Door and window temperature control method and system based on artificial intelligence

Abstract

The invention relates to the technical field of door and window temperature control, in particular to a door and window temperature control method and system based on artificial intelligence, and the method comprises the steps: collecting indoor and outdoor temperature, meteorological conditions and personnel activity data through a distributed sensor network, analyzing the temperature distribution characteristics through a convolutional neural network, and predicting the temperature diffusion trend through a long-short-term memory network. And a reinforcement learning algorithm is combined to optimize a door and window control strategy. According to the application, accurate management of indoor temperature distribution can be realized through deep analysis and dynamic adjustment of multi-dimensional data, and the energy utilization efficiency and the living comfort are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of door and window temperature control, and specifically relates to a door and window temperature control method and system based on artificial intelligence. Background Art

[0002] Smart building technology, which uses technology to improve indoor comfort and energy efficiency, is a key area of modern architectural research. Doors and windows, as key nodes in the interface between indoor and outdoor environments, have a significant impact on building energy consumption and the living experience through their temperature control functions. Currently, traditional door and window temperature control systems rely on simple on / off controls or fixed-mode ventilation adjustments, which exhibit limitations in complex and changing indoor and outdoor environments. These methods often lack dynamic temperature perception, making it difficult to adapt to real-time weather conditions or occupant activity. This can lead to uneven temperature distribution, high energy consumption, and insufficient comfort. The core challenge lies in accurately managing temperature diffusion, which manifests itself in the following aspects: First, the real-time collection and analysis of multi-point temperature data is limited, making it difficult to accurately capture dynamic changes in indoor temperature distribution. Second, the lack of an intelligent door and window opening and closing decision-making mechanism makes it difficult to dynamically optimize air flow based on predicted temperature diffusion paths. Finally, the system's limited adaptability to varying weather conditions and occupant distribution makes it difficult to consistently maintain ideal temperature control. To address the above issues, the focus of this research is how to use artificial intelligence technology to integrate multi-point temperature sensing, machine learning prediction, and dynamic door and window control to build a system that can adaptively adjust temperature diffusion according to changes in indoor and outdoor environments. Summary of the Invention

[0003] The purpose of the present invention is to address the above-mentioned deficiencies in the prior art and to provide a door and window temperature control method and system based on artificial intelligence.

[0004] The purpose of the present invention is achieved through the following technical solutions: A door and window temperature control method based on artificial intelligence, comprising the following steps:

[0005] S1, obtains indoor and outdoor multi-point temperature data, meteorological condition data and personnel activity data through a distributed sensor network, and generates a multidimensional dataset with a time stamp;

[0006] S2. Based on the multidimensional data set, a convolutional neural network is used to analyze the temperature distribution characteristics, extract the indoor temperature gradient, outdoor temperature difference and crowded areas, and obtain a spatial map of uneven temperature distribution;

[0007] S3. If there are local high or low temperature areas in the spatial map of uneven temperature distribution, the temperature diffusion trend in the next 30 minutes is predicted through the long short-term memory network to generate a temperature diffusion path;

[0008] S4. Process the temperature diffusion path and meteorological condition data through a reinforcement learning algorithm, calculate the door and window opening angles and ventilation duration, and obtain an energy consumption optimization control strategy;

[0009] S5. Based on the energy consumption optimization control strategy, the dynamic door and window control module is used to adjust the opening and closing status of each door and window to generate real-time air flow field distribution;

[0010] S6. If the deviation between the real-time air flow field distribution and the temperature diffusion path exceeds a preset threshold, the door and window opening and closing angles are re-optimized using a genetic algorithm to obtain an updated control instruction;

[0011] S7. Collect the adjusted indoor temperature data through the multi-point temperature sensing module to generate a new multidimensional data set, and repeat the above steps to determine the continuously optimized temperature diffusion management;

[0012] S8. Based on the external wind speed and humidity data obtained by the meteorological condition analysis module, a support vector machine algorithm is used to predict short-term meteorological changes, generate auxiliary control parameters, and update the energy consumption optimization control strategy;

[0013] S9. The infrared sensor data obtained by the personnel activity perception module is used to analyze the changes in densely populated areas, adjust the weights of the temperature diffusion paths, and obtain an adaptive control solution for dynamic environmental changes.

[0014] The present invention is further configured such that, in step S3, the formula of the long short-term memory network is Among them, T future (x, y, t) represents the predicted temperature at the location (x, y) at a certain moment in the future, T current (x,y) is the current temperature, ΔT i (x, y, t) is the temperature change caused by the i-th influencing factor, w i is the corresponding influence weight.

[0015] The present invention is further configured such that, in step S4, the formula for calculating the door and window opening and closing angles is θ=f RL (T diff ,M weather ); where θ is the door and window opening angle, T diff is the temperature difference between indoor and outdoor, M weather is the meteorological condition data; f RL It is the function mapping relationship of the reinforcement learning model.

[0016] The present invention is further configured such that, in step S4, the formula for calculating the ventilation time is t vent =g RL (T gradient ,P occupancy ); where t ventis the ventilation time, T gradient is the indoor temperature gradient, P occupancy is the population distribution density, g RL It is the function mapping relationship of the reinforcement learning model.

[0017] The present invention is further configured such that, in step S6, the genetic algorithm formula is F(θ)=α·E save +β·C comfort Among them, E save For energy saving effect rating, C comfort is the comfort score, α and β are weight coefficients.

[0018] The present invention is further configured such that, in step S8, the formula of the support vector machine algorithm is: Among them, K(x i ,x) is the kernel function, α i is the Lagrange multiplier, b is the bias term, y i is the training sample label.

[0019] The present invention is further configured to obtain indoor and outdoor multi-point temperature data, meteorological condition data, and personnel activity data through a distributed sensor network to generate a multidimensional data set containing a timestamp, including:

[0020] Synchronize data collected by different types of sensors deployed indoors and outdoors to ensure data time consistency;

[0021] Filter and denoise the raw data collected by the sensor to remove outliers and noise interference;

[0022] The processed data is sorted by timestamp and stored as a structured multidimensional cube.

[0023] The present invention is further configured to analyze temperature distribution characteristics using a convolutional neural network based on the multidimensional dataset, extract indoor temperature gradients, outdoor temperature differences, and crowded areas, and obtain a spatial mapping of uneven temperature distribution, including:

[0024] Inputting the multidimensional data set into the first layer of a convolutional neural network and performing spatial feature extraction on the data using a two-dimensional convolution kernel;

[0025] In the middle layer of the convolutional neural network, the pooling operation is used to reduce the dimension of the feature map and retain the main feature information;

[0026] In the last layer of the convolutional neural network, the output contains feature maps of temperature gradient, outdoor temperature difference, and crowded areas, and generates a spatial map of uneven temperature distribution.

[0027] An artificial intelligence-based door and window temperature control system, comprising:

[0028] The data acquisition module is used to obtain temperature data, meteorological condition data, and personnel activity data at multiple points indoors and outdoors through a distributed sensor network, and generate a multidimensional data set with a time stamp;

[0029] A temperature distribution analysis module is used to analyze temperature distribution characteristics using a convolutional neural network based on the multidimensional dataset, extract indoor temperature gradients, outdoor temperature differences, and areas with high population density, and obtain a spatial mapping of uneven temperature distribution;

[0030] A temperature diffusion prediction module is used to predict the temperature diffusion trend within the next thirty minutes through a long short-term memory network if there are local high temperature or low temperature areas in the spatial map of uneven temperature distribution, and generate a temperature diffusion path;

[0031] A control strategy optimization module is used to process the temperature diffusion path and meteorological condition data through a reinforcement learning algorithm, calculate the door and window opening and closing angles and ventilation duration, and obtain an energy consumption optimization control strategy;

[0032] A dynamic door and window control module is used to adjust the opening and closing status of each door and window according to the energy consumption optimization control strategy and generate real-time air flow field distribution;

[0033] a deviation optimization module, configured to re-optimize the door and window opening and closing angles using a genetic algorithm to obtain updated control instructions if the deviation between the real-time air flow field distribution and the temperature diffusion path exceeds a preset threshold;

[0034] The loop optimization module is used to collect the adjusted indoor temperature data through the multi-point temperature sensing module, generate a new multidimensional data set, and execute the above steps in a loop to determine the continuously optimized temperature diffusion management.

[0035] Beneficial effects of the present invention: The present invention collects indoor and outdoor temperature, weather and personnel activity data through a distributed sensor network, uses a convolutional neural network to analyze temperature distribution characteristics, a long and short-term memory network to predict temperature diffusion trends, and a reinforcement learning algorithm to optimize door and window control strategies. The present invention dynamically adjusts the opening and closing status of doors and windows, generates real-time air flow field distribution, and uses a genetic algorithm to optimize control instructions based on deviations. At the same time, the present invention uses a support vector machine to predict short-term weather changes, analyze changes in densely populated areas, and achieve adaptive control of dynamic environments. Through cyclic optimization, the present invention can sustainably improve indoor temperature distribution, improve energy utilization efficiency, and provide users with a comfortable and energy-saving intelligent temperature management solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The invention is further described with reference to the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the invention. A person skilled in the art can obtain other drawings based on the following drawings without making any creative effort.

[0037] Figure 1 is a flow chart of the present invention; DETAILED DESCRIPTION

[0038] The present invention is further described with reference to the following examples.

[0039] Depend on Figure 1 It can be seen that the embodiment of the present invention provides a door and window temperature control method based on artificial intelligence, comprising the following steps: S1, obtaining indoor and outdoor multi-point temperature data, meteorological condition data and personnel activity data through a distributed sensor network, and generating a multidimensional data set containing a timestamp; S2, using a convolutional neural network to analyze the temperature distribution characteristics based on the multidimensional data set, extracting the indoor temperature gradient, outdoor temperature difference and personnel-dense areas, and obtaining a spatial mapping of uneven temperature distribution; S3, if there are local high temperature or low temperature areas in the spatial mapping of uneven temperature distribution, the temperature diffusion trend within the next thirty minutes is predicted through a long short-term memory network, and a temperature diffusion path is generated; S4, processing the temperature diffusion path and meteorological condition data through a reinforcement learning algorithm, calculating the door and window opening and closing angles and ventilation time, and obtaining an energy consumption optimization control strategy; S5, according to the energy consumption optimization control Strategy, using the dynamic door and window control module to adjust the opening and closing status of each door and window to generate a real-time air flow field distribution; S6, if the deviation between the real-time air flow field distribution and the temperature diffusion path exceeds the preset threshold, the genetic algorithm is used to re-optimize the door and window opening and closing angles to obtain updated control instructions; S7, the adjusted indoor temperature data is collected through the multi-point temperature sensing module to generate a new multidimensional data set, and the above steps are executed in a loop to determine the continuously optimized temperature diffusion management; S8, based on the external wind speed and humidity data obtained by the meteorological condition analysis module, the support vector machine algorithm is used to predict short-term meteorological changes, generate auxiliary control parameters, and update the energy consumption optimization control strategy; S9, the infrared sensor data obtained by the personnel activity sensing module is used to analyze the changes in densely populated areas, adjust the weights of the temperature diffusion path, and obtain an adaptive control solution for dynamic environmental changes.

[0040] The system begins with data collection from a distributed sensor network. This network consists of multiple sensor nodes, each responsible for real-time monitoring of indoor and outdoor temperature, weather conditions, and human activity data. These sensors, including temperature sensors, humidity sensors, infrared sensors, and weather station equipment, collect data once per second, recording each set of data with a timestamp. For example, in a 100-square-meter smart building, 20 sensor nodes are deployed: 10 measure indoor temperature distribution, 5 monitor outdoor weather conditions, and 5 infrared sensors detect changes in areas with high occupancy. These sensor nodes transmit data via wireless communication protocols, forming a multidimensional dataset containing time series information. In a practical application, suppose the outdoor temperature at a certain moment is 30 degrees Celsius, while the indoor temperatures at different locations are 24, 26, and 28 degrees Celsius, respectively. At the same time, the infrared sensors detect that people are primarily concentrated in the northeast corner of the room. This data is then integrated, timestamped, and passed to the next module for processing.

[0041] Next, a convolutional neural network is used to extract features from the collected multidimensional data set to generate a spatial map of uneven temperature distribution. The design structure of the convolutional neural network includes an input layer, several convolutional layers, a pooling layer, and a fully connected layer. The input layer receives multidimensional data from the sensor network, the convolutional layer extracts local temperature gradient features through a sliding window, and the pooling layer performs dimensionality reduction on the features to reduce computational complexity. Finally, the fully connected layer outputs a two-dimensional spatial map that clearly shows the spatial distribution of the current indoor temperature. For example, in the above scenario, the convolutional neural network will generate a heat map showing that the temperature in the northeast corner is higher and the temperature in the southwest corner is lower. If there are obvious local high or low temperature areas in the spatial map, the system will further call the long short-term memory network to predict the temperature diffusion trend in the next 30 minutes. The formula for the long short-term memory network is Among them, T future (x, y, t) represents the predicted temperature at the location (x, y) at a certain moment in the future, T current (x,y) is the current temperature, ΔT i (x, y, t) is the temperature change caused by the i-th influencing factor (such as meteorological conditions or human activities), w i is the corresponding impact weight. For example, when the outdoor wind speed increases, ΔT i (x,y,t) will increase, resulting in a more significant change in the predicted temperature. By dynamically adjusting the weight w i ,The system can accurately predict the temperature diffusion path and provide a basis for subsequent control strategies.

[0042] After completing the prediction of the temperature diffusion path, the system uses the reinforcement learning algorithm to process the temperature diffusion path and meteorological condition data to calculate the optimal door and window opening angle and ventilation time. The formula for calculating the door and window opening angle is θ = f RL (T diff ,M weather ); where θ is the door and window opening angle, T diff is the temperature difference between indoor and outdoor, M weather is the meteorological condition data; f RL is the function mapping relationship of the reinforcement learning model. The formula for calculating the ventilation time is t vent =g RL (T gradient ,P occupancy ); where t vent is the ventilation time, T gradient is the indoor temperature gradient, P occupancy is the population distribution density, g RL is the function mapping relationship of the reinforcement learning model. For example, when the temperature difference between indoor and outdoor is large and the outdoor wind speed is low, the system will prioritize larger door and window opening angles to accelerate air circulation, while adjusting the ventilation time based on the indoor temperature gradient to ensure a more uniform temperature distribution. During this process, the reinforcement learning model gradually improves control accuracy through continuous trial and error and feedback optimization strategies. Assuming that the initial door and window opening angle is 30 degrees, after multiple iterations, the system ultimately determines the optimal opening angle to be 45 degrees and the ventilation time to be 15 minutes. This strategy can minimize energy consumption while ensuring comfort.

[0043] To further enhance the system's adaptive capabilities, when the deviation between the real-time air flow distribution and the temperature diffusion path exceeds a preset threshold, the system will re-optimize the door and window opening angles using a genetic algorithm. The genetic algorithm formula is F(θ) = α·E save +β·C comfort Among them, E save For energy saving effect rating, C comfort is the comfort score, and α and β are weighting coefficients. For example, in hot summer environments, the system may prefer to improve the energy-saving effect score, so α is set to 0.7 and β to 0.3. The genetic algorithm gradually optimizes the door and window opening angles until the maximum value of the objective function is reached through population initialization, crossover mutation, and selection operations. Assuming the initial population contains 10 individuals, each representing a door and window opening angle solution, after five generations of evolution, the system ultimately determines the optimal solution to be an opening angle of 50 degrees, with an energy-saving effect score of 0.9, a comfort score of 0.85, and an overall score of 0.88.

[0044] In addition, the system also uses the support vector machine algorithm to predict short-term meteorological changes and generate auxiliary control parameters to dynamically adjust the temperature diffusion path weight. The formula of the support vector machine algorithm is Among them, K(x i ,x) is the kernel function, α i is the Lagrange multiplier, b is the bias term, y i For example, if strong winds are predicted within the next hour, the system automatically adjusts the weights of the temperature diffusion paths, prioritizing the impact of wind speed on indoor temperature distribution. This allows the system to proactively respond to weather changes and avoid temperature fluctuations caused by external interference.

[0045] Finally, infrared sensors analyze changes in crowded areas to further optimize the weights of temperature diffusion paths. For example, if a person moves from the northeast corner to the northwest corner, the system dynamically adjusts the weights to better adapt the temperature diffusion path to the shift in occupancy. This mechanism effectively improves the system's flexibility and responsiveness, ensuring an efficient and comfortable intelligent temperature management solution in a variety of complex environments.

[0046] The artificial intelligence-based door and window temperature control system described in this embodiment includes: a data acquisition module for acquiring indoor and outdoor multi-point temperature data, meteorological condition data, and personnel activity data through a distributed sensor network to generate a multidimensional data set containing a timestamp; a temperature distribution analysis module for analyzing temperature distribution characteristics using a convolutional neural network based on the multidimensional data set, extracting indoor temperature gradients, outdoor temperature differences, and areas with dense personnel, and obtaining a spatial mapping of uneven temperature distribution; a temperature diffusion prediction module for predicting the temperature diffusion trend within the next thirty minutes through a long short-term memory network if there are local high-temperature or low-temperature areas in the spatial mapping of uneven temperature distribution, and generating a temperature diffusion path; a control strategy An optimization module is used to process the temperature diffusion path and meteorological condition data through a reinforcement learning algorithm, calculate the door and window opening and closing angles and ventilation time, and obtain an energy consumption optimization control strategy; a dynamic door and window control module is used to adjust the opening and closing status of each door and window according to the energy consumption optimization control strategy, and generate a real-time air flow field distribution; a deviation optimization module is used to re-optimize the door and window opening and closing angles through a genetic algorithm if the deviation between the real-time air flow field distribution and the temperature diffusion path exceeds a preset threshold, and obtain an updated control instruction; a loop optimization module is used to collect the adjusted indoor temperature data through a multi-point temperature sensing module, generate a new multidimensional data set, and cyclically execute the above steps to determine continuously optimized temperature diffusion management.

[0047] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the scope of protection of 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 essence and scope of the technical solutions of the present invention.

Claims

1. A door and window temperature control method based on artificial intelligence, characterized by: The following steps are involved: S1, obtains indoor and outdoor multi-point temperature data, meteorological condition data and personnel activity data through a distributed sensor network, and generates a multidimensional dataset with a time stamp; S2. Based on the multidimensional data set, a convolutional neural network is used to analyze the temperature distribution characteristics, extract the indoor temperature gradient, outdoor temperature difference and crowded areas, and obtain a spatial map of uneven temperature distribution; S3. If there are local high or low temperature areas in the spatial map of uneven temperature distribution, the temperature diffusion trend in the next 30 minutes is predicted through the long short-term memory network to generate a temperature diffusion path; S4. Process the temperature diffusion path and meteorological condition data through a reinforcement learning algorithm, calculate the door and window opening angles and ventilation duration, and obtain an energy consumption optimization control strategy; S5. Based on the energy consumption optimization control strategy, the dynamic door and window control module is used to adjust the opening and closing status of each door and window to generate real-time air flow field distribution; S6. If the deviation between the real-time air flow field distribution and the temperature diffusion path exceeds a preset threshold, the door and window opening and closing angles are re-optimized using a genetic algorithm to obtain an updated control instruction; S7. Collect the adjusted indoor temperature data through the multi-point temperature sensing module to generate a new multidimensional data set, and repeat the above steps to determine the continuously optimized temperature diffusion management; S8. Based on the external wind speed and humidity data obtained by the meteorological condition analysis module, a support vector machine algorithm is used to predict short-term meteorological changes, generate auxiliary control parameters, and update the energy consumption optimization control strategy; S9. The infrared sensor data obtained by the personnel activity perception module is used to analyze the changes in densely populated areas, adjust the weights of the temperature diffusion paths, and obtain an adaptive control solution for dynamic environmental changes.

2. The artificial intelligence-based door and window temperature control method according to claim 1, characterized in that: In step S3, the formula of the long short-term memory network is Among them, T future (x, y, t) represents the predicted temperature at the location (x, y) at a certain moment in the future, T current (x,y) is the current temperature, ΔT i (x, y, t) is the temperature change caused by the i-th influencing factor, w i is the corresponding influence weight.

3. The artificial intelligence-based door and window temperature control method according to claim 1, characterized in that: In step S4, the formula for calculating the door and window opening angle is θ = f RL (T diff ,M weather ); where θ is the door and window opening angle, T diff is the temperature difference between indoor and outdoor, M weather is the meteorological condition data; f RL It is the function mapping relationship of the reinforcement learning model.

4. The artificial intelligence-based door and window temperature control method according to claim 1, characterized in that: In step S4, the formula for calculating ventilation time is t vent =g RL (T gradient ,P occupancy ); where t vent is the ventilation time, T gradient is the indoor temperature gradient, P occupancy is the population distribution density, g RL It is the function mapping relationship of the reinforcement learning model.

5. The artificial intelligence-based door and window temperature control method according to claim 1, characterized in that: In step S6, the genetic algorithm formula is F(θ)=α·E save +β·C comfort ; Among them, E save For energy saving effect rating, C comfort is the comfort score, α and β are weight coefficients.

6. The artificial intelligence-based door and window temperature control method according to claim 1, characterized in that: In step S8, the formula of the support vector machine algorithm is: Among them, K(x i ,x) is the kernel function, α i is the Lagrange multiplier, b is the bias term, y i is the training sample label.

7. The artificial intelligence-based door and window temperature control method according to claim 1, characterized in that: The distributed sensor network acquires temperature data, meteorological conditions, and human activity data from multiple indoor and outdoor locations, generating a multidimensional dataset with time stamps, including: Synchronize data collected by different types of sensors deployed indoors and outdoors to ensure data time consistency; Filter and denoise the raw data collected by the sensor to remove outliers and noise interference; The processed data is sorted by timestamp and stored as a structured multidimensional cube.

8. The artificial intelligence-based door and window temperature control method according to claim 1, characterized in that: Based on the multidimensional dataset, a convolutional neural network is used to analyze the temperature distribution characteristics, extract the indoor temperature gradient, outdoor temperature difference, and crowded areas, and obtain a spatial map of uneven temperature distribution, including: Inputting the multidimensional data set into the first layer of a convolutional neural network and performing spatial feature extraction on the data using a two-dimensional convolution kernel; In the middle layer of the convolutional neural network, the pooling operation is used to reduce the dimension of the feature map and retain the main feature information; In the last layer of the convolutional neural network, the output contains feature maps of temperature gradient, outdoor temperature difference, and crowded areas, and generates a spatial map of uneven temperature distribution.

9. An artificial intelligence-based door and window temperature control system, characterized in that: include: The data acquisition module is used to obtain temperature data, meteorological condition data, and personnel activity data at multiple points indoors and outdoors through a distributed sensor network, and generate a multidimensional data set with a time stamp; A temperature distribution analysis module is used to analyze temperature distribution characteristics using a convolutional neural network based on the multidimensional dataset, extract indoor temperature gradients, outdoor temperature differences, and areas with high population density, and obtain a spatial mapping of uneven temperature distribution; A temperature diffusion prediction module is used to predict the temperature diffusion trend within the next thirty minutes through a long short-term memory network if there are local high temperature or low temperature areas in the spatial map of uneven temperature distribution, and generate a temperature diffusion path; A control strategy optimization module is used to process the temperature diffusion path and meteorological condition data through a reinforcement learning algorithm, calculate the door and window opening and closing angles and ventilation duration, and obtain an energy consumption optimization control strategy; A dynamic door and window control module is used to adjust the opening and closing status of each door and window according to the energy consumption optimization control strategy and generate real-time air flow field distribution; a deviation optimization module, configured to re-optimize the door and window opening and closing angles using a genetic algorithm to obtain updated control instructions if the deviation between the real-time air flow field distribution and the temperature diffusion path exceeds a preset threshold; The loop optimization module is used to collect the adjusted indoor temperature data through the multi-point temperature sensing module, generate a new multidimensional data set, and execute the above steps in a loop to determine the continuously optimized temperature diffusion management.

Images