A HVAC automatic control system with intelligent regulation

Abstract

The present invention discloses a heating, ventilation, and air conditioning (HVAC) automatic control system with intelligent regulation. First, a data acquisition module collects current environmental parameters, solar power generation, and user needs every time period T. Second, the collected data is input into a trained convolutional neural network, which outputs solar power generation, environmental parameter change trends, user needs, and control signals for future time periods. Third, a central controller sends a control signal to the HVAC system, adjusting the operating parameters of the air conditioning equipment according to the output power for the future time period, thereby automatically regulating indoor temperature and humidity. Finally, according to the instructions of the central controller, it selects whether to use solar power or grid power to drive the air conditioning equipment in the future time period. The present invention can use a convolutional neural network to dynamically adjust the operating mode and power of the air conditioning equipment according to the temperature and humidity requirements set by the user, achieving refined regulation of the indoor environment and improving user comfort.

Description

Technical Field

[0001] The present invention relates to the technical field of air conditioning, and in particular to a heating, ventilation and air conditioning automatic control system with intelligent regulation. Background Art

[0002] With rapid socioeconomic development and improved living standards, building energy consumption is increasingly accounting for a larger proportion of total energy consumption. Among these, HVAC systems, as the primary energy consumer in buildings, are attracting significant attention. Traditional HVAC systems typically operate in fixed modes, unable to dynamically adjust to environmental changes and user needs, resulting in energy waste and reduced user comfort. In recent years, the use of renewable energy sources such as solar energy has garnered widespread attention. As a clean, renewable energy source, solar energy is inexhaustible. Applying solar energy to HVAC systems not only reduces dependence on traditional energy sources but also reduces carbon emissions, offering significant environmental and economic benefits. However, solar energy is intermittent and unstable, significantly affected by factors such as sunlight intensity and weather conditions, posing challenges to the stability and reliability of solar power supply.

[0003] To improve the energy efficiency of HVAC systems and meet user comfort needs, intelligent control technologies are gradually being introduced into HVAC systems. Traditional control strategies, which often rely on rule-based methods, struggle to adapt to complex and changing environments and user needs. Simultaneously, with the development of artificial intelligence and big data technologies, machine learning methods, particularly neural network models, are being widely used in the fields of prediction and control. Neural networks possess powerful nonlinear mapping capabilities and can learn complex relationships from large amounts of historical data, providing support for system prediction and control.

[0004] In existing research, some scholars have attempted to apply recurrent neural networks such as long short-term memory networks (LSTM) to the prediction and control of HVAC systems. However, LSTM has problems such as high training complexity and low computational efficiency when processing long time series data. In contrast, convolutional neural networks (CNN) have advantages in feature extraction and computational efficiency, but traditional CNNs are mainly used in the field of image processing, and their direct application to time series data has limited effect. In addition, traditional activation functions, such as Sigmoid, Sigmoid, etc., cannot fully utilize the information of external environmental parameters. For example, in the prediction of solar power generation, light intensity is a key factor, but in the existing neural network structure, there is no direct connection between the activation function and light intensity, resulting in insufficient sensitivity of the model when dealing with light changes, affecting the prediction accuracy.

[0005] Existing intelligent air conditioning systems operate automatically in real time, causing the air conditioning frequency to fluctuate constantly, shortening the lifespan of the air conditioning hardware. They don't employ time-based control, which would extend the lifespan and save energy by maintaining constant air conditioning operation within a certain timeframe. To address these issues, it's necessary to design a new HVAC automatic control system that combines solar power and intelligent control technology to improve prediction accuracy and the system's energy efficiency. Summary of the Invention

[0006] In response to the above-mentioned problems mentioned in the prior art, the present invention provides a heating, ventilation and air conditioning automatic control system with intelligent control. First, the data acquisition module collects the current time environmental parameters, solar power generation, and user needs every time period T. Second, the collected current and historical environmental parameters, solar power generation, and user needs are input into a trained convolutional neural network, which outputs the solar power generation, environmental parameter change trends, user needs, and control signals for the future time period. Third, the central controller sends a control signal to the heating, ventilation and air conditioning system, and adjusts the operating parameters of the air conditioning equipment according to the output power of the future time period, thereby achieving automatic adjustment of the indoor temperature and humidity. Finally, according to the instructions of the central controller, it selects to use solar power or grid power supply in the future time period to drive the operation of the air conditioning equipment. The present invention can dynamically adjust the operating mode and power of the air conditioning equipment according to the temperature and humidity requirements set by the user, thereby achieving fine adjustment of the indoor environment and improving user comfort.

[0007] This application provides a heating, ventilation and air conditioning automatic control system with intelligent regulation, including the following modules:

[0008] The data acquisition module collects the current environmental parameters, solar power generation, and user needs every time period T; the environmental parameters include indoor and outdoor temperature, humidity, and light intensity, and the user needs include the temperature and humidity set by the user;

[0009] The calculation module inputs the collected current and historical environmental parameters, solar power generation, and user needs into the trained convolutional neural network. The convolutional neural network outputs the solar power generation, environmental parameter change trends, and user needs for future periods, and outputs the control signal for the future period. The control signal includes the operating mode and power of the air conditioning equipment. The convolutional neural network uses the light-modulated Sigmoid activation function f(x,L):

[0010]

[0011] Among them, x is the activation function input, k(L) is the slope function adjusted by the light intensity L, k0 is the basic slope; α is the adjustment coefficient, which controls the influence of light intensity on the slope; L max is the maximum light intensity;

[0012] The HVAC adjustment module, in which the central controller sends a control signal to the HVAC system, adjusts the operating parameters of the air conditioning equipment according to the power output in the future period, and realizes automatic regulation of indoor temperature and humidity;

[0013] The power supply selection module selects solar power or grid power supply to drive the air conditioning equipment in the future according to the instructions of the central controller.

[0014] Preferably, the solar power generation unit comprises a photovoltaic panel and a photovoltaic inverter for converting solar energy into electrical energy.

[0015] Preferably, the operating modes of the air-conditioning device include: a first mode controller controls the use of solar power to drive the air-conditioning device for cooling or heating; and a second mode controller controls switching to grid power.

[0016] Preferably, the data acquisition module includes a temperature sensor, a humidity sensor, a light intensity sensor and a user input interface device, and the user input interface device is used to obtain temperature and humidity parameters set by the user.

[0017] Preferably, the convolutional neural network includes a convolutional layer, a pooling layer and a fully connected layer.

[0018] Preferably, the calculation module further includes a data pre-processing unit for performing median filtering on the collected current and historical data.

[0019] Preferably, the HVAC adjustment module includes a variable frequency air conditioning device, which can dynamically adjust the cooling or heating power according to the control signal.

[0020] Preferably, the HVAC automatic control system further comprises an electric energy storage unit for storing electric energy when there is excess solar power generation and for providing auxiliary power supply when there is insufficient solar power supply.

[0021] Preferably, a current sensor and a voltage sensor are installed at the output end of the solar power generation unit, the current sensor is used to measure the output current of the solar power generation unit, and the voltage sensor is used to measure the output voltage of the solar power generation unit; the solar power generation amount is calculated.

[0022] Preferably, before the collected current and historical environmental parameters, solar power generation, and user needs are input into the trained convolutional neural network, the process also includes: first, performing missing value processing and outlier detection on the collected data to fill in or eliminate missing and abnormal data; second, aligning the collected data by timestamp to ensure the consistency of the data in the time dimension; finally, normalizing the data and converting the normalized data into a tensor format suitable for convolutional neural network input.

[0023] The present invention provides a heating, ventilation and air conditioning automatic control system with intelligent regulation, which can achieve the following beneficial technical effects:

[0024] 1. The present invention first collects the current environmental parameters, solar power generation, and user needs every time period T through a data acquisition module; secondly, the collected current and historical environmental parameters, solar power generation, and user needs are input into a trained convolutional neural network, and the convolutional neural network outputs the solar power generation, environmental parameter change trends, user needs, and control signals for future time periods; thirdly, the central controller sends a control signal to the HVAC system, and adjusts the operating parameters of the air-conditioning equipment according to the power output of the future time period, thereby achieving automatic adjustment of the indoor temperature and humidity; finally, according to the instructions of the central controller, it selects to use solar power or grid power in the future time period to drive the operation of the air-conditioning equipment. The present invention can dynamically adjust the operating mode and power of the air-conditioning equipment according to the temperature and humidity requirements set by the user, thereby achieving refined adjustment of the indoor environment, improving the user's comfort, and greatly enhancing the user experience.

[0025] 2. This invention uses a convolutional neural network to predict current and historical environmental parameters, solar power generation, and user demand. In particular, the introduction of a light-modulated Sigmoid activation function directly integrates light intensity into the neural network's activation process, enabling the model to dynamically adapt to changing lighting conditions. This significantly improves the accuracy of forecasts of future solar power generation and environmental parameter trends. This highly accurate prediction provides a reliable basis for optimizing air conditioning equipment operating strategies and ensuring efficient energy utilization.

[0026] 3. This invention uses the future time period control signal output by the calculation module to automatically adjust the air conditioner's operation within that time period, extending hardware life and saving power. The system dynamically adjusts the air conditioner's operating mode and power according to user-defined temperature and humidity requirements. The HVAC adjustment module utilizes variable-frequency air conditioners, enabling precise control of cooling or heating power, enabling refined adjustments to the indoor environment and enhancing user comfort.

[0027] 4. The power supply selection module intelligently selects between solar power and grid power based on predicted solar power generation and instructions from the central controller. When solar power is sufficient, solar power is prioritized, and when there is excess power, the energy storage unit stores the energy. When solar power is insufficient, the system automatically switches to grid power or uses stored energy to ensure continuous operation of the air conditioner. This flexible energy management strategy effectively reduces dependence on the grid and saves energy costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0029] Figure 1 This is a schematic diagram of the control process steps of a heating, ventilation and air conditioning automatic control system with intelligent regulation of the present invention;

[0030] Figure 2 This is a schematic diagram of a heating, ventilation and air conditioning automatic control system with intelligent regulation according to the present invention. DETAILED DESCRIPTION

[0031] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making any creative efforts shall fall within the scope of protection of the present invention.

[0032] Example 1:

[0033] In view of the above problems mentioned in the prior art, in order to solve the above technical problems, the present invention provides a heating, ventilation and air conditioning automatic control system with intelligent regulation; Figure 1 As shown, the steps are as follows: first, the data acquisition module collects the current time environmental parameters, solar power generation, and user needs every time period T; second, the collected current and historical environmental parameters, solar power generation, and user needs are input into the trained convolutional neural network, and the convolutional neural network outputs the solar power generation, environmental parameter change trend and user needs and control signals in the future time period; third, the central controller sends the control signal to the HVAC system, and adjusts the operating parameters of the air-conditioning equipment according to the power output of the future time period, thereby realizing automatic adjustment of the indoor temperature and humidity; finally, according to the instructions of the central controller, it selects to use solar power or grid power supply in the future time period to drive the operation of the air-conditioning equipment. The present invention can dynamically adjust the operating mode and power of the air-conditioning equipment according to the temperature and humidity requirements set by the user, thereby realizing fine adjustment of the indoor environment and improving the user's comfort.

[0034] The present invention provides a heating, ventilation and air conditioning automatic control system with intelligent regulation, as shown in the attached Figure 2The following hardware is shown: The data acquisition module includes sensor components: temperature sensors (indoor and outdoor), humidity sensors (indoor and outdoor), light intensity sensors, current sensors (installed at the output of the solar power generation unit), voltage sensors (installed at the output of the solar power generation unit), user input interfaces: touch screens, keypads, mobile terminal interfaces (such as mobile phone apps, tablets, etc.), data processing units: microcontrollers or embedded processors, A / D converters (for converting analog signals to digital signals), and communication interfaces (such as UART, SPI, I2C, etc.). The computing module includes: a central processing unit (CPU), a high-performance processor or industrial computer, and data storage (RAM and flash memory for storing data and models). The convolutional neural network processing unit includes: a GPU accelerator or dedicated neural network chip; model storage (for storing trained CNN models); data preprocessing unit: a digital signal processor (DSP), data buffers; communication interfaces: an Ethernet interface and wireless communication modules (such as Wi-Fi, ZigBee, LoRa, etc.). The HVAC control module includes: air conditioning equipment: a variable-frequency air conditioner main unit, indoor units, and outdoor units. Controller interface module: variable frequency drive, actuators (such as valves, motors, etc.), control signal receiving module (such as Modbus, BACnet, and other protocols). The power supply selection module includes: power switching switches including automatic transfer switches (ATS), relays, or contactors; the control unit includes a microcontroller, a voltage detection module, and a communication interface with the central controller. The solar power generation unit includes photovoltaic panels (solar panels), photovoltaic inverters, and DC distribution boxes. The energy storage unit (optional) includes a battery pack, charge and discharge controller, and energy management system (EMS); the communication network includes a local area network (LAN), a bus network (such as CAN bus, RS-485, etc.), and a wireless network (for remote monitoring and control).

[0035] Sensor connections: Temperature, humidity, and light intensity sensors are connected to the A / D converter or GPIO interface of the data processing unit via analog or digital signal lines. Outdoor sensors are installed on the exterior of the building, and indoor sensors are installed in the indoor area that needs to be monitored. Current and voltage sensors are installed at the output of the solar power generation unit to collect the generated current and voltage signals. They are connected to the A / D converter of the data processing unit via signal lines. The user input interface device is connected, and the touch screen and key panel are connected to the data processing unit via a serial communication interface (such as UART, I2C). The mobile terminal interface is connected to the central controller via a wireless communication module (such as a Wi-Fi module) to enable interaction with mobile devices.

[0036] The data acquisition module and the computation module are connected and transmitted as follows: The data processing unit sends the collected sensor data to the central controller via a communication interface (e.g., Ethernet, serial port, bus). This is done using either a wired connection (e.g., Ethernet, RS-485) or a wireless connection (e.g., Wi-Fi, ZigBee). The computation module internally connects the central controller and the convolutional neural network processing unit via a high-speed data bus (e.g., PCIe, SPI), ensuring rapid data transmission and processing. The data preprocessing unit, connected between the central controller and the convolutional neural network processing unit, preprocesses the data before passing it to the CNN model for prediction.

[0037] The control signal transmission between the calculation module and the HVAC control module is as follows: Based on the CNN prediction results, the central controller generates control signals and sends them to the HVAC control module via a communication interface. This can be done using an industrial control bus (such as Modbus or BACnet) or a digital / analog signal line. The controller interface module of the HVAC control module receives the control signals and adjusts the variable frequency drive and actuators to adjust the operating parameters of the air conditioning equipment. The control signal transmission between the calculation module and the power supply selection module is as follows: Based on the predicted solar power generation and user demand, the central controller sends instructions to the power supply selection module to control power switching. This connection can be done using a digital signal line or a communication bus. The power supply selection module switches between solar power and grid power by controlling an automatic transfer switch (ATS). The power connection between the solar power generation unit and the power supply selection module is as follows: the AC output of the photovoltaic inverter is connected to the input of the power supply selection module, and the output of the power supply selection module is connected to the power input of the air conditioning equipment. The energy storage unit is connected as follows: a battery bank is connected to the solar power generation unit and the power supply selection module via a charge and discharge controller. The energy management system (EMS) communicates with the central controller to coordinate the storage and release of energy. The communication network is connected as follows: the local area network (LAN) central controller, data processing unit, power supply selection module, energy management system, etc. are connected via Ethernet switches to form a LAN, enabling high-speed data communication. The wireless network is used to connect to mobile terminals (such as mobile phone apps) for remote monitoring and control. The central controller is equipped with a wireless communication module (such as Wi-Fi, 4G / 5G module).

[0038] The data acquisition module collects current environmental parameters, solar power generation, and user requirements at intervals of T. Environmental parameters include indoor and outdoor temperature, humidity, and light intensity, while user requirements include user-defined temperature and humidity. Indoor and outdoor temperature sensors use high-precision digital temperature sensors, such as the DS18B20. The indoor temperature sensor should be installed at a height within the room's active areas, away from air conditioning vents and direct sunlight, to ensure that the measured temperature represents the average indoor temperature.

[0039] Outdoor temperature sensor: Installed in a cool place on the exterior wall of the building, protected from direct sunlight and rain, to ensure that the measured temperature reflects the actual outdoor temperature. The sensor interface is connected to the data processing unit via a single bus (One-Wire) protocol. The data acquisition cycle collects temperature data every time period ηT (for example, every 5 minutes). The indoor and outdoor humidity acquisition sensor model uses a digital temperature and humidity sensor, such as DHT22 or SHT31. The indoor humidity sensor and the indoor temperature sensor are installed in the same location to ensure the consistency of temperature and humidity data. The outdoor humidity sensor is installed together with the outdoor temperature sensor and is waterproof and dustproof (such as adding a protective cover). The sensor interface is connected via I 2 The sensor is connected to the data processing unit via an I2C or SPI interface. The data acquisition cycle collects humidity data every time period ηT. The sensor model uses a light intensity sensor, such as BH1750 or TSL2561. It is installed on the roof or an unobstructed exterior wall, with the sensor facing the sky to avoid obstructions affecting the measurement. The sensor interface is connected via I2C or SPI interface. 2 The C interface is connected to the data processing unit. The data acquisition cycle collects light intensity data every time period ηT. The current sensor uses a Hall effect current sensor, such as the ACS712 (select an appropriate model based on the current range). The voltage sensor uses a voltage divider circuit or a voltage sensor module, such as the ZMPT101B.

[0040] Installed at the output of a solar power generation unit (photovoltaic inverter), it measures the output current and voltage. The sensor interface is connected to the A / D converter channel of the data processing unit via an analog signal line. The data acquisition cycle collects current and voltage data every time period ηT. Solar power generation and instantaneous power calculations are performed based on the collected current I and voltage U: instantaneous power P = U × I. Power generation calculation: Multiply the instantaneous power by the sampling interval Δt (i.e., time period ηT) to obtain the power generation E = P × Δt for that time period. Cumulative power generation: The power generation for each time period is accumulated to obtain the total power generation for the specified period.

[0041] The user input interface is as follows: the touch screen display is a 7-inch color LCD touch screen, which is installed in a convenient location indoors. The mobile terminal application develops a mobile phone APP that supports iOS and Android systems. Users can remotely set the temperature and humidity parameters through their mobile phones. The user enters the temperature and humidity setting interface through the touch screen or mobile phone APP and enters the desired indoor temperature and humidity values ​​(for example, temperature 24°C, humidity 50%). After the setting is completed, the user clicks the "Confirm" button, and the parameters are saved and sent to the data processing unit. The touch screen input sends the user-set parameters to the data processing unit through the serial port (such as UART) or USB interface; the user-set parameters are sent to the central controller through a wireless network (such as Wi-Fi or Bluetooth), and then passed to the data processing unit; the user parameters are updated in real time after setting, and the data processing unit obtains the latest user needs within each time period T.

[0042] The calculation module inputs the collected current and historical environmental parameters, solar power generation, and user needs into the trained convolutional neural network. The convolutional neural network outputs the solar power generation, environmental parameter change trends, and user needs for future periods, and outputs the control signal for the future period. The control signal includes the operating mode and power of the air conditioning equipment. The convolutional neural network uses the light-modulated Sigmoid activation function f(x,L):

[0043]

[0044] Among them, x is the activation function input, k(L) is the slope function adjusted by the light intensity L, k0 is the basic slope; α is the adjustment coefficient, which controls the influence of light intensity on the slope; L max is the maximum light intensity;

[0045] In some embodiments, data input and preprocessing of environmental parameters include indoor and outdoor temperature, humidity, and light intensity. Current and historical solar power generation data are also included. User requirements include user-defined target temperature and humidity. Data cleaning addresses missing values ​​and outliers to ensure data integrity and accuracy. Data synchronization aligns data with different sampling frequencies and timestamps to form a unified time series dataset. Normalization scales the data for each feature to the same numerical range (e.g., [0, 1]), eliminating dimensional differences and facilitating model training and prediction. Feature engineering constructs time series samples, using data from a specific time window (e.g., the past 24 hours) as model input to capture temporal dependencies and trends in the data. Input tensor construction organizes preprocessed data into multidimensional arrays (tensors), with each sample containing data from multiple dimensions and time steps. The time dimension reflects the length of the time series. Feature dimensions include environmental parameters, solar power generation, and user requirements. Batch processing groups multiple samples into batches to improve model training and prediction efficiency.

[0046] The architecture of a convolutional neural network includes: an input layer that receives preprocessed multidimensional input tensors. A one-dimensional convolutional layer (1D Convolutional Layer): Performs convolution operations along the time axis to extract local features and patterns in the time series. Convolutional kernels (filters): Multiple convolution kernels are applied simultaneously to the input data to learn different feature representations. The activation layer uses a light-modulated Sigmoid activation function: Based on the traditional Sigmoid activation function, it incorporates light intensity information to dynamically adjust the activation function's response with light intensity. A pooling layer (max pooling or average pooling): Reduces the dimensionality of the convolutional layer output, retaining important features and reducing computational effort. A fully connected layer expands the convolutional and pooled features into a one-dimensional vector, which is then passed through the fully connected layer to further extract high-level features. The output layer has the appropriate number of output nodes based on the prediction task. For regression tasks (such as predicting future power generation and environmental parameters), a linear activation function can be used; for classification tasks (such as determining operating mode), a softmax activation function can be used.

[0047] Implementation of the light-modulated Sigmoid activation function. The traditional Sigmoid activation function maps the input to the (0,1) interval, but its shape is fixed and cannot be adjusted according to changes in the external environment. In this embodiment, the light intensity is used as a dynamic parameter to adjust the slope of the activation function so that the sensitivity of the activation function to the input changes with the change of light intensity. The slope adjustment dynamically calculates the slope parameter of the activation function based on the current light intensity. When the light intensity is high, the slope of the activation function is increased to make the model more sensitive to input changes and adapt to the data characteristics under strong light conditions. When the light intensity is low, the slope of the activation function is reduced to prevent noise amplification and stabilize the output of the model.

[0048] In some embodiments, the base slope sets a default slope value to ensure that the model can work properly even without light modulation. The adjustment coefficient determines the degree of influence of light intensity on the slope based on experience or through model training. The activation function is applied to the output of each convolutional layer so that the nonlinear characteristics of the entire network are dynamically adjusted with the light intensity. The training data uses historical environmental parameters, solar power generation and user demand data to construct a training set. For regression tasks, the loss function uses the mean square error (MSE) as the loss function to measure the difference between the predicted value and the true value. The optimization algorithm uses an optimization algorithm with an adaptive learning rate, such as Adam or RMSProp, to accelerate the convergence of the model. Training process: Forward propagation input data is calculated through the network layer by layer to obtain the predicted output. The loss calculation calculates the prediction error according to the loss function. Backward propagation: Calculate the gradient and update the model parameters, including the slope parameters of the convolution kernel and the activation function. Involvement of light intensity In each training sample, light intensity is used as a parameter that affects the activation function and participates in the model calculation together with the input data.

[0049] In some embodiments, input data, including current and historical environmental parameters, solar power generation, and user demand, is collected in real time and preprocessed before being input into the model. The model predicts solar power generation for future periods: This predicts power generation trends over the next period. Environmental parameter trends predict future changes in indoor and outdoor temperature, humidity, and light intensity. User demand predicts potential changes in user demand (if the user has set a regularity). Control signal generation calculates the operating mode and power setpoint for the air conditioner for the next period based on the prediction results. The operating mode is selected based on solar power availability, with solar power being used whenever possible when solar power generation is sufficient. Mode 1: When the predicted solar power generation is sufficient to meet the air conditioner's power requirements, solar power is selected to operate the air conditioner for cooling or heating. Mode 2: When the predicted solar power generation is insufficient, grid power is switched to ensure normal operation of the air conditioner. Power setpoint calculation calculates the required cooling or heating power for the air conditioner based on the user-set temperature and humidity target values ​​and the predicted environmental parameters. The power setpoint is optimized based on energy costs and energy-saving strategies to avoid energy waste. The central controller sends a transmission mode control signal to the HVAC adjustment module and power supply selection module. The communication protocol uses standard industrial communication protocols such as Modbus, BACnet, or custom protocols to ensure reliable transmission of control signals. The HVAC control module adjusts the operating parameters of the air conditioning equipment based on the received operating mode and power setpoint. The power supply selection module switches the power source according to the instructions to ensure stable operation of the air conditioning equipment.

[0050] In some embodiments, the specific structure of the convolutional neural network is as follows: input layer, input data: pre-processed current and historical environmental parameters, solar power generation, user demand. Data format: multi-dimensional tensor, usually a three-dimensional data structure, including the number of samples, time steps and feature dimensions. The convolution layer includes a one-dimensional convolution layer, which performs a convolution operation on the input data along the time axis to extract local features and patterns in the time series. The convolution kernel sets multiple one-dimensional convolution kernels, each of which has a certain length (such as 3, 5, 7, etc.) to capture time features of different scales. Step size: usually set to 1 to ensure that the convolution operation does not skip any time point. Padding: Select "same" or "different" padding method as needed to control the output size.

[0051] The activation function uses the light-modulated Sigmoid activation function to control the influence of light intensity on the slope. The maximum light intensity is used for normalization. The slope of the activation function is dynamically adapted to changes in light intensity, enhancing the model's sensitivity to light changes. The function of the maximum pooling layer is to reduce the dimensionality of the output of the convolution layer, extract the most significant features, and reduce computational complexity. Pooling window size: such as 2 or 3, depending on the data characteristics. Step size: the same as the pooling window size. Fully connected layer, the first fully connected layer inputs the feature vector after convolution and pooling. The number of neurons is set as needed, such as 128 or 256 neurons. The activation function uses the Sigmoid activation function to introduce nonlinear characteristics. The second fully connected layer further extracts high-level features to prevent overfitting. The activation function is Sigmoid or other nonlinear activation functions.

[0052] The output layer outputs a continuous value of the predicted solar power generation for future time periods, using either a linear activation function or no activation function. Environmental parameter trend predictions include predicted values ​​for indoor and outdoor temperature, humidity, and light intensity. User demand predictions can be added to the corresponding output nodes if changes in user demand need to be predicted. Recommendations for air conditioning equipment operating mode and power: Based on the prediction results, threshold judgments or simple logical rules are used to generate operating modes (such as using solar power or grid power) and power setpoints.

[0053] Data preprocessing and data collection collect current and historical environmental parameters, solar power generation, and user demand from the data acquisition module. Missing value processing fills missing data, typically using forward filling or mean filling. Outlier detection identifies and processes anomalous data points to prevent them from adversely affecting the model. Data normalization scales all feature values ​​to the same range (e.g., [0, 1]) to eliminate dimensionality differences. Time series sample construction sets a time window (e.g., the past 24 hours) and combines consecutive time point data into samples. Input tensor construction organizes the preprocessed data into a tensor format suitable for CNN input, with a shape of (number of samples, time step, feature dimension). Input to the CNN model passes the input tensor to the trained CNN model. In the forward propagation convolution operation, the convolution layer performs a one-dimensional convolution on the input data to extract time series features. Dynamic slope calculation calculates the slope k(L) of the activation function based on the light intensity L of the current sample. Activation value calculation applies a light-modulated Sigmoid activation function to the convolution result to obtain a nonlinear transformed output. The pooling layer performs dimensionality reduction on the activated features and extracts key features. Feature vector expansion and flattening flatten the output of the pooling layer into a one-dimensional feature vector for processing by the fully connected layer. The fully connected layer performs linear transformation and nonlinear activation on the input feature vector to further extract high-level features. The activation function, Sigmoid, is used to improve the model's expressiveness. The predicted value calculation output layer calculates the future solar power generation, environmental parameter trends, and user demand forecasts based on the features of the fully connected layer. Operating mode and power setting: Logical judgment determines the operating mode of the air conditioner (solar power or grid power) based on the predicted solar power generation and user demand. Power calculation calculates the required cooling or heating power based on environmental parameters and user demand. Control signal generation: The control signal composition includes the operating mode and power setting value of the air conditioner. This control signal is transmitted to the HVAC control module and power supply selection module.

[0054] The HVAC adjustment module, in which the central controller sends a control signal to the HVAC system, adjusts the operating parameters of the air conditioning equipment according to the power output in the future period, and realizes automatic regulation of indoor temperature and humidity;

[0055] In some embodiments, the central controller generates a control signal based on the prediction results of the calculation module (convolutional neural network), including the operating mode and power setting value of the air-conditioning equipment. The communication interface has multiple communication methods, such as Ethernet interface, RS-485, wireless communication module (Wi-Fi, ZigBee, etc.). The variable frequency air-conditioning equipment supports cooling and heating functions and has adjustable cooling / heating power. The air-conditioning controller is built into the air-conditioning equipment and is responsible for receiving external control signals and adjusting the operating parameters of the air-conditioning. The communication interface module is used to communicate with the central controller and supports standard communication protocols. The communication protocol adopts standard industrial communication protocols, such as Modbus, BACnet or custom protocols to ensure the reliability and compatibility of data transmission. Wired connection uses RS-485 bus or Ethernet cable, which is suitable for occasions with short distances or stable environments. Wireless connection uses Wi-Fi, ZigBee or other wireless communication methods, which are suitable for occasions where wiring is difficult or flexible deployment is required.

[0056] Control Signal Generation and Result Acquisition: The central controller obtains the convolutional neural network's prediction results from the computing module, including future environmental parameter changes, solar power generation, and user demand. Operation Mode Selection: Mode 1: When the predicted solar power generation is sufficient to meet the air conditioning equipment's power requirements, solar power is selected to reduce reliance on the grid. Mode 2: When the predicted solar power generation is insufficient, grid power is switched to ensure normal operation of the air conditioning equipment. Power Setpoint Calculation: Based on the predicted indoor and outdoor temperatures and humidity, as well as the user-set target temperature and humidity, the required cooling or heating power is calculated. Energy efficiency and energy-saving strategies are considered to optimize the power setpoint to avoid excessive cooling or heating. Control Signal Transmission: Instruction Packaging: The central controller encapsulates the operating mode and power setpoint into control instructions and packages them according to the communication protocol format. Signal Transmission: Wiredly, control instructions are sent to the communication interface module of the HVAC control module via the RS-485 bus or Ethernet. Wirelessly, control instructions are sent to the wireless receiving module of the air conditioning equipment via the wireless communication module. Air Conditioning Equipment Adjustment: Control Command Reception: The air conditioning controller receives control instructions from the communication interface module. Operating parameter adjustment and mode switching: The air conditioner's operating mode, such as cooling, heating, ventilation, or standby, is switched according to instructions. Power regulation adjusts the compressor speed and fan air volume to precisely control cooling or heating power. The air conditioner controller uses a feedback mechanism to feed back information such as the device's operating status and current power to the central controller for system monitoring and further optimization. Indoor environment adjustment and temperature and humidity control: The air conditioner adjusts the indoor air temperature and humidity based on the set power and operating mode. Through precise control, the indoor temperature and humidity gradually approach the target values ​​set by the user. Continuous monitoring and adjustment: The data acquisition module continuously monitors changes in indoor and outdoor environmental parameters and user needs. Dynamic adjustment: The central controller periodically (e.g., every time period ηT) updates prediction results and control strategies. Based on the latest data and prediction results, the air conditioner's operating parameters are adjusted in real time to ensure indoor comfort.

[0057] The power selection module, based on instructions from the central controller, selects solar power or grid power for future periods to operate the air conditioning equipment. The power selection module's main components include: The automatic transfer switch (ATS): This core device switches between solar power and grid power. The control unit, with a built-in microcontroller, receives instructions from the central controller and controls the ATS's operating status. The voltage and current detection module monitors the voltage, current, and other parameters of the power supply line in real time to ensure power supply safety. The communication interface is used to communicate with the central controller and supports standard communication protocols. Solar power generation units, such as photovoltaic modules and solar panels, generate direct current (DC). PV inverters convert DC power to AC to match the power supply requirements of the air conditioning equipment. DC distribution boxes aggregate and protect DC circuits. The utility access point receives AC power from the power company. The AC distribution box manages and protects the AC circuits. The variable-frequency air conditioning equipment and the loads of the power selection module require a stable AC power supply.

[0058] In some embodiments, the power switching mechanism uses a dual-power input power selection module whose input is connected to both the output of the solar power generation unit (AC power after the inverter) and the grid. The single-load output is connected to the air conditioner, ensuring that power can be supplied to the air conditioner when either power source is operating. Priority is set: solar power is prioritized when solar power generation is sufficient, saving energy costs. Grid backup automatically switches to grid power when solar power is insufficient or unavailable, ensuring continuous operation of the air conditioner. Based on the predictions from the calculation module, the central controller determines the power supply method to use for the future period and sends the instruction to the power selection module. The power selection module responds to the instruction received from the central controller via a communication interface. Status determination checks the current solar power generation and grid power status to ensure the normal availability of the power source. Switching is executed, switching to solar power: If solar power generation is normal and the instruction requires solar power, the ATS switches to solar power. Switching to grid power: If solar power generation is insufficient or the instruction requires grid power, the ATS switches to grid power. Safety protection prevents power interruptions or surges during the switching process, ensuring the safe operation of the air conditioner.

[0059] In some embodiments, the central controller sends a command that includes the desired power supply method (solar or grid) and the effective time (immediately or at a future time period). The power supply selection module receives the command via a communication interface (such as RS-485, Ethernet, or wireless communication). It parses the command to confirm the power supply method and execution time. The solar power generation status detection module monitors the output voltage and current of the solar power generation unit in real time through the voltage and current detection module to determine the availability of solar power. The grid power supply status module detects the voltage and frequency of the grid power supply to ensure that the mains power is available. The device status assessment evaluates the availability and stability of the current power source based on the detection results. The switching condition judgment determines whether solar power is available and the central controller requires solar power, then prepares to switch to solar power. If solar power is unavailable or insufficient, or if the central controller requires grid power, then prepares to switch to grid power. The seamless power supply selection module controls the ATS to switch power, ensuring uninterrupted power during the switching process. The ATS is capable of connecting to two power sources simultaneously for a short period of time, avoiding power interruptions. The switching time control ensures that the switching operation is completed within tens of milliseconds, with no significant impact on the operation of the air conditioning equipment. After the switch is completed, the power supply status is updated and the current power source is recorded. Feedback on switching results and power supply status is sent to the central controller. Electrical isolation for safety and protection ensures physical separation between solar power and grid power, preventing reverse power flow or short circuits. Overvoltage and overcurrent protection, equipped with circuit breakers, fuses, and other protective devices, prevents damage to equipment caused by power supply anomalies. If a power supply anomaly or switching failure is detected, the power supply selection module promptly sends an alarm to the central controller and maintenance personnel.

[0060] In some embodiments, the solar power generation unit includes a photovoltaic panel and a photovoltaic inverter for converting solar energy into electrical energy.

[0061] In some embodiments, monocrystalline silicon photovoltaic panels utilize high-efficiency monocrystalline silicon solar panels, which offer advantages such as high conversion efficiency, stable performance, and long lifespan. Each photovoltaic panel has a rated power of 300 watts. Based on the power requirements of the air conditioning equipment and the available installation area, 20 photovoltaic panels are installed, for a total installed capacity of 6 kilowatts (20 panels x 300 watts). The photovoltaic panels are installed on the roof of a building or other unobstructed area to ensure maximum solar radiation. Based on the local latitude and solar altitude, the panels are installed at an optimal tilt angle (e.g., 30°) to maximize solar radiation. The panels face due south (in the Northern Hemisphere) to ensure maximum sunlight throughout the day. Electrical connections utilize a series connection to connect multiple photovoltaic panels in series to increase the output voltage. For example, connecting 10 photovoltaic panels in series produces a DC voltage of approximately 300 V. A parallel connection connects multiple series groups in parallel to increase the output current. For example, two series groups can be connected in parallel to meet higher power requirements. A DC combiner box is used to aggregate the output current of the photovoltaic panels, providing overcurrent protection and lightning protection.

[0062] In some embodiments, a grid-connected photovoltaic inverter is selected that has grid-connected functionality and can convert the DC power generated by the photovoltaic panels into AC power synchronized with the grid for use in air conditioning equipment. The rated output power of the inverter should match the total power of the photovoltaic panels; here, an inverter with a rated power of 6 kW is selected. The DC input voltage range of the inverter should accommodate the total voltage of the photovoltaic panels connected in series (e.g., 200V to 500V). Maximum power point tracking (MPPT) is a built-in MPPT function within the inverter that tracks the optimal operating point of the photovoltaic panels in real time, improving energy conversion efficiency. DC-to-AC conversion converts the DC power generated by the photovoltaic panels into standard AC power (e.g., 220V, 50Hz), meeting the power supply requirements of the air conditioning equipment. The AC power output by the grid-connected synchronous inverter is synchronized with the grid voltage and frequency, ensuring safe grid-connected operation. Overload protection automatically limits output when the output power exceeds the rated power of the inverter to prevent equipment damage. Short-circuit protection quickly cuts off output when a short circuit is detected, ensuring system safety. To prevent islanding, the inverter automatically disconnects from the grid during a power outage, preventing harm to grid maintenance personnel. The PV panels are connected to the inverter via DC cables, with the DC power output from the combiner box connected to the inverter's DC input. Cable selection should be based on the current and transmission distance, with appropriate DC cable specifications selected to minimize transmission losses. The inverter is connected to the power selection module, and the inverter's AC output is connected to the module's solar input via an AC cable. The power selection module selects solar power when sufficient solar power is available and switches to the grid when insufficient solar power is available. The PV panels, mounting brackets, and inverter must all be reliably grounded to prevent leakage and electric shock. Lightning protection devices, such as lightning rods and surge protectors, should be installed to protect the system from lightning damage.

[0063] In some embodiments, the operating modes of the air-conditioning equipment include: a first mode controller controls the use of solar power to drive the air-conditioning equipment for cooling or heating; a second mode controller controls the switch to grid power. The variable frequency air-conditioning host supports cooling and heating functions, has a high energy efficiency ratio, and can adjust the output power according to load requirements. The indoor unit and outdoor unit have a standard split air-conditioning structure, the indoor unit is responsible for regulating the indoor air, and the outdoor unit performs heat exchange. The central controller, as the core control unit of the system, is responsible for determining the operating mode and power supply method of the air-conditioning equipment based on predictions and real-time data. The communication interface between the central controller and the air-conditioning equipment supports standard protocols such as Modbus, BACnet, etc. The automatic transfer switch (ATS) is responsible for switching between solar power supply and grid power supply to ensure that the air-conditioning equipment has a continuous and stable power supply.

[0064] The first mode uses solar power. The system selects this mode when the predicted solar power generation is sufficient to meet the power requirements of the air conditioner. The central controller for power switching controls the ATS through the power selection module to switch to solar power, directly supplying the output of the solar power generation unit to the air conditioner. The air conditioner operates under solar power, providing cooling or heating, and regulating indoor temperature and humidity. Based on real-time changes in solar power generation, the central controller dynamically adjusts the power output of the air conditioner to ensure a balance between supply and demand. This fully utilizes clean energy, reduces grid power consumption, and lowers carbon emissions. Using self-generated power reduces electricity costs, especially during periods of high electricity prices.

[0065] The second mode switches to grid power. This mode is selected when solar power generation is insufficient to meet the AC unit's power needs, or when solar power is unavailable at night or on rainy days. The central controller, through the power selection module, controls the ATS to switch to grid power, ensuring continuous operation of the AC unit. The AC unit operates under grid power, continuing to cool or heat and maintain indoor comfort. Based on electricity pricing information and energy strategies, the central controller adjusts the AC unit's operating parameters to achieve energy-efficient operation. Grid power ensures the AC unit operates normally in all conditions, meeting user comfort needs. The system can flexibly switch between solar and grid power according to actual conditions, optimizing energy usage.

[0066] The central controller acquires data, including real-time solar power generation, environmental parameters, and user demand. Predictive calculations utilize convolutional neural networks to predict solar power generation and air conditioning load demand for future periods. Based on the predicted results, the central controller determines whether to operate in the first or second mode. The power selection command sends a power switching instruction to the power selection module, specifying whether to use solar power or the grid. The air conditioning control command sends operating mode and power setting instructions to the air conditioning equipment to ensure that the air conditioning equipment operates as expected. Synchronous control ensures that power switching is synchronized with the air conditioning equipment's operating mode switching to prevent equipment failure or abnormal operation due to power supply fluctuations. Real-time monitoring continuously monitors solar power generation, air conditioning equipment operating status, and indoor and outdoor environmental parameters. Power regulation dynamically adjusts the operating power of the air conditioning equipment based on power supply capacity and load demand to avoid overload or power shortage. Mode switching conditions: Switching to the first mode: When solar power generation recovers and reaches a certain threshold, the central controller can switch back to solar power. Switching to the second mode: When solar power generation drops to a point where it can no longer meet air conditioning demand, the central controller promptly switches to the grid.

[0067] In some embodiments, the data acquisition module includes a temperature sensor, a humidity sensor, a light intensity sensor, and a user input interface device, the user input interface device being used to obtain user-defined temperature and humidity parameters. The temperature sensor measures indoor and outdoor temperatures, providing information about the thermal conditions of the current environment. The indoor temperature sensor is installed in the center of the room, away from air conditioning vents and direct sunlight, to ensure that the measured temperature represents the average indoor temperature. The outdoor temperature sensor is installed in a cool, shady location on the exterior wall of a building, away from direct sunlight and rain, to obtain accurate outdoor temperatures, helping the system determine the cooling or heating needs of the air conditioner and optimize energy use.

[0068] Humidity sensors monitor indoor and outdoor humidity levels to ensure a comfortable and healthy indoor environment. Indoor humidity sensors, located adjacent to indoor temperature sensors, provide data on indoor air humidity. Outdoor humidity sensors, installed outdoors, collect external humidity information to help regulate fresh air systems and assist in adjusting the dehumidification or humidification functions of air conditioners to meet user humidity requirements.

[0069] Light intensity sensors detect ambient light intensity in real time, capturing solar radiation levels. They are typically installed on rooftops or in unobstructed locations to ensure measurement accuracy. Solar power generation prediction provides input parameters for solar power generation units, helping to predict future power generation capacity. Modulating neural network activation functions uses light intensity data to modulate the activation function of convolutional neural networks, improving the model's response to changes in light intensity. User input interface devices capture user-set temperature and humidity parameters, reflecting personalized comfort needs. Touchscreen panels are installed indoors, allowing users to directly enter or adjust temperature and humidity settings. Mobile applications allow users to remotely set and view temperature and humidity parameters via their phones or tablets. Data functions transmit user settings to a central controller, which then adjusts the operation of the air conditioning equipment based on these parameters.

[0070] In some embodiments, the convolutional neural network includes a convolutional layer, a pooling layer, and a fully connected layer. The specific structure of the convolutional neural network (CNN). The convolutional layer is responsible for extracting features from the input data. By performing a convolution operation on the input data, local correlation features are extracted to capture the spatiotemporal patterns of data such as environmental parameters, solar power generation, and user needs. The pooling layer is usually placed after the convolutional layer to reduce the dimension of the feature map, reduce the number of parameters and the amount of computation, and prevent overfitting. Common pooling operations include maximum pooling and average pooling. The fully connected layer integrates the high-level features extracted by the previous convolutional layer and pooling layer, and outputs prediction results, such as future environmental parameter change trends and control signals for air-conditioning equipment. The input layer input data format is preprocessed current and historical environmental parameters, solar power generation, and user needs, organized into a multidimensional array (tensor) that is suitable for the input requirements of the neural network. The convolution layer slides the convolution kernel (filter) on the input data, performs a convolution operation, and extracts local features. The convolution kernel size is selected according to the characteristics of the time series data (such as 1×3, 1×5). The number of convolution kernels is set to multiple convolution kernels to learn different features of the data. The activation function uses a nonlinear activation function, such as Sigmoid, to introduce nonlinear characteristics and improve the network's expressive power.

[0071] The pooling layer downsamples the output of the convolutional layer, reducing the data dimension while retaining key features. The pooling method, max pooling, takes the maximum value within the pooling window, retaining the most significant features. The pooling window size is selected appropriately based on the data characteristics (e.g., 2 or 3). The fully connected layer expands the feature map after convolution and pooling into a one-dimensional vector. This is passed through several fully connected layers to comprehensively extract features and output a prediction result. The number of neurons in one or more fully connected layers can be adjusted based on model complexity (e.g., 128 or 256). Activation functions such as sigmoid or sigmoid are commonly used. The output layer predicts future solar power generation, environmental parameter trends, user demand, and control signals for air conditioning equipment. The activation function is selected based on the task type, such as linear activation (for regression) or softmax (for classification). Data input: Preprocessed data is fed into the convolutional neural network. Feature extraction: The convolutional layer extracts local features, while the pooling layer reduces dimensionality. Feature integration: The fully connected layer comprehensively extracts high-level features. Output prediction: The output layer generates prediction results and control signals. Improve forecast accuracy: Deep learning captures complex data patterns, improving the accuracy of forecasts of future environmental changes and energy demand. Based on the forecast results, control signals are generated to output optimized air conditioning equipment operating modes and power settings.

[0072] The data preprocessing unit uses median filtering to eliminate noise and outliers, preventing them from adversely affecting model training and prediction. Current and historical data on environmental parameters, solar power generation, and user demand are collected from sensors and user interfaces. Median filtering is a nonlinear signal processing technique that replaces the central value with the median value within a window to eliminate sudden noise and outliers. The window size is selected based on the data sampling rate and noise characteristics, determining an appropriate window length (e.g., 3, 5, or 7 data points). Sliding window processing applies a sliding window across the entire data sequence, calculating the median value for each data point. Data replacement replaces the data point at the center of the window with the calculated median value. This effectively removes noise and effectively suppresses random noise and spikes. Edge feature preservation eliminates the blurring of data edges and abrupt points that occurs with mean filtering, preserving valid data information. Other preprocessing steps include: Missing value processing: Missing data is handled through interpolation, filling, or deletion. Outlier detection uses statistical methods, such as the triple standard deviation method, to identify and address anomalous data. Data alignment and synchronization: Data from different sources are aligned by timestamp to ensure temporal consistency. Data normalization scales data to the same dimensional range (e.g., between 0 and 1) to improve model training stability. Converting to tensor format organizes preprocessed data into a multidimensional array format suitable for convolutional neural network input. Based on the model's input requirements, data is structured into tensors containing information such as batch, time step, and feature dimensions. Improving model performance: Median filtering and other preprocessing methods improve data quality and ensure the accuracy and reliability of model input. Accelerating model training through normalization and data format conversion helps improve training efficiency and promote rapid model convergence.

[0073] In some embodiments, the HVAC regulation module of a variable-frequency air conditioner includes: The variable-frequency air conditioner is equipped with a variable-frequency compressor and a variable-frequency fan, capable of dynamically adjusting cooling or heating power based on load demand. Variable-frequency technology controls the operating speed of the compressor and fan by varying the power supply frequency, adjusting the motor speed. This avoids the energy waste caused by the frequent start-up and shutdown of traditional air conditioners, improving energy efficiency. It can precisely adjust cooling or heating power based on actual demand to maintain stable indoor temperature and humidity. The control module receives control signals from a central controller, including operating mode and power setpoint. Operating mode switching switches between cooling, heating, ventilation, and other modes based on commands. Power regulation adjusts compressor speed and fan air volume to achieve refined power control. The communication interface supports standard communication protocols and interfaces, such as Modbus, BACnet, and RS-485. Data exchange enables real-time exchange of data and commands with the central controller, ensuring timely and accurate control. Status monitoring provides real-time monitoring of the air conditioner's operating status, power consumption, fault information, and other information. Information feedback transmits monitoring data to the central controller to assist in optimizing control strategies. Improve user comfort by dynamically adjusting cooling or heating power to maintain indoor temperature and humidity within set ranges, providing a comfortable indoor environment. Optimize energy utilization by adjusting air conditioning power based on actual load demand and energy supply (such as solar power generation) to achieve energy-saving operation. Improve system responsiveness. The rapid response capability of variable-frequency air conditioning equipment ensures that the system can promptly adapt to changes in the environment and demand.

[0074] In some embodiments, the calculation module further includes a data preprocessing unit for performing median filtering on the collected current and historical data.

[0075] In some embodiments, the HVAC adjustment module includes a variable frequency air conditioning device capable of dynamically adjusting cooling or heating power according to a control signal.

[0076] In some embodiments, the HVAC automated control system also includes an energy storage unit for storing energy during periods of excess solar power generation and providing auxiliary power when solar power is insufficient. The energy storage unit is incorporated into the HVAC automated control system to optimize energy efficiency and ensure stable system operation even when solar power is insufficient. The energy storage unit primarily consists of a battery pack, a charge-discharge controller, and an energy management system (EMS). It stores energy during periods of excess solar power generation and provides auxiliary power to the air conditioning equipment when needed. The battery pack utilizes high-energy-density, long-life lithium-ion or lithium iron phosphate batteries. These batteries offer high charge-discharge efficiency, long cycle life, and low maintenance costs, making them suitable for large-scale energy storage applications. The total capacity of the battery pack is determined by the power requirements and backup power duration of the air conditioning equipment. For example, if the battery is required to support four hours of operation of the air conditioning equipment when solar power is insufficient and the air conditioning power is 2 kW, a battery pack of at least 8 kWh is required. The charge-discharge controller manages the battery's charging and discharging processes, ensuring optimal battery operation and extending its service life. When solar power generation exceeds the AC unit's real-time power demand, the charge controller directs the excess energy to the battery bank for charging. When solar power is insufficient and auxiliary power is required, the controller directs the battery bank to supply power to the AC unit, balancing the power supply difference.

[0077] The protection mechanism includes overcharge, over-discharge, overcurrent, and short-circuit protection functions to ensure battery and system safety. The energy management system (EMS) monitors and manages the operating status of the energy storage unit in real time, including parameters such as battery charge, voltage, current, and temperature. Intelligent scheduling optimizes charging and discharging strategies based on the central controller's instructions and system operating status. For example, charging occurs when electricity prices are low or sunlight is abundant, and discharging occurs during peak hours or when sunlight is insufficient. The communication interface exchanges data with the central controller and the power selection module to collaboratively formulate energy management strategies. The energy storage unit, in coordination with the power selection module, serves as a third power source. Under the instructions of the central controller, the power selection module can flexibly switch between solar power, grid power, and energy storage unit power, ensuring continuous operation of the air conditioning equipment. The system can operate in grid-connected mode (connected to the grid) or off-grid mode (relying entirely on solar power and batteries) as needed, enhancing energy flexibility.

[0078] In some embodiments, a current sensor and a voltage sensor are installed at the output of the solar power generation unit. The current sensor measures the output current of the solar power generation unit, and the voltage sensor measures the output voltage of the solar power generation unit; the solar power generation amount is then calculated. Regarding sensor selection and installation, the current sensor uses a Hall-effect current sensor or a shunt-type current sensor. Hall-effect sensors offer advantages such as non-contact measurement, high accuracy, and a wide bandwidth, making them suitable for measuring both DC and AC current. Installed at the DC output of the solar power generation unit, immediately before the input of the photovoltaic inverter, they measure the DC current output of the photovoltaic panel. The voltage sensor uses a resistor divider type or a photoelectric isolation type. The resistor divider type is simple and reliable, while the photoelectric isolation type offers improved safety and anti-interference capabilities. Corresponding to the current sensor, it is installed at the DC output of the photovoltaic panel to measure the output voltage of the photovoltaic system. A reasonable sampling frequency (e.g., once per second or higher) is set to ensure that dynamic changes in current and voltage are captured. The signal transmission sensor transmits the collected analog signal via a shielded signal cable to the data acquisition module, where it enters the A / D converter for digitization. Filtering filters the collected current and voltage signals to remove noise and interference, improving measurement accuracy. Calibration and calibration calibrate the sensors, compensate for system errors, and ensure the accuracy of the measured data. For power generation calculation and instantaneous power calculation, at each sampling moment, the measured voltage U(t) and current I(t) are multiplied to obtain the instantaneous power P(t) = U(t) × I(t). Instantaneous power reflects the power generation capacity of the solar power generation unit at a specific moment and can be used for real-time monitoring. Cumulative power generation calculation integrates the instantaneous power over time, accumulating the power generation increment ΔE(t) = P(t) × Δt within each sampling interval to obtain the total power generation E = ∑ΔE(t). Cumulative power generation is used to evaluate energy production over time and provide data support for energy management and analysis. Local storage includes memory in the data processing unit to store real-time and historical power and power generation data. Data backup regularly backs up data to a central controller or cloud to prevent data loss. Communication methods include wired (such as RS-485, CAN bus) or wireless (such as Wi-Fi, ZigBee) communication, sending data to the central controller or computing module. Standard communication protocols (such as Modbus, TCP / IP) are used to ensure data transmission reliability and compatibility.

[0079] In some embodiments, before inputting the collected current and historical environmental parameters, solar power generation, and user demands into a trained convolutional neural network, the following steps are performed: first, missing value processing and outlier detection are performed on the collected data to fill in or remove missing and outlier data; second, the collected data is aligned by timestamp to ensure data consistency in the temporal dimension; and finally, the data is normalized and converted into a tensor format suitable for convolutional neural network input. Before inputting the collected current and historical environmental parameters, solar power generation, and user demands into a trained convolutional neural network, the data needs to be preprocessed to ensure data quality and consistency of the model input. First, missing value processing and outlier detection are performed on the collected data. Missing data points can be filled using methods such as linear interpolation, mean filling, or nearest neighbor filling. Detected outliers are identified using statistical methods such as standard deviation screening or boxplot analysis, and then corrected or removed as appropriate to prevent the abnormal data from interfering with model training. Second, data from different sources is aligned by timestamp to ensure synchronization and consistency across the temporal dimension across the datasets. This step requires addressing time differences caused by varying sampling frequencies or transmission delays, potentially employing techniques such as time windowing and data resampling. Finally, the cleaned and aligned data is normalized to scale features of different dimensions and numerical ranges to the same scale, such as mapping values ​​to a range of 0 to 1. This standardization helps accelerate model convergence and improve prediction accuracy. The normalized data is converted into a tensor format suitable for convolutional neural network input, typically a multidimensional array, to meet the model's input data structure requirements.

[0080] The present invention provides a heating, ventilation and air conditioning automatic control system with intelligent regulation, which can achieve the following beneficial technical effects:

[0081] 1. The present invention first collects the current environmental parameters, solar power generation, and user needs every time period T through a data acquisition module; secondly, the collected current and historical environmental parameters, solar power generation, and user needs are input into a trained convolutional neural network, and the convolutional neural network outputs the solar power generation, environmental parameter change trends, user needs, and control signals for future time periods; thirdly, the central controller sends a control signal to the HVAC system, and adjusts the operating parameters of the air-conditioning equipment according to the power output of the future time period, thereby achieving automatic adjustment of the indoor temperature and humidity; finally, according to the instructions of the central controller, it selects to use solar power or grid power in the future time period to drive the operation of the air-conditioning equipment. The present invention can dynamically adjust the operating mode and power of the air-conditioning equipment according to the temperature and humidity requirements set by the user, thereby achieving refined adjustment of the indoor environment, improving the user's comfort, and greatly enhancing the user experience.

[0082] 2. This invention uses a convolutional neural network to predict current and historical environmental parameters, solar power generation, and user demand. In particular, the introduction of a light-modulated Sigmoid activation function directly integrates light intensity into the neural network's activation process, enabling the model to dynamically adapt to changing lighting conditions. This significantly improves the accuracy of forecasts of future solar power generation and environmental parameter trends. This highly accurate prediction provides a reliable basis for optimizing air conditioning equipment operating strategies and ensuring efficient energy utilization.

[0083] 3. This invention uses the future time period control signal output by the calculation module to automatically adjust the air conditioner's operation within that time period, extending hardware life and saving power. The system dynamically adjusts the air conditioner's operating mode and power according to user-defined temperature and humidity requirements. The HVAC adjustment module utilizes variable-frequency air conditioners, enabling precise control of cooling or heating power, enabling refined adjustments to the indoor environment and enhancing user comfort.

[0084] 4. The power supply selection module intelligently selects between solar power and grid power based on predicted solar power generation and instructions from the central controller. When solar power is sufficient, solar power is prioritized, and when there is excess power, the energy storage unit stores the energy. When solar power is insufficient, the system automatically switches to grid power or uses stored energy to ensure continuous operation of the air conditioner. This flexible energy management strategy effectively reduces dependence on the grid and saves energy costs.

[0085] The above is a detailed introduction to a heating, ventilation and air conditioning automatic control system with intelligent control. Specific examples are used herein 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 ideas of the present invention. At the same time, for those skilled in the art, according to the ideas and methods of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

Claims

1. A heating, ventilation and air conditioning automatic control system with intelligent regulation, characterized in that: Includes the following modules: The data acquisition module collects the current environmental parameters, solar power generation, and user needs every time period T; the environmental parameters include indoor and outdoor temperature, humidity, and light intensity, and the user needs include the temperature and humidity set by the user; The calculation module inputs the collected current and historical environmental parameters, solar power generation, and user needs into the trained convolutional neural network. The convolutional neural network outputs the solar power generation, environmental parameter change trends, and user needs in the future period, and outputs the control signal for the future period. The control signal includes the operating mode and power of the air-conditioning equipment. The convolutional neural network uses the Sigmoid activation function modulated by light. : ； ； in, is the activation function input, With light intensity The slope function of the adjustment, is the base slope; is the adjustment coefficient, which controls the influence of light intensity on the slope; is the maximum light intensity; The HVAC adjustment module, in which the central controller sends a control signal to the HVAC system, adjusts the operating parameters of the air conditioning equipment according to the power output in the future period, and realizes automatic regulation of indoor temperature and humidity; The power supply selection module selects solar power or grid power supply to drive the air conditioning equipment in the future according to the instructions of the central controller.

2. The system according to claim 1, wherein: A solar power generation unit consists of photovoltaic panels and photovoltaic inverters, which are used to convert solar energy into electrical energy.

3. The system according to claim 1 or 2, characterized in that The operating modes of the air-conditioning device include: a first mode controller controls the use of solar power to drive the air-conditioning device to cool or heat; a second mode controller controls the switch to grid power.

4. The system according to claim 1, wherein: The data acquisition module includes a temperature sensor, a humidity sensor, a light intensity sensor and a user input interface device, and the user input interface device is used to obtain temperature and humidity parameters set by the user.

5. The system according to claim 1, wherein: The convolutional neural network includes a convolutional layer, a pooling layer and a fully connected layer.

6. The system according to claim 1, wherein: The calculation module also includes a data pre-processing unit for performing median filtering on the collected current and historical data.

7. The system according to claim 1, wherein: The HVAC adjustment module includes a variable frequency air conditioning device that can dynamically adjust the cooling or heating power according to the control signal.

8. The system according to claim 1, wherein: The HVAC automatic control system further includes an electric energy storage unit for storing electric energy when there is excess solar power generation and for providing auxiliary power supply when there is insufficient solar power supply.

9. The system according to claim 2, wherein: A current sensor and a voltage sensor are installed at the output end of the solar power generation unit. The current sensor is used to measure the output current of the solar power generation unit, and the voltage sensor is used to measure the output voltage of the solar power generation unit; the solar power generation amount is calculated.

10. The system according to claim 1, wherein: Before inputting the collected current and historical environmental parameters, solar power generation, and user needs into the trained convolutional neural network, the following steps are also included: first, missing value processing and outlier detection are performed on the collected data to fill in or eliminate missing and abnormal data; second, the collected data is aligned by timestamp to ensure the consistency of the data in the time dimension; finally, the data is normalized and converted into a tensor format suitable for convolutional neural network input.

Images