A non-working user air conditioner regulation modeling method considering power grid interaction and user comfort

Abstract

The application discloses a non-industrial user air conditioner regulation modeling method considering power grid interaction and user comfort, and relates to the field of air conditioner regulation, and the method comprises the following steps: acquiring state data of an air conditioner and indoor and outdoor conditions by using a monitoring sensor, combining neural network technology to build an air conditioner power prediction model after preprocessing the state data; acquiring air conditioner load demand based on the air conditioner power prediction model, and combining the air conditioner load demand with a power grid demand response signal to design a demand response mechanism; establishing an optimal control strategy model with user comfort as a target based on the demand response mechanism, and performing partitioned solution on the optimal control strategy model by using a simulated annealing algorithm, and determining air conditioner regulation results according to a solution result. Through real-time data analysis and a prediction model, system operation parameters are dynamically adjusted to realize energy efficiency maximization, cost minimization, and to enhance the adaptability and response capability of the system to power grid demand changes.

Description

Technical Field

[0001] This invention relates to the field of air conditioning control, and more specifically, to a modeling method for air conditioning control of non-industrial users that takes into account grid interaction and user comfort. Background Technology

[0002] With the continuous growth of energy demand and the increasing severity of environmental problems, building energy consumption optimization has become an important area of ​​energy conservation and emission reduction. As a major component of building energy consumption, air conditioning systems account for a considerable proportion of the total building energy consumption. Traditional air conditioning control strategies often adopt fixed temperature settings and simple start-stop control, lacking dynamic response to changes in indoor and outdoor environments and user needs, resulting in energy waste and low operating efficiency.

[0003] Meanwhile, in existing technologies, the control of air conditioning systems mainly relies on preset control logic and manual adjustments by the user. These methods often cannot respond in a timely manner to changes in the indoor and outdoor environment and the actual needs of the user. For example, when the outdoor temperature changes, a fixed temperature setting may not be able to guarantee indoor comfort, and it will also cause unnecessary energy waste.

[0004] For example, Chinese patent CN202410993166.1 discloses an operation management method and system for a high-efficiency and energy-saving air conditioning room. This method utilizes the collection of historical and real-time data, but it does not take into account the demand response of the power grid when applied. It cannot alleviate the pressure on the power grid by adjusting the operating mode during peak load periods. Therefore, the existing intelligent control strategy still has some limitations, including insufficient analysis and utilization of big data, incomplete consideration of user behavior and comfort, and an imperfect coordination mechanism with the power grid demand response.

[0005] No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention

[0006] To address the problems in related technologies, this invention proposes a non-industrial user air conditioning control modeling method that considers grid interaction and user comfort, in order to overcome the aforementioned technical problems existing in the current related technologies.

[0007] Therefore, the specific technical solution adopted by the present invention is as follows:

[0008] A non-industrial user air conditioning control modeling method considering grid interaction and user comfort, the non-industrial user air conditioning control modeling method includes:

[0009] The system uses monitoring sensors to acquire status data of the air conditioner and the indoor and outdoor environments, and after preprocessing the status data, it combines neural network technology to build an air conditioner power prediction model.

[0010] The air conditioning load demand is obtained based on the air conditioning power prediction model, and the demand response mechanism is designed by combining the air conditioning load demand with the power grid demand response signal.

[0011] An optimal control strategy model with user comfort as the objective is established based on the demand response mechanism. The simulated annealing algorithm is used to solve the optimal control strategy model in partitions, and the air conditioning control result is determined based on the solution results.

[0012] Preferably, the method involves acquiring air conditioner and indoor / outdoor status data using monitoring sensors, and after preprocessing the status data, constructing an air conditioner power prediction model using neural network technology, including:

[0013] The system uses smart meters to monitor the power consumption of air conditioners during operation and uses temperature and humidity sensors to obtain indoor and outdoor temperature and humidity data at the air conditioner installation location, thus obtaining status data of the air conditioner and the indoor and outdoor environments.

[0014] Data standardization processing technology is used to unify the time format and decimals of the state data, and density clustering analysis technology is used to divide the state data into core points, boundary points and noise points;

[0015] Noise points are removed, core points and boundary points are retained, and the data acquisition cycle is used to detect the data loss of the state data after noise point removal;

[0016] Interpolation is performed on the state data to address data gaps. Preprocessed state data is obtained based on the interpolation results, and an air conditioning power prediction model is constructed using neural network technology.

[0017] Preferably, the state data is interpolated based on the missing data, and the preprocessed state data is obtained based on the interpolation result. This is then combined with neural network technology to construct an air conditioning power prediction model, including:

[0018] Based on the analysis of missing data, the location and quantity of missing data are determined. Then, linear interpolation is used to fill in the missing state data according to the analysis results, resulting in preprocessed state data.

[0019] By using statistical analysis and data mining techniques, the feature set that has the greatest impact on air conditioner power is selected from the state data, and the feature set is combined with the radial basis neural network to construct an air conditioner power prediction model;

[0020] The air conditioner power prediction model is trained using the filtered state data, and the model parameters are adjusted to obtain the trained air conditioner power prediction model.

[0021] Preferably, the demand response mechanism is designed by obtaining the air conditioning load demand based on the air conditioning power prediction model and combining the air conditioning load demand with the power grid demand response signal, including:

[0022] Based on the air conditioning power prediction model, the air conditioning power under temperature and humidity conditions is analyzed, and the air conditioning load demand in the future time period is analyzed based on the air conditioning power and energy efficiency ratio. The load control system is used to obtain the grid interaction demand response.

[0023] By combining air conditioning load demand with grid interaction demand response, the corresponding power change values ​​are generated when the air conditioning set temperature changes, the fan speed changes, and the operating mode changes. Based on the power change values, a demand response mechanism is designed.

[0024] Preferably, an optimal control strategy model with user comfort as the objective is established based on a demand response mechanism, and the optimal control strategy model is solved by partitioning using a simulated annealing algorithm. The air conditioning control results are determined based on the solution results, including:

[0025] Based on user subjective thermal sensation analysis, the degree of user comfort satisfaction with the air conditioner's working state under the demand response mechanism is analyzed, and an objective function is constructed based on user comfort satisfaction and demand response benefit coefficient.

[0026] Machine learning techniques are used to select a model framework, and demand response constraints and objective functions are defined to construct an optimal control strategy model with user comfort as the goal.

[0027] Simulated annealing algorithm is used as an optimization technique to find new solutions to the objective function. Monte Carlo method is combined to calculate the acceptance probability of the new solution. Based on the calculation results, the optimal solution is found to determine the air conditioning control result.

[0028] Preferably, based on user subjective thermal perception analysis, the objective function is constructed according to the user's comfort satisfaction with the air conditioner's operating status under the demand response mechanism, and the objective function is based on the user's comfort satisfaction and the demand response benefit coefficient.

[0029] The analysis of indoor relative temperature and air temperature under the demand response mechanism, the analysis of air flow speed based on air temperature, and the calculation of user subjective thermal sensation by combining human metabolic rate and human mechanical work.

[0030] Based on the user's subjective thermal sensation and satisfaction analysis formula, the user's comfort and satisfaction with the air conditioner's working status under the demand response mechanism is obtained.

[0031] Based on the analysis of demand response time and actual average response load, the demand response benefit coefficient under different types is analyzed, and an objective function is constructed based on the demand response benefit coefficient and user comfort and satisfaction.

[0032] The preferred formula for calculating user subjective heat is:

[0033] ;

[0034] In the formula, PMV represents the user's subjective thermal sensation, e represents the natural constant, M represents the human metabolic rate, W represents the human mechanical work, and P... a t represents the water vapor pressure of the air surrounding the human body. a f represents air temperature. cl t represents the ratio of the actual surface area of ​​a person when clothed to the surface area of ​​the naked body. cl The temperature of the outer surface of clothing is represented by t. r Indicates radiation temperature, h c This represents the convective heat transfer coefficient.

[0035] Preferably, the demand response benefit coefficients under different types are analyzed based on demand response time and actual average response load, and an objective function is constructed based on the demand response benefit coefficients and user comfort and satisfaction, including:

[0036] Based on the demand response mechanism, the demand response time and actual average response load are analyzed. The maximum allowable unit price of adjustable load is obtained based on the actual average response load, and the demand response revenue coefficient under different types is calculated.

[0037] The thermal comfort weighting coefficient is calculated based on the number of people that can be accommodated in the air-conditioned application area and the actual number of people. The objective function is constructed by combining the demand response benefit coefficient and the user comfort satisfaction. The demand response benefit coefficients under different types include peak shaving and valley filling demand response benefit coefficients and real-time demand response benefit coefficients.

[0038] The preferred formula for calculating the peak-shaving and valley-filling demand response benefit coefficient is as follows:

[0039] ;

[0040] The formula for calculating the real-time demand response benefit factor is:

[0041] ;

[0042] In the formula, This represents the demand response benefit coefficient for peak shaving and valley filling. R represents the real-time demand response benefit coefficient. final T represents the actual average response load. dr F represents the demand response time. dr K represents the maximum permissible unit price for adjustable load. check R represents the calibration coefficient. check F represents the verification capacity. r This indicates a capacity subsidy.

[0043] Preferably, simulated annealing is used as an optimization technique to find new solutions to the objective function, and Monte Carlo methods are combined to calculate the acceptance probability of the new solutions. Based on the calculation results, the optimal solution is found to determine the air conditioning control results, including:

[0044] The objective function is defined to include both energy consumption and comfort as optimization objectives, and the weights of the optimization objectives are adjusted according to the number of people and environmental conditions in the air-conditioned application area.

[0045] After changing the initial parameters of the simulated annealing algorithm, the initial parameters of the air conditioner are set to obtain the initial solution. At each temperature, the initial solution is perturbed by the simulated annealing algorithm, and the temperature setting and fan speed of the air conditioner are changed to obtain new candidate solutions.

[0046] Calculate the difference between the objective function value of the new candidate solution and the objective function value of the initial solution, and evaluate whether the new candidate solution is better than the initial solution based on the difference result;

[0047] Based on the evaluation results and Monte Carlo calculations, the acceptance probability of new candidate solutions is obtained, and the optimal solution is found according to the acceptance probability. The optimal solution is then used to determine the number and mode of air conditioning operation in different areas to obtain the air conditioning control results.

[0048] The beneficial effects of this invention are as follows:

[0049] 1. This invention uses real-time data analysis and prediction models to dynamically adjust system operating parameters to maximize energy efficiency, minimize costs, and enhance the system's adaptability and responsiveness to changes in grid demand, thereby promoting the intelligent and green development of building energy systems.

[0050] 2. This invention collects real-time operating data of the air conditioning system through sensors and smart instruments, and analyzes and processes the data to train an air conditioning power prediction model to predict future energy demand. Based on the prediction results and the demand response signal of the power grid, an adaptive demand response mechanism is formulated to dynamically adjust the operating mode of the air conditioning system. Furthermore, an optimized control algorithm is used to optimize system parameters while meeting comfort and economic benefits, ensuring that the air conditioning system can adapt to environmental changes and user needs, and achieve long-term efficient and intelligent operation. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart of a non-industrial user air conditioning control modeling method that considers grid interaction and user comfort according to an embodiment of the present invention. Detailed Implementation

[0053] To further illustrate the various embodiments, the present invention provides accompanying drawings, which are part of the disclosure of the present invention. These drawings are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these drawings, those skilled in the art should be able to understand other possible implementation methods and the advantages of the present invention.

[0054] According to an embodiment of the present invention, a non-industrial user air conditioning control modeling method that considers grid interaction and user comfort is provided.

[0055] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figure 1 As shown, according to an embodiment of the present invention, a non-industrial user air conditioning control modeling method considering grid interaction and user comfort includes:

[0056] Step S1: Use monitoring sensors to acquire the status data of the air conditioner and the indoor and outdoor environments, and after preprocessing the status data, construct an air conditioner power prediction model by combining neural network technology.

[0057] Step S2: Obtain the air conditioning load demand based on the air conditioning power prediction model, and combine the air conditioning load demand with the power grid demand response signal to design a demand response mechanism;

[0058] Step S3: Based on the demand response mechanism, establish an optimal control strategy model with user comfort as the objective, and use the simulated annealing algorithm to solve the optimal control strategy model in partitions, and determine the air conditioning control result based on the solution result.

[0059] In this embodiment, when using monitoring sensors to acquire the status data of the air conditioner and the indoor and outdoor environments, and after preprocessing the status data, a power prediction model for the air conditioner is constructed using neural network technology. This involves using smart meters to monitor the air conditioner's power during operation, and using temperature and humidity sensors to acquire the indoor and outdoor temperatures and humidity at the air conditioner's installation location, thus obtaining the status data of the air conditioner and the indoor and outdoor environments. Data standardization processing technology is used to unify the time format and decimal values ​​of the status data. Density clustering analysis is used to divide the status data into core points, boundary points, and noise points. Noise points are removed, while core points and boundary points are retained. The data acquisition cycle is used to detect data loss in the status data after noise point removal. Interpolation processing is performed on the status data based on the data loss. Preprocessed status data is obtained based on the interpolation results, and a power prediction model for the air conditioner is constructed using neural network technology.

[0060] It needs to be explained that in the process of dividing state data into core points, boundary points, and noise points based on density clustering analysis, the local density of the state data can be calculated to obtain density peak points as cluster centers, thereby achieving rapid density-based partitioning of the state data and obtaining state data of different densities. Clustering is then performed on this basis to achieve the partitioning of core points, boundary points, and noise points. Specifically, the implementation steps are as follows:

[0061] Step 1: Input state data, calculate the distance matrix between any state data points, and calculate the local density of each state data point based on the distance matrix and the cutoff distance.

[0062] ;

[0063] In the formula, v i Let d represent the local density of the i-th state data point, j represent the j-th state data point, and d represent the local density of the i-th state data point. ij d represents the Euclidean distance between the i-th state data point and the j-th state data point. c This represents the cutoff distance, and since the impact of the cutoff distance is relatively small, the value of this cutoff distance is the average of the adjacent data points within the state data.

[0064] Step 2: Calculate the minimum distance and decision value of the state data points based on their local density, and determine whether the difference between the decision values ​​is greater than their mean. If so, mark the corresponding state data point as the cluster center. The formula for calculating the minimum distance of the i-th state data point is:

[0065] ;

[0066] There exists an extreme case where the local density of state data i reaches its maximum, then β at that point... i The value is defined as:

[0067] ;

[0068] In the formula, D represents the state dataset, v j This represents the local density of the j-th state data point.

[0069] The decision value α for the i-th state data point i The calculation formula is:

[0070] ;

[0071] Step 3: Sort the remaining state data points in descending order of their local density, and assign them to the nearest cluster center in turn, generating density state datasets of different densities. Then, determine the parameter E corresponding to the density state dataset. psi (Neighborhood radius) and MinPtsi (minimum points);

[0072] All state data points in the density state dataset are marked as unvisited. Select any unvisited state data point and determine whether the neighborhood of the state data point with the neighborhood radius contains the minimum number of state data points. If so, mark it as a core point and establish a new cluster C. Add all the minimum number of points in the neighborhood to the new cluster C. Otherwise, mark it as a noise point.

[0073] Step 4: For each unvisited state data point in the new cluster C, calculate whether its neighborhood contains at least the minimum number of state data points. If so, add the state data points in the neighborhood that are not assigned to any cluster to cluster F and mark them as boundary points.

[0074] Repeat step four to continue with the unvisited state data points in the new cluster C until no new state data points are added to cluster F. Repeat steps three to four until all state data points are added to a cluster or marked as noise points. Merge different densities and output the clustering results of the density state dataset.

[0075] In this embodiment, when interpolating state data based on data missing conditions, obtaining preprocessed state data based on the interpolation results, and constructing an air conditioner power prediction model using neural network technology, the location and quantity of missing data can be analyzed based on the data missing conditions. Linear interpolation techniques are then used to fill in the missing state data based on the analysis results, resulting in preprocessed state data. Statistical analysis and data mining techniques are used to select the feature set that has the greatest impact on air conditioner power from the state data, and this feature set is combined with a radial basis function neural network to construct an air conditioner power prediction model. The selected state data is then used to train the air conditioner power prediction model, and the model parameters are adjusted to obtain the trained air conditioner power prediction model.

[0076] It should be explained that when analyzing the location and quantity of missing data based on the data missing situation, and using linear interpolation technology to fill in the missing state data according to the analysis results, we can first establish a star-shaped and snowflake-shaped spatial multidimensional database structure to clarify the distribution characteristics of the state data. Then, after parameter initialization, we can achieve consistent data dimension partitioning to improve data quality. Next, we can use basis functions to establish a Lagrange interpolation polynomial, introduce the idea of ​​normalization to ensure that the values ​​fluctuate within a certain range, avoid Runge phenomenon, and generate a new interpolation polynomial. The result of the interpolation polynomial calculation is the state data to fill in the missing positions.

[0077] In the process of establishing a spatial multidimensional database structure, the topological structure of the state data is analyzed first. This helps to understand the distribution characteristics and storage methods of the state data, making interpolation processing more convenient. The topological structure of the state data can be described by a star schema.

[0078] ;

[0079] If the data has a very large number of dimensions and a complex distribution, the star schema described above cannot meet the needs of state data processing and will affect data interpolation analysis. Assuming there are four dimension tables, a snowflake schema database model can be used in this case.

[0080] ;

[0081] In the formula, Let e(a) represent the set of dimension tables for state data, where e(a) represents a dimension tables and n represents the total number of dimension tables.

[0082] Simultaneously, when establishing the Lagrange interpolation polynomial using basis functions, the essence of the Lagrange algorithm is to calculate a polynomial through multiple iterations, which can obtain the interpolation exactly at the calibration point. The specific interpolation process is as follows:

[0083] For example, there are k+1 state data points (X0, Y0)...(X... k Y k If two random state data points are not identical, then the Lagrange interpolation polynomial is:

[0084] ;

[0085] In the formula, S j (X) represents an odd function, Y j This represents the data position corresponding to the j-th state data point, where j∈[0, k], as follows:

[0086] ;

[0087] Based on the Lagrange interpolation polynomial, a weight normalization approach is introduced to map state data points to a fixed interval. This normalization method ensures that the basis function values ​​are within the (0, 1) interval, reducing polynomial instability and guaranteeing calibration accuracy within the ideal range. The new normalized Lagrange interpolation polynomial is then as follows:

[0088] ;

[0089] In the formula, T j (X) represents the new basis function, y j If the normalized state data is represented, then the result of the normalized new Lagrange interpolation polynomial calculation is the state data to fill the missing positions.

[0090] In this embodiment, when obtaining air conditioning load demand based on the air conditioning power prediction model and designing a demand response mechanism by combining the air conditioning load demand with the grid demand response signal, the air conditioning power under temperature and humidity conditions can be analyzed based on the air conditioning power prediction model, and the air conditioning load demand in the future time period can be analyzed based on the air conditioning power and energy efficiency ratio. The grid interaction demand response can be obtained using the load control system. The air conditioning load demand and the grid interaction demand response can be combined to generate the power change value corresponding to the change in air conditioning set temperature, wind speed and working mode, and the demand response mechanism can be designed based on the power change value.

[0091] In this embodiment, an optimal control strategy model with user comfort as the objective is established based on the demand response mechanism. Simulated annealing is used to solve the optimal control strategy model in partitions. When determining the air conditioning control result based on the solution results, the user's subjective thermal sensation analysis can be used to assess user comfort satisfaction with the air conditioning operation under the demand response mechanism. An objective function is constructed based on user comfort satisfaction and the demand response benefit coefficient. Machine learning techniques are used to select a model framework, and demand response constraints are defined and integrated with the objective function to construct an optimal control strategy model with user comfort as the objective. Simulated annealing is used as an optimization technique to find new solutions to the objective function, and Monte Carlo methods are combined to calculate the acceptance probability of the new solution. Based on the calculation results, the optimal solution is found to determine the air conditioning control result.

[0092] In this embodiment, when analyzing user comfort satisfaction with the air conditioner's operation under the demand response mechanism based on subjective thermal sensation, and constructing an objective function based on user comfort satisfaction and demand response benefit coefficient, the system analyzes indoor relative temperature and air temperature under the demand response mechanism, analyzes airflow velocity based on air temperature, and calculates user subjective thermal sensation by combining human metabolic rate and human mechanical work. The system obtains user comfort satisfaction with the air conditioner's operation under the demand response mechanism based on the user subjective thermal sensation and satisfaction analysis formula. It then analyzes the demand response benefit coefficient under different types based on demand response time and actual average response load, and constructs an objective function based on the demand response benefit coefficient and user comfort satisfaction.

[0093] In this embodiment, when analyzing the demand response benefit coefficients under different types based on demand response time and actual average response load, and constructing an objective function based on the demand response benefit coefficients and user comfort satisfaction, the demand response time and actual average response load can be analyzed based on the demand response mechanism. The maximum allowable unit price of adjustable load can be obtained based on the actual average response load, and the demand response benefit coefficients under different types can be calculated. The thermal comfort weight coefficient can be calculated based on the accommodating number of people in the air-conditioned application area and the actual number of people. The objective function can be constructed by combining the demand response benefit coefficients and user comfort satisfaction. The demand response benefit coefficients under different types include peak shaving and valley filling demand response benefit coefficients and real-time demand response benefit coefficients.

[0094] In this embodiment, when using simulated annealing as an optimization technique to find new solutions to the objective function, and combining it with Monte Carlo calculations to determine the acceptance probability of the new solutions, and then using the calculation results to find the optimal solution and determine the air conditioning control results, the objective function can be defined to include both energy consumption and comfort as optimization objectives. The weights of the optimization objectives can be adjusted according to the number of people and environmental conditions in the air conditioning application area. After changing the initial parameters of the simulated annealing algorithm, the initial parameters of the air conditioning are set to obtain an initial solution. At each temperature, the simulated annealing algorithm is used to perturb the initial solution, changing the air conditioning temperature setting and fan speed to obtain new candidate solutions. The difference between the objective function value of the new candidate solution and the objective function value of the initial solution is calculated, and the difference results are used to evaluate whether the new candidate solution is better than the initial solution. Based on the evaluation results and Monte Carlo calculations, the acceptance probability of the new candidate solution is calculated, and the optimal solution is found based on the acceptance probability. The optimal solution is then used to determine the number and operation mode of air conditioning in different areas to obtain the air conditioning control results.

[0095] To facilitate understanding of the above technical solutions of the present invention, the following provides a detailed description of the operation of the present invention in actual practice.

[0096] Step 1: Air Conditioning Data Acquisition and Preprocessing

[0097] Temperature and humidity sensors and smart meters are installed in the air conditioning system to monitor key parameters such as indoor and outdoor temperature, humidity, and air conditioning power in real time to obtain status data. The operation data of the air conditioning system is collected through sensors to ensure the real-time and accuracy of the data. At the same time, the status data is standardized, including unifying the time format, unifying the units of similar data, and retaining decimals.

[0098] Simultaneously, clustering is performed based on density to divide data points into core points, boundary points, and noise points. Noise points are usually considered outliers, and these outliers are removed. Furthermore, detection is performed according to the data collection cycle. After missing values ​​are detected, the corresponding interpolation method is selected according to the data type to fill them in until there are no outliers in the state data.

[0099] Step 2: Building the Air Conditioner Power Prediction Model

[0100] Through statistical analysis and data mining methods, the features that have the greatest impact on air conditioner power are screened from the state data, including indoor and outdoor temperature and humidity. Based on the feature dataset, an air conditioner power prediction model based on radial basis function (RBF) neural network is constructed. The model input is a feature vector, including set temperature, ambient temperature, wind speed, working mode, and usage time. The output is air conditioner power. At the same time, historical data is used to train the model and adjust parameters to optimize model performance. The trained model is used to predict the air conditioner power demand in the future time period based on temperature, humidity and time information.

[0101] The load control system acquires grid demand response and analyzes air conditioning load demand in the future based on power demand and energy efficiency ratio. This yields power change values ​​corresponding to changes in air conditioning set temperature, fan speed, and operating mode. Furthermore, by incorporating human presence sensor information and power change values ​​resulting from different air conditioning control methods, a demand response mechanism is established to complete demand response in stages and zones.

[0102] First, all air conditioners in unoccupied areas are shut down to meet a portion of the demand response power. The remaining required control power is then allocated to air conditioners in other areas. Based on the subsequent optimization strategy, the number of air conditioners requiring control and the control method are calculated. Before or after the demand response, users can define or modify control rules and weights. When executing the next demand response, constraints will be added or modified accordingly to meet user needs.

[0103] Step 3: Utilize machine learning to develop optimization strategies to optimize the operating parameters of the air conditioning system while satisfying both comfort and economic benefits.

[0104] (1) PDD calculation:

[0105] The formula for calculating the user's subjective heat is:

[0106] ;

[0107] The formula for calculating the water vapor pressure of the air surrounding the human body is:

[0108] ;

[0109] The formula for calculating the ratio of the actual surface area of ​​a clothed body to the surface area of ​​a naked body is:

[0110] ;

[0111] The formula for calculating the temperature of the outer surface of clothing is:

[0112] ;

[0113] The expression for the convective heat transfer coefficient is:

[0114] ;

[0115] In the formula, PMV represents the user's subjective thermal sensation, e represents the natural constant, M represents the human metabolic rate, W represents the human mechanical work, and P... a t represents the water vapor pressure of the air surrounding the human body. a f represents air temperature. cl t represents the ratio of the actual surface area of ​​a person when clothed to the surface area of ​​the naked body. cl The temperature of the outer surface of clothing is represented by t. r Indicates radiation temperature, h c I represents the convective heat transfer coefficient. cl φ represents the thermal resistance of the clothing, and φ represents the indoor relative humidity.

[0116] PPD (Predicted Percentage of Dissatisfaction) refers to the predicted percentage of dissatisfaction, which is based on PMV (Prognostics and Values). It represents the percentage of people expected to feel dissatisfied in a specific environment, and the calculation formula is as follows:

[0117] ;

[0118] The relationship between PMV and PPD can be described as a bell curve relationship: the closer the PMV value is to 0 (neither cold nor hot), the lower the PPD value, indicating that most people feel comfortable; while the further the PMV value deviates from 0 (whether positive or negative), the higher the PPD value, indicating that the percentage of people who feel dissatisfied increases.

[0119] (2) Objective function:

[0120] ;

[0121] ;

[0122] In the formula, λ i This represents the thermal comfort weighting coefficient, which is adjusted according to the number of people in each area; n i N represents the actual number of people in the area. i Indicates the maximum number of people the area can accommodate; S dr This refers to the revenue generated from demand response. The revenue generated from different types of demand response (peak shaving and valley filling demand response revenue versus real-time demand response revenue) varies, as follows:

[0123] The formula for calculating the peak-shaving and valley-filling demand response benefit coefficient is as follows:

[0124] ;

[0125] The formula for calculating the real-time demand response benefit factor is:

[0126] ;

[0127] In the formula, This represents the demand response benefit coefficient for peak shaving and valley filling. R represents the real-time demand response benefit coefficient. final T represents the actual average response load. dr F represents the demand response time. dr K represents the maximum permissible unit price for adjustable load. check R represents the calibration coefficient. check F represents the verification capacity. r This indicates a capacity subsidy.

[0128] (3) Demand response constraints:

[0129] Demand response constraints include peak-shaving demand response, valley-filling demand response, and real-time demand response. Peak-shaving demand response is defined as follows:

[0130] ;

[0131] The valley filling demand response is as follows:

[0132] ;

[0133] Real-time demand response is as follows:

[0134] ;

[0135] In the formula, Indicates the maximum load during the response period. Indicates the baseline maximum load. Indicates the baseline average load. R represents the average load during the response period. resp This indicates the response capability confirmation value. Indicates the minimum load during the response period. This indicates the minimum baseline load.

[0136] During demand response, the power grid treats the entire industrial park as a whole, and the baseline average load and the average load during the response period should be based on the grid interaction power. The calculation method is as follows:

[0137] ;

[0138] In the formula, The typical daily load is shown in Table 1 below:

[0139] Table 1 Typical Daily Load Calculation

[0140]

[0141] (4) Solving using the simulated annealing algorithm:

[0142] The simulated annealing algorithm is used to solve the optimal control strategy in different regions. The specific implementation steps are as follows:

[0143] 1. Set the initial parameters of the simulated annealing algorithm, including initial temperature (T), cooling rate, final temperature, etc. In the optimization of air conditioning system parameters, the objective function is a dual-level optimization objective of energy consumption and comfort, and the weights of the two optimization objectives are dynamically adjusted according to the number of people and environmental conditions in different areas.

[0144] 2. The algorithm starts by setting an initial solution, i.e., the initial parameter settings of the air conditioning system. At each temperature, new candidate solutions are generated by perturbing the current solution. This perturbation mainly involves changing parameters such as the air conditioning temperature setting and fan speed. For each newly generated candidate solution, the difference between its objective function value and the objective function value of the current solution is calculated to evaluate whether the new solution is better than the current solution. According to the Metropolis (Monte Carlo) criterion, the probability of accepting the new solution is calculated using the following formula:

[0145] ;

[0146] In the formula, P represents the probability of accepting the new solution, and E new E represents the objective function value of the new solution. old Let T represent the objective function value of the current solution, and let T represent the current temperature. If the objective function value of the new solution is better than that of the current solution, then the new solution is accepted. If the objective function value of the new solution is worse, there is still a certain probability that the new solution will be accepted. This probability gradually decreases as the temperature decreases. If the new solution is accepted, it will become the new current solution; otherwise, the current solution will remain unchanged.

[0147] (3) After each iteration, the temperature is reduced at a predetermined cooling rate. As the temperature decreases, the probability of accepting a poor solution gradually decreases, and the algorithm gradually converges to the optimal solution. When the temperature is reduced to the termination temperature or the preset termination conditions are met, including the number of iterations and the improvement of the objective function value being less than the minimum threshold, the algorithm stops. Finally, the algorithm outputs the optimal solution found in the search process, namely the number of air conditioners and the operating mode in different areas, so as to achieve the goal of the best overall comfort when participating in the grid demand response.

[0148] In summary, by utilizing the above-mentioned technical solutions of this invention, the present invention dynamically adjusts system operating parameters through real-time data analysis and prediction models to maximize energy efficiency, minimize costs, and enhance the system's adaptability and responsiveness to changes in grid demand, thereby promoting the intelligent and green development of building energy systems. This invention collects real-time operating data of the air conditioning system through sensors and smart meters, analyzes and processes the data to train an air conditioning power prediction model to predict future energy demand. Based on the prediction results and grid demand response signals, an adaptive demand response mechanism is formulated to dynamically adjust the operating mode of the air conditioning system. Furthermore, an optimized control algorithm is used to optimize system parameters while meeting comfort and economic benefits, ensuring that the air conditioning system can adapt to environmental changes and user needs, achieving long-term efficient and intelligent operation.

[0149] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A non-industrial user air conditioning control modeling method considering grid interaction and user comfort, characterized in that, The modeling method for air conditioning control for non-industrial users includes: The system uses monitoring sensors to acquire status data of the air conditioner and the indoor and outdoor environments, and after preprocessing the status data, it combines neural network technology to build an air conditioner power prediction model. Based on the air conditioning power prediction model, the air conditioning power under temperature and humidity conditions is analyzed, and the air conditioning load demand in the future time period is analyzed based on the air conditioning power and energy efficiency ratio. The load control system is used to obtain the grid interaction demand response. The power change value is generated by combining the air conditioning load demand with the power grid interaction demand response, and the corresponding power change value is generated when the air conditioning set temperature, fan speed and operating mode change. The demand response mechanism is designed based on the power change value. Based on user subjective thermal sensation analysis, the degree of user comfort satisfaction with the air conditioner's working state under the demand response mechanism is analyzed, and an objective function is constructed based on user comfort satisfaction and demand response benefit coefficient. Machine learning techniques are used to select a model framework, and demand response constraints and objective functions are defined to construct an optimal control strategy model with user comfort as the goal. Simulated annealing algorithm is used as an optimization technique to find new solutions to the objective function. Monte Carlo method is combined to calculate the acceptance probability of the new solution. Based on the calculation results, the optimal solution is found to determine the air conditioning control result.

2. The non-industrial user air conditioning control modeling method considering grid interaction and user comfort according to claim 1, characterized in that, The process of acquiring air conditioner and indoor / outdoor status data using monitoring sensors, and then constructing an air conditioner power prediction model by combining the preprocessed status data with neural network technology includes: The system uses smart meters to monitor the power consumption of air conditioners during operation and uses temperature and humidity sensors to obtain indoor and outdoor temperature and humidity data at the air conditioner installation location, thus obtaining status data of the air conditioner and the indoor and outdoor environments. Data standardization processing technology is used to unify the time format and decimals of the state data, and density clustering analysis technology is used to divide the state data into core points, boundary points and noise points; Noise points are removed, core points and boundary points are retained, and the data acquisition cycle is used to detect the data loss of the state data after noise point removal; Interpolation is performed on the state data to address data gaps. Preprocessed state data is obtained based on the interpolation results, and an air conditioning power prediction model is constructed using neural network technology.

3. The non-industrial user air conditioning control modeling method considering grid interaction and user comfort according to claim 2, characterized in that, The process of interpolating the state data based on missing data, obtaining preprocessed state data based on the interpolation results, and constructing an air conditioner power prediction model using neural network technology includes: Based on the analysis of missing data, the location and quantity of missing data are determined. Then, linear interpolation is used to fill in the missing state data according to the analysis results, resulting in preprocessed state data. By using statistical analysis and data mining techniques, the feature set that has the greatest impact on air conditioner power is selected from the state data, and the feature set is combined with the radial basis neural network to construct an air conditioner power prediction model; The air conditioner power prediction model is trained using the filtered state data, and the model parameters are adjusted to obtain the trained air conditioner power prediction model.

4. The non-industrial user air conditioning control modeling method considering grid interaction and user comfort according to claim 1, characterized in that, The user comfort satisfaction with the air conditioner's operating status under the demand response mechanism, based on subjective thermal sensation analysis, includes the following objective function constructed according to user comfort satisfaction and demand response benefit coefficient: The analysis of indoor relative temperature and air temperature under the demand response mechanism, the analysis of air flow speed based on air temperature, and the calculation of user subjective thermal sensation by combining human metabolic rate and human mechanical work. Based on the user's subjective thermal sensation and satisfaction analysis formula, the user's comfort and satisfaction with the air conditioner's working status under the demand response mechanism is obtained. Based on the analysis of demand response time and actual average response load, the demand response benefit coefficient under different types is analyzed, and an objective function is constructed based on the demand response benefit coefficient and user comfort and satisfaction.

5. The non-industrial user air conditioning control modeling method considering grid interaction and user comfort according to claim 4, characterized in that, The formula for calculating the user's subjective thermal sensation is: ； In the formula, PMV represents the user's subjective thermal sensation, e represents the natural constant, M represents the human metabolic rate, W represents the human mechanical work, and P... a t represents the water vapor pressure of the air surrounding the human body. a f represents air temperature. cl t represents the ratio of the actual surface area of ​​a person when clothed to the surface area of ​​the naked body. cl The temperature of the outer surface of clothing is represented by t. r Indicates radiation temperature, h c This represents the convective heat transfer coefficient.

6. The non-industrial user air conditioning control modeling method considering grid interaction and user comfort according to claim 5, characterized in that, The analysis of demand response benefit coefficients under different types based on demand response time and actual average response load, and the construction of an objective function based on the demand response benefit coefficients and user comfort and satisfaction, include: Based on the demand response mechanism, the demand response time and actual average response load are analyzed. The maximum allowable unit price of adjustable load is obtained based on the actual average response load, and the demand response revenue coefficient under different types is calculated. The thermal comfort weighting coefficient is calculated based on the number of people that can be accommodated in the air-conditioned application area and the actual number of people. The objective function is constructed by combining the demand response benefit coefficient and the user comfort satisfaction. The demand response benefit coefficients under different types include peak shaving and valley filling demand response benefit coefficients and real-time demand response benefit coefficients.

7. The non-industrial user air conditioning control modeling method considering grid interaction and user comfort according to claim 6, characterized in that, The formula for calculating the peak shaving and valley filling demand response benefit coefficient is as follows: ； The formula for calculating the real-time demand response benefit coefficient is as follows: ； In the formula, This represents the demand response benefit coefficient for peak shaving and valley filling. R represents the real-time demand response benefit coefficient. final T represents the actual average response load. dr F represents the demand response time. dr K represents the maximum permissible unit price for adjustable load. check R represents the calibration coefficient. check F represents the verification capacity. r This indicates a capacity subsidy.

8. The non-industrial user air conditioning control modeling method considering grid interaction and user comfort according to claim 7, characterized in that, The process of using simulated annealing as an optimization technique to find new solutions to the objective function, and combining this with Monte Carlo calculations to determine the acceptance probability of the new solutions, and then finding the optimal solution based on the calculation results to determine the air conditioning control outcome includes: The objective function is defined to include both energy consumption and comfort as optimization objectives, and the weights of the optimization objectives are adjusted according to the number of people and environmental conditions in the air-conditioned application area. After changing the initial parameters of the simulated annealing algorithm, the initial parameters of the air conditioner are set to obtain the initial solution. At each temperature, the initial solution is perturbed by the simulated annealing algorithm, and the temperature setting and fan speed of the air conditioner are changed to obtain new candidate solutions. Calculate the difference between the objective function value of the new candidate solution and the objective function value of the initial solution, and evaluate whether the new candidate solution is better than the initial solution based on the difference result; Based on the evaluation results and Monte Carlo calculations, the acceptance probability of new candidate solutions is obtained, and the optimal solution is found according to the acceptance probability. The optimal solution is then used to determine the number and mode of air conditioning operation in different areas to obtain the air conditioning control results.

Images