Parallel control system and method of air conditioner, air conditioner and storage medium
By combining the parallel control system and the LSTM model, the problems of slow response and high energy consumption of traditional air conditioners were solved, and rapid response and precise control of environmental changes were achieved.
Patent Information
- Application Number
- CN202511049635.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-29
- Publication Date
- 2025-09-26
AI Technical Summary
Traditional air-conditioning control systems lack multi-dimensional environmental perception, resulting in slow response, low control accuracy and high energy consumption.
A parallel control system is adopted to obtain indoor and outdoor environmental parameters through the data acquisition module, and FPGA is used for parallel preprocessing and standardization. The LSTM model is combined to perform multi-dimensional feature extraction and joint decision-making, generate accurate control strategies, and adjust the air conditioning operation through PWM and GPIO units.
It achieves rapid response to environmental changes, improves control accuracy and reduces energy consumption.
Smart Images
Figure CN120702086A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of air conditioners, and in particular relates to a parallel control system, method, air conditioner, and storage medium of an air conditioner. Background Art
[0002] Traditional air conditioning control systems typically rely on single-dimensional environmental data or independent analysis and control of multiple environmental data. These systems are slow to respond to environmental changes and lack multi-dimensional environmental perception or comprehensive consideration of both indoor and outdoor environmental factors. For example, an air conditioner is equipped with three sensors, each capturing one data stream, which is then processed sequentially. This control approach can lead to low control accuracy and an inability to effectively cope with complex and changing indoor and outdoor environments. Furthermore, control strategies based on single-sensor data often lead to unnecessary energy consumption, impacting the system's energy efficiency.
[0003] The above content is only used to assist in understanding the technical solution of the present invention and does not constitute an admission that the above content is prior art. Summary of the Invention
[0004] The purpose of the present invention is to provide a parallel control system, method, air conditioner, storage medium and computer program product for an air conditioner, so as to solve the problems of slow response of the air conditioner to environmental changes in related solutions, lack of multi-dimensional perception of the environment and comprehensive consideration of indoor and outdoor environmental factors, resulting in low control accuracy and high energy consumption, so as to achieve rapid response through comprehensive perception of indoor and outdoor environments and parallel processing, generate accurate strategies through multi-dimensional comprehensive analysis, realize rapid response to environmental changes, effectively improve control accuracy and reduce energy consumption.
[0005] The present invention provides a parallel control system for an air conditioner, comprising: a data acquisition module for acquiring indoor environmental parameters and outdoor environmental parameters; a parallel preprocessing module for performing parallel preprocessing on the indoor environmental parameters and the outdoor environmental parameters to eliminate data noise; a data standardization module for unifying the formats of the indoor environmental parameters and the outdoor environmental parameters after the parallel preprocessing to form a standardized data set; a parallel feature extraction module for performing feature analysis on the parameters included in the standardized data set to obtain dynamic change characteristics of the parameters; a joint decision-making module for performing multi-dimensional joint analysis on the dynamic change characteristics to generate a control strategy for the air conditioner; and a control module for adjusting the operating parameters and working mode of the air conditioner according to the control strategy.
[0006] In some implementations, the parallel preprocessing module utilizes parallel logic resources within the FPGA to perform parallel preprocessing on the indoor environmental parameters and the outdoor environmental parameters using a mean filtering method.
[0007] In some implementations, the parallel feature extraction module includes multiple parallel computing units, and the parallel computing units are constructed based on FPGA hardware.
[0008] In some embodiments, the joint decision module includes an LSTM model, and the LSTM model implements joint analysis of multi-dimensional data through FPGA hardware parallel acceleration.
[0009] In some embodiments, when processing data, the LSTM model defines each operation as an independent state machine, executes fixed-point operations of each LSTM unit in parallel through the arithmetic logic unit of the FPGA, and stores the model data in the built-in distributed RAM of the FPGA.
[0010] In some embodiments, the control module includes a PWM unit and a GPIO unit, the PWM unit is used to adjust the compressor speed and fan speed of the air conditioner, and the GPIO unit is used to switch between cooling mode and heating mode.
[0011] Matching the above method, the present invention provides a parallel control method for an air conditioner on the other hand, including: obtaining indoor environmental parameters and outdoor environmental parameters; performing parallel preprocessing on the indoor environmental parameters and the outdoor environmental parameters to eliminate data noise; unifying the format of the indoor environmental parameters and the outdoor environmental parameters after parallel preprocessing to form a standardized data set; performing feature analysis on the parameters contained in the standardized data set to obtain dynamic change characteristics of the parameters; performing multi-dimensional joint analysis on the dynamic change characteristics to generate a control strategy for the air conditioner; and adjusting the operating parameters and working mode of the air conditioner according to the control strategy.
[0012] In accordance with the above method, the present invention provides an air conditioner on another aspect, comprising: the above-mentioned parallel control system for the air conditioner.
[0013] In accordance with the above method, the present invention further provides a storage medium, which includes a stored program, wherein when the program is running, the device where the storage medium is located is controlled to execute the above-mentioned parallel control method for air conditioners.
[0014] The solution of the present invention uses a data acquisition module to acquire indoor and outdoor environmental parameters. A parallel preprocessing module eliminates data noise. A data normalization module forms a standardized data set. A parallel feature extraction module then extracts dynamic parameter change characteristics. Finally, a joint decision-making module generates a control strategy, which the control module uses to adjust air conditioner operation. This solves the problems of slow response, low control accuracy, and high energy consumption associated with traditional air conditioners, improving the air conditioner's response to environmental changes, enhancing control accuracy, and reducing energy consumption.
[0015] Other features and advantages of the present invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the present invention.
[0016] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 Schematic diagram of the structure of an embodiment of a parallel control system for an air conditioner according to the present invention;
[0018] Figure 2 1 is a flow chart of an embodiment of a parallel control method for an air conditioner according to the present invention;
[0019] Figure 3 Schematic diagram of hierarchical division of parallel control system;
[0020] Figure 4 Schematic diagram of the working process of the parallel control system;
[0021] Figure 5 A schematic diagram of the FPGA parallel processing architecture;
[0022] Figure 6 Schematic diagram of the LSTM implementation principle;
[0023] Figure 7 Schematic diagram of the hardware structure of the parallel control system. DETAILED DESCRIPTION
[0024] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with specific embodiments of the present invention and corresponding drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0025] According to an embodiment of the present invention, a parallel control system of an air conditioner is provided, such as Figure 1 The parallel control system of the air conditioner may include: a data acquisition module, a parallel processing module, a data standardization module, a parallel feature extraction module, a joint decision module, and a control module.
[0026] The data acquisition module is used to obtain indoor and outdoor environmental parameters. Indoor parameters include temperature, humidity, carbon dioxide concentration, etc. Outdoor parameters include temperature, humidity, UV index, air pressure, etc.
[0027] The data acquisition module includes indoor sensors and an Ethernet interface. The indoor sensors collect indoor environmental parameters, while the Ethernet interface connects to a weather server to obtain outdoor environmental parameters. Indoor sensors are deployed indoors to monitor indoor environmental parameters in real time. The Ethernet interface, based on the Ethernet protocol, connects the device to the internet or a local area network for data communication. The system connects to the weather server through the Ethernet interface to obtain outdoor environmental data, for example, by exchanging data with the server using the TCP / IP protocol.
[0028] A parallel preprocessing module is used to perform parallel preprocessing on the indoor and outdoor environmental parameters to eliminate data noise, thereby shortening the data processing cycle and ensuring the accuracy of the input data. Preprocessing methods include sliding window filtering, adaptive threshold filtering, Kalman filtering, wavelet transform denoising, and data interpolation and completion. For indoor parameters, sliding window filtering and adaptive threshold filtering are primarily used to suppress high-frequency jitter and interference from human activity. For outdoor parameters, wavelet transform and Kalman filtering are primarily used to separate complex environmental noise from actual changes. The two channels operate independently, and the processing results are simultaneously output to the data normalization module to avoid delays caused by congestion in a single channel.
[0029] In some implementations, the parallel preprocessing module utilizes parallel logic resources within the FPGA to perform parallel preprocessing on the indoor environmental parameters and the outdoor environmental parameters using a mean filtering method.
[0030] Air conditioning control needs to respond quickly to environmental changes. Traditional CPU serial processing cannot meet the real-time filtering requirements of multiple parameters. FPGA parallel computing can reduce processing delays to milliseconds, ensuring timely adjustment of control strategies.
[0031] FPGAs, as semiconductor devices with programmable internal logic circuits, support a large number of computing units operating in parallel. They offer strong parallel processing capabilities and low latency, making them suitable for real-time data processing. Parallel logic resources, a combination of programmable logic units and interconnect resources within the FPGA, are used to implement parallel algorithms. Programmable logic units include lookup tables (LUTs) and registers. The system decomposes the mean filtering algorithm into multiple subtasks, assigning them to different logic units for simultaneous execution. The mean filtering algorithm smoothes the signal and suppresses random noise by calculating the average value of data within a sliding window.
[0032] The data preprocessing process of the parallel preprocessing module is as follows: indoor and outdoor environmental parameters are input into the FPGA in parallel through the GPIO interface; the window size of the mean filtering algorithm is set according to the characteristics of different parameters. The window size affects the filtering effect. A larger window has a stronger smoothing effect but reduces the response speed; the FPGA assigns the filtering task of each parameter to an independent logic unit (such as a DSP module) for parallel execution; the filtered data is aligned by timestamp and output in parallel to the data normalization module.
[0033] Mean filtering improves the signal-to-noise ratio of sensor data and reduces misjudgments caused by noise. FPGA parallel processing also improves the real-time performance of the system.
[0034] The data normalization module is used to unify the format of the indoor and outdoor environmental parameters after parallel preprocessing to form a standardized data set. This normalization eliminates data format differences between different sources and dimensions, providing standardized input for subsequent feature extraction and analysis. For example, temperature and humidity can be mapped to a unified numerical range.
[0035] The parallel feature extraction module is used to perform feature analysis on the parameters contained in the standardized data set to obtain the dynamic change characteristics of the parameters, such as temperature change rate, humidity change trend, CO2 concentration fluctuation, etc., so that the air conditioner can quickly capture the dynamic changes of the road environment and improve the real-time decision-making.
[0036] In some implementations, the parallel feature extraction module includes multiple parallel computing units, and the parallel computing units are constructed based on FPGA hardware.
[0037] Air conditioning control needs to quickly respond to environmental changes, such as sudden changes in CO2 concentration caused by the entry and exit of people. Traditional CPU serial processing and feature extraction takes approximately 50-100ms, while FPGA parallel processing can compress this time to 5-10ms. An FPGA's parallel computing unit is an independent logic module that implements specific computational functions and can process multiple data channels simultaneously. Its types include but are not limited to basic units such as adders, multipliers, and shift registers, as well as integrated digital signal processing modules.
[0038] The feature extraction process of the parallel feature extraction module is as follows: the standardized data set is transmitted in parallel to each operation unit through the FPGA internal bus; each operation unit synchronously executes the feature extraction algorithm, such as temperature change rate, humidity fluctuation trend, carbon dioxide concentration gradient, etc.; the characteristic values output by each operation unit are aligned according to the timestamp to form a multidimensional feature vector, which is transmitted to the joint decision-making module, thereby improving the system's data processing speed and the system's response speed to environmental changes.
[0039] The joint decision-making module is used to perform multi-dimensional joint analysis on the dynamic change characteristics and generate a control strategy for the air conditioner.
[0040] In some embodiments, the joint decision module includes an LSTM model, and the LSTM model implements joint analysis of multi-dimensional data through FPGA hardware parallel acceleration.
[0041] Air conditioning environmental parameters exhibit distinct time-series characteristics, and LSTM effectively captures these time-series dependencies. Compared to traditional machine learning methods, LSTM offers higher accuracy in predicting future environmental trends. The LSTM model is a specialized recurrent neural network (RNN) that uses a gating mechanism (input gate, forget gate, and output gate) to address the vanishing gradient problem of traditional RNNs and excels at processing time-series data. Compared to CPUs / GPUs, the LSTM model's matrix operations are mapped to hardware resources such as FPGA DSP modules and BRAM (block random access memory), enabling parallel computing and delivering higher energy efficiency and lower inference latency. Joint analysis of multi-dimensional data simultaneously considers multiple environmental parameters (such as temperature, humidity, and human activity) and their interrelationships to generate comprehensive decisions, such as triggering cooling and ventilation strategies when both temperature and CO2 concentrations are detected.
[0042] The LSTM model workflow consists of input layer processing, LSTM cell execution, and output layer decision-making. The input layer processes the multidimensional feature vector output by the parallel feature extraction module and maps the input features to the hidden state space of the LSTM cell through the embedding layer. The LSTM cell execution involves a forget gate, an input gate, a cell state update, and an output gate. The forget gate determines which information in the previous hidden state should be discarded; the input gate determines which information in the current input should be added to the cell state; the cell state update combines the outputs of the forget and input gates to update the cell state; and the output gate determines which information in the current cell state should be output. The output layer decision-making involves mapping the LSTM hidden state to the control policy space through a fully connected layer.
[0043] The implementation principle of the LSTM model is as follows Figure 6As shown, data is first cleaned to remove outliers and noise. The cleaned data is then scaled to a uniform range to facilitate model training. A time series dataset is constructed, using data from the past 10 minutes to predict environmental parameter changes over the next 5 minutes. The dataset can be extended or shortened depending on the severity of environmental changes. Next, an LSTM model is designed. The input layer is defined to receive normalized environmental parameter data; the LSTM layer is defined as a stack of multiple LSTM units to capture long-term dependencies. The output layer contains the predicted future time series data. The parallel computing power of the FPGA is leveraged to accelerate the LSTM inference process. Finally, the decision logic is defined as follows: the air conditioner's cooling / heating intensity is adjusted based on the predicted indoor and outdoor temperatures and UV index; the air conditioner's dehumidification function is adjusted based on the predicted indoor and outdoor humidity; and the air conditioner's ventilation rate is adjusted based on the predicted indoor and outdoor CO2 concentrations and UV index.
[0044] Compared to traditional PID control, the LSTM model offers higher control accuracy and can effectively reduce system energy consumption through predictive control (such as pre-emptive temperature adjustment). The LSTM model can learn from different users' usage habits to achieve personalized control and improve the user experience.
[0045] In some embodiments, when processing data, the LSTM model defines each operation as an independent state machine, executes fixed-point operations of each LSTM unit in parallel through the arithmetic logic unit of the FPGA, and stores the model data in the built-in distributed RAM of the FPGA.
[0046] An independent state machine abstracts each operational step of the LSTM model (such as the input gate and forget gate calculations) into an independent state, each with clear input, output, and transition conditions. The state machine controls the pipeline execution of the LSTM units, ensuring a clear data flow and no redundant computations. The arithmetic logic unit (ALU) is a hardware unit within the FPGA that performs basic arithmetic and logic operations. It is used to execute fixed-point arithmetic operations in parallel for the small LSTM units. Fixed-point arithmetic converts floating-point numbers to integer representation, scaling the values by setting the decimal point position. Compared to floating-point arithmetic, fixed-point arithmetic consumes fewer resources and is faster. The FPGA's built-in distributed RAM is a small block of random access memory distributed between logic units, providing low access latency.
[0047] The FPGA implementation of the LSTM model involves state machine design, fixed-point number conversion, ALU parallel computation, and distributed RAM data storage. State machine design decomposes each LSTM time step into multiple independent states. For example, state 1 reads input data and the previous hidden state; state 2 performs the parallel computation of the input gate, forget gate, output gate, and candidate cell states; state 3 updates the cell state and hidden state; and state 4 outputs and stores the current result. Fixed-point number conversion quantizes 32-bit floating-point parameters to 16-bit fixed-point numbers. For example, the floating-point number 2.75 is represented as 00000010.11000000. ALU parallel computation decomposes the LSTM matrix operations into multiple ALU-performable basic operations, including input gate calculation, forget gate calculation, output gate calculation, and candidate cell state calculation. Multiple ALU units synchronize these computations, with each ALU processing a portion of the data for a different gate. Distributed RAM data storage includes weight storage and intermediate result storage. Weight storage refers to storing the quantized weight matrix in distributed RAM; intermediate result storage refers to temporarily storing the output of each state machine in distributed RAM during the calculation process for use in subsequent states.
[0048] Through parallel computing and state machine optimization, the LSTM model achieves faster inference speed on FPGAs, improving processing efficiency. Furthermore, fixed-point arithmetic combined with distributed RAM storage reduces model resource usage and lowers hardware costs.
[0049] A control module is used to adjust the operating parameters and working mode of the air conditioner according to the control strategy.
[0050] In some embodiments, the control module includes a PWM unit and a GPIO unit, the PWM unit is used to adjust the compressor speed and fan speed of the air conditioner, and the GPIO unit is used to switch between cooling mode and heating mode.
[0051] The PWM unit, a pulse width modulation unit, achieves digital control of analog signals by adjusting the duty cycle of the pulse signal, precisely controlling the speed of the compressor motor and fan. The PWM unit controls the compressor speed as follows: the LSTM model outputs the target compressor speed, and the PWM unit generates a pulse signal with a corresponding duty cycle to drive the compressor motor. The fan control process is similar to compressor control: the fan PWM signal duty cycle is adjusted based on the LSTM output to achieve wind speed regulation.
[0052] The GPIO unit is a general-purpose input / output (GPIO) unit. It can be configured by the user program as a digital input or output interface. It transmits high and low level signals to switch between cooling and heating modes. For example, if the GPIO unit outputs a high level, the four-way valve is in the first state, allowing the refrigerant discharged from the compressor to flow to the outdoor heat exchanger, achieving cooling. If the GPIO unit outputs a low level, the four-way valve is in the second state, allowing the refrigerant discharged from the compressor to flow to the indoor heat exchanger, achieving heating.
[0053] In some embodiments, the architecture of the parallel control system is as follows: Figure 3 As shown in the figure, the system consists of a data acquisition layer, an FPGA parallel processing layer, a decision layer, and an execution layer. The data acquisition layer is responsible for collecting various environmental information; the FPGA parallel processing layer efficiently processes the collected data; the decision layer generates reasonable control strategies based on the processed data; and the execution layer converts decision instructions into actual air conditioning operation operations.
[0054] In some embodiments, the parallel control system works as follows: Figure 4 As shown, it includes: steps 1 to 6.
[0055] Step 1: Environmental data collection. The system obtains indoor and outdoor environmental parameters through sensors and Ethernet.
[0056] Step 2: Data filtering, which performs noise reduction and filtering on the sensor data input.
[0057] Step 3: Protocol parsing and format conversion: the system performs protocol parsing and data format conversion on the input data of each sensor.
[0058] Step 4: Data gradient analysis, threshold judgment, and feature extraction. A parallel multiplier array is used to perform gradient analysis, threshold judgment, and other feature extraction on temperature, humidity, CO2 concentration, and other data, and quickly calculate the rate of change of temperature, humidity, and CO2 concentration.
[0059] Step 5: Lightweight neural network. Based on the lightweight neural network, FPGA hardware parallel acceleration is used to realize the joint analysis of multi-dimensional data such as temperature, humidity, and CO2 concentration change rate to generate the optimal control strategy.
[0060] Step 6: PWM adjustment and GPIO port control. The FPGA adjusts the compressor speed and fan speed through the PWM module according to the decision results, and uses the GPIO port to control the solenoid valve to achieve cooling / heating mode switching.
[0061] The multi-source heterogeneous data acquisition module can be set to execute steps 1 and 2, and the dynamic control output module can be set to execute step 6. Steps 3 to 5 are executed by the FPGA parallel processing core architecture.
[0062] In some embodiments, the FPGA parallel processing architecture is as follows: Figure 5 As shown, the system integrates multiple sensors. Temperature and humidity sensors are deployed indoors to accurately measure indoor temperature and humidity, providing data for regulating the indoor temperature and humidity balance. The temperature data is Tn, and the humidity data is Hn. A CO2 concentration sensor monitors indoor air quality in real time, providing CO2 concentration data C. To acquire outdoor environmental data, the system connects to a meteorological server via an Ethernet interface to quickly obtain environmental data such as outdoor temperature Tw, humidity Hw, and UV index UV. All sensors are connected to the FPGA's GPIO interface via serial ports.
[0063] The system data preprocessing unit makes full use of the parallel logic resources inside the FPGA and performs filtering and noise reduction on the sensor data Tn, Hn, and C. The filtering adopts the mean filtering method to effectively eliminate the delays and errors generated during the data transmission process and ensure the accuracy of the data.
[0064] The filtered data undergoes protocol parsing and format conversion, converting the raw data collected by different sensors into a unified format for subsequent processing. The parsed and converted data is recorded as Tn*, Hn*, and C*. The system integrates the indoor and outdoor environmental data and outputs one channel of data.
[0065] The FPGA feature extraction parallel array deploys multiple parallel computing modules to perform feature extraction operations such as gradient analysis and threshold determination for different types of data, such as temperature and humidity. Taking temperature data as an example, the use of a parallel multiplier array enables rapid calculation of the temperature change rate, thereby capturing dynamic temperature trends. FPGA parallel processing can simultaneously calculate multiple environmental parameters, significantly improving data processing efficiency and reducing processing time.
[0066] The decision acceleration engine incorporates a lightweight neural network and utilizes FPGA hardware parallel acceleration technology to perform joint analysis of multi-dimensional data, including current changes in temperature, humidity, CO2 concentration, and UV index. It also uses an LSTM model to predict future trends in each parameter. Combining current and future environmental parameters, it automatically weights parameters, enabling intelligent control of air conditioning functions such as cooling / heating intensity and wind speed. For example, the system can operate in a 24°C airflow mode during the day and a 28°C airflow mode at night for sleeping. This data is stored and analyzed autonomously by the machine. Because temperature varies over time and environment, while the mode remains constant, temperature is a more sensitive parameter, and its weighting is automatically determined to be greater than the mode weight.
[0067] Based on the decision logic's output, the FPGA in the dynamic control output module precisely regulates the compressor speed and fan speed via the PWM module. By adjusting the PWM signal's duty cycle, the system flexibly controls the compressor's operating intensity and the fan's air volume, achieving precise control of indoor temperature and airflow. Furthermore, the GPIO port controls the solenoid valve, enabling rapid switching between cooling and heating modes to accommodate varying environmental requirements. The predicted values output by the LSTM model guide the system's automatic parameter control. For example, if the predicted value output indicates that the temperature weight is greater than the mode weight, the system prioritizes compressor speed over fan speed. Alternatively, if the user prefers cooling at low temperatures during the day and fan air at night, the system increases the compressor speed during the day and the fan speed at night.
[0068] This solution uses an FPGA architecture to parallelize the processing of multi-dimensional perception data from indoor and outdoor environments, achieving decoupling and rapid analysis of multiple data impacts. By leveraging LSTM models and FPGA hardware parallel acceleration, it enables joint analysis of multi-dimensional data, generating optimal control strategies for the compressor and solenoid valves. This enables rapid and precise control of the air conditioner, improving user comfort while optimizing operational strategies and reducing energy consumption, resulting in significant economic and environmental benefits.
[0069] In some embodiments, the hardware structure of the parallel control system is as follows: Figure 7 As shown in the figure, the hardware uses FPGA as the core controller, connects to temperature sensor, humidity sensor, CO2 sensor and Ethernet weather server, and outputs control compressor and solenoid valve.
[0070] The technical solution of this embodiment uses a data acquisition module to acquire indoor and outdoor environmental parameters. A parallel preprocessing module eliminates data noise, and a data normalization module generates a standardized data set. A parallel feature extraction module then extracts dynamic parameter change characteristics. Finally, a joint decision-making module generates a control strategy, which the control module uses to adjust air conditioner operation. This solves the problems of traditional air conditioners, such as slow response, low control accuracy, and high energy consumption, by improving the air conditioner's response to environmental changes, enhancing control accuracy, and reducing energy consumption.
[0071] According to an embodiment of the present invention, a parallel control method for an air conditioner corresponding to the parallel control system of the air conditioner is also provided. Figure 2 FIG2 is a flow chart of an embodiment of the method of the present invention. The parallel control method of the air conditioner may include: steps S110 to S160.
[0072] In step S110 , indoor environment parameters and outdoor environment parameters are acquired.
[0073] Specifically, indoor sensors directly collect indoor parameters to ensure real-time data. Outdoor parameters are obtained through Ethernet connections to meteorological servers, supplementing macro-environmental information. This approach ensures the accuracy of indoor data while avoiding the high cost of building in-house outdoor monitoring equipment, achieving a balance between comprehensive data and cost-effectiveness.
[0074] In step S120, the indoor environment parameters and the outdoor environment parameters are pre-processed in parallel to eliminate data noise.
[0075] Preprocessing relies on FPGA hardware, leveraging its internal parallel logic resources to simultaneously apply mean filtering to indoor and outdoor parameters. For example, a sliding window mean calculation is performed on both temperature and humidity data to eliminate noise caused by sensor jitter or transmission interference. Parallel processing improves the efficiency of multi-parameter filtering, and mean filtering smoothes fluctuations while preserving data trends, ensuring data accuracy.
[0076] In step S130, the indoor environment parameters and outdoor environment parameters after parallel preprocessing are formatted in a unified manner to form a standardized data set, eliminating dimensional differences between different parameters so that the data meet the format requirements of subsequent model analysis.
[0077] In step S140 , characteristic analysis is performed on the parameters included in the standardized data set to obtain dynamic change characteristics of the parameters.
[0078] The analysis process relies on parallel computing units built on the FPGA. Multiple units work synchronously, calculating multi-dimensional features such as the temperature change rate and CO2 concentration gradient. Parallel analysis enables feature extraction speeds of milliseconds, enabling faster capture of sudden environmental changes than serial processing.
[0079] In step S150, a multi-dimensional joint analysis is performed on the dynamic change characteristics to generate a control strategy for the air conditioner.
[0080] The analysis relies on the LSTM model, which excels at capturing the temporal dependencies of parameters and achieves fast inference through FPGA hardware parallel acceleration. Each model operation is defined as an independent state machine, and fixed-point operations are performed in parallel using the FPGA's arithmetic logic unit. Data is stored in the built-in distributed RAM. For example, the LSTM input gate and forget gate parameters are calculated simultaneously, shortening single-step inference time. The LSTM's timing analysis capabilities, combined with FPGA hardware acceleration, enable the control strategy to accurately predict environmental changes, such as predicting temperature increases, while also delivering real-time decision output, addressing the lag inherent in traditional control.
[0081] In step S160, the operating parameters and working mode of the air conditioner are adjusted according to the control strategy.
[0082] The PWM unit precisely adjusts the compressor and fan speeds for precise control of temperature and humidity. The GPIO unit controls the solenoid valve to switch between cooling and heating modes, quickly responding to environmental needs, such as switching to heating when the outdoor temperature drops. PWM regulation enables stepless speed change, which is more energy-efficient than traditional fixed-frequency control. GPIO control shortens mode switching response time, improving the user experience.
[0083] Since the processing and functions implemented by the method of this embodiment basically correspond to the embodiments, principles and examples of the aforementioned methods, for any details not fully described in this embodiment, please refer to the relevant descriptions in the aforementioned embodiments and will not be repeated here.
[0084] The technical solution of this invention eliminates data noise from indoor and outdoor environmental parameters, unifies their format to form a standardized data set, then performs feature analysis to identify dynamic parameter change characteristics. Finally, based on these characteristics, a control strategy is generated to regulate air conditioner operation. This solves the problems of slow response, low control accuracy, and high energy consumption associated with traditional air conditioners, improving their response speed to environmental changes, enhancing control accuracy, and reducing energy consumption.
[0085] According to an embodiment of the present invention, an air conditioner corresponding to a parallel control system for an air conditioner is also provided. The air conditioner may include: the parallel control system for an air conditioner described above.
[0086] Since the processing and functions implemented by the air conditioner of this embodiment basically correspond to the embodiments, principles and examples of the aforementioned method, for any details not fully described in this embodiment, please refer to the relevant descriptions in the aforementioned embodiments and will not be repeated here.
[0087] The technical solution of this invention uses a data acquisition module to acquire indoor and outdoor environmental parameters. A parallel preprocessing module eliminates data noise, and a data normalization module forms a standardized data set. A parallel feature extraction module then extracts dynamic parameter change characteristics. Finally, a joint decision-making module generates a control strategy, which the control module uses to adjust air conditioner operation. This solves the problems of traditional air conditioners, such as slow response, low control accuracy, and high energy consumption, by improving the air conditioner's response to environmental changes, enhancing control accuracy, and reducing energy consumption.
[0088] According to an embodiment of the present invention, a storage medium corresponding to a parallel control system of an air conditioner is also provided, wherein the storage medium includes a stored program, wherein when the program is running, the device where the storage medium is located is controlled to execute the above-mentioned parallel control method of the air conditioner.
[0089] Since the processing and functions implemented by the storage medium of this embodiment basically correspond to the embodiments, principles and examples of the aforementioned method, for any details not fully described in this embodiment, please refer to the relevant descriptions in the aforementioned embodiments and will not be repeated here.
[0090] The technical solution of this invention eliminates data noise from indoor and outdoor environmental parameters, unifies their format to form a standardized data set, then performs feature analysis to identify dynamic parameter change characteristics. Finally, based on these characteristics, a control strategy is generated to regulate air conditioner operation. This solves the problems of slow response, low control accuracy, and high energy consumption associated with traditional air conditioners, improving their response speed to environmental changes, enhancing control accuracy, and reducing energy consumption.
[0091] In summary, it is easy for those skilled in the art to understand that, under the premise of no conflict, the above-mentioned advantageous methods can be freely combined and superimposed.
[0092] The foregoing description is merely an embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of the claims.
Claims
1. A parallel control system for air conditioning, characterized in that: include: Data acquisition module, used to obtain indoor environmental parameters and outdoor environmental parameters; A parallel preprocessing module, configured to perform parallel preprocessing on the indoor environment parameters and the outdoor environment parameters to eliminate data noise; A data standardization module is used to unify the formats of the indoor environment parameters and outdoor environment parameters after parallel preprocessing to form a standardized data set; A parallel feature extraction module is used to perform feature analysis on the parameters contained in the standardized data set to obtain dynamic change characteristics of the parameters; a joint decision-making module, configured to perform a multi-dimensional joint analysis on the dynamic change characteristics and generate a control strategy for the air conditioner; A control module is used to adjust the operating parameters and working mode of the air conditioner according to the control strategy.
2. The parallel control system for air conditioners according to claim 1, characterized in that: The parallel preprocessing module utilizes the parallel logic resources in the FPGA to perform parallel preprocessing on the indoor environment parameters and the outdoor environment parameters by adopting the mean filtering method.
3. The parallel control system for air conditioners according to claim 1, characterized in that: The parallel feature extraction module includes multiple parallel operation units, and the parallel operation units are constructed based on FPGA hardware.
4. The parallel control system for air conditioners according to claim 1, characterized in that: The joint decision-making module includes an LSTM model, and the LSTM model realizes joint analysis of multi-dimensional data through FPGA hardware parallel acceleration.
5. The parallel control system for air conditioners according to claim 4, characterized in that: When processing data, the LSTM model defines each operation as an independent state machine, executes fixed-point operations of each LSTM cell in parallel through the arithmetic logic unit of the FPGA, and stores the model data in the built-in distributed RAM of the FPGA.
6. The parallel control system for air conditioners according to claim 1, characterized in that: The control module includes a PWM unit and a GPIO unit. The PWM unit is used to adjust the compressor speed and fan speed of the air conditioner, and the GPIO unit is used to switch between cooling mode and heating mode.
7. A parallel control method for air conditioners, characterized in that: include: Get indoor and outdoor environmental parameters: performing parallel preprocessing on the indoor environment parameters and the outdoor environment parameters to eliminate data noise; Unifying the formats of the indoor environmental parameters and the outdoor environmental parameters after parallel preprocessing to form a standardized data set; Performing feature analysis on the parameters contained in the standardized data set to obtain dynamic change characteristics of the parameters; Performing a multi-dimensional joint analysis on the dynamic change characteristics to generate a control strategy for the air conditioner; The operating parameters and working mode of the air conditioner are adjusted according to the control strategy.
8. An air conditioner, characterized in that: include: The parallel control system for an air conditioner according to any one of claims 1 to 6.
9. A storage medium, characterized in that: The storage medium includes a stored program, wherein when the program is executed, the device where the storage medium is located is controlled to execute the parallel control method for air conditioners according to claim 7 .