Method and device for controlling dehumidification heat compensation of constant-temperature and constant-humidity air conditioning unit

Abstract

The invention provides a control method and device for dehumidification heat compensation of a constant-temperature and constant-humidity air conditioning unit, and the method comprises the steps: firstly monitoring indoor and outdoor temperature and humidity data sets in a current time period T, and obtaining a historical temperature and humidity data set; then, the indoor temperature prediction value and the indoor humidity prediction value are input into the trained convolutional neural network model, and after the convolutional neural network model outputs the indoor temperature prediction value and the indoor humidity prediction value, the indoor temperature prediction value and the indoor humidity prediction value are compared with a preset temperature range and a preset humidity range; when the temperature predicted value or the humidity predicted value exceeds the upper limit or the lower limit of the corresponding preset range, a humidity adjusting mode or a temperature adjusting mode is started, and when the temperature predicted value or the humidity predicted value is within the preset range, the system enters a high-precision fine adjustment mode; and finally, the display displays the current temperature and humidity. An improved convolutional neural network model is adopted, temperature and humidity predicted values are calculated in real time and compared with a preset range, and a humidity adjustment mode or a temperature adjustment mode or a high-precision fine adjustment mode is selected to be started; the temperature and humidity adjusting accuracy is greatly improved, and the user experience is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of air conditioning, and in particular to a control method and device for dehumidification heat compensation of a constant temperature and humidity air conditioning unit. Background Art

[0002] Constant temperature and humidity air conditioning units are common environmental control equipment in modern buildings and are widely used in places where temperature and humidity need to be strictly controlled, such as laboratories, museums, data centers, clean rooms, etc. Its basic function is to maintain a constant indoor environment by adjusting the temperature and humidity of the air. However, with the advancement of building technology and the improvement of users' requirements for environmental comfort, traditional constant temperature and humidity air conditioning units often face many technical challenges when dealing with complex and changeable environmental factors.

[0003] First, the control of temperature and humidity is interrelated. During the dehumidification process, the temperature usually decreases, so a corresponding thermal compensation function is required to maintain the stability of the temperature. Traditional air-conditioning units mostly use simple temperature or humidity control systems to adjust the dehumidification and heating functions respectively. However, with the increasing complexity of the environment, such as changes in the number of people indoors and the influence of outdoor meteorological conditions, a single control method has been difficult to meet the precise requirements of constant temperature and humidity. Especially in the case of large fluctuations in temperature and humidity, traditional air-conditioning systems may be difficult to respond quickly, resulting in environmental instability and affecting the functional requirements of the use site. Secondly, existing air-conditioning systems usually rely on static control methods and fail to make full use of modern intelligent technologies for predictive adjustment. For example, traditional air-conditioning units usually simply perform feedback control by real-time detection of current temperature and humidity data, lacking foresight for future temperature and humidity changes. Therefore, when environmental conditions change suddenly, the system responds slowly, which easily leads to fluctuations in the temperature and humidity of the environment, and it is difficult to ensure long-term constant temperature and humidity effects. In addition, with the development of intelligent buildings, dynamic factors such as the number of people indoors have an increasing impact on environmental control. How to pre-adjust the air-conditioning system through intelligent algorithms to meet the needs of future changes has become an urgent problem to be solved in the industry.

[0004] In recent years, intelligent control technology has been gradually applied to constant temperature and humidity air-conditioning units. By introducing artificial intelligence and deep learning technology, the air-conditioning system can predict future changes in temperature and humidity based on historical data and current status, thereby adjusting the system's operating parameters in advance. However, the existing intelligent air-conditioning control system still has many limitations. For example, most current intelligent control systems are based only on simple neural networks or traditional PID control, and fail to fully consider complex environmental factors, such as the dynamic changes in the number of people indoors and fluctuations in external climate conditions. In addition, existing systems usually rely only on static environmental data and cannot dynamically adjust control strategies, resulting in the inability to achieve refined control in practical applications. In addition, existing sound quality enhancement systems usually lack a comprehensive analysis of the user's hearing characteristics and fail to fully consider the impact of the ear canal structure on audio transmission, resulting in less than ideal sound quality enhancement effects. In addition, the existing constant temperature and humidity air conditioners have fewer considerations in terms of adjustment, and do not make comprehensive considerations based on historical indoor and outdoor temperature and humidity data. When the outdoor temperature and humidity are high, the power required for indoor adjustment is higher, and no future prediction values ​​are determined based on the number of people indoors. The temperature and humidity of the air conditioner are adjusted in advance by determining future prediction values, so that the indoor temperature and humidity are kept in a smaller fluctuation range. In addition, the existing technology does not involve a fine adjustment mode for comprehensive adjustment, resulting in a significant reduction in the efficiency of temperature and humidity adjustment. In view of the limitations of existing air conditioner temperature and humidity adjustment, a new solution is urgently needed for the refined automated real-time temperature and humidity adjustment processing method to improve the processing efficiency and accuracy, so as to improve user satisfaction. Summary of the invention

[0005] In view of the above problems mentioned in the prior art, the present invention provides a control method and device for dehumidification heat compensation of a constant temperature and humidity air conditioning unit. The method first records the temperature and humidity data set X within the current period T of indoor and outdoor c and historical temperature and humidity dataset X h ; Secondly, the temperature and humidity dataset X c , indoor population and historical temperature and humidity dataset X h Input to the trained convolutional neural network model, the convolutional neural network model outputs the indoor temperature prediction value T f and humidity prediction value H f ; Then, compare with the preset temperature range and humidity range; when the temperature prediction value T f Or humidity prediction value H fWhen the upper or lower limit of the corresponding preset range is exceeded, the humidity adjustment mode or temperature adjustment mode is started. When it is within the preset range, the system enters the high-precision fine-tuning mode; finally, the display shows the current temperature and humidity. This application adopts an improved convolutional neural network model to calculate the temperature and humidity prediction values ​​in real time and compare them with the preset range, and choose to start the humidity adjustment mode or temperature adjustment mode or high-precision fine-tuning mode; it greatly improves the accuracy and efficiency of air conditioning temperature and humidity adjustment, and greatly increases the user experience.

[0006] The present application provides a control method for dehumidification heat compensation of a constant temperature and humidity air conditioning unit, comprising the steps of:

[0007] S1: Use humidity sensors and temperature sensors to monitor the relative humidity and air temperature of the indoor environment in real time, and record the temperature and humidity data set X for the current period T indoors and outdoors c and historical temperature and humidity dataset X h , X c ={T c , H c}, X h ={T h , H h}, where T c is the indoor and outdoor temperature data vector in the current period T, H c is the indoor and outdoor humidity data vector in the current period T, T h is the historical indoor and outdoor temperature data vector, H h Historical indoor and outdoor humidity data vectors;

[0008] S2: The temperature and humidity data set X for the current period T of indoor and outdoor c , indoor population and historical temperature and humidity dataset X h Input to the trained convolutional neural network model, the convolutional neural network model outputs the predicted indoor temperature value T at the future time t0 f and humidity prediction value H f ;

[0009] S3: predicted indoor temperature value T at future time t0 f and humidity prediction value H f Compare with the preset temperature range and humidity range respectively. When the temperature prediction value T f Or humidity prediction value H f When the upper or lower limit of the corresponding preset range is exceeded, the humidity adjustment mode or the temperature adjustment mode is activated;

[0010] S4: When the predicted temperature and humidity values ​​at the future time t0 are within the preset range, the system enters the high-precision fine-tuning mode and automatically adjusts the humidity and temperature so that the ambient temperature and humidity fluctuation ranges do not exceed the set thresholds respectively;

[0011] S5: The display shows the current temperature and humidity, and the humidity sensor and temperature sensor continue to monitor in real time.

[0012] Preferably, the system enters a high-precision fine-tuning mode, and automatically adjusts the humidity and temperature, including: S41: calculating the temperature error ΔT=T f -T m And humidity error ΔH=H f -H m , where T m is the current indoor temperature, H m is the current indoor humidity; S42: calculate the current heating power P according to the temperature error ΔT and the humidity error ΔH h and dehumidification power P d , in, is the temperature proportional gain, is the humidity proportional gain; S43: repeat step S41 until the temperature error ΔT and the humidity error ΔH are both 0.

[0013] Preferably, the temperature proportional gain at time t Among them, α temperature is the adaptive coefficient, K p0 is the initial proportional gain of temperature, is the rate of change of temperature error; humidity proportional gain at time t Among them, β is the humidity adaptive coefficient, K p1 is the initial proportional gain of humidity, is the rate of change of humidity error at time t.

[0014] Preferably, the trained convolutional neural network model adopts an improved Swish activation function f(x,N):

[0015]

[0016] Among them, x is the input of the convolutional layer, N is the number of people in the room, σ is the adjustment parameter, and e is the base of the natural logarithm.

[0017] Preferably, the number of people indoors is obtained by capturing the current indoor video image data in real time using a camera installed indoors, and using a YOLO target detection model to identify the number of people in the indoor video image to obtain the number of people N indoors.

[0018] The present application also provides a control device for dehumidification heat compensation of a constant temperature and humidity air conditioning unit, comprising:

[0019] Acquisition module: Humidity sensor and temperature sensor monitor the relative humidity and air temperature of the indoor environment in real time, and record the temperature and humidity data set X within the current period T indoors and outdoorsc and historical temperature and humidity dataset X h , X c ={T c , H c}, X h ={T h , H h}, where T c is the indoor and outdoor temperature data vector in the current period T, H c is the indoor and outdoor humidity data vector in the current period T, T h is the historical indoor and outdoor temperature data vector, H h Historical indoor and outdoor humidity data vectors;

[0020] Processing module: The temperature and humidity data set X for the current period T of indoor and outdoor c , indoor population and historical temperature and humidity dataset X h Input to the trained convolutional neural network model, the convolutional neural network model outputs the predicted indoor temperature value T at the future time t0 f and humidity prediction value H f ;

[0021] Adjustment module: indoor temperature prediction value T at future time t0 f and humidity prediction value H f Compare with the preset temperature range and humidity range respectively. When the temperature prediction value T f Or humidity prediction value H f When the upper or lower limit of the corresponding preset range is exceeded, the humidity adjustment mode or the temperature adjustment mode is activated;

[0022] Fine-tuning module: When the predicted temperature and humidity values ​​at the future time t0 are within the preset range, the system enters the high-precision fine-tuning mode and automatically adjusts the humidity and temperature so that the ambient temperature and humidity fluctuation ranges do not exceed the set thresholds respectively;

[0023] Display module: The display shows the current temperature and humidity, and the humidity sensor and temperature sensor continue to monitor in real time.

[0024] Preferably, the system enters a high-precision fine-tuning mode, and automatically adjusts the humidity and temperature, including: S41: calculating the temperature error ΔT=T f -T m And humidity error ΔH=H f -H m , where T m is the current indoor temperature, H m is the current indoor humidity; S42: calculate the current heating power P according to the temperature error ΔT and the humidity error ΔH h and dehumidification power P d, in, is the temperature proportional gain, is the humidity proportional gain; S43: repeat step S41 until the temperature error ΔT and the humidity error ΔH are both 0.

[0025] Preferably, the temperature proportional gain at time t Among them, α temperature is the adaptive coefficient, K p0 is the initial proportional gain of temperature, is the rate of change of temperature error; humidity proportional gain at time t Among them, β is the humidity adaptive coefficient, K p1 is the initial proportional gain of humidity, is the rate of change of humidity error at time t.

[0026] Preferably, the trained convolutional neural network model adopts an improved Swish activation function f(x,N):

[0027]

[0028] Among them, x is the input of the convolutional layer, N is the number of people in the room, σ is the adjustment parameter, and e is the base of the natural logarithm.

[0029] Preferably, the number of people indoors is obtained by capturing the current indoor video image data in real time using a camera installed indoors, and using a YOLO target detection model to identify the number of people in the indoor video image to obtain the number of people N indoors.

[0030] The present invention provides a control method and device for dehumidification heat compensation of a constant temperature and humidity air conditioning unit, which can achieve the following beneficial technical effects:

[0031] 1. The present invention first records the temperature and humidity data set X within the current period T of indoor and outdoor c and historical temperature and humidity dataset X h ; Secondly, the temperature and humidity dataset X c , indoor population and historical temperature and humidity dataset X h Input to the trained convolutional neural network model, the convolutional neural network model outputs the indoor temperature prediction value T f and humidity prediction value H f ; Then, compare with the preset temperature range and humidity range; when the temperature prediction value T f Or humidity prediction value H fWhen the upper or lower limit of the corresponding preset range is exceeded, the humidity adjustment mode or temperature adjustment mode is started. When it is within the preset range, the system enters the high-precision fine-tuning mode; finally, the display shows the current temperature and humidity. This application adopts an improved convolutional neural network model to calculate the temperature and humidity prediction values ​​in real time and compare them with the preset range, and choose to start the humidity adjustment mode or temperature adjustment mode or high-precision fine-tuning mode; it greatly improves the accuracy and efficiency of air conditioning temperature and humidity adjustment, and greatly increases the user experience.

[0032] 2. The trained convolutional neural network model of the present invention adopts an improved Swish activation function. By introducing the number of people N in the room into the activation function, the system can dynamically perceive the changes in the indoor environment, especially the impact of the number of people on temperature and humidity. As the number of people increases, the indoor temperature and humidity usually change significantly. Traditional control methods usually only adjust according to the real-time data of the temperature and humidity sensor, and cannot respond in advance to the temperature and humidity fluctuations caused by the change in the number of people. By taking the number of people as an input parameter, the system can dynamically adjust the heating power and dehumidification power according to the current number of people in the room, improving the adaptability and response speed to complex environments. The system can quickly adapt to and predict the changes in temperature and humidity in advance, avoiding the lagging adjustment problem of traditional systems. The accuracy of the system's temperature and humidity control is significantly improved. When facing smaller environmental fluctuations, it can also accurately adjust the temperature and humidity through changes in the number of people to maintain a constant indoor environment.

[0033] 3. The high-precision fine-tuning mode of the present invention automatically adjusts humidity and temperature. By calculating the temperature error and humidity error, the heating power and dehumidification power at the current moment are calculated according to the temperature error and humidity error. The temperature proportional gain and humidity proportional gain are considered in the calculation process, which greatly improves the fine adjustment. Through the setting of high-precision mode and temperature adjustment mode and humidity adjustment mode, the problem of over-adjustment or under-adjustment is avoided, and the overall operation stability is improved.

[0034] 4. The present invention records the temperature and humidity data set X within the current period T of indoor and outdoor c and historical temperature and humidity data sets, the convolutional neural network model can analyze the changing patterns of temperature and humidity based on current and historical data, and predict the future trends of temperature and humidity. This combination not only relies on the current environmental conditions, but also takes into account the laws of past changes, thereby improving the accuracy of temperature and humidity predictions. Traditional air-conditioning systems are usually controlled based on immediate temperature and humidity feedback, which can easily lead to frequent start and stop of equipment, increase equipment wear, and shorten the service life of equipment. By using current and historical temperature and humidity data, the system can predict future temperature and humidity changes in advance and take appropriate measures to avoid frequent start and stop. For example, when historical data shows that the ambient temperature and humidity have periodic fluctuations in a short period of time, the system can intervene in advance to make smoother adjustments and reduce the frequency of overreaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. 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 creative work.

[0036] Figure 1 It is a schematic diagram of the steps of a control method for dehumidification heat compensation of a constant temperature and humidity air conditioning unit of the present invention;

[0037] Figure 2 It is a schematic diagram of a control device for dehumidification heat compensation of a constant temperature and humidity air conditioning unit of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] Embodiment 1:

[0040] In view of the above problems mentioned in the prior art, in order to solve the above technical problems, as shown in the attached Figure 1 As shown: This application provides a control method for dehumidification heat compensation of a constant temperature and humidity air conditioning unit, comprising the steps of:

[0041] S1: Use humidity sensors and temperature sensors to monitor the relative humidity and air temperature of the indoor environment in real time, and record the temperature and humidity data set X for the current period T indoors and outdoors c and historical temperature and humidity dataset X h , X c ={T c , H c}, X h ={T h , H h}, where T c is the indoor and outdoor temperature data vector in the current period T, H c is the indoor and outdoor humidity data vector in the current period T, T h is the historical indoor and outdoor temperature data vector, H hHistorical indoor and outdoor humidity data vectors; the constant temperature and humidity air conditioning unit used in the present invention includes the following modules: the humidity sensor is used to monitor the relative humidity indoors and outdoors in real time, and the data unit is %RH (relative humidity percentage); the temperature sensor is used to monitor the indoor and outdoor air temperature in real time, and the data unit is ℃; the data acquisition unit is responsible for recording the temperature and humidity data in the current time period T, and storing these data as data vectors; the unit is also responsible for recording the temperature and humidity data in the historical time period; the convolutional neural network processing unit predicts the temperature and humidity at future times by using the current and historical temperature and humidity data. The system collects temperature and humidity data in a time period T of every 10 minutes. The humidity sensor and temperature sensor in the system will monitor the indoor and outdoor temperature and humidity conditions respectively in each time period, and record these data as data vectors. The system also needs to record and save the historical temperature and humidity data within a certain period of time in the past, so that the subsequent convolutional neural network model can predict the temperature and humidity at future times.

[0042] S2: The temperature and humidity data set X for the current period T of indoor and outdoor c , indoor population and historical temperature and humidity dataset X h Input to the trained convolutional neural network model, the convolutional neural network model outputs the predicted indoor temperature value T at the future time t0 f and humidity prediction value H f ; Input the above data set and the number of people N indoors into the trained convolutional neural network model. The convolutional neural network extracts features from these input data through the convolution layer, such as the changing trends of temperature and humidity, the mutual influence between indoor and outdoor, etc. The convolution layer performs convolution operations on the temperature and humidity data to extract features, such as temperature fluctuations and humidity change trends; the fully connected layer combines the features extracted by the convolution layer with the indoor number data to generate a comprehensive feature representation. Output the predicted temperature and humidity values ​​at a future moment. After being processed by the convolutional neural network, the system can output the predicted temperature and humidity values ​​at a certain moment in the future:

[0043] The convolutional neural network (CNN) in the present invention is mainly composed of a convolutional layer, a pooling layer, a fully connected layer and an output layer. Its core function is to extract valuable features from input temperature and humidity data and environmental factors such as the number of people indoors, and then predict the temperature and humidity at future times.

[0044] The convolution layer is the foundation of the network and is used to extract local features from the input data. Each convolution layer uses multiple convolution kernels to process the input data. The convolution kernels slide on the input data to extract the feature information of local changes. The main function of the convolution layer is to capture short-term or local change patterns in the input data. For example, features such as short-term fluctuations in temperature and humidity and the impact of the number of people in the room on the environment can be identified through convolution operations. The depth of the convolution layer is usually shallow. The first few layers are responsible for extracting basic features, such as the basic change trend of temperature and humidity. With the increase of convolution layers, the extracted features are gradually abstracted and can capture higher-level relationships, such as the long-term temperature and humidity change patterns and the mutual influence between different input data. The pooling layer is used to reduce the dimension of the features extracted by the convolution layer, reduce the computational complexity and retain key features. Common pooling operations include maximum pooling and average pooling. Usually, maximum pooling is used to select the maximum value in a certain area, thereby retaining the main features and reducing the impact of noise on the prediction. Through the pooling layer, the system can extract the core feature information, reduce redundant data, and reduce the computational burden of the model. The pooling layer can enhance the robustness of the network so that small input changes will not lead to drastic changes in the output prediction, thereby improving the generalization ability of the model. After passing through several convolutional layers and pooling layers, the convolutional neural network inputs the extracted features into the fully connected layer for comprehensive processing. After flattening all feature vectors, the fully connected layer performs high-level nonlinear combinations through the connection of multiple layers of neurons. Through the fully connected layer, the network can integrate different input features together to form a comprehensive understanding of future temperature and humidity changes. The fully connected layer is responsible for generating the final prediction results, that is, the temperature and humidity at the future moment. The fully connected layer acts like a predictor, responsible for making a comprehensive prediction of the temperature and humidity at the future moment based on all the extracted features. The output layer is the last layer of the CNN, which is used to generate the predicted values ​​of temperature and humidity at the future moment. The output layer usually uses a linear activation function to ensure that the output prediction value is a continuous value and conforms to the actual range of temperature and humidity.

[0045] The training process of CNN is as follows: Before network training, the system requires a large amount of historical temperature and humidity data, indoor population data, and environmental condition data. These data are divided into training sets and validation sets for training and optimizing convolutional neural networks. The training set is used to learn the model, and the validation set is used to evaluate the generalization ability of the model. During the training process, the input data first passes through the convolution layer, pooling layer, and fully connected layer, and then passes to the output layer to generate the predicted value of temperature and humidity. This process is called forward propagation, and the convolutional neural network generates a prediction result based on the current input data. The output result of the network is compared with the actual historical temperature and humidity data, and the prediction error is calculated through the loss function. Commonly used loss functions include mean square error (MSE), which can measure the difference between the predicted value and the actual value. According to the output result of the loss function, the system performs back propagation and adjusts the weight parameters in the network. The optimizer (such as the Adam optimizer) updates the weights in the convolution kernel and the fully connected layer layer by layer by calculating the gradient of the loss function relative to the weight of each layer until the value of the loss function is minimized. This process is repeated so that the model can gradually learn the law of temperature and humidity changes. Throughout the training process, the model continuously learns through multiple iterations and continuously adjusts weights to reduce prediction errors. The training process is usually divided into multiple epochs, that is, the model learns multiple times on the entire training dataset to ensure that the model can learn stable features from different samples.

[0046] After the convolutional neural network training is completed, the model can be used to predict the temperature and humidity in the future in real time. The system inputs the temperature and humidity data of the current period, the number of people indoors, etc. into the trained convolutional neural network. The convolution layer processes the input data through the convolution kernel to extract local features, and the pooling layer reduces the data dimension and retains the main features. The fully connected layer combines the temperature and humidity according to the features extracted by the convolution layer. The model can identify the trend of future temperature and humidity changes under the current environmental conditions from the extracted features. Finally, the model generates the predicted temperature and humidity values ​​for the future at the output layer, and the system decides whether to adjust the temperature and humidity based on these prediction results. Through this prediction process, the convolutional neural network can perceive the changes in temperature and humidity in the future environment in advance, thereby helping the air conditioning system to adjust in advance, avoiding drastic fluctuations in temperature and humidity, and ensuring the stability and comfort of the environment.

[0047] S3: predicted indoor temperature value T at future time t0 f and humidity prediction value H f Compare with the preset temperature range and humidity range respectively. When the temperature prediction value T f Or humidity prediction value H f When the upper or lower limit of the corresponding preset range is exceeded, the humidity adjustment mode or the temperature adjustment mode is activated;

[0048] In some embodiments, when the constant temperature and humidity air conditioning unit is running, the system will set a target temperature and humidity range according to environmental requirements. For example: temperature range: 22°C to 26°C; humidity range: 40%RH to 60%RH; this target range is set according to the specific needs of the site. For example, in places with high environmental requirements such as laboratories or museums, the control of temperature and humidity needs to be very precise.

[0049] After processing the temperature and humidity data of the current period, the historical temperature and humidity data, and the number of people indoors, the convolutional neural network generates the predicted temperature and humidity values ​​for the future. For example: the predicted temperature value for the next 10 minutes: 27°C; the predicted humidity value for the next 10 minutes: 65%RH; these predicted values ​​indicate that the system predicts the future temperature and humidity by learning the changing trends in the environment. If the system does not take any adjustment measures, the future temperature and humidity may deviate from the preset target range. The system compares the future temperature and humidity prediction values ​​output by the convolutional neural network with the preset temperature and humidity range: Temperature comparison: The future temperature prediction value is 27°C, which exceeds the upper limit of the target temperature range (26°C), so the system needs to adjust the temperature. Humidity comparison: The future humidity prediction value is 65%RH, which exceeds the upper limit of the target humidity range (60%RH), and the system needs to adjust the humidity.

[0050] According to the situation that the temperature and humidity forecast values ​​exceed the target range, the system starts the corresponding adjustment mode: because the future temperature forecast value (27℃) exceeds the upper limit of the preset range (26℃), the system determines that the environment is about to become overheated. Therefore, the air conditioning unit will start the cooling mode. The specific steps are as follows: The system starts the cooling function in advance according to the predicted temperature value to reduce the indoor temperature. By adjusting the refrigerant flow and wind speed, the system gradually reduces the temperature to within the target range. Monitor the indoor temperature changes. If it drops to the target range (such as 25℃), the cooling power is gradually reduced to keep the temperature stable.

[0051] The future humidity forecast is 65%RH, which exceeds the upper limit of the target humidity range (60%RH), indicating that the environment is about to become too humid. The system will start the dehumidification mode, and the specific steps are as follows: The air conditioning unit starts the dehumidification function to adjust the environment by reducing the humidity in the air. The system may increase the operating efficiency of the cooling coil to promote condensation of moisture in the air, thereby reducing humidity. When the humidity drops to within the target range (such as 58%RH), the system will gradually reduce the dehumidification function to maintain humidity stability.

[0052] By using the predictive ability of convolutional neural networks, the system can perceive future environmental changes in advance and start the adjustment mode before the temperature and humidity exceed the preset range, avoiding the lag of the "passive" adjustment of traditional air-conditioning systems. The advantages of this pre-adjustment include: because the system starts the adjustment in advance, the temperature and humidity fluctuations are controlled within a smaller range, ensuring the constancy of the environment. Pre-adjustment can avoid excessive adjustment or frequent start and stop of the system, optimize energy use, and improve the operating efficiency of the air-conditioning system. The system adjusts the temperature and humidity in advance to avoid the discomfort caused by sudden changes in the environment, especially in places with high requirements for temperature and humidity (such as laboratories, archives, etc.), which can protect equipment, cultural relics or samples from changes in temperature and humidity. By predicting future temperature and humidity through convolutional neural networks and combining the preset temperature and humidity range, the system can take adjustment measures in advance according to the prediction results to avoid the occurrence of excessively high or low temperature and humidity. This method greatly improves the intelligence level of constant temperature and humidity air-conditioning units.

[0053] S4: When the predicted temperature and humidity values ​​at the future time t0 are within the preset range, the system enters the high-precision fine-tuning mode and automatically adjusts the humidity and temperature so that the ambient temperature and humidity fluctuation ranges do not exceed ±0.1°C and ±1%, respectively;

[0054] S5: The display shows the current temperature and humidity, and the humidity sensor and temperature sensor continue to monitor in real time.

[0055] In some embodiments, the system enters a high-precision fine-tuning mode, and automatically adjusts humidity and temperature, including: S41: calculating the temperature error ΔT=T f -T m And humidity error ΔH=H f -H m , where T m is the current indoor temperature, H m is the current indoor humidity; S42: calculate the current heating power P according to the temperature error ΔT and the humidity error ΔH h and dehumidification power P d , in, is the temperature proportional gain, is the humidity proportional gain; S43: repeat step S41 until the temperature error ΔT and the humidity error ΔH are both 0.

[0056] In some embodiments, the temperature proportional gain at time t Among them, α temperature is the adaptive coefficient, K p0 is the initial proportional gain of temperature, is the rate of change of temperature error; humidity proportional gain at time t Among them, β is the humidity adaptive coefficient, Kp1 is the initial proportional gain of humidity, is the rate of change of humidity error at time t.

[0057] In some embodiments, the trained convolutional neural network model uses an improved Swish activation function f(x,N):

[0058]

[0059] Among them, x is the input of the convolutional layer, N is the number of people in the room, σ is the adjustment parameter, and e is the base of the natural logarithm.

[0060] In some embodiments, the number of people in the room is captured in real time by a camera installed indoors, and the number of people in the indoor video image is identified by a YOLO target detection model to obtain the number of people in the indoor video image. This embodiment describes in detail how to capture video image data in real time by a camera installed indoors, and use the YOLO target detection model to identify the number of people in the video, thereby determining the number of people N in the current room, and inputting the number information into the air-conditioning system for intelligent control. In this embodiment, the system installs a high-definition camera at a key position indoors (such as a ceiling or a corner) to ensure that the camera can cover most of the activity area. The camera captures dynamic video images indoors in real time at a high frame rate (such as 15 to 30 frames per second). The images captured by the camera contain the activities of the people indoors, and the system extracts static image data from these video frames for subsequent processing. For example: the camera captures an image containing multiple people, and the image shows that there are several people active or working indoors. The system uses this image as the input of the subsequent YOLO detection model.

[0061] The YOLO target detection model is a target detection technology based on deep learning. It can quickly and accurately identify multiple target objects (such as people, animals, objects, etc.) in an image and locate the positions of these targets at the same time. In this embodiment, the YOLO model is trained to detect the category of "people", that is, the model can identify the specific position of each person from the captured image and calculate the total number of people. The specific operation steps are as follows: the camera passes the captured static image to the YOLO model for processing at intervals (such as every second or every few seconds). The YOLO model analyzes the image, quickly identifies the position of each person, and marks the bounding box of each target person on the image. Through the identified bounding box, the system can determine how many people are in the image. At this time, the number of people in the room returned by the system is the number of people in the current image. The YOLO model processes a static image and successfully detects that there are 5 bounding boxes in the image, indicating that there are 5 people in the image, so the system sets the number of people N in the room at the current moment to 5.

[0062] The system will update the number of people in the room in real time based on the images captured by the camera and the detection results of the YOLO model. Since people may move in or out of the room, the system needs to continuously monitor the video data and continuously update the number of people in the room. For example, every 1 second, the camera captures a frame of image, and the YOLO model analyzes each frame of image to obtain the latest number of people. If there are 6 bounding boxes in the image captured by the system within 1 second, the system will update the number of people in the room to 6. If there are fewer people in the next captured image, and there are only 4 bounding boxes, the system will update the number of people in the room to 4. In this way, the system can realize dynamic monitoring of the number of people in the room and ensure that the latest personnel information is obtained in real time. After obtaining the number of people in the room N, the system inputs this data into the air conditioning control system as an important parameter for adjusting temperature and humidity. The change of the number of people in the room directly affects the temperature and humidity in the room. For example, when the number of people in the room increases, the system will predict that the increase in people will cause the temperature to rise, so it will increase the cooling power in advance to avoid excessive temperature. When the number of people in the room decreases, the system will reduce the cooling power to prevent over-cooling. By acquiring and processing indoor population data in real time, the system can dynamically adjust the temperature and humidity according to the changes in the number of people, ensuring that the environment always maintains a constant temperature and humidity. By using the YOLO target detection model, this system can accurately identify and calculate the number of people indoors in real time, and adjust the operating parameters of the air-conditioning system according to the changes in the number of people. Compared with the traditional static setting method, the dynamic detection method based on YOLO improves the intelligence level of the system, ensuring that the air-conditioning system responds to environmental changes in a timely manner, thereby improving the accuracy and efficiency of temperature and humidity adjustment.

[0063] The present application also provides a control device for dehumidification heat compensation of a constant temperature and humidity air conditioning unit, such as Figure 2As shown, it includes: the structure of the control device mainly includes the following key components, each of which works together to achieve precise control of indoor temperature and humidity: humidity sensor, temperature sensor, camera, data acquisition module, processing module (controller), dehumidification device, heater, display module, execution module. The humidity sensor and temperature sensor monitor the temperature and humidity data indoors and outdoors in real time. The humidity sensor and temperature sensor are installed indoors and outdoors, connected to the data acquisition module by wire or wireless, and transmit the monitored temperature and humidity data to the data acquisition module. The camera captures the image data of the room in real time and identifies the number of people in the room through the YOLO target detection model. The camera is installed on the ceiling or corner of the room, and transmits the captured image data to the processing module through the network or wired method. The processing module analyzes the number of people according to the image data and combines temperature and humidity adjustment. The data acquisition module receives and summarizes the data transmitted by the humidity sensor and temperature sensor. The data acquisition module connects the humidity sensor and the temperature sensor, and transmits the temperature and humidity data to the processing module by wire or wireless method. As the control center of the entire device, the processing module (controller) is responsible for receiving and analyzing the temperature, humidity and number of people data, predicting future temperature and humidity changes, and deciding whether to start the dehumidification or temperature adjustment mode. The processing module receives the temperature and humidity data from the data acquisition module and the image data from the camera, processes the data through an algorithm, and controls the operation of the dehumidification device and the heater.

[0064] Dehumidifier and heater The dehumidifier is used to reduce the moisture in the air, and the heater is used to adjust the temperature. The dehumidifier and the heater are connected to the processing module respectively, and the processing module controls the opening, closing or power adjustment of the dehumidifier and the heater according to the predicted temperature and humidity changes. The display module is used to display the current temperature and humidity and the system operation status. The display module is connected to the processing module to display the data calculated by the processing module and the system operation status. The execution module receives the control signal of the processing module and actually controls the start, stop and adjustment of the heater and the dehumidifier. The execution module is connected to the processing module, executes the control command of the processing module, and connects the dehumidifier and the heater at the same time.

[0065] Humidity sensors are installed indoors and outdoors to monitor the relative humidity of the environment in real time. The sensors transmit the collected humidity data to the data acquisition module, and then to the processing module through the data acquisition module. The change in indoor humidity is the main basis for starting the dehumidification mode. Temperature sensors are installed indoors and outdoors to monitor the temperature changes of the environment in real time. The output signal of the temperature sensor is transmitted to the processing module through the data acquisition module. The indoor and outdoor temperature data will help the processing module decide whether to start the heating or cooling function to ensure that the temperature remains within the preset range.

[0066] The camera is installed at a high position in the room, and the YOLO target detection model is used to identify and count the number of people in the room in real time. The image data captured by the camera is analyzed by the processing module, and the identified number of people in the room is combined with the temperature and humidity adjustment. When the number of people in the room increases, the processing module will adjust the temperature and humidity in advance according to the change in the number of people to prevent drastic fluctuations in temperature and humidity. The data acquisition module serves as an interface between the sensor and the processing module, which aggregates all data from the humidity sensor and the temperature sensor and transmits the data to the processing module. It is the key part of the entire system responsible for data transmission and collection.

[0067] The processing module is the core of the system and is responsible for analyzing all data. It receives temperature and humidity data and number of people data, and predicts the temperature and humidity changes at future moments through the built-in convolutional neural network model. The processing module compares the predicted temperature and humidity with the preset temperature and humidity range. If it exceeds the preset range, it issues a command to control the dehumidifier or heater. The dehumidifier is controlled by the processing module and is used to reduce the humidity in the air. When the processing module predicts that the humidity may exceed the set range in the future, the dehumidifier is started to control the humidity within an appropriate range. The dehumidifier is directly connected to the execution module and is controlled by the instructions of the processing module. The heater is used to heat the indoor air to prevent the temperature from being too low. When the processing module predicts that the temperature will be lower than the set range, the heater is started to keep the temperature stable. The heater is also directly controlled by the execution module and adjusts the temperature according to the prediction results of the processing module. The display module is used to display the temperature and humidity data of the current environment, the operating status of the system and other related information. Users can view the current environmental conditions and the working status of the air conditioning unit through the display module. The execution module is the interface between the processing module and the heater and dehumidifier. The control signal sent by the processing module is first transmitted to the execution module, and then the execution module controls the start and stop of the dehumidification device and the heater or adjusts the working power.

[0068] Humidity sensors, temperature sensors, and cameras monitor environmental data in real time, and the data is transmitted to the processing module through the data acquisition module. The processing module uses a convolutional neural network model to analyze current data and predict future changes in temperature and humidity. If the predicted value exceeds the preset range, the processing module will decide whether to start dehumidification or temperature adjustment. The processing module issues instructions through the execution module to control the dehumidification device or heater to ensure that the indoor temperature and humidity remain within the set range. The system continuously monitors changes in temperature and humidity, adjusts the working status of the heating and dehumidification equipment in real time, and displays the current environmental information through the display module.

[0069] Acquisition module: Humidity sensor and temperature sensor monitor the relative humidity and air temperature of the indoor environment in real time, and record the temperature and humidity data set X within the current period T indoors and outdoors c and historical temperature and humidity dataset X h , Xc ={T c , H c}, X h ={T h , H h}, where T c is the indoor and outdoor temperature data vector in the current period T, H c is the indoor and outdoor humidity data vector in the current period T, T h is the historical indoor and outdoor temperature data vector, H h Historical indoor and outdoor humidity data vectors;

[0070] Processing module: The temperature and humidity data set X for the current period T of indoor and outdoor c , indoor population and historical temperature and humidity dataset X h Input to the trained convolutional neural network model, the convolutional neural network model outputs the predicted indoor temperature value T at the future time t0 f and humidity prediction value H f ;

[0071] Adjustment module: indoor temperature prediction value T at future time t0 f and humidity prediction value H f Compare with the preset temperature range and humidity range respectively. When the temperature prediction value T f Or humidity prediction value H f When the upper or lower limit of the corresponding preset range is exceeded, the humidity adjustment mode or the temperature adjustment mode is activated;

[0072] Fine-tuning module: When the predicted temperature and humidity values ​​at the future time t0 are within the preset range, the system enters the high-precision fine-tuning mode and automatically adjusts the humidity and temperature so that the ambient temperature and humidity fluctuation ranges do not exceed ±0.1°C and ±1%, respectively;

[0073] Display module: The display shows the current temperature and humidity, and the humidity sensor and temperature sensor continue to monitor in real time.

[0074] In some embodiments, the system enters a high-precision fine-tuning mode, and automatically adjusts humidity and temperature, including: S41: calculating the temperature error ΔT=T f -T m And humidity error ΔH=H f -H m , where T m is the current indoor temperature, H m is the current indoor humidity; S42: calculate the current heating power P according to the temperature error ΔT and the humidity error ΔH h and dehumidification power P d , in, is the temperature proportional gain, is the humidity proportional gain; S43: repeat step S41 until the temperature error ΔT and the humidity error ΔH are both 0.

[0075] In some embodiments, the temperature proportional gain at time t Among them, α temperature is the adaptive coefficient, K p0 is the initial proportional gain of temperature, is the rate of change of temperature error; humidity proportional gain at time t Among them, β is the humidity adaptive coefficient, K p1 is the initial proportional gain of humidity, is the rate of change of humidity error at time t.

[0076] In some embodiments, the trained convolutional neural network model uses an improved Swish activation function f(x,N):

[0077]

[0078] Among them, x is the input of the convolutional layer, N is the number of people in the room, σ is the adjustment parameter, and e is the base of the natural logarithm.

[0079] In some embodiments, the number of people indoors is obtained by capturing the current indoor video image data in real time using a camera installed indoors, and using a YOLO target detection model to identify the number of people in the indoor video image to obtain the number of people N indoors.

[0080] The present invention provides a control method and device for dehumidification heat compensation of a constant temperature and humidity air conditioning unit, which can achieve the following beneficial technical effects:

[0081] 1. The present invention first records the temperature and humidity data set X within the current period T of indoor and outdoor c and historical temperature and humidity dataset X h ; Secondly, the temperature and humidity dataset X c , indoor population and historical temperature and humidity dataset X h Input to the trained convolutional neural network model, the convolutional neural network model outputs the indoor temperature prediction value T f and humidity prediction value H f ; Then, compare with the preset temperature range and humidity range; when the temperature prediction value T f Or humidity prediction value H fWhen the upper or lower limit of the corresponding preset range is exceeded, the humidity adjustment mode or temperature adjustment mode is started. When it is within the preset range, the system enters the high-precision fine-tuning mode; finally, the display shows the current temperature and humidity. This application adopts an improved convolutional neural network model to calculate the temperature and humidity prediction values ​​in real time and compare them with the preset range, and choose to start the humidity adjustment mode or temperature adjustment mode or high-precision fine-tuning mode; it greatly improves the accuracy and efficiency of air conditioning temperature and humidity adjustment, and greatly increases the user experience.

[0082] 2. The trained convolutional neural network model of the present invention adopts an improved Swish activation function. By introducing the number of people N in the room into the activation function, the system can dynamically perceive the changes in the indoor environment, especially the impact of the number of people on temperature and humidity. As the number of people increases, the indoor temperature and humidity usually change significantly. Traditional control methods usually only adjust according to the real-time data of the temperature and humidity sensor, and cannot respond in advance to the temperature and humidity fluctuations caused by the change in the number of people. By taking the number of people as an input parameter, the system can dynamically adjust the heating power and dehumidification power according to the current number of people in the room, improving the adaptability and response speed to complex environments. The system can quickly adapt to and predict the changes in temperature and humidity in advance, avoiding the lagging adjustment problem of traditional systems. The accuracy of the system's temperature and humidity control is significantly improved. When facing smaller environmental fluctuations, it can also accurately adjust the temperature and humidity through changes in the number of people to maintain a constant indoor environment.

[0083] 3. The high-precision fine-tuning mode of the present invention automatically adjusts humidity and temperature. By calculating the temperature error and humidity error, the heating power and dehumidification power at the current moment are calculated according to the temperature error and humidity error. The temperature proportional gain and humidity proportional gain are considered in the calculation process, which greatly improves the fine adjustment. Through the setting of high-precision mode and temperature adjustment mode and humidity adjustment mode, the problem of over-adjustment or under-adjustment is avoided, and the overall operation stability is improved.

[0084] 4. The present invention records the temperature and humidity data set X within the current period T of indoor and outdoor c and historical temperature and humidity data sets, the convolutional neural network model can analyze the changing patterns of temperature and humidity based on current and historical data, and predict the future trends of temperature and humidity. This combination not only relies on the current environmental conditions, but also takes into account the laws of past changes, thereby improving the accuracy of temperature and humidity predictions. Traditional air-conditioning systems are usually controlled based on immediate temperature and humidity feedback, which can easily lead to frequent start and stop of equipment, increase equipment wear, and shorten the service life of equipment. By using current and historical temperature and humidity data, the system can predict future temperature and humidity changes in advance and take appropriate measures to avoid frequent start and stop. For example, when historical data shows that the ambient temperature and humidity have periodic fluctuations in a short period of time, the system can intervene in advance to make smoother adjustments and reduce the frequency of overreaction.

[0085] The above is a detailed introduction to a control method and device for dehumidification heat compensation of a constant temperature and humidity air-conditioning unit. This article uses specific examples to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the core idea of ​​the present invention; at the same time, for general technical personnel in this field, according to the ideas and methods of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A control method for dehumidification heat compensation of a constant temperature and humidity air conditioning unit, characterized in that: Includes steps: S1: Use humidity sensors and temperature sensors to monitor the relative humidity and air temperature of the indoor environment in real time, and record the temperature and humidity data set X for the current period T indoors and outdoors c and historical temperature and humidity dataset X h , X c ={T c , H c }, X h ={T h , H h }, where T c is the indoor and outdoor temperature data vector in the current period T, H c is the indoor and outdoor humidity data vector in the current period T, T h is the historical indoor and outdoor temperature data vector, H h Historical indoor and outdoor humidity data vectors; S2: The temperature and humidity data set X for the current period T of indoor and outdoor c , indoor population and historical temperature and humidity dataset X h Input to the trained convolutional neural network model, the convolutional neural network model outputs the predicted indoor temperature value T at the future time t0 f and humidity prediction value H f ; S3: predicted indoor temperature value T at future time t0 f and humidity prediction value H f Compare with the preset temperature range and humidity range respectively. When the temperature prediction value T f Or humidity prediction value H f When the upper or lower limit of the corresponding preset range is exceeded, the humidity adjustment mode or the temperature adjustment mode is activated; S4: When the predicted temperature and humidity values ​​at the future time t0 are within the preset range, the system enters the high-precision fine-tuning mode and automatically adjusts the humidity and temperature so that the ambient temperature and humidity fluctuation ranges do not exceed the set thresholds respectively; S5: The display shows the current temperature and humidity, and the humidity sensor and temperature sensor continue to monitor in real time.

2. A control method for dehumidification heat compensation of a constant temperature and humidity air conditioning unit as claimed in claim 1, characterized in that: The system enters a high-precision fine-tuning mode to automatically adjust humidity and temperature, including: S41: calculating the temperature error ΔT=T f -T m And humidity error ΔH=H f -H m , where T m is the current indoor temperature, H m is the current indoor humidity; S42: calculate the current heating power P according to the temperature error ΔT and the humidity error ΔH h and dehumidification power P d , in, is the temperature proportional gain, is the humidity proportional gain; S43: repeat step S41 until the temperature error ΔT and the humidity error ΔH are both 0.

3. A control method for dehumidification heat compensation of a constant temperature and humidity air conditioning unit as claimed in claim 2, characterized in that: The proportional gain of temperature at time t Among them, α temperature is the adaptive coefficient, K p0 is the initial proportional gain of temperature, is the rate of change of temperature error; humidity proportional gain at time t Among them, β is the humidity adaptive coefficient, K p1 is the initial proportional gain of humidity, is the rate of change of humidity error at time t.

4. A control method for dehumidification heat compensation of a constant temperature and humidity air conditioning unit as claimed in claim 1, characterized in that: The trained convolutional neural network model adopts the improved Swish activation function f(x,N): Among them, x is the input of the convolutional layer, N is the number of people in the room, σ is the adjustment parameter, and e is the base of the natural logarithm.

5. A control method for dehumidification heat compensation of a constant temperature and humidity air conditioning unit as claimed in claim 1, characterized in that: The number of people in the room is determined by capturing the current indoor video image data in real time using a camera installed indoors, and using a YOLO target detection model to identify the number of people in the indoor video image to obtain the number of people N in the room.

6. A control device for dehumidification heat compensation of a constant temperature and humidity air conditioning unit, characterized in that: include: Acquisition module: Humidity sensor and temperature sensor monitor the relative humidity and air temperature of the indoor environment in real time, and record the temperature and humidity data set X within the current period T indoors and outdoors c and historical temperature and humidity dataset X h , X c ={T c , H c }, X h ={T h , H h }, where T c is the indoor and outdoor temperature data vector in the current period T, H c is the indoor and outdoor humidity data vector in the current period T, T h is the historical indoor and outdoor temperature data vector, H h Historical indoor and outdoor humidity data vectors; Processing module: The temperature and humidity data set X for the current period T of indoor and outdoor c , indoor population and historical temperature and humidity dataset X h Input to the trained convolutional neural network model, the convolutional neural network model outputs the predicted indoor temperature value T at the future time t0 f and humidity prediction value H f ; Adjustment module: indoor temperature prediction value T at future time t0 f and humidity prediction value H f Compare with the preset temperature range and humidity range respectively. When the temperature prediction value T f Or humidity prediction value H f When the upper or lower limit of the corresponding preset range is exceeded, the humidity adjustment mode or the temperature adjustment mode is activated; Fine-tuning module: When the predicted temperature and humidity values ​​at the future time t0 are within the preset range, the system enters the high-precision fine-tuning mode and automatically adjusts the humidity and temperature so that the ambient temperature and humidity fluctuation ranges do not exceed the set thresholds respectively; Display module: The display shows the current temperature and humidity, and the humidity sensor and temperature sensor continue to monitor in real time.

7. A control device for dehumidification heat compensation of a constant temperature and humidity air conditioning unit as claimed in claim 6, characterized in that: The system enters a high-precision fine-tuning mode to automatically adjust humidity and temperature, including: S41: calculating the temperature error ΔT=T f -T m And humidity error ΔH=H f -H m , where T m is the current indoor temperature, H m is the current indoor humidity; S42: calculate the current heating power P according to the temperature error ΔT and the humidity error ΔH h and dehumidification power P d , in, is the temperature proportional gain, is the humidity proportional gain; S43: repeat step S41 until the temperature error ΔT and the humidity error ΔH are both 0.

8. A control device for dehumidification heat compensation of a constant temperature and humidity air conditioning unit as claimed in claim 6, characterized in that: The proportional gain of temperature at time t Among them, α temperature is the adaptive coefficient, K p0 is the initial proportional gain of temperature, is the rate of change of temperature error; humidity proportional gain at time t Among them, β is the humidity adaptive coefficient, K p1 is the initial proportional gain of humidity, is the rate of change of humidity error at time t.

9. A control device for dehumidification heat compensation of a constant temperature and humidity air conditioning unit as claimed in claim 6, characterized in that: The trained convolutional neural network model adopts the improved Swish activation function f(x,N): Among them, x is the input of the convolutional layer, N is the number of people in the room, σ is the adjustment parameter, and e is the base of the natural logarithm.

10. A control device for dehumidification heat compensation of a constant temperature and humidity air conditioning unit as claimed in claim 6, characterized in that: The number of people in the room is determined by capturing the current indoor video image data in real time using a camera installed indoors, and using a YOLO target detection model to identify the number of people in the indoor video image to obtain the number of people N in the room.

Images