An AI-powered intelligent AC / DC home appliance energy management system

Abstract

This invention discloses an AI-powered intelligent AC / DC home appliance energy management system, comprising four main modules: a multi-level collaborative monitoring module integrating AC and DC home appliance monitoring units, temperature and humidity sensors, a central processing unit, data storage, energy consumption statistics, demand response analysis, and display units via an RS485 bus; a dynamic energy prediction module consisting of dataset creation, data preprocessing, and individual prediction units; a flexible execution control module utilizing IGBT and PWM control chips, combined with voltage detection and adjustment circuits and an AI optimization unit, to achieve dynamic adjustment of the 300V-400V DC bus voltage; and an energy efficiency evaluation module covering green electricity percentage, economical operating electricity bills, comfort evaluation, and carbon emission calculation. The system achieves unified energy management and flexible control of AC and DC home appliances.

Description

Technical Field

[0001] This invention belongs to the field of smart home technology, specifically relating to an AI-powered intelligent AC / DC home appliance energy management system. Background Technology

[0002] As the global energy structure accelerates its transformation towards cleaner and lower-carbon energy, the efficient utilization of renewable energy and the deep decarbonization of the building sector have become key pathways to achieving the "dual carbon" goal. Residential buildings, as a crucial end-user of energy consumption, require electrification and DC power conversion not only as a vital support for enhancing the local utilization of renewable energy, but also as a strategic breakthrough for promoting green and low-carbon development of residential buildings and contributing to ecological civilization construction. Currently, residential energy systems are still dominated by AC power supply. However, with the integration of a high proportion of renewable energy, traditional AC systems face problems such as low power conversion efficiency, high harmonic interference, and poor compatibility with distributed photovoltaic / energy storage devices, making it difficult to meet the future demands for flexible interaction and efficient operation of residential energy systems. In contrast, DC power supply systems, with their advantages of improved energy efficiency, simplified equipment structure, and enhanced source-storage-load synergy, are considered a core technology direction for building zero-carbon homes. However, their large-scale promotion is still constrained by the lack of core hardware equipment and systematic management solutions.

[0003] Currently, DC technology for residential buildings faces two core challenges: First, the market lacks industrially viable energy management systems that match DC home appliances. Although DC energy storage appliances (such as DC air conditioners and integrated photovoltaic-storage appliances) are considered key to improving energy efficiency and flexibility, their technical requirements in areas such as wide voltage adaptability, dynamic power response, and interoperability differ fundamentally from traditional AC appliances, and a unified intelligent energy management approach has not yet been established. Second, there is a lack of systematic solutions for real-time forecasting, multi-device coordinated control, accurate metering, and energy efficiency evaluation in DC residential energy systems. For example, DC systems need to cope with multiple uncertainties such as fluctuations in photovoltaic output, dynamic switching of energy storage charging and discharging, and random changes in load demand. However, existing energy management systems are mostly designed based on steady-state AC scenarios, making it difficult to achieve closed-loop optimization of the entire "prediction-monitoring-execution-evaluation" chain in DC scenarios, resulting in limited system efficiency and user-side responsiveness. Summary of the Invention

[0004] To address the shortcomings of the existing technologies, this application provides an AI-powered intelligent AC / DC home appliance energy management system.

[0005] This application proposes an AI-powered intelligent AC / DC home appliance energy management system, which includes a multi-level collaborative monitoring module, a dynamic energy prediction module, a flexible execution control module, and an energy efficiency evaluation module. The multi-level collaborative monitoring module is used to collect real-time voltage, current, and power data of AC and DC household appliances, as well as ambient temperature and humidity data, via RS485 bus; and connects to a local LCD screen via SPI interface to dynamically display the power consumption, energy consumption trend, and device response status of the appliances, and record historical monitoring data. The dynamic energy prediction module is used to predict weather based on LSTM neural network, predict grid peak and off-peak demand based on ARIMA model, predict photovoltaic power generation based on random forest algorithm, and predict user electricity consumption habits based on cluster analysis; the model parameters and intermediate data are stored in the flash memory and static random access memory of the central processing unit, and the prediction periods are 1 hour, 6 hours, 2 hours and 24 hours respectively. The flexible execution control module is used to communicate with DC home appliances via RS485 bus, adjust the DC bus voltage in real time, and use IGBT module and PWM control chip for closed-loop voltage regulation, based on bus voltage regulation and AI optimization for flexible control of DC home appliances. The energy efficiency evaluation module is used to calculate the percentage of green electricity, time-of-use electricity bills, PMV comfort level and carbon emissions, and displays a circular progress bar, bar chart and dynamic curve on the LCD screen through the SPI interface.

[0006] In some embodiments, the multi-level collaborative monitoring module includes an AC appliance monitoring unit, a DC appliance monitoring unit, a data transmission interface, a temperature and humidity sensor, a data transmission line, a central processing unit, a data storage unit, an energy consumption statistics unit, a demand response analysis unit, and a display unit connected in parallel via an RS485 bus. The AC appliance monitoring unit includes an electromagnetic induction current transformer, a voltage sampling and power calculation chip, which is used to collect voltage, current and power data of AC appliances in real time. The DC appliance monitoring unit includes a Hall effect current sensor, a differential voltage sampler, and a microcontroller, used to collect current, voltage, and power data of DC appliances. The data transmission interface uses an RS485 bus to transmit the acquired data to the data processing layer. The temperature and humidity sensor uses a combination of capacitive humidity sensing element and thermistor to measure indoor ambient temperature and humidity. The data transmission line converts analog signals into digital signals, and the data transmission line is connected to the data processing layer via an RS485 bus. The LCD screen of the display unit is connected to the central processing unit via an SPI interface; The data storage unit includes a storage chip and an SD card interface. The storage chip has a dedicated cache area for storing the input data and intermediate calculation results of the LSTM weather forecasting model, the ARIMA power grid demand forecasting model, and the random forest photovoltaic forecasting model. The dedicated cache area is divided into four physically isolated storage partitions, including the first partition for caching real-time and historical data from weather sensors; the second partition for caching historical data on grid load and photovoltaic power generation; the third partition for storing the calculation results of the intermediate layers of the LSTM neural network; and the fourth partition for storing the output results of the decision tree of the random forest algorithm. The energy consumption statistics unit uses a sliding window algorithm to accumulate home appliance energy consumption data; The demand response analysis unit is used to calculate the demand response time based on the wavelet transform algorithm.

[0007] In some embodiments, both the AC appliance monitoring unit and the DC appliance monitoring unit integrate a power calculation chip and are connected in parallel using an RS485 bus topology. The output of the temperature and humidity sensor is connected to an RS485 signal converter to convert the analog signal into a differential digital signal for transmission.

[0008] In some embodiments, the RS485 bus pins of the AC appliance monitoring unit, DC appliance monitoring unit, and energy consumption statistics unit are defined as follows: Pin 1: A+ is defined as the positive terminal of the differential signal; Pin 2: B- is defined as the negative terminal of the differential signal; Pin 3: GND is defined as signal ground; Pin 4: VCC is defined as the 5V power supply terminal; The RS485 interface of the central processing unit is connected in parallel with the AC home appliance monitoring unit, DC home appliance monitoring unit and energy consumption statistics unit in a bus topology. The central processing unit, as the master device, provides differential signals and 5V power to the slave devices through pins 1-4.

[0009] In some embodiments, the SPI interface pins of the display unit are defined as follows: Pin 5: SCLK is defined as the clock signal line; Pin 6: MOSI is defined as master output / slave input line; Pin 7: MISO is defined as the master input / slave output line; Pin 8: CS is defined as the chip select signal line; Pin 9: GND is defined as signal ground; The central processing unit is directly connected to the LCD screen point-to-point via the SPI interface. The central processing unit communicates with the LCD screen via pin 8, and transmits synchronous clock and bidirectional data signals via pins 5-9.

[0010] In some embodiments, the I2C interface pins of the data storage unit are defined as follows: Pin 10: SDA is defined as the Central Processing Unit data line; Pin 11: SCL is defined as the central processing unit clock line; Pin 12: GND is defined as signal ground; Pin 13: VCC is defined as the 3.3V power supply terminal; The SPI pins of the SD card interface are defined as follows: Pin 14: SCLK, 15: MOSI, 16: MISO, 17: CS, 18: GND, 19: VCC; The central processing unit (CPU) connects to the memory chip in a single master-slave topology via pins 10-13 of the I2C interface. The CPU accesses the slave device through a unique address. It connects to the SD card in a point-to-point topology via pins 14-19 of the SPI interface. The CPU communicates via strobe pin 17.

[0011] In some embodiments, the PCIe interface pins of the demand response analysis unit are defined as follows: Pin 24: CLK is defined as the clock signal; Pins 25: RX+ and 26: RX- are defined as receive differential pairs; Pins 27: TX+ and 28: TX- are defined as transmit differential pairs; Pin 29: GND is defined as signal ground; Pin 30: VCC is defined as the 12V power supply terminal; The central processing unit is directly connected to the demand response analysis unit via a PCIe bus.

[0012] In some embodiments, the dynamic energy prediction module includes a dataset establishment unit, a data preprocessing unit, and a single-item prediction unit; The dataset creation unit is used to acquire and construct weather datasets, power grid datasets, photovoltaic power generation datasets, and user electricity consumption datasets. The data preprocessing unit is used to clean and standardize the multi-source data collected by the dataset establishment unit. The single-item prediction unit is used to predict weather changes, power demand during peak and off-peak hours of the power grid, photovoltaic power generation, and users' future electricity consumption habits within a specific future time period. It compares the deviation between the real-time monitoring of actual energy data and the predicted data, and provides feedback to the user when the deviation exceeds a set threshold.

[0013] In some embodiments, the flexible execution control module includes a DC bus voltage regulation circuit that communicates with the central processing unit via a remote terminal unit protocol. The circuit includes a voltage detection circuit and a voltage regulation circuit to achieve dynamic regulation of the 300V-400V bus voltage. The working process of the voltage regulation circuit is as follows: The central processing unit outputs 12-bit digital voltage commands to the DAC converter via GPIO; The analog signal output from the DAC is amplified by an operational amplifier and then isolated from the input PWM controller by an optocoupler. The PWM controller adjusts the duty cycle to drive the IGBT module, while the bus voltage is fed back to the ADC pin of the central processing unit through the voltage divider circuit to form a closed-loop control. The flexible execution control module incorporates a deep learning neural network algorithm. Based on preprocessed data, it establishes a multi-objective optimization model for energy saving and comfort, and dynamically adjusts the bus voltage according to real-time data. Each DC appliance has a built-in communication module and voltage control interface.

[0014] In some embodiments, the energy efficiency evaluation module further includes: The green electricity percentage calculation unit is used to connect the green electricity metering module and the total electricity metering module via RS485 bus, and execute division instructions to calculate the real-time green electricity percentage. The carbon emission calculation unit is used to read the preset carbon emission factor of the corresponding regional power grid from flash memory and perform a multiplication instruction with the energy consumption data to generate the carbon emission amount.

[0015] The beneficial effects of this plan are: This research focuses on the core technological bottlenecks of AI-powered intelligent AC / DC home appliance energy management systems. It revolves around four key aspects: multi-level collaborative supervision, dynamic energy forecasting, flexible execution control, and energy efficiency evaluation. By combining AI for dynamic energy forecasting and flexible voltage regulation, it participates in grid demand-side response, improving energy utilization efficiency. It is compatible with both AC and DC home appliances, adapting to the uncertainties of DC systems and addressing the lack of a matching management system for DC home appliances. It achieves multi-level collaborative supervision and intelligent feedback, providing convenient and comprehensive energy management information and optimization suggestions. It facilitates the consumption of renewable energy, quantifies carbon emissions and energy-saving effects, and promotes green and low-carbon development. Finally, it constructs an integrated intelligent management system, breaking through core technological bottlenecks and providing a DC residential energy management system solution. Attached Figure Description

[0016] Figure 1 This is a system principle block diagram of this solution.

[0017] Figure 2 This is a schematic diagram of data transmission in this scheme.

[0018] In the diagram: 1. Multi-level collaborative monitoring module; 11. Central processing unit; 12. DC appliance monitoring unit; 13. AC appliance monitoring unit; 14. Data storage unit; 15. Energy consumption statistics unit; 16. Demand response analysis unit; 17. Display unit; 2. Dynamic energy prediction module; 21. Data set establishment unit; 22. Data preprocessing unit; 23. Single prediction unit; 3. Flexible execution control module; 31. Voltage detection circuit; 32. Voltage regulation circuit; 33. AI optimization unit; 4. Energy efficiency evaluation module; 41. Green electricity percentage calculation unit; 42. Economic operation electricity bill calculation unit; 43. Comfort evaluation unit; 44. Carbon emission calculation unit. Detailed Implementation

[0019] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein; rather, these embodiments are provided so that a more thorough understanding of the invention can be achieved and that the full scope of the invention can be conveyed to those skilled in the art.

[0020] This application proposes an AI-powered intelligent AC / DC home appliance energy management system, such as... Figure 1-2 As shown, it includes a multi-level collaborative supervision module 1, a dynamic energy prediction module 2, a flexible execution control module 3, and an energy efficiency evaluation module 4; The multi-level collaborative monitoring module 1 is used to collect real-time voltage, current and power data of AC and DC home appliances, as well as ambient temperature and humidity data, via RS485 bus; and to connect to a local LCD screen via SPI interface to dynamically display the power, energy consumption trend and equipment response status of home appliances, and record historical monitoring data. In some embodiments, the multi-level collaborative monitoring module 1 includes a DC home appliance monitoring unit 13, a data transmission interface, a temperature and humidity sensor, a data transmission line, a central processing unit 11, a data storage unit 14, an energy consumption statistics unit 15, a demand response analysis unit 16, and a display unit 17 connected in parallel via an RS485 bus. The DC appliance monitoring unit 13 includes an electromagnetic induction current transformer, a voltage sampling and power calculation chip, which is used to collect voltage, current and power data of AC appliances in real time. The DC home appliance monitoring unit 12 includes a Hall effect current sensor, a differential voltage sampler, and a microcontroller, used to collect current, voltage, and power data of DC home appliances; The data transmission interface uses an RS485 bus to transmit the acquired data to the data processing layer. The temperature and humidity sensor uses a combination of capacitive humidity sensing element and thermistor to measure indoor ambient temperature and humidity. The data transmission line converts analog signals into digital signals, and the data transmission line is connected to the data processing layer via an RS485 bus. The LCD screen of the display unit 17 is connected to the central processing unit 11 via an SPI interface. The LCD screen dynamically displays the power consumption of each appliance, ambient temperature and humidity, displays daily / weekly / monthly energy consumption trends in a bar chart, and shows the current response status (peak / flat / valley) of each appliance in indicator lights. It also includes a fault warning light. The local display screen updates real-time data every 2 seconds and refreshes statistical charts every 10 minutes. When an abnormal power consumption is detected (exceeding 120% of the rated power), the red indicator light in the alarm zone flashes.

[0021] The data storage unit 14 includes a storage chip and an SD card interface. The storage chip is divided into a dedicated cache area for storing the input data and intermediate calculation results of the LSTM weather forecasting model, the ARIMA power grid demand forecasting model, and the random forest photovoltaic forecasting model. The dedicated cache area is divided into four physically isolated storage partitions, including the first partition for caching real-time and historical data from weather sensors; the second partition for caching historical data on grid load and photovoltaic power generation; the third partition for storing the calculation results of the intermediate layers of the LSTM neural network; and the fourth partition for storing the output results of the decision tree of the random forest algorithm. The energy consumption statistics unit 15 uses a sliding window algorithm to accumulate home appliance energy consumption data; The demand response analysis unit 16 is used to calculate the demand response time based on the wavelet transform algorithm.

[0022] In some embodiments, both the DC appliance monitoring unit 13 and the DC appliance monitoring unit 12 integrate a power calculation chip and are connected in parallel using an RS485 bus topology. The output of the temperature and humidity sensor is connected to an RS485 signal converter to convert the analog signal into a differential digital signal for transmission.

[0023] In some embodiments, the RS485 bus pins of the DC appliance monitoring unit 13, the DC appliance monitoring unit 12, and the energy consumption statistics unit 15 are defined as follows: Pin 1: A+ is defined as the positive terminal of the differential signal; Pin 2: B- is defined as the negative terminal of the differential signal; Pin 3: GND is defined as signal ground; Pin 4: VCC is defined as the 5V power supply terminal; The RS485 interface of the central processing unit 11 is connected in parallel with the DC home appliance monitoring unit 13, the DC home appliance monitoring unit 12 and the energy consumption statistics unit 15 in a bus topology. The central processing unit 11, as the master device, provides differential signals and 5V power to the slave devices through pins 1-4.

[0024] In some embodiments, the SPI interface pins of the display unit 17 are defined as follows: Pin 5: SCLK is defined as the clock signal line; Pin 6: MOSI is defined as master output / slave input line; Pin 7: MISO is defined as the master input / slave output line; Pin 8: CS is defined as the chip select signal line; Pin 9: GND is defined as signal ground; The central processing unit 11 is directly connected to the LCD screen point-to-point via the SPI interface. The central processing unit 11 selects the display screen for communication via pin 8, and transmits synchronous clock and bidirectional data signals via pins 5-9. During data transmission, the CS signal is used to select the LCD screen to communicate with.

[0025] In some embodiments, the I2C interface pins of the data storage unit 14 are defined as follows: Pin 10: SDA is defined as the data line for the central processing unit 11; Pin 11: SCL is defined as the clock line for the central processing unit 11; Pin 12: GND is defined as signal ground; Pin 13: VCC is defined as the 3.3V power supply terminal; The SPI pins of the SD card interface are defined as follows: Pin 14: SCLK, 15: MOSI, 16: MISO, 17: CS, 18: GND, 19: VCC; The central processing unit 11 connects to the memory chip in a single master-slave topology via pins 10-13 of the I2C interface. The central processing unit 11 accesses the slave device through a unique address. It connects to the SD card in a point-to-point topology via pins 14-19 of the SPI interface. The central processing unit 11 communicates via pin 17.

[0026] Among them, SDA (data line): used for data transmission, pin number 10 on the central processing unit 11 side, and pin number 1 on the memory chip side; SCL (clock line): provides a synchronous clock signal. The pin number on the central processing unit 11 side is 11, and the corresponding pin number on the memory chip side is 2. GND: Signal ground, pin number 12 on the central processing unit 11 side, and pin number 3 on the memory chip side; VCC (3.3V): This supplies power to the memory chip. The pin number on the central processing unit 11 side is 13, and the corresponding pin number on the memory chip side is 4. SD card interface (SPI interface) pin definitions: SCLK (clock line): Synchronization clock signal, pin number 14 on the central processing unit 11 side, and pin number 1 on the SD card interface side; MOSI (Master Output, Slave Input): Data transmission line, pin number 15 on the central processing unit 11 side, and pin number 2 on the SD card interface side; MISO (Master Input, Slave Output): Data receive line, pin number 16 on the central processing unit 11 side, and pin number 3 on the SD card interface side; CS (Chip Select Line): Chip select signal, pin number 17 on the central processing unit 11 side, and pin number 4 on the SD card interface side; GND: Signal ground, pin number 18 on the central processing unit 11 side, and pin number 5 on the SD card interface side; VCC (3.3V): Powers the SD card. Pin number 19 is on the central processing unit 11 side, and pin number 6 is on the SD card interface side. The memory chip (I2C interface) adopts a multi-master-slave topology, but in this system, only the central processing unit 11 acts as the master device, and the memory chip acts as the slave device, communicating through a unique address. The SD card interface (SPI interface) adopts a point-to-point topology. The central processing unit 11 is directly connected to the SD card interface via the SPI bus, and communication with the SD card is controlled by the CS signal.

[0027] In some embodiments, the PCIe interface pins of the demand response analysis unit 16 are defined as follows: Pin 24: CLK is defined as the clock signal; Pins 25: RX+ and 26: RX- are defined as receive differential pairs; Pins 27: TX+ and 28: TX- are defined as transmit differential pairs; Pin 29: GND is defined as signal ground; Pin 30: VCC is defined as the 12V power supply terminal; The central processing unit 11 is directly connected to the demand response analysis unit 16 via a PCIe bus.

[0028] Among them, CLK (clock signal): provides a high-speed clock, pin number 24 on the side of central processing unit 11, and pin number 1 on the side of demand response analysis unit 16; RX (Receive Differential Pair): Receives differential data signals. The pin numbers on the central processing unit 11 side are 25 (RX+) and 26 (RX-), and the corresponding pin numbers on the demand response analysis unit 16 side are 2 (RX+) and 3 (RX-). TX (Transmit Differential Pair): Transmits differential data signals. The pin numbers on the central processing unit 11 side are 27 (TX+) and 28 (TX-), and the corresponding pin numbers on the demand response analysis unit 16 side are 4 (TX+) and 5 (TX-). GND: Signal ground, pin number 29 on the central processing unit 11 side, and pin number 6 on the demand response analysis unit 16 side; VCC (12V): Powers the demand response analysis unit 16. The pin number on the central processing unit 11 side is 30, and the corresponding pin number on the demand response analysis unit 16 side is 7. A point-to-point topology is adopted, and the central processing unit 11 and the demand response analysis unit 16 are directly connected via the PCIe bus.

[0029] The dynamic energy prediction module 2 is used to predict weather based on LSTM neural network, predict grid peak and off-peak demand based on ARIMA model, predict photovoltaic power generation based on random forest algorithm, and predict user electricity consumption habits based on cluster analysis; the model parameters and intermediate data are stored in the flash memory and static random access memory of the central processing unit, and the prediction periods are 1 hour, 6 hours, 2 hours and 24 hours respectively. In some embodiments, the dynamic energy prediction module 2 includes a dataset establishment unit 21, a data preprocessing unit 22, and a single-item prediction unit 23; The dataset creation unit 21 is used to acquire and construct weather datasets, power grid datasets, photovoltaic power generation datasets, and user electricity consumption datasets. Weather dataset: Real-time and future weather data are obtained through meteorological satellite receiving devices, meteorological station data interfaces and third-party meteorological service APIs, including temperature sensors, humidity sensors, wind speed sensors, light intensity sensors and weather condition recognition devices.

[0030] Power Grid Dataset: Connects to the power grid dispatch center via a power data communication interface to obtain historical power demand data during peak and off-peak periods, real-time power grid load data, and power grid generation plan data.

[0031] Photovoltaic power generation dataset: Deployed in distributed photovoltaic systems, including photovoltaic panel temperature sensors, inverter efficiency monitoring modules, and power generation data acquisition devices, used to establish historical power generation data and equipment operating status data.

[0032] User electricity consumption dataset: Through the power / energy consumption acquisition module of the home appliance energy consumption data acquisition layer of the AI-HEMS system, real-time power consumption, voltage, current and other data of each appliance in the user's home are collected. At the same time, it is equipped with a user information input terminal to collect subjective data such as home equipment update information and changes in daily routines filled in by the user.

[0033] The data preprocessing unit 22 is used to clean and standardize the multi-source data collected by the dataset establishment unit 21. The collected multi-source data is cleaned, including removing redundancy and filling in missing values. Interpolation algorithms and regression models are used to fill in missing data. Data of different types and magnitudes are converted into a unified format and range, including analog-to-digital conversion, data normalization, and data format conversion.

[0034] The single-item prediction unit 23 is used to predict weather changes, power demand during peak and off-peak hours of the power grid, photovoltaic power generation, and users' future electricity consumption habits within a specific future time period. It compares the deviation between the real-time monitoring of actual energy data and the predicted data, and provides feedback to the user when the deviation exceeds a set threshold.

[0035] Weather Forecasting: Based on the LSTM neural network deep learning algorithm, combined with historical weather data and real-time meteorological conditions, this system predicts weather changes within a specific future time period. A central processing unit (CPU) is employed. 512KB of flash memory is allocated to store model parameters such as weights and biases of the LSTM neural network. 64KB of static random access memory (SRAM) is allocated in a K4S561632 chip to cache input data such as historical weather data and real-time meteorological conditions. 128KB of SRAM is allocated to store intermediate results during the LSTM neural network computation. 16KB of SRAM is allocated to store the predicted weather changes for the specific future time period. The computation cycle is set to one hour, and the LSTM neural network algorithm is burned into the CPU's flash memory as a binary executable program.

[0036] Peak-valley demand forecasting for the power grid: An ARIMA time series model is used, combined with historical peak-valley demand data, real-time grid load, and influencing factors such as holidays and special events, to predict future power demand during peak-valley periods. A central processing unit (CPU) is employed. 256KB of flash memory is allocated to store parameters such as the order and coefficients of the ARIMA model, as well as some important historical peak-valley demand data. 32KB of static random access memory (SRAM) is allocated in the K4S561632 chip to cache real-time grid load, holiday, and special event data; 64KB of SRAM is allocated to store intermediate variables and temporary results during the ARIMA model calculation; and 8KB of SRAM is allocated to store the predicted future peak-valley demand results. The ARIMA time series model is run every 6 hours to predict peak-valley demand. Photovoltaic power generation prediction: Utilizing the random forest machine learning algorithm, this method comprehensively considers weather forecast results, photovoltaic power plant equipment parameters, and historical power generation data to predict photovoltaic power output. A central processing unit (CPU) 11 is selected. 384KB of flash memory is allocated to store parameters such as the decision tree structure and node splitting thresholds of the random forest model. 48KB of static random access memory (SRAM) is allocated in the K4S561632 chip to store input information such as weather forecast results, photovoltaic power plant equipment parameters, and historical power generation data. 96KB of SRAM is allocated to store the output results of each decision tree during the random forest algorithm calculation and intermediate data from the integrated calculation. 12KB of SRAM is allocated to store the predicted photovoltaic power output. The random forest machine learning algorithm is run every 2 hours to predict photovoltaic power generation. User habit prediction: This method employs clustering analysis and association rule mining to analyze energy consumption data related to user household appliance usage, identify user electricity consumption patterns, and combine this with subjective information provided by users to predict future electricity consumption habits. The central processing unit (CPU) allocates 320KB of flash memory to store the clustering analysis and association rule mining algorithms and related model parameters. It also allocates 64KB of static random access memory (SRAM) on the K4S561632 chip to cache user energy consumption data, 16KB of flash memory to store subjective information such as user habits and electricity preferences, and 128KB of SRAM to store intermediate results and the final user electricity consumption habit prediction results. The calculation cycle is set to 24 hours, running the clustering analysis and association rule mining algorithms once a day to predict user electricity consumption habits.

[0037] By comparing the deviation between real-time monitored actual energy data and predicted data, the central processing unit 11 triggers a model parameter update process when the deviation exceeds a set threshold. The data required for updating the parameters of each individual prediction model is retrieved from the temporary storage area of ​​the static random access memory, and the updated parameters are written to the model parameter storage area in the flash memory. Simultaneously, the prediction results are transmitted to the LCD screen via the SPI interface for display and feedback to the user.

[0038] The flexible execution control module 3 is used to communicate with DC home appliances via RS485 bus, adjust the DC bus voltage in real time, and use IGBT module and PWM control chip for closed-loop voltage regulation, based on bus voltage regulation and AI optimization for flexible control of DC home appliances. In some embodiments, the flexible execution control module 3 includes a DC bus voltage regulation circuit that communicates with the central processing unit 11 via a remote terminal unit protocol. The circuit includes a voltage detection circuit 31 and a voltage regulation circuit 32, enabling dynamic regulation of the 300V-400V bus voltage. The working process of the voltage regulation circuit 32 is as follows: The central processing unit 11 outputs a 12-bit digital voltage command to the DAC converter via GPIO; The analog signal output from the DAC is amplified by an operational amplifier and then isolated from the input PWM controller by an optocoupler. The PWM controller adjusts the duty cycle to drive the IGBT module, while the bus voltage is fed back to the ADC pin of the central processing unit 11 through the voltage divider circuit to form a closed-loop control. Voltage detection circuit 31: It uses a high-precision voltage sensor to monitor the DC bus voltage in real time, converts the collected voltage signal into a digital signal and transmits it to the central control unit (as the core control module of the system, used to receive, process data and send control commands).

[0039] Voltage regulation circuit 32: Dynamically regulates the DC bus voltage according to instructions from the central control unit. Based on DC power distribution design standards, the adjustable voltage range is 300V - 400V. The operating power of DC appliances is controlled by changing the output voltage. Infineon IGBT module FF450R12ME4 and TI UCC28950 are selected as the PWM control chip.

[0040] The flexible execution control module 3 incorporates a deep learning neural network algorithm. Based on preprocessed data, it establishes a multi-objective optimization model for energy saving and comfort, and dynamically adjusts the bus voltage according to real-time data. Each DC appliance has a built-in communication module and voltage control interface.

[0041] It also includes an AI optimization unit 33. When dynamically adjusting the bus voltage, the AI ​​optimization unit 33 first generates a target voltage command, which is output through the GPIO interface of the central processing unit 11. This command is converted into an analog signal by the built-in 12-bit DAC, amplified by an operational amplifier, isolated by an optocoupler, and then input to the PWM controller. The controller adjusts the duty cycle accordingly, and the pulse signal is amplified by the IGBT driver chip to control the IGBT's on and off states. Simultaneously, the bus voltage is sampled and fed back to the central processing unit 11. When the deviation exceeds the threshold, the voltage is recalculated and adjusted, forming a closed-loop control that achieves precise voltage regulation within the 300V-400V range. A built-in deep learning neural network algorithm is used to establish a multi-objective optimization model for energy saving and comfort based on pre-processed data. The bus voltage is dynamically adjusted according to real-time data. Each DC appliance has a built-in communication module and voltage control interface, allowing it to communicate with the system and adjust its operating status according to the received voltage control commands. This maximizes energy savings while meeting user comfort requirements.

[0042] The energy efficiency evaluation module 4 is used to calculate the percentage of green electricity, time-of-use electricity bills, PMV comfort level and carbon emissions, and displays a circular progress bar, bar chart and dynamic curve on the LCD screen through the SPI interface.

[0043] In some embodiments, the energy efficiency evaluation module 4 further includes: The green electricity percentage calculation unit 41 is used to connect the green electricity metering module and the total electricity metering module via RS485 bus, and execute division instructions to calculate the real-time green electricity percentage. The central processing unit 11 reads green electricity usage data from the green electricity metering module via the RS485 bus, and simultaneously reads total electricity consumption data from the total electricity consumption metering module via the same bus. The processor calls a division instruction to divide the green electricity usage data by the total electricity consumption data, and temporarily stores the result in register R0. It then executes a multiplication instruction to multiply the result in register R0 by 100 to obtain the green electricity percentage value. Simultaneously, it updates the trend data hourly and stores it in a designated partition on the SD card.

[0044] It also includes an economic operation electricity bill calculation unit 42 and a comfort evaluation unit 43. The economic operation electricity bill calculation unit 42 transmits real-time electricity price data to the central processing unit 11 via an Ethernet module. Energy consumption data is sent to the processor via an RS485 bus from the energy data acquisition module. The processor, according to the time-of-use billing rules, divides different time periods in its internal registers and performs multiplication operations on the energy consumption data of each time period with the corresponding electricity price to obtain the electricity cost for each time period. At 2:00 AM each day, the processor executes an accumulation instruction to add up the costs of each time period for the daily bill; at 3:00 AM every Sunday, it accumulates the daily bills for the week to obtain the weekly bill; and at 4:00 AM on the last day of each month, it accumulates the daily bills for the month to obtain the monthly bill. The bill data is sent to the LCD screen for display via the SPI interface.

[0045] Comfort evaluation unit 43: The environmental parameter acquisition module transmits environmental data such as temperature, humidity, and wind speed to the central processing unit 11 via an RS485 bus. The processor reads the parameters of the PMV comfort evaluation algorithm and the human comfort model data from the flash memory. The processor compares the PMV value with preset thresholds (e.g., -0.5 to 0.5 is comfortable, -1 to -0.5 and 0.5 to 1 are relatively comfortable, and the rest are uncomfortable) to determine the comfort level. The comfort level data is transmitted to the LCD screen for display via the SPI interface.

[0046] The carbon emission calculation unit 44 is used to read the regional power grid carbon emission factor from flash memory and perform a multiplication instruction with energy consumption data to generate carbon emission amount.

[0047] The central processing unit 11 reads energy consumption data from the energy consumption metering module via the RS485 bus and reads regional power grid carbon emission factor data from flash memory. It executes a multiplication instruction to multiply the energy consumption data and carbon emission factor data to obtain the carbon emission amount. The processor summarizes the carbon emission data daily and stores it in a designated folder on the SD card. Weekly, it calls the trend analysis subroutine to generate carbon emission trend curve data based on historical data. The carbon emission data and trend data are sent to the LCD screen for user viewing via the SPI interface. The LCD screen includes a green electricity percentage display area, which displays the proportion and trend of green electricity use using a circular progress bar and line graph; an economic operation electricity bill display area, which compares electricity expenditures at different times using a bar chart and presents detailed bill information in a list; a comfort display area, which displays comfort levels and environmental data using a star rating and an environmental parameter dashboard (an environmental parameter acquisition module for multi-level collaborative monitoring); and a carbon emission display area, which uses a dynamic curve to show changes in carbon emissions and provides comparative data to illustrate the emission reduction effect. The local display refreshes the statistical charts every 10 minutes, and the data is uploaded to the mobile APP via WiFi module. Users can switch data views, synchronize data once a day, and remotely view historical data.

[0048] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0049] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0050] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this disclosure.

[0051] In the embodiments provided in this disclosure, it should be understood that the disclosed apparatus / computer devices and methods can be implemented in other ways. For example, the apparatus / computer device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. Multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0052] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0053] Furthermore, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0054] If an integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program may include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. A computer-readable medium may include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in a computer-readable medium may be appropriately added to or subtracted according to the requirements of legislation and patent practice in a jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media may not include electrical carrier signals and telecommunication signals.

[0055] The above are merely preferred embodiments of the present invention. It should be noted that any modifications and improvements made by those skilled in the art without departing from the present technical solution should also be considered to fall within the scope of protection claimed by the present solution.

Claims

1. An AI-powered intelligent AC / DC home appliance energy management system, characterized in that: It includes a multi-level collaborative supervision module, a dynamic energy forecasting module, a flexible execution control module, and an energy efficiency evaluation module; The multi-level collaborative monitoring module is used to collect real-time voltage, current, and power data of AC and DC household appliances, as well as ambient temperature and humidity data, via RS485 bus; and connects to a local LCD screen via SPI interface to dynamically display the power consumption, energy consumption trend, and device response status of the appliances, and record historical monitoring data. The dynamic energy prediction module is used to predict weather based on LSTM long short-term memory neural network, predict grid peak and off-peak demand based on ARIMA model, predict photovoltaic power generation based on random forest algorithm, and predict user electricity consumption habits based on cluster analysis; the model parameters and intermediate data are stored in the flash memory and static random access memory of the central processing unit, and the prediction periods are 1 hour, 6 hours, 2 hours and 24 hours respectively. The flexible execution control module is used to communicate with DC home appliances via RS485 bus, adjust the DC bus voltage in real time, and use IGBT module and PWM control chip for closed-loop voltage regulation, based on bus voltage regulation and AI optimization for flexible control of DC home appliances. The energy efficiency evaluation module is used to calculate the percentage of green electricity, time-of-use electricity bills, PMV comfort level and carbon emissions, and displays a circular progress bar, bar chart and dynamic curve on the LCD screen through the SPI interface.

2. The system of claim 1, wherein: The multi-level collaborative monitoring module includes an AC home appliance monitoring unit, a DC home appliance monitoring unit, a data transmission interface, a temperature and humidity sensor, a data transmission line, a central processing unit, a data storage unit, an energy consumption statistics unit, a demand response analysis unit, and a display unit connected in parallel via an RS485 bus. The AC appliance monitoring unit includes an electromagnetic induction current transformer, a voltage sampling and power calculation chip, which is used to collect voltage, current and power data of AC appliances in real time. The DC appliance monitoring unit includes a Hall effect current sensor, a differential voltage sampler, and a microcontroller, used to collect current, voltage, and power data of DC appliances. The data transmission interface uses an RS485 bus to transmit the acquired data to the data processing layer. The temperature and humidity sensor uses a combination of capacitive humidity sensing element and thermistor to measure indoor ambient temperature and humidity. The data transmission line converts analog signals into digital signals, and the data transmission line is connected to the data processing layer via an RS485 bus. The LCD screen of the display unit is connected to the central processing unit via an SPI interface; The data storage unit includes a storage chip and an SD card interface. The storage chip has a dedicated cache area for storing the input data and intermediate calculation results of the LSTM weather forecasting model, the ARIMA power grid demand forecasting model, and the random forest photovoltaic forecasting model. The dedicated cache area is divided into four physically isolated storage partitions, including the first partition for caching real-time and historical data from weather sensors; the second partition for caching historical data on grid load and photovoltaic power generation; the third partition for storing the calculation results of the intermediate layers of the LSTM neural network; and the fourth partition for storing the output results of the decision tree of the random forest algorithm. The energy consumption statistics unit uses a sliding window algorithm to accumulate home appliance energy consumption data; The demand response analysis unit is used to calculate the demand response time based on the wavelet transform algorithm.

3. The system of claim 2, wherein: Both the AC and DC home appliance monitoring units integrate power calculation chips and are connected in parallel using an RS485 bus topology. The output of the temperature and humidity sensor is connected to an RS485 signal converter to convert the analog signal into a differential digital signal for transmission.

4. The system of claim 3, wherein: The RS485 bus pin definitions for the AC appliance monitoring unit, DC appliance monitoring unit, and energy consumption statistics unit are as follows: Pin 1: A+ is defined as the positive terminal of the differential signal; Pin 2: B- is defined as the negative terminal of the differential signal; Pin 3: GND is defined as signal ground; Pin 4: VCC is defined as the 5V power supply terminal; The RS485 interface of the central processing unit is connected in parallel with the AC home appliance monitoring unit, DC home appliance monitoring unit and energy consumption statistics unit in a bus topology. The central processing unit, as the master device, provides differential signals and 5V power to the slave devices through pins 1-4.

5. The system of claim 4, wherein: The SPI interface pins of the display unit are defined as follows: Pin 5: SCLK is defined as the clock signal line; Pin 6: MOSI is defined as master output / slave input line; Pin 7: MISO is defined as the master input / slave output line; Pin 8: CS is defined as the chip select signal line; Pin 9: GND is defined as signal ground; The central processing unit is directly connected to the LCD screen point-to-point via the SPI interface. The central processing unit communicates with the LCD screen via pin 8, and transmits synchronous clock and bidirectional data signals via pins 5-9.

6. The system of claim 5, wherein: The I2C interface pins of the data storage unit are defined as follows: Pin 10: SDA is defined as the Central Processing Unit data line; Pin 11: SCL is defined as the central processing unit clock line; Pin 12: GND is defined as signal ground; Pin 13: VCC is defined as the 3.3V power supply terminal; The SPI pins of the SD card interface are defined as follows: Pin 14: SCLK, 15: MOSI, 16: MISO, 17: CS, 18: GND, 19: VCC; The central processing unit (CPU) connects to the memory chip in a single master-slave topology via pins 10-13 of the I2C interface. The CPU accesses the slave device through a unique address. It connects to the SD card in a point-to-point topology via pins 14-19 of the SPI interface. The CPU communicates via strobe pin 17.

7. The system of claim 6, wherein: The PCIe interface pins of the demand response analysis unit are defined as follows: Pin 24: CLK is defined as the clock signal; Pins 25: RX+ and 26: RX- are defined as receive differential pairs; Pins 27: TX+ and 28: TX- are defined as transmit differential pairs; Pin 29: GND is defined as signal ground; Pin 30: VCC is defined as the 12V power supply terminal; The central processing unit is directly connected to the demand response analysis unit via a PCIe bus.

8. The system of claim 7, wherein: The dynamic energy prediction module includes a dataset creation unit, a data preprocessing unit, and a single-item prediction unit. The dataset creation unit is used to acquire and construct weather datasets, power grid datasets, photovoltaic power generation datasets, and user electricity consumption datasets. The data preprocessing unit is used to clean and standardize the multi-source data collected by the dataset establishment unit. The single-item prediction unit is used to predict weather changes, power demand during peak and off-peak hours of the power grid, photovoltaic power generation, and users' future electricity consumption habits within a specific future time period. It compares the deviation between the real-time monitoring of actual energy data and the predicted data, and provides feedback to the user when the deviation exceeds a set threshold.

9. The system according to claim 8, characterized in that: The flexible execution control module includes a DC bus voltage regulation circuit that communicates with the central processing unit via a remote terminal unit protocol. This circuit includes a voltage detection circuit and a voltage regulation circuit, enabling dynamic regulation of the 300V-400V bus voltage. The working process of the voltage regulation circuit is as follows: The central processing unit outputs 12-bit digital voltage commands to the DAC converter via GPIO; The analog signal output from the DAC is amplified by an operational amplifier and then isolated from the input PWM controller by an optocoupler. The PWM controller adjusts the duty cycle to drive the IGBT module, while the bus voltage is fed back to the ADC pin of the central processing unit through the voltage divider circuit to form a closed-loop control. The flexible execution control module incorporates a deep learning neural network algorithm. Based on preprocessed data, it establishes a multi-objective optimization model for energy saving and comfort, and dynamically adjusts the bus voltage according to real-time data. Each DC appliance has a built-in communication module and voltage control interface.

10. The system according to claim 9, characterized in that: The energy efficiency evaluation module also includes: The green electricity percentage calculation unit is used to connect the green electricity metering module and the total electricity metering module via RS485 bus, and execute division instructions to calculate the real-time green electricity percentage. The carbon emission calculation unit is used to read the regional power grid carbon emission factor from flash memory and perform a multiplication instruction with energy consumption data to generate carbon emission amounts.

Images