Central air conditioner control method and system and storage medium
Through multiple sensors to acquire data and build deep learning models and optimization algorithms, the problem of difficulty in achieving accurate temperature and humidity adjustment in complex environments is solved, and the control accuracy and energy saving of the air conditioning system are improved.
Patent Information
- Application Number
- CN202411761048.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-03
- Publication Date
- 2025-03-18
AI Technical Summary
It is difficult for existing air conditioning systems to achieve accurate adjustment of indoor temperature and humidity under complex environmental conditions, resulting in reduced indoor comfort and greater energy consumption.
By arranging a variety of sensors (such as temperature sensors, humidity sensors, human body sensing sensors, and light sensors), obtain reference data, perform preprocessing and feature extraction, build deep learning models and optimization algorithms, and optimize the operating parameters of the air-conditioning system to achieve more accurate control.
It realizes more precise control of the central air-conditioning system, improves indoor comfort, and improves the energy saving and automatic adjustment capabilities of the system by optimizing energy consumption.
Smart Images
Figure CN119642376A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of air conditioning control, and more specifically, to a central air conditioning control method, system and storage medium. Background Art
[0002] With the improvement of modern buildings and living standards, people's requirements for indoor air quality (IAQ) and temperature and humidity comfort are constantly increasing. Traditional air conditioning systems (HVAC, Heating, Ventilation and Air Conditioning) mainly focus on temperature regulation, and it is difficult to accurately control humidity at the same time. Especially in high humidity areas or under environmental changes in different seasons, traditional systems are prone to problems such as insufficient humidity regulation and excessive energy consumption. In the past, the design of HVAC systems often ignored the problem of humidity regulation, especially in areas with high humidity in summer, where humidity is often difficult to control; in winter, the air may be too dry. Existing air conditioning systems are difficult to achieve precise regulation of the indoor environment under complex environmental conditions, resulting in reduced indoor comfort and high energy consumption during system operation. In addition, with the continuous improvement of national requirements for carbon emissions and energy efficiency, the development of energy-saving technologies in the construction field is particularly urgent. The heat pump fresh air system combines the advantages of heat pump technology and fresh air system, and can provide cooling, heating, dehumidification and fresh air ventilation functions at the same time. It is particularly suitable for air conditioning needs in high humidity and high load environments. However, how to accurately control indoor temperature and humidity under complex load conditions and how to improve the system's energy efficiency and automatic adjustment capabilities have become important research directions in the industry.
[0003] The prior art discloses a multi-sensor fusion intelligent central air-conditioning control system based on deep learning, including: a central air-conditioning outdoor unit, a central controller deploying a deep learning model, and a central air-conditioning indoor unit, a wind speed and direction regulator, a temperature sensor, a humidity sensor and a radar sensor configured in each working area; the present invention realizes real-time monitoring of the temperature, humidity and number of people in the working area through the collaborative work of multiple temperature sensors, humidity sensors and radar sensors, and generates a comprehensive environmental index based on these data, and intelligently controls the central air-conditioning system through a deep learning model to optimize the energy efficiency and user comfort of the air-conditioning system. However, the system only intelligently controls the central air-conditioning system through a deep learning model. Limited by the limitations of the deep learning model itself, the control of the air-conditioning operation is limited. Summary of the invention
[0004] The purpose of the present invention is to disclose a central air-conditioning control method, system and storage medium with better performance.
[0005] In order to achieve the above object, the present invention provides a central air conditioning control method, comprising:
[0006] S1: Arrange multiple sensors and obtain reference data through multiple sensors;
[0007] S2: preprocess the reference data to obtain preprocessed data;
[0008] S3: Extract features from the preprocessed data to obtain feature data; build a deep learning model, train the deep learning model with the feature data to obtain a parameter prediction model; obtain predicted operating parameters through the parameter prediction model;
[0009] S4: Based on the predicted operating parameters, define the optimization objective function, and optimize the optimization objective function through the optimization algorithm; to obtain the final operating parameters;
[0010] S5: Control the central air conditioner according to the final operating parameters.
[0011] Further, step S1 includes:
[0012] The multiple sensors include at least one of the following: a temperature sensor, a humidity sensor, a human body sensor, and a light sensor;
[0013] The reference data includes at least one of the following: indoor temperature, humidity, light intensity, personnel distribution, and the operation data of the air-conditioning system includes at least one of the following: chilled water temperature, cooling water temperature, fan speed.
[0014] Furthermore, the preprocessing in step S2 includes:
[0015] S2.1: Data cleaning:
[0016] Handling missing data: Delete rows or columns containing missing values; Use mean, median, mode, interpolation, KNN, and regression methods to fill in; For different categories of data, use a specific marker value to indicate missingness;
[0017] Handling duplicate data: Use the drop_duplicates() method in Pandas to remove duplicate data;
[0018] Handling outliers: Use box plots to identify outliers; Delete outliers after identifying them;
[0019] S2.2: Data conversion:
[0020] Data type conversion: Convert data columns to the correct data type as needed;
[0021] Standardization: Normalization adjusts the data to a standard normal distribution with a mean of 0 and a variance of 1.
[0022] Furthermore, step S3 includes: using TensorFlow as a machine learning framework to build an RNN model as a deep learning model.
[0023] Furthermore, the RNN model includes: an input layer, a hidden layer, and an output layer;
[0024] in
[0025] The state update of the hidden layer:
[0026] h t =f(W hh h t-1 +W xh x t +b h )
[0027] h t is the hidden layer state at time step;
[0028] h t-1 is the hidden layer state at the previous time step;
[0029] x t is the input of the current time step;
[0030] W hh is the weight matrix from hidden layer to hidden layer;
[0031] W xh is the weight matrix from the input layer to the hidden layer;
[0032] b h is the bias term of the hidden layer;
[0033] f is the activation function, which is a nonlinear function;
[0034] t is the time step;
[0035] Output layer calculation:
[0036] y t is the output of the time step;
[0037] W hy is the weight matrix from the hidden layer to the output layer;
[0038] b y is the bias term of the output layer;
[0039] g is the activation function of the output layer, which is a linear activation function;
[0040] The following is the time step iteration process of RNN in sequence data processing:
[0041] Time step 1: h1 =f(W hh h 0 +W xh x 1 +b h );
[0042] Time step 2: h 2 =f(W hh h 1 +W xh x 2 +b h );
[0043] Time step m: h t =f(W hh h t-1 +W xh x t +b h ).
[0044] Further, in step S4, the objective function includes: F = αE + β (1-C) + γ (1-L)
[0045] in:
[0046] E: is the total energy consumption of the central air conditioning system, obtained by calculating the electrical energy consumed during the operation of the system;
[0047] C: is the indoor comfort index
[0048] L: is the predicted value of equipment life;
[0049] α, β, and γ are the weight coefficients corresponding to energy consumption, comfort, and equipment life, respectively.
[0050] Further, in step S4, optimizing by using an optimization algorithm includes: optimizing by using a simulated annealing algorithm:
[0051] initialization:
[0052] Choose an initial solution x 0 ;
[0053] Set the initial temperature T 0 and the rate of temperature drop;
[0054] Set the maximum number of iterations or stop condition;
[0055] Iteration process:
[0056] At each temperature T, the solution is optimized step by step by following the steps:
[0057] Generate neighborhood solutions: Randomly generate a new candidate solution x from the current solution new, the candidate solution should be a neighborhood solution of the current solution; the neighborhood solution is obtained through small changes;
[0058] Calculate the energy difference: Calculate the difference between the objective function and the candidate solution, set as:
[0059] ΔE=f(x new )-f(x t )
[0060] Where f(x) represents the quality of the solution;
[0061] Acceptance criteria: Decide whether to accept the candidate solution x based on the energy difference ΔE and temperature T new :
[0062] If ΔE≤0, that is, the new solution is better, directly accept the new solution and update the current solution to x new ;
[0063] If ΔE>0, that is, the new solution is worse, then the probability Accept new interpretations;
[0064] Stop condition: When the temperature drops to a predetermined low value, the iteration stops; or when the number of iterations exceeds the set maximum value, the iteration stops;
[0065] Output the optimal solution: At the end of the algorithm, output the current optimal solution or the objective function value corresponding to the current solution.
[0066] Furthermore, step S6: through the user interface, the operating status of the air-conditioning system is monitored and adjusted in real time to achieve remote management and maintenance.
[0067] In addition, the present invention also provides a central air-conditioning control system, comprising:
[0068] Data module: deploy multiple sensors and obtain reference data through multiple sensors;
[0069] Preprocessing module: preprocess the reference data to obtain preprocessed data;
[0070] Model module: extract features from preprocessed data to obtain feature data; build a deep learning model, train the deep learning model with feature data to obtain a parameter prediction model; obtain predicted operating parameters through the parameter prediction model;
[0071] Optimization module: Define the optimization objective function according to the predicted operating parameters, and optimize the optimization objective function through the optimization algorithm to obtain the final operating parameters;
[0072] Control module: Control the central air conditioner according to the final operating parameters.
[0073] In addition, the present invention provides a central air-conditioning control storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned central air-conditioning control method is implemented.
[0074] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:
[0075] The present invention extracts features from preprocessed data to obtain feature data; and constructs a deep learning model, trains the deep learning model through the feature data to obtain a parameter prediction model; obtains predicted operating parameters through the parameter prediction model; then defines an optimization objective function based on the predicted operating parameters, and optimizes the optimization objective function through an optimization algorithm; and obtains final operating parameters; the parameters predicted by the deep learning model can be further optimized to improve the operating performance of the central air conditioner. BRIEF DESCRIPTION OF THE DRAWINGS
[0076] Figure 1 This is a flow chart of a central air conditioning control method according to Embodiment 1;
[0077] Figure 2 This is a block diagram of a central air conditioning control system described in Example 3; DETAILED DESCRIPTION
[0078] The drawings are for illustrative purposes only and should not be construed as limiting the present patent;
[0079] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.
[0080] Embodiment 1:
[0081] This embodiment provides Figure 1 A central air conditioning control method is shown, comprising:
[0082] S1: Arrange multiple sensors and obtain reference data through multiple sensors;
[0083] S2: preprocess the reference data to obtain preprocessed data;
[0084] S3: Extract features from the preprocessed data to obtain feature data; build a deep learning model, train the deep learning model with the feature data to obtain a parameter prediction model; obtain predicted operating parameters through the parameter prediction model;
[0085] S4: Based on the predicted operating parameters, define the optimization objective function, and optimize the optimization objective function through the optimization algorithm; to obtain the final operating parameters;
[0086] S5: Control the central air conditioner according to the final operating parameters.
[0087] This embodiment extracts features from the preprocessed data to obtain feature data; and constructs a deep learning model, trains the deep learning model through the feature data to obtain a parameter prediction model; obtains predicted operating parameters through the parameter prediction model; then defines an optimization objective function based on the predicted operating parameters, and optimizes the optimization objective function through an optimization algorithm; and obtains final operating parameters; the parameters predicted by the deep learning model can be further optimized to improve the operating performance of the central air conditioner.
[0088] Embodiment 2:
[0089] This embodiment further discloses on the basis of the first embodiment:
[0090] The multiple sensors include at least one of the following: a temperature sensor, a humidity sensor, a human body sensor, and a light sensor;
[0091] The reference data includes at least one of the following: indoor temperature, humidity, light intensity, personnel distribution, and the operation data of the air-conditioning system includes at least one of the following: chilled water temperature, cooling water temperature, fan speed.
[0092] The preprocessing in step S2 includes:
[0093] S2.1: Data cleaning:
[0094] Handling missing data: Delete rows or columns containing missing values; Use mean, median, mode, interpolation, KNN, and regression methods to fill in; For different categories of data, use a specific marker value to indicate missingness;
[0095] Handling duplicate data: Use the drop_duplicates() method in Pandas to remove duplicate data;
[0096] Handling outliers: Use box plots to identify outliers; Delete outliers after identifying them;
[0097] S2.2: Data conversion:
[0098] Data type conversion: Convert data columns to the correct data type as needed;
[0099] Standardization: Normalization adjusts the data to a standard normal distribution with a mean of 0 and a variance of 1.
[0100] Step S3 includes: using TensorFlow as a machine learning framework to build an RNN model as a deep learning model.
[0101] The RNN model includes: input layer, hidden layer and output layer;
[0102] in
[0103] The state update of the hidden layer:
[0104] h t =f(W hh h t-1 +W xh x t +b h )
[0105] h t is the hidden layer state at time step;
[0106] h t-1 is the hidden layer state at the previous time step;
[0107] x t is the input of the current time step;
[0108] W hh is the weight matrix from hidden layer to hidden layer;
[0109] W xh is the weight matrix from the input layer to the hidden layer;
[0110] b h is the bias term of the hidden layer;
[0111] f is the activation function, which is a nonlinear function;
[0112] t is the time step;
[0113] Output layer calculation:
[0114] y t is the output of the time step;
[0115] W hy is the weight matrix from the hidden layer to the output layer;
[0116] b y is the bias term of the output layer;
[0117] g is the activation function of the output layer, which is a linear activation function;
[0118] The following is the time step iteration process of RNN in sequence data processing:
[0119] Time step 1: h 1 =f(W hh h 0 +W xh x 1 +b h );
[0120] Time step 2: h 2 =f(W hh h 1 +Wxh x 2 +b h );
[0121] Time step m: h t =f(W hh h t-1 +W xh x t +b h ).
[0122] In step S4, the objective function includes: F = αE + β (1-C) + γ (1-L)
[0123] in:
[0124] E: is the total energy consumption of the central air conditioning system, obtained by calculating the electrical energy consumed during the operation of the system;
[0125] C: is the indoor comfort index
[0126] L: is the predicted value of equipment life;
[0127] α, β, and γ are the weight coefficients corresponding to energy consumption, comfort, and equipment life, respectively.
[0128] In step S4, optimizing by using an optimization algorithm includes: optimizing by using a simulated annealing algorithm:
[0129] initialization:
[0130] Choose an initial solution x 0 ;
[0131] Set the initial temperature T 0 and the rate of temperature drop;
[0132] Set the maximum number of iterations or stop condition;
[0133] Iteration process:
[0134] At each temperature T, the solution is optimized step by step by following the steps:
[0135] Generate neighborhood solutions: Randomly generate a new candidate solution x from the current solution new , the candidate solution should be a neighborhood solution of the current solution; the neighborhood solution is obtained through small changes;
[0136] Calculate the energy difference: Calculate the difference between the objective function and the candidate solution, set as:
[0137] ΔE=f(x new )-f(x t )
[0138] Where f(x) represents the quality of the solution;
[0139] Acceptance criteria: Decide whether to accept the candidate solution x based on the energy difference ΔE and temperature T new :
[0140] If ΔE≤0, that is, the new solution is better, directly accept the new solution and update the current solution to x new ;
[0141] If ΔE>0, that is, the new solution is worse, then the probability Accept new interpretations;
[0142] Stop condition: When the temperature drops to a predetermined low value, the iteration stops; or when the number of iterations exceeds the set maximum value, the iteration stops;
[0143] Output the optimal solution: At the end of the algorithm, output the current optimal solution or the objective function value corresponding to the current solution.
[0144] Step S6: Through the user interface, the operating status of the air-conditioning system is monitored and adjusted in real time to achieve remote management and maintenance.
[0145] The main advantages of using simulated annealing include the following:
[0146] 1. Global search capability
[0147] One of the biggest advantages of simulated annealing is its powerful global search capability. During the optimization process, the algorithm is able to jump out of the local optimal solution and avoid falling into the local minimum (or maximum). This is because the simulated annealing algorithm allows a certain degree of "jumping" during the search process, that is, accepting some inferior solutions, which can help the algorithm explore a wider solution space. Especially when the search space is very large or complex, simulated annealing can effectively avoid falling into the local optimum.
[0148] 2. Not dependent on gradient information
[0149] The simulated annealing algorithm does not rely on the gradient information of the objective function. This allows it to handle non-differentiable or discontinuous objective functions. For some complex combinatorial optimization problems or black box optimization problems, simulated annealing is a very useful tool.
[0150] 3. Strong robustness
[0151] The simulated annealing algorithm is highly robust and can adapt to different types of problems. Whether the problem is linear, nonlinear, or has many constraints, simulated annealing can effectively solve it by properly adjusting the temperature parameters and acceptance probability function.
[0152] 4. Simple and easy to implement
[0153] The implementation of the simulated annealing algorithm is relatively simple. The core idea is easy to understand, and the implementation process is not complicated. You only need to set a few basic elements such as the initial temperature, temperature cooling strategy, neighborhood structure, acceptance criteria, etc. to quickly build an effective optimization model.
[0154] 5. Strong adaptability
[0155] The temperature drop process of simulated annealing can be adjusted in a variety of ways, which is flexible. For example, the temperature can be adjusted in a linear, exponential, or more complex adaptive manner. Through different adjustment strategies, the algorithm can perform better in different scenarios.
[0156] 6. Avoid premature convergence
[0157] Because the "acceptance probability" used in simulated annealing allows occasional acceptance of poor solutions, the algorithm is prevented from converging prematurely. As the temperature decreases, the algorithm's search becomes more and more refined, and when approaching the optimal solution, the algorithm will focus more on local optimization. Therefore, simulated annealing has a good balance, which allows for a wide range of searches while focusing on refining the accuracy of the solution in the final stage.
[0158] This embodiment extracts features from the preprocessed data to obtain feature data; and constructs a deep learning model, trains the deep learning model through the feature data to obtain a parameter prediction model; obtains predicted operating parameters through the parameter prediction model; then defines an optimization objective function based on the predicted operating parameters, and optimizes the optimization objective function through an optimization algorithm; and obtains final operating parameters; the parameters predicted by the deep learning model can be further optimized to improve the operating performance of the central air conditioner.
[0159] Embodiment three:
[0160] This embodiment provides Figure 2 A central air conditioning control system is shown, comprising:
[0161] Data module: deploy multiple sensors and obtain reference data through multiple sensors;
[0162] Preprocessing module: preprocess the reference data to obtain preprocessed data;
[0163] Model module: extract features from preprocessed data to obtain feature data; build a deep learning model, train the deep learning model with feature data to obtain a parameter prediction model; obtain predicted operating parameters through the parameter prediction model;
[0164] Optimization module: Define the optimization objective function according to the predicted operating parameters, and optimize the optimization objective function through the optimization algorithm to obtain the final operating parameters;
[0165] Control module: Control the central air conditioner according to the final operating parameters.
[0166] This embodiment extracts features from the preprocessed data to obtain feature data; and constructs a deep learning model, trains the deep learning model through the feature data to obtain a parameter prediction model; obtains predicted operating parameters through the parameter prediction model; then defines an optimization objective function based on the predicted operating parameters, and optimizes the optimization objective function through an optimization algorithm; and obtains final operating parameters; the parameters predicted by the deep learning model can be further optimized to improve the operating performance of the central air conditioner.
[0167] Embodiment 4:
[0168] This embodiment provides a central air conditioning control storage medium, on which a computer program is stored. When the computer program is executed by a processor, the central air conditioning control method is implemented.
[0169] This embodiment extracts features from the preprocessed data to obtain feature data; and constructs a deep learning model, trains the deep learning model through the feature data to obtain a parameter prediction model; obtains predicted operating parameters through the parameter prediction model; then defines an optimization objective function based on the predicted operating parameters, and optimizes the optimization objective function through an optimization algorithm; and obtains final operating parameters; the parameters predicted by the deep learning model can be further optimized to improve the operating performance of the central air conditioner.
[0170] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.
Claims
1. A central air conditioning control method, characterized in that: include: S1: Arrange multiple sensors and obtain reference data through multiple sensors; S2: preprocess the reference data to obtain preprocessed data; S3: extracting features from the preprocessed data to obtain feature data; A deep learning model is constructed, and the deep learning model is trained through feature data to obtain a parameter prediction model; through the parameter prediction model, the predicted operating parameters are obtained; S4: Based on the predicted operating parameters, define the optimization objective function, and optimize the optimization objective function through the optimization algorithm; to obtain the final operating parameters; S5: Control the central air conditioner according to the final operating parameters.
2. A central air conditioning control method according to claim 1, characterized in that: Step S1 includes: The multiple sensors include at least one of the following: a temperature sensor, a humidity sensor, a human body sensor, and a light sensor; The reference data includes at least one of the following: indoor temperature, humidity, light intensity, personnel distribution, and the operation data of the air-conditioning system includes at least one of the following: chilled water temperature, cooling water temperature, fan speed.
3. A central air conditioning control method according to claim 1, characterized in that: The preprocessing in step S2 includes: S2.1: Data cleaning: Handling missing data: Delete rows or columns containing missing values; Use mean, median, mode, interpolation, KNN, and regression methods to fill in; For different categories of data, use a specific marker value to indicate missingness; Handling duplicate data: Use the drop_duplicates() method in Pandas to remove duplicate data; Handling outliers: Use box plots to identify outliers; Delete outliers after identifying them; S2.2: Data conversion: Data type conversion: Convert data columns to the correct data type as needed; Standardization: Normalization adjusts the data to a standard normal distribution with a mean of 0 and a variance of 1.
4. A central air conditioning control method according to claim 1, characterized in that: Step S3 includes: using TensorFlow as a machine learning framework to build an RNN model as a deep learning model.
5. A central air conditioning control method according to claim 4, characterized in that: The RNN model includes: input layer, hidden layer and output layer; in The state update of the hidden layer: h t =f(W hh h t-1 +W xh x t +b h ) h t : is the hidden layer state at time step; h t-1 is the hidden layer state at the previous time step; x t is the input of the current time step; W hh is the weight matrix from hidden layer to hidden layer; W xh is the weight matrix from the input layer to the hidden layer; b h is the bias term of the hidden layer; f is the activation function, which is a nonlinear function; t is the time step; Output layer calculation: yt is the output of the time step; W hy is the weight matrix from the hidden layer to the output layer; b y is the bias term of the output layer; g is the activation function of the output layer, which is a linear activation function; The following is the time step iteration process of RNN in sequence data processing: Time step 1: h1 = f(W hh h0+W xh x1+b h ); Time step 2: h2 = f(W hh h1+W xh x2+b h ); Time step m: h t =f(W hh h t-1 ++W xh x t +b h ).
6. A central air conditioning control method according to claim 1, characterized in that: In step S4, the objective function includes: F = αE + β (1-C) + γ (1-L) in: E: is the total energy consumption of the central air conditioning system, obtained by calculating the electrical energy consumed during the operation of the system; C: is the indoor comfort index L: is the predicted value of equipment life; α, β, and γ are the weight coefficients corresponding to energy consumption, comfort, and equipment life, respectively.
7. A central air conditioning control method according to claim 1, characterized in that: In step S4, optimizing by using an optimization algorithm includes: optimizing by using a simulated annealing algorithm: initialization: Select an initial solution x0; Set the initial temperature T0 and the temperature drop rate; Set the maximum number of iterations or stop condition; Iteration process: At each temperature T, the solution is optimized step by step by following the steps: Generate neighborhood solutions: Randomly generate a new candidate solution x from the current solution new , the candidate solution should be a neighborhood solution of the current solution; the neighborhood solution is obtained through small changes; Calculate the energy difference: Calculate the difference between the objective function and the candidate solution, set as: ΔE=f(x new )-f(x t ) Among them, f(x) represents the quality of the solution; Acceptance criteria: Decide whether to accept the candidate solution x based on the energy difference ΔE and temperature T new : If ΔE≤0, that is, the new solution is better, directly accept the new solution and update the current solution to x new ; If ΔE>0, that is, the new solution is worse, then the probability Accept new interpretations; Stop condition: When the temperature drops to a predetermined low value, the iteration stops; or when the number of iterations exceeds the set maximum value, the iteration stops; Output the optimal solution: At the end of the algorithm, output the current optimal solution or the objective function value corresponding to the current solution.
8. A central air conditioning control method according to claim 1, characterized in that: The system also includes step S6: monitoring and adjusting the operating status of the air-conditioning system in real time through the user interface to achieve remote management and maintenance.
9. A central air conditioning control system, characterized in that: include: Data module: deploy multiple sensors and obtain reference data through multiple sensors; Preprocessing module: preprocess the reference data to obtain preprocessed data; Model module: extract features from preprocessed data to obtain feature data; A deep learning model is constructed, and the deep learning model is trained through feature data to obtain a parameter prediction model; through the parameter prediction model, the predicted operating parameters are obtained; Optimization module: Define the optimization objective function according to the predicted operating parameters, and optimize the optimization objective function through the optimization algorithm to obtain the final operating parameters; Control module: Control the central air conditioner according to the final operating parameters.
10. A central air conditioning control storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, a central air conditioning control method as described in any one of claims 1 to 8 is implemented.