Central air conditioner energy-saving control method based on AI self-adaptive adjustment

Abstract

The invention discloses a central air conditioner energy-saving control method based on AI self-adaptive adjustment, and relates to the technical field of intelligent control, and the method comprises the following steps: collecting environmental data, equipment operation data and energy consumption data of a central air conditioner in real time, preprocessing multi-modal data, constructing a physical constraint equation in combination with a thermodynamic law, and generating a multi-modal data set; generating a multi-target optimization control instruction according to the optimization weight, solving an optimal equipment parameter combination through a Pareto frontier algorithm, and transmitting the optimal equipment parameter combination to a central air conditioner actuator; actual data after instruction execution are collected, the deviation degree of the energy-saving efficiency and the comfort degree is calculated, physical information neural network parameters are updated through a meta-learning framework, and thermodynamic partial differential equation coefficients are adjusted. According to the method, by constructing a multi-modal physical information fusion framework and a dynamic closed-loop optimization system, the comprehensive regulation and control capability of the central air conditioner in a complex building environment is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, and in particular to an energy-saving control method for a central air conditioner based on AI adaptive regulation. Background Art

[0002] In recent years, intelligent optimization and control technology for central air conditioning has gradually evolved from a single data-driven model to a multimodal fusion approach. Traditional methods primarily rely on statistical learning of historical data or physical modeling based on thermodynamic equations, but both have significant limitations. Data-driven models lack explicit embedding of physical laws, resulting in predictions that easily deviate from thermodynamic constraints under complex operating conditions. Purely physical models, on the other hand, struggle to adapt to dynamic environmental changes and exhibit high computational complexity. With the rise of physical information neural networks, some studies have attempted to incorporate partial differential equations as soft constraints into deep learning frameworks. However, these approaches still suffer from rigid weight allocation and insufficient causal reasoning in dynamic multi-objective optimization scenarios, making it impossible to effectively balance the real-time conflict between energy conservation and comfort.

[0003] The core shortcoming of existing technologies lies in the inadequate integration of data-driven approaches with physical laws, which limits the effectiveness of multi-objective collaborative optimization. Physical constraints are superimposed a posteriori in the form of loss functions, making it difficult to ensure that predictions strictly conform to thermodynamic equations. A causal inference mechanism is missing, making it impossible to quantify the combined effects of key variables on energy consumption and comfort. Dynamic weight allocation relies on empirical rules, and optimization models are lagging in updates, making them unable to adapt online to dynamic disturbances such as equipment aging and environmental changes. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a central air-conditioning energy-saving control method based on AI adaptive adjustment to solve the problems of prediction deviation and poor dynamic adaptability caused by insufficient integration of data-driven models and physical laws in the multi-objective optimization control of central air-conditioning.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a central air conditioning energy-saving control method based on AI adaptive regulation, which includes real-time collection of central air conditioning environmental data, equipment operation data, and energy consumption data, preprocessing multimodal data, and constructing physical constraint equations based on the laws of thermodynamics to generate a multimodal data set;

[0008] The multimodal data set is input into a physical information neural network. Through a deep learning model embedded with thermodynamic partial differential equation constraints, the thermodynamic state and cooling load demand of the building in the future period are predicted and the prediction results are output.

[0009] Based on the prediction results, a causal graph model is constructed to identify key variables that affect energy consumption and comfort, quantify the causal effects of key variables, and dynamically allocate optimization weights for energy saving and comfort;

[0010] Generate multi-objective optimization control instructions based on optimization weights, solve the optimal equipment parameter combination through the Pareto frontier algorithm, and transmit it to the central air conditioning actuator;

[0011] The actual data after the instruction execution is collected, the energy-saving efficiency and comfort deviation are calculated, and the physical information neural network parameters are updated and the coefficients of the thermodynamic partial differential equation are adjusted through the meta-learning framework.

[0012] As a preferred solution of the central air-conditioning energy-saving control method based on AI adaptive adjustment described in the present invention, wherein: the generation of a multimodal data set, the specific steps are as follows:

[0013] Perform timestamp alignment and sampling frequency normalization on multimodal data, use generative adversarial networks to complete missing data, and output preprocessed multimodal data;

[0014] Construct physical constraint equations based on the equipment characteristics of central air conditioners and the laws of thermodynamics;

[0015] The preprocessed multimodal data are fused with the physical constraint equations to generate a multimodal dataset containing physical laws.

[0016] As a preferred solution of the central air-conditioning energy-saving control method based on AI adaptive adjustment described in the present invention, the physical constraint equation includes the heat conduction equation and the energy conservation equation.

[0017] As a preferred solution of the central air-conditioning energy-saving control method based on AI adaptive adjustment described in the present invention, wherein: the output prediction result is specifically performed in the following steps:

[0018] Input the multimodal dataset into the spatiotemporal encoder and physical encoder of the physical information neural network to extract spatiotemporal features and generate implicit physical features;

[0019] The fusion layer of the input graph neural network integrates the spatiotemporal features and implicit physical features through a dynamic adaptive weight function to perform integral residual projection on the thermodynamic partial differential equation in the spatiotemporal domain to generate physically consistent spatiotemporal joint features;

[0020] Based on the spatiotemporal joint characteristics, a multi-scale prediction of the building's overall cooling load demand and the spatial distribution of the building's internal thermodynamic state are performed. The building's overall cooling load demand prediction is output through a fully connected layer, and the spatial distribution of the building's internal thermodynamic state is output through a U-Net structure.

[0021] The cooling load demand forecast value and the spatial distribution of the thermodynamic state are input into the normalized mapping layer to generate the forecast results.

[0022] As a preferred solution of the central air-conditioning energy-saving control method based on AI adaptive adjustment described in the present invention, the specific steps of dynamically allocating the optimization weights of energy saving and comfort are as follows:

[0023] Extract causal candidate variables from the prediction results and construct a causal graph model using the improved PC algorithm;

[0024] Based on the causal graph model, the key variables that affect energy consumption and comfort are identified. A dual machine learning method is used to calculate the joint causal effect coefficients of the key variables on energy consumption and comfort. The KL divergence function is used to quantify the distribution difference between historical data and real-time data.

[0025] According to the joint causal effect coefficient and distribution difference, a dynamic weight function driven by information entropy is constructed to dynamically allocate the optimization weights of energy saving and comfort.

[0026] As a preferred solution of the central air-conditioning energy-saving control method based on AI adaptive adjustment described in the present invention, the key variables refer to variables that have a causal impact on the energy consumption and comfort of the central air-conditioning identified by the causal graph model;

[0027] The key variables include energy consumption related variables, comfort related variables and environmental covariates.

[0028] As a preferred solution of the central air-conditioning energy-saving control method based on AI adaptive adjustment described in the present invention, wherein: the transmission to the central air-conditioning actuator is carried out in the following specific steps:

[0029] Decompose the multi-objective optimization control instructions into multiple sub-optimization problems, assign dynamic weights to each sub-problem and define the device parameter search space;

[0030] The Pareto frontier algorithm is used to solve the sub-optimization problems in parallel, screen the non-dominated solution set that meets all constraints, and generate candidate equipment parameter combinations;

[0031] Select the optimal device parameter combination from the candidate device parameter combinations based on the real-time load rate;

[0032] The optimal device parameter combination is encoded into a control instruction compatible with the central air-conditioning actuator protocol and transmitted to the central air-conditioning actuator.

[0033] As a preferred solution of the central air-conditioning energy-saving control method based on AI adaptive adjustment described in the present invention, the specific steps of updating the physical information neural network parameters and adjusting the thermodynamic partial differential equation coefficients are as follows:

[0034] Collect actual energy consumption data, temperature field distribution data, and equipment status data from central air-conditioning actuators and sensors, and eliminate sensor noise through adaptive Kalman filtering;

[0035] Based on the filtered actual energy consumption data and the cooling load demand forecast value, the dynamic energy-saving efficiency index is calculated, and the spatial weighted comfort deviation is calculated through the deviation between the temperature field distribution data and the preset value of the target temperature field distribution;

[0036] The dynamic energy-saving efficiency index and comfort deviation are input into the meta-learning framework, and the parameters of the physical information neural network are updated through cross-task gradient descent, and the coefficients of the thermodynamic partial differential equation are optimized simultaneously.

[0037] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the central air-conditioning energy-saving control method based on AI adaptive adjustment as described in the first aspect of the present invention is implemented.

[0038] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the central air-conditioning energy-saving control method based on AI adaptive adjustment as described in the first aspect of the present invention.

[0039] The beneficial effects of the present invention are as follows: by constructing a multimodal physical information fusion framework and a dynamic closed-loop optimization system, the comprehensive control capability of central air conditioning in complex building environments is improved. Based on real-time acquisition of multi-source data and embedding of thermodynamic constraints, it can accurately predict the thermodynamic state of the building and the cooling load demand, ensuring that the prediction results strictly follow the laws of energy conservation and heat conduction. Through causal reasoning, the interaction effects between key variables are identified, the coupling relationship between energy consumption and comfort is dynamically quantified, and the optimization weights are adaptively allocated to fully tap the energy-saving potential. Combined with the Pareto frontier algorithm, a global optimal equipment parameter combination is generated to improve the collaborative control efficiency of actuators such as compressors and air valves, avoiding resource waste or performance imbalance caused by local optimization. Through the closed-loop update mechanism driven by meta-learning, it is possible to perceive the deviation between the actual operating data and the predicted results in real time, autonomously adjust the neural network parameters and thermodynamic equation coefficients, and continuously adapt to dynamic disturbances such as seasonal changes and equipment aging. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0041] Figure 1 The figure is a flow chart of the central air-conditioning energy-saving control method based on AI adaptive adjustment.

[0042] Figure 2 Flowchart for multimodal data processing.

[0043] Figure 3 Flowchart of dynamic weight allocation and optimization control.

[0044] Figure 4 Flowchart for adaptive optimal device parameter update. DETAILED DESCRIPTION

[0045] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0046] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0047] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0048] Reference Figures 1 to 4 , is an embodiment of the present invention, which provides a central air-conditioning energy-saving control method based on AI adaptive adjustment, comprising the following steps:

[0049] S1. Collect environmental data, equipment operation data, and energy consumption data of central air conditioners in real time, pre-process the multimodal data, construct physical constraint equations based on the laws of thermodynamics, and generate a multimodal data set.

[0050] Furthermore, the multimodal data is timestamp aligned and the sampling frequency normalized, and the missing data is supplemented using a generative adversarial network to output the preprocessed multimodal data.

[0051] Specifically, through a unified reference clock, the timestamps of the central air conditioning system's environmental data, equipment operation data, and energy consumption data are converted to standard UTC format. Then, linear interpolation is used to synchronize data with different sampling frequencies to the same time interval. For example, if the sampling frequency of environmental data is 1Hz, that of equipment operation data is 0.5Hz, and that of energy consumption data is 0.2Hz, all data is interpolated to 1Hz. Sampling frequency normalization is achieved through resampling techniques. High-frequency data is downsampled using mean pooling, and low-frequency data is upsampled using cubic spline interpolation, ensuring that all modal data have the same temporal resolution.

[0052] It should be noted that the generative adversarial network consists of a generator and a discriminator. The generator receives multimodal data with missing values ​​and outputs completed data; the discriminator distinguishes between completed data and real complete data; the training data of the generative adversarial network must include artificially simulated random missing patterns to cover actual scenarios.

[0053] Preprocessed multimodal data is stored as a structured array in time series, containing three fields: environmental data, equipment operation data, and energy consumption data. The data dimensions within each field are consistent, and timestamps are strictly aligned. Data is saved in HDF5 or NPZ format. For example, in the output file, the temperature, humidity, and wind speed of environmental data are arranged at a 1Hz frequency. The compressor power and air valve opening of equipment operation data are aligned with the instantaneous power of energy consumption data on the same time axis. Missing values ​​are flagged as 0, and completed values ​​are flagged as 1.

[0054] Construct physical constraint equations based on the equipment characteristics of central air conditioners and the laws of thermodynamics;

[0055] It should be noted that equipment characteristics refer to the inherent physical parameters and operating rules of various components of central air conditioners, such as compressors, air valves, and heat exchangers, the power-cooling capacity curve of the compressor, and the nonlinear relationship between the air valve opening and air volume.

[0056] The laws of thermodynamics refer to the principles of conservation of energy and heat transfer used when constructing physical constraint equations. For example, the input electrical energy is equal to the sum of the cooling capacity and heat loss, and the heat flux density is proportional to the temperature gradient.

[0057] Specifically, the construction of physical constraint equations is based on the actual operating characteristics of central air conditioning and classical thermodynamic principles. Control conditions are formed by analyzing the correlation between preprocessed environmental data, equipment operation data, and energy consumption data. The temperature field distribution in the environmental data establishes an energy transfer relationship with parameters such as compressor power and damper opening in the equipment operation data. Energy consumption data is used to verify energy conservation. The first law of thermodynamics ensures total energy balance, and the principle of heat conduction reflects the dynamic changes in the temperature field. The heat transfer efficiency of each zone is dynamically adjusted based on the real-time wind speed and damper status, and temperature gradient changes are constrained and optimized using historical operating data.

[0058] Construct the physical constraint equation, the expression is:

[0059]

[0060] Where T represents the temperature field, t represents the current time, Represents the partial derivative of the temperature field T with respect to the current time t, It is represented by the Nabla operator, κ represents the dynamic thermal conductivity, x represents the building space position, κ(t,x) represents the dynamic thermal conductivity of the building space position x and the current time t, It is represented as the gradient vector of the temperature field, i is the discretization region index, and N is the total number of discretization regions. Indicates the accumulation of discretized N regions, h i (t) represents the dynamic heat transfer coefficient of the i-th discretized region, A i represents the heat transfer area of ​​the i-th discretization region, T′ represents the external ambient temperature, and β represents the gradient suppression coefficient.

[0061] It should be noted that the dynamic heat transfer coefficient is calibrated by the convection heat transfer theory and the equipment operation data to achieve real-time mapping of the heat transfer efficiency. For example, when the natural convection is 1 to 10 W / (m2·K), the opening is 0%-50%, h i (t)∈[0,20], air valve closed or low wind speed, forced convection: 10~100W / (m2·K), opening degree is 50%-100%h i (t)∈[20,40], the air valve is fully open; the gradient suppression coefficient is determined by historical data optimization and numerical stability analysis, for example, the optimal β for summer conditions is 1.2, and the optimal β for winter conditions is 0.8.

[0062] The preprocessed multimodal data are fused with the physical constraint equations to generate a multimodal dataset containing physical laws.

[0063] Specifically, preprocessed multimodal data is integrated with physical constraint equations through a data-physics coupling method. The temperature field distribution from the environmental data and parameters such as compressor power and damper opening from the equipment operation data are substituted into the physical constraint equations to verify the relationship between energy conservation and heat conduction. Energy consumption data is used to calculate residuals and correct equation coefficients. The resulting multimodal dataset contains both the original observations and the physical equation verification results at each time step. For example, when the damper opening is 80% and the compressor power is 15kW, if the residual between the calculated temperature field gradient of 0.5°C / m and the measured value of 0.48°C / m is less than 5%, the data is marked as physically consistent.

[0064] It should be noted that raw observations refer to original data records that are directly collected from sensors or devices and preprocessed but not corrected by physical equations, including environmental data observations, equipment operation data observations, and energy consumption data observations.

[0065] The physical constraint equations include the heat conduction equation and the energy conservation equation.

[0066] It should be noted that the thermal diffusivity in the heat conduction equation is decomposed into a composite parameter of thermal conductivity, density, and specific heat capacity. Thermal conductivity is dynamically adjusted based on compressor efficiency and ambient humidity. The thermal resistance in the energy conservation equation is determined by the damper opening and wind speed, while the specific heat capacity depends on the heat exchanger material properties. The heat transfer coefficient has a nonlinear relationship with the damper opening and wind speed.

[0067] S2. Input the multimodal dataset into the physical information neural network. By embedding a deep learning model constrained by thermodynamic partial differential equations, the thermodynamic state and cooling load demand of the building in the future period are predicted and the prediction results are output.

[0068] Furthermore, the multimodal dataset and parameterized physical constraint equations are input into the spatiotemporal encoder and physical encoder of the physical information neural network to extract spatiotemporal features and generate implicit physical features.

[0069] Specifically, the spatiotemporal encoder processes the environmental data, equipment operation data and energy consumption data in the multimodal data through a three-dimensional convolutional network, and extracts spatiotemporal features with time correlation and spatial correlation; the physical encoder inputs the parameterized physical constraint equation parameters such as thermal resistance, heat capacity and heat transfer coefficient into the fully connected network to generate implicit physical features that characterize the laws of thermodynamics.

[0070] The spatiotemporal features and implicit physical features are input into the fusion layer of the graph neural network. The integral residual projection of the thermodynamic partial differential equation in the spatiotemporal domain is performed through a dynamic adaptive weight function to generate physically consistent spatiotemporal joint features.

[0071] Specifically, the fusion layer of the graph neural network receives the spatiotemporal features extracted by the spatiotemporal encoder and the implicit physical features generated by the physical encoder. The dynamic adaptive weight function assigns weight coefficients according to the spatiotemporal features and implicit physical features, and projects the weighted features with the integral residual term of the thermodynamic partial differential equation. The spatiotemporal features and physical features interact through the attention mechanism to generate physically consistent spatiotemporal joint features.

[0072] Dynamic adaptive weight function, the expression is:

[0073]

[0074] Where λ(t) represents the dynamic weight, exp represents the exponential function, η represents the decay rate coefficient, t0 represents the starting time, and Ψ represents the deviation between the temperature field and the external ambient temperature. Expressed as the gradient modulus of the bias field, It represents the integration from the starting time t0 to the current time t, and dt represents the time integral variable.

[0075] It should be noted that the decay rate coefficient is dynamically adjusted through offline optimization and online adaptation by fusing historical temperature gradient data with multimodal data of control instructions, for example, in a shopping mall (with sudden changes in passenger flow): η∈[0.1,5].

[0076] The integral residual projection function is expressed as:

[0077]

[0078] Where, It is expressed as physical constraint residual, j represents sensor index, S represents the total number of sensors, It represents the accumulation of the total number of sensors, t′ represents the future time (for example, the next 5 seconds), Expressed as the deviation of the temperature field from the external ambient temperature For the current time The partial derivative of , α represents the thermal diffusion coefficient, is represented by the Laplace operator of the deviation field, γ represents the KL divergence regularization coefficient, KL(p||q) represents the difference between the probability distribution p of the temperature field output by the predicted physical information neural network and the statistical distribution q of the temperature field measured by the sensor, p represents the probability distribution of the temperature field output by the predicted physical information neural network, and q represents the statistical distribution of the temperature field measured by the sensor.

[0079] It should be noted that the thermal diffusion coefficient is obtained through material parameters. For example, when the material is concrete, α = 1.0, and the KL divergence regularization coefficient is calibrated by back propagation of the error between the predicted distribution and the actual distribution. For example, in an office building (stable working condition): γ = 0.5.

[0080] Based on the spatiotemporal joint characteristics, a multi-scale prediction of the building's overall cooling load demand and the spatial distribution of the building's internal thermodynamic state are performed. The building's overall cooling load demand prediction is output through a fully connected layer, and the spatial distribution of the building's internal thermodynamic state is output through a U-Net structure.

[0081] Specifically, the physically consistent spatiotemporal joint features are input into a multi-scale prediction structure, which adopts a three-layer fully connected network structure. The fully connected layer maps the global features into the predicted value of the overall cooling load demand of the building. The spatiotemporal joint features are input into the U-Net structure, which gradually restores the spatial resolution through the encoder-decoder architecture and jump connections, and outputs the spatial distribution of thermodynamic states such as the temperature field and humidity field inside the building.

[0082] The cooling load demand forecast value and the spatial distribution of the thermodynamic state are input into the normalized mapping layer to generate the forecast results.

[0083] Specifically, the cooling load demand forecast value is scaled to the [0, 1] interval through Min-Max normalization, the thermodynamic state distribution is normalized using Z-score, and the normalized values ​​are linearly combined to generate the final forecast result.

[0084] S3. Construct a causal graph model based on the prediction results, identify the key variables that affect energy consumption and comfort, quantify the causal effects of key variables, and dynamically allocate the optimization weights of energy saving and comfort.

[0085] Furthermore, causal candidate variables are extracted from the prediction results, and a causal graph model is constructed using the improved PC algorithm;

[0086] Specifically, the energy consumption-related variables, comfort-related variables, and environmental covariates in the prediction results are input into the improved PC algorithm as candidate causal variables. A causal graph model is constructed through conditional independence tests and causal direction determination. The improved PC algorithm uses a conditional independence test based on information entropy to screen for intervariable dependencies and uses the time priority criterion to determine causal direction.

[0087] Based on the causal graph model, the key variables that affect energy consumption and comfort are identified. A dual machine learning method is used to calculate the joint causal effect coefficients of the key variables on energy consumption and comfort. The KL divergence function is used to quantify the distribution difference between historical data and real-time data.

[0088] Specifically, a causal graph model was used to identify key variables influencing energy consumption and comfort. Variables with direct causal edges between the input and output variables and the target variable were screened, including compressor power, damper opening, and outdoor temperature. A dual machine learning approach was used to calculate the joint causal effect coefficients of key variables. The first layer of machine learning fitted the conditional expectation of the cooling load forecast and the temperature field distribution with respect to the key variables. The second layer of machine learning calculated partial derivatives based on the residuals and sought the expectation, yielding the joint causal effect coefficients of compressor power on energy consumption and comfort. The KL divergence function was used to quantify the differences between the historical and real-time data distributions.

[0089] It should be noted that the target variable is a predefined optimization target, which is obtained by directly collecting or calculating indicators through sensors; variables with direct causal edges refer to variables that are directly connected to the target variable through directed edges in the causal graph model, including equipment operation data and environmental data. The identified key variables are high-influence variables screened from the direct causal variables, such as those with in-degree greater than 2.

[0090] The causal effect coefficient of key variables is calculated using dual machine learning, and the expression is:

[0091]

[0092] V∈{P,f,T out};

[0093] Where, represents the joint causal effect coefficient of key variables on energy consumption and comfort, Q represents the predicted cooling load, V represents the key variable set, Expressed as the mean of global comfort deviation, represents the partial derivative of the predicted cooling load Q with respect to the key variable set V, Expressed as the mean of global comfort deviation The partial derivative of the key variable set V, P represents the compressor power, f represents the frequency, T out Expressed as the external temperature, Expressed as the mathematical expectation operator.

[0094] According to the joint causal effect coefficient and distribution difference, a dynamic weight function driven by information entropy is constructed to dynamically allocate the optimization weights of energy saving and comfort.

[0095] Specifically, based on the quantitative results of the joint causal effect coefficient and KL divergence, the joint causal effect coefficient measures the causal effect strength of key variables on energy consumption and comfort. For example, for every 1kW increase in compressor power, energy consumption increases by 5% and comfort decreases by 0.3%. The KL divergence quantifies the difference in the distribution between historical and real-time data. For example, the KL divergence between real-time data and historical benchmarks under summer conditions is 0.8. A dynamic weighting function is constructed in conjunction with the variable information entropy. The dynamic weighting function uses the Softmax function to normalize the product of the joint causal effect coefficient and the information entropy to generate dynamic optimization weights for energy conservation and comfort targets.

[0096] Construct a dynamic weight function driven by information entropy, the expression is:

[0097]

[0098] In the formula, λ′(t) represents the dynamic optimization weight of the key variable set V, H represents the information entropy of the key variable set V, k represents the key variable index, K represents the total number of key variables, Indicates the accumulation of the total number of key variables K, Expressed as the joint causal effect coefficient of the kth key variable on energy consumption and comfort, V k Denoted as the kth key variable.

[0099] Key variables refer to the variables identified through the causal graph model that have a causal impact on central air conditioning energy consumption and comfort.

[0100] It should be noted that the causal graph model uses an improved PC algorithm to screen key variables with causal relationships with the target variable. Causal effects are verified through conditional independence tests and causal direction determination. For example, the causal effect coefficient between compressor power and cooling load is 0.6, and the causal effect coefficient between air valve opening and temperature field gradient is 0.3. The strength of the causal edge is quantified using partial correlation coefficients or mutual information, and only variables with an effect strength exceeding 0.2 are retained as key variables.

[0101] Key variables include energy consumption related variables, comfort related variables and environmental covariates.

[0102] It should be noted that energy consumption-related variables include equipment operating parameters and energy efficiency indicators; comfort-related variables include thermodynamic state and uniformity indicators; environmental covariates include external environment and building characteristics;

[0103] Equipment operating parameters include compressor power, refrigerant flow, fan frequency and water pump power;

[0104] Energy efficiency indicators include instantaneous energy efficiency ratio, cooling capacity and condenser heat load;

[0105] Thermodynamic state includes the spatial distribution of temperature field, the spatial distribution of humidity field, temperature gradient and local wind speed;

[0106] Uniformity indicators include the standard deviation of temperature difference in each area, PMV index and PPD;

[0107] The external environment includes outdoor temperature, outdoor humidity, solar radiation intensity and wind speed;

[0108] Building characteristics include envelope thermal resistance, window area ratio, and occupancy density.

[0109] S4. Generate multi-objective optimization control instructions based on the optimization weights, solve the optimal equipment parameter combination through the Pareto frontier algorithm, and transmit it to the central air-conditioning actuator.

[0110] Furthermore, the multi-objective optimization control instructions are decomposed into multiple sub-optimization problems, each of which is assigned a dynamic weight and defines the device parameter search space;

[0111] Specifically, the multi-objective optimization control instructions are decomposed into energy-saving optimization sub-problems and comfort optimization sub-problems.

[0112] It should be noted that the energy-saving optimization sub-problem takes the cooling load demand forecast value as the objective function, and the equipment parameter search space is based on the statistical range of historical operating data to limit the operating range of compressor power, air valve opening, etc.

[0113] The comfort optimization sub-problem takes the temperature field gradient as the objective function, and the equipment parameter search space limits the adjustment range of fan frequency, refrigerant flow, etc.

[0114] A dynamic weighting function, driven by information entropy, assigns energy-saving and comfort-related weights. The overall optimization objective is the sum of the weighted cooling load error and temperature gradient error. The upper and lower bounds of the equipment parameter search space are determined using quantile statistics of historical operating data to eliminate interference from extreme operating conditions and ensure that optimization instructions are generated within the feasible domain.

[0115] The Pareto frontier algorithm is used to solve the sub-optimization problems in parallel, screen the non-dominated solution set that meets all constraints, and generate candidate equipment parameter combinations;

[0116] Specifically, the Pareto Frontier algorithm solves the two subproblems in parallel and generates a non-dominated solution set using the NSGA-II algorithm. The non-dominated solution set is selected based on the following criteria: it is not dominated by other solutions in terms of both energy conservation and comfort objectives and satisfies the device parameter search space constraints. Candidate device parameter combinations are extracted from the non-dominated solution set, and representative parameter combinations are selected using a congestion ranking and uniform sampling strategy.

[0117] Select the optimal device parameter combination from the candidate device parameter combinations based on the real-time load rate;

[0118] Specifically, the real-time load rate is determined by the ratio of the current cooling load demand forecast to the equipment's rated capacity. Candidate equipment parameter combinations are matched to the energy-saving and comfort-optimization weights assigned by the dynamic weighting function based on the range of the real-time load rate. A weighted summation method is used to comprehensively score and rank the achievement of each candidate combination's energy-saving and comfort-target goals. The scoring process simultaneously verifies whether the candidate combination meets the temperature field gradient residual requirements of the physical constraint equation and the equipment parameter search space restrictions. The parameter combination with the highest score that meets all constraints is selected as the optimal solution. The output ensures that the selected equipment parameter combination achieves a balance between minimizing energy consumption and minimizing comfort deviation under real-time load rate conditions.

[0119] The optimal device parameter combination is encoded into a control instruction compatible with the central air-conditioning actuator protocol and transmitted to the central air-conditioning actuator.

[0120] Specifically, parameters such as compressor power, damper opening, and fan frequency in the optimal equipment parameter combination are converted into binary instruction codes according to the register address and data format defined by the central air conditioner actuator protocol. The instruction codes are then encapsulated into communication protocol frames after appending a checksum. The encoded control instructions are transmitted to the central air conditioner actuator via the industrial communication protocol, using standard communication parameters. Before execution, the instructions are verified for compatibility with the physical constraint equations. If the verification fails, a rollback mechanism is triggered, and a suboptimal parameter combination is selected. The issued control instructions strictly match the function code and address mapping table defined by the actuator protocol, ensuring that the actuator accurately interprets and executes the operation.

[0121] S5. Collect the actual data after the instruction is executed, calculate the energy-saving efficiency and comfort deviation, update the physical information neural network parameters and adjust the thermodynamic partial differential equation coefficients through the meta-learning framework.

[0122] Furthermore, actual energy consumption data, temperature field distribution data, and equipment status data are collected from central air-conditioning actuators and sensors, and sensor noise is eliminated through adaptive Kalman filtering;

[0123] Specifically, data is transmitted to the data processing unit via Modbus RTU or MQTT protocols. Actual energy consumption data includes instantaneous and cumulative compressor power values. Temperature distribution data covers temperature, humidity, and wind speed in each area. Equipment status data records damper opening, fan frequency, and refrigerant flow. An adaptive Kalman filter dynamically adjusts the process noise covariance matrix and the observation noise covariance matrix based on the statistical characteristics of sensor noise to smooth out abnormal fluctuations in the temperature distribution data.

[0124] Based on the filtered actual energy consumption data and the cooling load demand forecast value, the dynamic energy-saving efficiency index is calculated, and the spatial weighted comfort deviation is calculated through the deviation between the temperature field distribution data and the preset value of the target temperature field distribution;

[0125] It should be noted that the dynamic energy-saving efficiency index is calculated by the integral ratio of the filtered actual energy consumption data and the cooling load demand forecast value, and the spatial weighted comfort deviation is calculated by taking the square root of the weighted sum of the squares of the deviations between the temperature of each area in the temperature field distribution data and the preset target temperature. The weight coefficient is allocated according to the area and personnel density of the area.

[0126] Specifically, the dynamic energy-saving efficiency index is calculated as follows:

[0127]

[0128] Where ξ(t) represents the dynamic energy-saving efficiency index, P′ represents the actual compressor power, and σ represents the normalization coefficient. Expressed as the L2 norm square of the temperature field gradient after execution.

[0129] It should be noted that the normalization coefficient is obtained based on the statistical distribution calibration of the historical temperature field gradient, σ∈[0.1,1.0].

[0130] Calculate the spatial weighted comfort deviation, the expression is:

[0131]

[0132] Where D(t) represents the spatial weighted comfort deviation, s represents the region index, and S represents the total number of regions. Indicates the accumulation of the total number of S regions, w s Expressed as the weight of the sth region, It is represented as the temperature field gradient after execution in the sth region, and Ψ′ is represented as the deviation of the preset value of the target temperature field distribution.

[0133] The dynamic energy-saving efficiency index and comfort deviation are input into the meta-learning framework, and the parameters of the physical information neural network are updated through cross-task gradient descent, and the coefficients of the thermodynamic partial differential equation are optimized simultaneously.

[0134] Specifically, the meta-learning framework jointly updates the weights of the physical information neural network and the dynamic thermal conductivity and heat transfer coefficients in the thermodynamic partial differential equations through a cross-task optimization mechanism. A dynamic energy-saving efficiency index serves as the energy consumption optimization target, while a spatially weighted comfort deviation serves as the comfort optimization target. The two are combined using weights assigned by a dynamic weight function to form a multi-task joint loss function. During the cross-task gradient descent process, the spatiotemporal encoder parameters of the physical information neural network and the coefficients of the thermodynamic partial differential equations are adjusted synchronously. The updated thermal conductivity and heat transfer coefficients are calculated based on the temperature field residual and energy conservation constraints, ensuring that the equation parameters dynamically adapt to real-time operating conditions.

[0135] This embodiment also provides a computer device, which is suitable for the central air conditioning energy-saving control method based on AI adaptive adjustment, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the central air conditioning energy-saving control method based on AI adaptive adjustment proposed in the above embodiment.

[0136] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0137] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the central air-conditioning energy-saving control method based on AI adaptive adjustment proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, disk or optical disk.

[0138] In summary, the present invention improves the comprehensive control capability of central air conditioning in complex building environments by: constructing a multimodal physical information fusion framework and a dynamic closed-loop optimization system. Based on real-time acquisition of multi-source data and embedding of thermodynamic constraints, it can accurately predict the thermodynamic state of the building and the cooling load demand, ensuring that the prediction results strictly follow the laws of energy conservation and heat conduction. Through causal reasoning, the interaction effects between key variables are identified, the coupling relationship between energy consumption and comfort is dynamically quantified, and the optimization weights are adaptively allocated to fully tap the energy-saving potential. Combined with the Pareto frontier algorithm, a global optimal equipment parameter combination is generated to improve the collaborative control efficiency of actuators such as compressors and air valves, avoiding resource waste or performance imbalance caused by local optimization. Through the closed-loop update mechanism driven by meta-learning, it is possible to perceive the deviation between the actual operating data and the predicted results in real time, autonomously adjust the neural network parameters and thermodynamic equation coefficients, and continuously adapt to dynamic disturbances such as seasonal changes and equipment aging.

[0139] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A central air conditioning energy-saving control method based on AI adaptive regulation, characterized by: include, Collect environmental data, equipment operation data, and energy consumption data of central air conditioners in real time, pre-process the multimodal data, and construct physical constraint equations based on the laws of thermodynamics to generate a multimodal data set. The multimodal data set is input into a physical information neural network. Through a deep learning model embedded with thermodynamic partial differential equation constraints, the thermodynamic state and cooling load demand of the building in the future period are predicted and the prediction results are output. Based on the prediction results, a causal graph model is constructed to identify key variables that affect energy consumption and comfort, quantify the causal effects of key variables, and dynamically allocate optimization weights for energy saving and comfort; Generate multi-objective optimization control instructions based on optimization weights, solve the optimal equipment parameter combination through the Pareto frontier algorithm, and transmit it to the central air conditioning actuator; The actual data after the instruction execution is collected, the energy-saving efficiency and comfort deviation are calculated, and the physical information neural network parameters are updated and the coefficients of the thermodynamic partial differential equation are adjusted through the meta-learning framework.

2. The central air conditioning energy-saving control method based on AI adaptive regulation according to claim 1, characterized in that: The specific steps for generating a multimodal dataset are as follows: Perform timestamp alignment and sampling frequency normalization on multimodal data, use generative adversarial networks to complete missing data, and output preprocessed multimodal data; Construct physical constraint equations based on the equipment characteristics of central air conditioners and the laws of thermodynamics; The preprocessed multimodal data are fused with the physical constraint equations to generate a multimodal dataset containing physical laws.

3. The central air conditioning energy-saving control method based on AI adaptive regulation according to claim 2, characterized in that: The physical constraint equations include the heat conduction equation and the energy conservation equation.

4. The central air conditioning energy-saving control method based on AI adaptive regulation according to claim 3, characterized in that: The specific steps of outputting the prediction results are as follows: Input the multimodal dataset into the spatiotemporal encoder and physical encoder of the physical information neural network to extract spatiotemporal features and generate implicit physical features; The spatiotemporal features and implicit physical features are input into the fusion layer of the graph neural network. The integral residual projection of the thermodynamic partial differential equation in the spatiotemporal domain is performed through a dynamic adaptive weight function to generate physically consistent spatiotemporal joint features. Based on the spatiotemporal joint characteristics, a multi-scale prediction is performed on the predicted value of the building's overall cooling load demand and the spatial distribution of the building's internal thermodynamic state. The predicted value of the building's overall cooling load demand is output through a fully connected layer, and the spatial distribution of the building's internal thermodynamic state is output through a U-Net structure. The cooling load demand forecast value and the spatial distribution of the thermodynamic state are input into the normalized mapping layer to generate the forecast results.

5. The central air conditioning energy-saving control method based on AI adaptive regulation according to claim 4, characterized in that: The specific steps of dynamically allocating the optimization weights of energy saving and comfort are as follows: Extract causal candidate variables from the prediction results and construct a causal graph model using the improved PC algorithm; Based on the causal graph model, the key variables that affect energy consumption and comfort are identified. A dual machine learning method is used to calculate the joint causal effect coefficients of the key variables on energy consumption and comfort. The KL divergence function is used to quantify the distribution difference between historical data and real-time data. According to the joint causal effect coefficient and distribution difference, a dynamic weight function driven by information entropy is constructed to dynamically allocate the optimization weights of energy saving and comfort.

6. The central air conditioning energy-saving control method based on AI adaptive regulation according to claim 5, characterized in that: The key variables are variables identified by the causal graph model that have a causal impact on central air conditioning energy consumption and comfort; The key variables include energy consumption related variables, comfort related variables and environmental covariates.

7. The central air conditioning energy-saving control method based on AI adaptive regulation according to claim 6, characterized in that: The transmission to the central air conditioning actuator is carried out in the following specific steps: Decompose the multi-objective optimization control instructions into multiple sub-optimization problems, assign dynamic weights to each sub-problem and define the device parameter search space; The Pareto frontier algorithm is used to solve the sub-optimization problems in parallel, screen the non-dominated solution set that meets all constraints, and generate candidate device parameter combinations; Select the optimal device parameter combination from the candidate device parameter combinations based on the real-time load rate; The optimal device parameter combination is encoded into a control instruction compatible with the central air-conditioning actuator protocol and transmitted to the central air-conditioning actuator.

8. The central air conditioning energy-saving control method based on AI adaptive regulation according to claim 7, characterized in that: The specific steps of updating the physical information neural network parameters and adjusting the thermodynamic partial differential equation coefficients are as follows: Collect actual energy consumption data, temperature field distribution data, and equipment status data from central air-conditioning actuators and sensors, and eliminate sensor noise through adaptive Kalman filtering; Based on the filtered actual energy consumption data and the cooling load demand forecast value, the dynamic energy-saving efficiency index is calculated, and the spatial weighted comfort deviation is calculated through the deviation between the temperature field distribution data and the preset value of the target temperature field distribution; The dynamic energy-saving efficiency index and comfort deviation are input into the meta-learning framework, and the parameters of the physical information neural network are updated through cross-task gradient descent, and the coefficients of the thermodynamic partial differential equation are optimized simultaneously.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the central air-conditioning energy-saving control method based on AI adaptive adjustment according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the central air-conditioning energy-saving control method based on AI adaptive adjustment according to any one of claims 1 to 8 are implemented.

Images