The constraint conditions refer to a set of conditions that limit the value range of the decision variables during the optimization process, including both the "geometric constraints" that limit the value range of the decision variables themselves and the "correlation constraints" affected by the decision variables.

[0038] In one or more preferred embodiments, the number of operating precision air conditioners N , the wind speed ratio V are used as decision variables, then the objective function is as shown in formula (7), and the constraint conditions are as shown in formula (8).

[0039] min P = f’ ( N , V ) (7); (8); In the formula: N is the number of operating precision air conditioners, V is the wind speed ratio, C is the cold point temperature, H is the hot point temperature, R is the return air temperature.

[0040] S5. Solve the energy-saving optimization model by using an improved PSO algorithm based on dynamically adjusting the inertia weight and the compression factor, and obtain the optimal operating parameter combination when the precision air conditioner operates with the minimum energy consumption under the constraint conditions.

[0041] S5.1. The improved PSO algorithm based on dynamically adjusting the inertia weight and the compression factor is specifically as follows: Improve the PSO algorithm based on formulas (9) and (10): (9); (10); In the formula: w is the inertia weight, w initial is the initial inertia weight, w final is the final inertia weight, iter is the number of iterations, maxiter is the maximum number of iterations, c is the compression factor, c initial is the initial compression factor, c final is the final compression factor.

[0042] S5.2. Solve the energy-saving optimization model, including the following steps: 1) Input the modeling data during the optimization time period into the energy-saving optimization model.

[0043] 2) Determine the objective function and various constraint conditions.

[0044] 3) Set the initial parameters of the improved PSO algorithm.

[0045] The specific process is as Figure 4 shown, including the following steps: S5.2.1. Initialize the decision variables; S5.2.2. Initialize the population parameters; S5.2.3. Dynamically update the inertia weight w and the compression factor c ; S5.2.4. Update the velocity and position of each particle, and calculate the fitness value; S5.2.5. Obtain the historical optimal solution of each particle p best ; S5.2.6. Evaluate each particle to obtain the global optimal solution g best ; S5.2.7. Judge whether the preset end condition is satisfied (in a specific embodiment, it is set to: the fitness value changes less than 0.01 for 20 consecutive times or the iteration stops after 200 iterations). If not, return to step S5.2.3; if so, jump to step S5.2.8; S5.2.8. Output the optimal operation parameter combination of the precision air conditioner and the objective function value.

[0046] S5.3. Set the parameters of the precision air conditioner according to the best operation parameter combination obtained from the solution result to achieve energy-saving optimization.

[0047] In one or more preferred embodiments, the objective function and various constraint conditions are determined as shown in Equations (7) and (8). The specific initial parameters for setting the improved PSO algorithm are as follows: the number of particles is 50, the cognitive factor and social factor are 1.5, the maximum and minimum weights are 0.9 and 0.4, the maximum and minimum compression factors are 1.0 and 0.1, and the maximum number of iterations is 200. Update the particle positions and velocities, calculate the fitness values, and stop the iteration after the fitness values change less than 0.01 for 20 consecutive times or the number of iterations reaches 200 times.

[0048] In another embodiment of the present application, a precision air-conditioning energy-saving optimization system based on an improved PSO is provided. The system includes a data preprocessing module, an energy consumption feature engineering module, an energy consumption prediction module, and an energy-saving optimization module. The data preprocessing module is used to preprocess the historical operation data of the collected precision air conditioner. The preprocessing includes outlier removal, missing value imputation, and data normalization. The energy consumption feature engineering module is used to calculate the correlation of the influencing factors of the operation energy consumption of the precision air conditioner by using linear analysis methods and nonlinear analysis methods respectively. The energy consumption prediction module is used to establish an energy consumption prediction model for the precision air conditioner according to the correlation calculation results of the energy consumption feature engineering module, and optimize the hyperparameters of the energy consumption prediction model. The energy-saving optimization module is used to determine the objective function, decision variables, and constraint conditions, and establish an energy-saving optimization model for the precision air conditioner; solve the energy-saving optimization model based on an improved PSO algorithm with dynamically adjusted inertia weight and compression factor, and obtain the optimal operation parameter combination for the precision air conditioner to operate with the minimum energy consumption under the constraint conditions.

[0049] It should be noted here that the systems provided in the above embodiments are only illustrated by the division of the above functional modules. In actual applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure is divided into different functional modules to complete all or part of the functions described above. This system is an improved PSO-based precision air-conditioning energy-saving optimization method applied to the above embodiments.

[0050] In another embodiment of the present application, a storage medium is further provided, storing a program, which when executed by a processor, implements an improved PSO-based precision air-conditioning energy-saving optimization method, specifically: Preprocess the historical operation data of the collected precision air conditioner. The preprocessing includes outlier removal, missing value imputation, and data normalization. Perform energy consumption feature engineering on the precision air conditioner, including calculating the correlation of the influencing factors of the operation energy consumption of the precision air conditioner by using linear analysis methods and nonlinear analysis methods respectively. Based on the correlation calculation results of energy consumption feature engineering, an energy consumption prediction model of a precision air conditioner is established, and the hyperparameters of the energy consumption prediction model are optimized; Determine the objective function, decision variables and constraints, and establish an energy-saving optimization model of the precision air conditioner; Based on an improved PSO algorithm that dynamically adjusts the inertia weight and compression factor, solve the energy-saving optimization model to obtain the optimal operation parameter combination when the precision air conditioner operates with the minimum energy consumption under the constraints.

[0051] It should be understood that each part of the present application can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits with logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits with appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0052] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications made without departing from the spirit and principle of the present invention shall be equivalent replacement methods and are all included in the protection scope of the present invention.

Claims

1. A precision air - conditioner energy - saving optimization method based on improved PSO, characterized in that, It includes the following steps: Preprocess the historical operation data of the air-conditioning terminal equipment collected, and the preprocessing includes outlier removal, missing value imputation, and data normalization; Conduct energy consumption feature engineering for the air-conditioning terminal equipment, including calculating the correlation of the influencing factors of the operating energy consumption of the air-conditioning terminal equipment respectively by using linear analysis methods and non-linear analysis methods; Establish an energy consumption prediction model for the air-conditioning terminal equipment according to the correlation calculation results of the energy consumption feature engineering, and optimize the hyperparameters of the energy consumption prediction model; Determine the objective function, decision variables, and constraint conditions, and establish an energy-saving optimization model for the air-conditioning terminal equipment; Solve the energy-saving optimization model based on an improved PSO algorithm with dynamically adjusted inertia weight and compression factor, and obtain the optimal operation parameter combination when the air-conditioning terminal equipment operates with the minimum energy consumption under the constraint conditions.

2. The precision air conditioner energy-saving optimization method based on improved PSO according to claim 1, characterized in that The linear analysis method uses the Pearson correlation coefficient, specifically: ； Wherein: cov ( X , Y ) is a random variable X and Y covariance; and are respectively X and Y standard deviations; ranges from [-1, 1].

3. The precision air conditioner energy-saving optimization method based on improved PSO according to claim 1, characterized in that The non-linear analysis method uses LightGBM feature importance, specifically: Calculate the split gain of feature g 0 in each tree, and average the results of all trees to obtain the relative importance, as shown in the following formula: ； Wherein: N’ is the number of trees; T i’ is the i’ th tree; is the relative importance result; Feature g The importance of 0 on each tree is as follows: ； In the formula: is the degree of loss decline after splitting; L is the number of leaf nodes; L -1 is the number of non-leaf nodes; v t is t a node-related variable.

4. The precision air conditioner energy-saving optimization method based on improved PSO according to claim 1, characterized in that The establishment of the energy consumption prediction model for the air-conditioning terminal equipment according to the correlation calculation results of the energy consumption feature engineering is specifically: Divide the preprocessed data set into a training set, a validation set, and a test set according to a set ratio; The parameters that are not less than their respective thresholds in the calculation results of both the linear analysis method and the non - linear analysis method in the energy consumption feature engineering are used as energy consumption features, expressed as: { n The x 1, x 2,…, x i ,… x n}; Establish an energy consumption prediction model for the air-conditioning terminal equipment, as shown in the following formula: P = f ( x 1, x 2,…, x i ,… x n )； In the formula, P is the operating energy consumption of the air-conditioning terminal equipment, x i and i ∈{1, 2, …, n} is the i th energy consumption characteristic; f is the functional relationship between the operating energy consumption of the air-conditioning terminal equipment and the energy consumption characteristics.

5. The precision air conditioner energy-saving optimization method based on improved PSO according to claim 1, characterized in that Optimize the hyperparameters of the energy consumption prediction model, specifically: Optimize the hyperparameters of the energy consumption prediction model using Bayesian optimization combined with cross-validation, including the kernel function type Kernel , penalty parameter C , insensitive loss coefficient epsilon and kernel function width gamma . The specific optimization steps are as follows: Construct the objective function, taking the coefficient of determination of the model R 2 as the target value, and using the four hyperparameters of the energy consumption prediction model Kernel , C , epsilon , gamma as the inputs of the function, with the output target being R 2 the maximum; Define the hyperparameter optimization range. The hyperparameter optimization ranges of the 4 hyperparameters of the energy consumption prediction model are as shown in the following formula: ； In the formula, rbf represents the radial basis kernel function, linear represents the linear kernel function, ploy represents the polynomial kernel function, sigmoid represents the hyperbolic tangent kernel function.

6. The precision air conditioner energy-saving optimization method based on improved PSO according to claim 1, wherein The determination of the objective function, decision variables, and constraint conditions, and the establishment of the energy-saving optimization model for the air-conditioning terminal equipment are specifically: Taking the minimum operating energy consumption of the air-conditioning terminal equipment as the goal, the objective function is as shown in the following formula: min P = f’ ( y 1, y 2,…, y j ,… y m ); Wherein: P is the operating energy consumption of the air-conditioning terminal equipment; y i and j ∈{1, 2, …, m} is the j th decision variable; f’ is the functional relationship between the operating energy consumption of the air-conditioning terminal equipment and the decision variable; The constraint conditions include geometric constraints that limit the value range of the decision variables themselves, and correlation constraints affected by the decision variables.

7. The precision air conditioner energy-saving optimization method based on improved PSO according to claim 1, characterized in that The improved PSO algorithm based on dynamically adjusting the inertia weight and compression factor is specifically: ； ； Wherein: w is the inertia weight, w initial is the initial inertia weight, w final is the final inertia weight, iter is the number of iterations, maxiter is the maximum number of iterations, c is the compression factor, c initial is the initial compression factor, c final is the final compression factor.

8. The precision air conditioner energy-saving optimization method based on the improved PSO according to claim 7, characterized in that Solve the energy-saving optimization model, specifically: Input the modeling data during the optimization period into the energy-saving optimization model; Determine the objective function and various constraint conditions; Set the initial parameters of the improved PSO algorithm, the number of particles is 50, the cognitive factor and social factor are 1.5, the maximum and minimum weights are 0.9 and 0.4, the maximum and minimum compression factors are 1.0 and 0.1, and the maximum number of iterations is 200; update the particle position and speed, calculate the fitness value, and stop iterating after the fitness value changes less than 0.01 for 20 consecutive times or the number of iterations reaches 200 times.

9. The precision air - conditioning energy - saving optimization system based on improved PSO is characterized in that, Applied to the precision air-conditioning energy-saving optimization method based on the improved PSO described in any one of claims 1-8, including a data preprocessing module, an energy consumption feature engineering module, an energy consumption prediction module, and an energy-saving optimization module; The data preprocessing module is used to preprocess the historical operation data of the air-conditioning terminal equipment collected, and the preprocessing includes outlier removal, missing value imputation, and data normalization; The energy consumption feature engineering module is used to calculate the correlation of the influencing factors of the operating energy consumption of the air-conditioning terminal equipment by using the linear analysis method and the non-linear analysis method respectively; The energy consumption prediction module is used to establish an energy consumption prediction model of the air-conditioning terminal equipment according to the correlation calculation result of the energy consumption feature engineering module, and optimize the hyperparameters of the energy consumption prediction model; The energy-saving optimization module is used to determine the objective function, decision variables and constraint conditions, and establish an energy-saving optimization model of the air-conditioning terminal equipment; An improved PSO algorithm based on dynamically adjusting the inertia weight and compression factor is used to solve the energy-saving optimization model, and the optimal operation parameter combination when the air-conditioning terminal equipment operates with the minimum energy consumption under the constraint conditions is obtained.

10. A storage medium stores a program, characterized in that: When the program is executed by the processor, the precision air-conditioning energy-saving optimization method based on the improved PSO described in any one of claims 1-8 is realized.

Images