Device and method for applying portable modular edge AI to smart home system

Abstract

The invention relates to the technical field of smart home, in particular to a device and method for applying a portable modular edge AI to a smart home system. The device is characterized in that the device comprises a portable outer shell and a PCB multi-layer stack modular board card; the system also comprises a main control circuit board card, an edge calculation module, and one or more expansion circuit board system modules. The main control circuit board card comprises an MCU processor, a GPU graphics processor, a DSP digital signal processor, a power management module, a user interaction module, a storage module, a control module, a sensor module and a self-adaptive communication module. The edge calculation module comprises an NPU localization model reasoning chip, a built-in lightweight AI model, a privacy desensitization module and a storage device; the devices are connected and integrated to form a complete system which is used for realizing localized data processing and equipment control in a smart home system, and can execute complex tasks, disperse heat, reduce the size and cost and improve the intelligence and automation of the system under the condition of low power consumption.

Description

Technical Field

[0003] The present invention relates to the field of smart home technology, and in particular to a device and method for applying a portable modular edge AI in a smart home system. Background Art

[0005] With the development of Internet of Things technology, smart home systems are becoming more and more popular. Traditional smart home systems usually rely on cloud servers for data processing and control. With the development of artificial intelligence technology, there are more and more edge AI (also known as edge computing) applications based on artificial intelligence. Edge AI refers to running AI applications on terminal devices. Even without a network connection, the computing power on the terminal device can smoothly run various AIs such as large models. Edge AI applications are increasingly used in smart home devices such as smart lighting control, smart appliances, voice assistants, and security cameras. These applications require a large number of sensors and controllers to be deployed in the home. By processing data locally, these devices can quickly respond to user commands, detect anomalies, and automatically perform tasks without relying on cloud connections. Terminals and edge platforms use local data to retrain the basic models sent from the cloud.

[0006] However, most existing smart home systems rely on cloud servers for data processing and control, or use devices with large size, volume, and structure, which have problems such as low flexibility and maintainability, high latency, insufficient privacy protection, and dependence on network connections. Summary of the invention

[0008] The purpose of the present invention is to provide a device and method for applying portable modular edge AI in a smart home system, which can perform data processing and device control locally, perform complex tasks under low power consumption conditions, disperse heat, reduce size and volume, improve the intelligence and automation of the system, reduce dependence on cloud servers, and is suitable for portable mobile or Internet of Things nodes or other resource-constrained scenarios.

[0009] A device for applying portable modular edge AI in a smart home system, characterized in that it includes: a portable outer shell, inside which is installed a PCB multi-layer stacked modular board; the PCB multi-layer stacked modular board includes: a main control circuit board, an edge computing module, and one or more expansion circuit board system modules; the main control circuit board includes an MCU processor, a GPU graphics processor, a DSP digital signal processor, a power management module, a user interaction module, a storage module, a control module, a sensor module, and an adaptive communication module; the edge computing module includes an NPU localized model inference chip, a built-in lightweight AI model, a privacy desensitization module, and a storage device; the above-mentioned devices are connected and integrated into a complete system for realizing localized data processing and equipment control in a smart home system, and can perform complex tasks under low power consumption conditions, disperse heat, reduce size and cost, and improve the intelligence and automation of the system.

[0010] The portable outer shell includes an upper shell and a lower shell, which are movably connected by a card to form a hollow accommodating cavity. The upper shell is provided with a handle, and the lower shell is provided with a heat dissipation vent, an antenna and other related device interfaces (including a power switch, a power interface, a network connection interface, a serial port, a USB port, and other interfaces), which is suitable for portable mobile or Internet of Things nodes or other resource-constrained scenarios.

[0011] The PCB multi-layer stacked modular board is composed of multiple single-layer or double-layer panels stacked together, with the layers isolated by insulating materials (such as FR-4) and electrical connections achieved through pre-drilling and electroplating processes. Circuits with different functions are modularized, and each module is independently designed and integrated into a multi-layer PCB. The modules are interconnected through internal wiring or connectors; including the MCU processor on the main control circuit board, which is responsible for executing computer instructions, processing data in computer software, and controlling the computer's calculations, storage, and input and output devices; the GPU graphics processor focuses on graphics processing and parallel computing, and supports visually intensive tasks such as three-dimensional animation, graphic design, and video editing; the DSP digital signal processor supports high-speed, real-time transformation and extraction of information, and can extract information from various noisy and interfering environments. , and transformed into a programmable form that is easy for people or machines to use; a power management module that supports battery power and external power supply; a user interaction module that includes a touch screen and a voice input and output interface, supports a voice assistant function, and can interact with users through natural language processing technology; a storage module that is used to store local AI models and data; a control module that is used to coordinate the operation of each module and execute AI algorithms, a sensor module that includes at least one of a temperature sensor, a humidity sensor, a light sensor, and a motion sensor; an adaptive communication module that supports multiple wireless communication protocols, including Wi-Fi, Bluetooth, Zigbee, and 5G; and multiple expansion circuit boards (suitable for application scenarios that require more resources), the main control circuit board is connected with the above-mentioned module system functions to integrate a complete system.

[0012] The edge computing module is a highly integrated hardware system that processes the data of smart home devices locally and generates control instructions, including: NPU localized model inference chip: responsible for executing the inference task of the AI ​​model, supporting a network speed of up to 20 Gb / s and a memory bandwidth of 132 GB / s, features: high performance, low latency, suitable for real-time data processing and inference tasks, scalability: supports deployment through cluster expansion to meet larger-scale computing needs; built-in lightweight AI model: preset AI model for local data processing and inference, features: lightweight design, suitable for running on resource-constrained edge devices, reducing dependence on cloud computing; privacy desensitization module: desensitizes sensitive data to ensure user privacy security, features: locally processes data to avoid sensitive information from being transmitted to the cloud, and enhances data security; storage device: function: used to store locally processed data and models, features: high-speed reading and writing, support for large-capacity storage, and ensure the continuity and stability of data processing; and other related components: edge connector: used to connect external devices and sensors, smart gateway: realize communication and data exchange between devices; sensor interface: receives data collected by sensors, and supports multiple sensor types.

[0013] Its control process consists of the following steps: 1. Data acquisition: The sensor collects environmental data (such as temperature, humidity, light, etc.) through the sensor interface, and the DSP digital signal processor receives the sensor data and performs preliminary signal processing. 2. Data transmission: The processed data is transmitted to the edge computing module through the main control circuit board, and the data enters the NPU localized model inference chip through the edge connector; 3. Data inference: The NPU chip uses the built-in lightweight AI model to infer the data and generate control instructions; the privacy desensitization module desensitizes sensitive data to ensure privacy security; data storage: The processed data and control instructions are stored in the storage device to ensure data traceability and system stability; 4. Control instruction execution: The control module sends control signals to smart home devices through the smart gateway according to the generated instructions; the adaptive communication module ensures the stable transmission of control signals and supports multiple communication protocols; 5. System monitoring and feedback: The main control circuit board monitors the operating status of each module in real time to ensure the stability and reliability of the system. Through the adaptive communication module, the system can communicate with the external network to achieve remote monitoring and management.

[0014] The edge computing module realizes localized data processing and control of smart home devices through highly integrated hardware design and optimized control flow. Its core components, NPU localized model inference chip and built-in lightweight AI model, ensure efficient data inference capability, while the privacy desensitization module and storage device ensure data security and stability. By working in collaboration with the DSP digital signal processor and the main control circuit board, the edge computing module can efficiently process sensor data and generate control instructions, and ultimately achieve precise control of smart home devices through the intelligent gateway and adaptive communication module.

[0015] The sensor module is the core component of the smart home system, responsible for the collection, processing and transmission of environmental data, including: MCU controller: as the core of the module, MCU is responsible for collecting data from various sensors and transmitting the data to the main control system through the communication interface; signal processing converter: responsible for converting the analog signal collected by the sensor into a digital signal for processing by the MCU. The signal processing converter usually includes an analog-to-digital converter (ADC) and necessary filtering circuits to ensure the accuracy and stability of the signal; sensor array: including various types of sensors, such as thermistors, thermocouples, humidity sensors, infrared sensors, gas sensors, etc., responsible for monitoring parameters such as temperature, humidity, motion, and gas concentration in the environment; communication interface: used to transmit the processed data to the central control system. The communication interface includes I2C, SPI, UART, etc., and supports connection with the adaptive communication module; power management module: responsible for providing a stable power supply for the entire sensor module to ensure that the module can still work normally in low power consumption mode; edge computing module: used to perform simple data processing and logic operations locally, reduce dependence on the central control system, and improve the response speed and reliability of the system.

[0016] Its control process is: 1. Data acquisition: The sensor array collects environmental data in real time, such as temperature, humidity, motion status, gas concentration, etc., and these data are output in the form of analog signals; 2. Signal conversion: The signal processing converter converts the analog signal output by the sensor into a digital signal, and performs necessary filtering and amplification processing to ensure the accuracy of the data; 3. Data processing: After receiving the digital signal, the MCU controller performs preliminary data processing, such as data calibration, outlier filtering, etc.; the edge computing module 22 can perform simple logical operations locally, such as determining whether the temperature exceeds the threshold, detecting whether there is movement, etc.; 4. Data transmission: The processed data is transmitted to the central control system through the communication interface; the MCU controller is responsible for data packaging and transmission to ensure the integrity and real-time nature of the data; 5. System feedback: The central control system performs corresponding automated operations based on the received data, such as adjusting the air-conditioning temperature, starting the ventilation system, triggering security alarms, etc.

[0017] The adaptive communication module includes an MCU microcontroller: a core processing unit responsible for executing communication protocols, data processing and device control; a wireless communication chip: supporting multiple wireless protocols (such as Wi-Fi, Zigbee, Z-Wave, Bluetooth, etc.) to achieve communication between devices; an antenna: used to send and receive wireless signals; a power management unit: providing a stable power supply and managing energy consumption; a sensor interface: connecting to sensors such as temperature, humidity, and light to collect environmental data; a memory: storing firmware, configuration data, and temporary data; an I / O interface: connecting to other devices or modules, supporting wireless communication with smart home devices and cloud servers (dynamically switching local / cloud decision logic), and achieving communication and coordination between smart home devices, such as: protocol adaptation: the module can automatically identify and switch between different Communication protocols ensure the compatibility of devices in different network environments; data acquisition and processing: environmental data is collected through the sensor interface, processed by the microcontroller, and sent to other devices or the cloud through the wireless communication chip; device control: receiving instructions from users or other devices, and controlling the operating status of home appliances through the I / O interface; energy consumption management: the power management unit optimizes energy consumption and extends device life, especially in battery-powered devices; network ad hoc: supports ad hoc network technology, automatically establishes and maintains network topology, and ensures stable communication between devices; secure communication: adopts encryption and authentication mechanisms to ensure the security of data transmission and prevent unauthorized access; the adaptive communication module integrates multiple communication protocols and intelligent algorithms to achieve interconnection and intelligent control between devices, thereby improving the user experience and energy efficiency of the smart home system.

[0018] The privacy desensitizing module includes a storage device responsible for local data processing, which is used to temporarily store original data and desensitized data, and is connected to an adaptive communication module 23, supporting Wi-Fi, Bluetooth, Zigbee, etc., for communication between devices and with the cloud; and is connected to a DSP digital signal processor 213: obtaining original data from sensors and cameras: identifying sensitive information in the data, such as faces, license plates, etc.; desensitizing sensitive information, such as blurring, encrypting, etc., storing desensitized data and transmitting it to the cloud or other devices.

[0019] A method for applying portable modular edge AI in a smart home system, characterized in that the system management, artificial intelligence algorithm processing including storage of local AI models and data and the operating system AIOS of smart home applications running on the main control circuit board card and the edge computing module device are as follows:

[0020] 1. The user interaction module receives user input and sends instructions through the control module;

[0021] 2. The sensor module collects environmental data and sends it to the MCU processor for processing;

[0022] 3. The MCU processor processes data and communicates with smart home devices through an adaptive communication module;

[0023] 4. The NPU localized model inference chip in the edge computing module performs localized AI inference and handles complex tasks;

[0024] 5. The GPU graphics processor and the DSP digital signal processor process graphics and signal data respectively;

[0025] 6. The power management module manages the power supply of the entire system to ensure low power consumption operation;

[0026] 7. The storage module and storage device store the processed data and AI model;

[0027] 8. The expansion circuit board system module provides additional functional expansion;

[0028] And follow these steps:

[0029] S1: The operating system connects the cloud and the device end, connects the home devices through Wifi access, Bluetooth gateway, infrared access and Zigbee, and applies for cloud authorization from the cloud through configuration requests, so that the modular edge AI can achieve interconnection and resource configuration in multiple scenarios and update authentication information. If the cloud authentication fails, the device authentication access management is directly provided to the edge computing module 22; the basic model sent from the cloud is retrained using local data, and the smart home detection model and feature extraction model are loaded into the AI ​​chip of the edge computing module to identify the input information;

[0030] S2: local data preprocessing (noise reduction, feature extraction). The edge processing module supports dynamic model switching. When it detects that the network bandwidth is higher than the threshold, it automatically downloads a high-precision AI model to replace the local lightweight model.

[0031] S3: Edge model reasoning (dynamic loading of lightweight AI models), multi-device collaborative decision-making (direct communication between devices to avoid cloud transit), the edge computing module performs AI model reasoning locally, reducing dependence on cloud servers;

[0032] S4: Optimize the control strategy of smart home devices through machine learning algorithms, and implement the method steps of applying modular edge AI in a smart home system as described in any one of claims 1 to 7 when the operating system AIOS of the smart home application is executed by the processor.

[0033] Beneficial effects: The devices and methods implementing the present invention have high-density integration: multi-layer stacking modularization allows more functional modules to be integrated in a limited space, thereby improving circuit density; performance improvement: by optimizing signal paths and reducing interference, circuit performance is improved, and it is particularly suitable for high-frequency and high-speed applications; design flexibility: modular design facilitates functional expansion and upgrading, and enhances design flexibility and maintainability; thermal management optimization: multi-layer structure contributes to more effective thermal management, improving system stability and lifespan; cost-effectiveness: modular design can reduce subsequent maintenance and upgrade costs, and is more economical in the long run. It can perform data processing and device control locally, reduce dependence on cloud servers, improve system intelligence and automation, and is suitable for portable mobile or IoT nodes or other resource-constrained scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a front view schematic diagram of the portable device of the present invention;

[0036] Figure 2 A schematic diagram of the modular edge AI structure of a portable device of the present invention;

[0037] Figure 3 It is a schematic diagram of the system framework flow of the device method of the present invention;

[0038] Figure 4 It is a schematic diagram of the method steps of the present invention.

[0039] Markings in the accompanying drawings:

[0040] A portable modular edge AI application device 100 in a smart home system, a portable outer shell 10, an upper shell 11, a lower shell 12, a handle 13, a heat dissipation vent 14, an antenna 15, a related device interface 16, a PCB multi-layer stacked modular board 20, a main control circuit board card 21, an MCU processor 211, a GPU graphics processor 212, a DSP digital signal processor 213, a power management module 214, a user interaction module 215, a storage module 216, a control module 217, a sensor module 218, an adaptive communication module 219, an edge computing module 22, an NPU localized model inference chip 221, an edge connector 222, a built-in lightweight AI model 223, a privacy desensitization module 224, a smart gateway 225, a storage device 226, other related interface components 227, and an expansion circuit board system module 23. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] like Figure 1 As shown, a portable modular edge AI application device 100 in a smart home system is characterized in that it includes: a portable outer shell 10, which is internally equipped with a PCB multi-layer stacked modular board 20; the PCB multi-layer stacked modular board 20 includes: a main control circuit board 21, an edge computing module 22, and one or more expansion circuit board system modules 23; the main control circuit board 21 includes an MCU processor 211, a GPU graphics processor 212, a DSP digital signal processor 213, a power management module 214, a user interaction module 215, a storage module 216, a control module 217, a sensor module 218, and an adaptive communication module 219; the edge computing module 22 includes an NPU localized model inference chip 221, a built-in lightweight AI model 222, a privacy desensitization module 223, and a storage device 224; the above-mentioned devices are connected and integrated into a complete system for realizing localized data processing and device control in a smart home system, which can perform complex tasks under low power consumption conditions, disperse heat, reduce size and cost, and improve the intelligence and automation of the system.

[0044] like Figure 1 As shown, the portable outer shell 10 includes an upper shell 11 and a lower shell 12, and the upper and lower shells are movably connected by a card to form a hollow accommodating cavity. The upper shell 11 is provided with a handle 13, and the lower shell is provided with a heat dissipation vent 14, an antenna 15 and other related device interfaces 16 (including a power switch, a power interface, a network connection interface, a serial port, a USB port, and other interfaces), which are suitable for portable mobile or Internet of Things nodes or other resource-constrained scenarios.

[0045] like Figure 2As shown, the PCB multi-layer stacked modular board 20 is formed by stacking multiple single-layer or double-layer panels, the layers are isolated by insulating materials (such as FR-4), and electrical connections are achieved through pre-drilling and electroplating processes. Circuits with different functions are modularized, each module is independently designed and integrated into a multi-layer PCB, and the modules are interconnected through internal wiring or connectors; including an MCU processor 211 on the main control circuit board 21, which is responsible for executing computer instructions, processing data in computer software, and controlling the computer's calculations, storage, and input and output devices; a GPU graphics processor 212, which focuses on graphics processing and parallel computing, and supports visually intensive tasks such as three-dimensional animation, graphic design, and video editing; a DSP digital signal processor 213 supports high-speed, real-time transformation and extraction of information, and can extract information from various noise and interference environments. And transformed into a programmable form that is easy for people or machines to use; a power management module 214, which supports battery power supply and external power supply; a user interaction module 215, including a touch screen and a voice input and output interface, supports a voice assistant function, and can interact with users through natural language processing technology; a storage module 216, used to store local AI models and data; a control module 217, used to coordinate the operation of each module and execute AI algorithms, a sensor module 218, including at least one of a temperature sensor, a humidity sensor, a light sensor and a motion sensor; an adaptive communication module 219, supporting multiple wireless communication protocols, including Wi-Fi, Bluetooth, Zigbee and 5G; and multiple expansion circuit boards (suitable for application scenarios that require more resources), the main control circuit board is connected with the above-mentioned module system functions to integrate a complete system.

[0046] like Figure 3As shown, according to claim 1, the portable modular edge AI application device in the smart home system is characterized in that the edge computing module 22 is a highly integrated hardware system that locally processes the data of smart home devices and generates control instructions including: NPU localized model inference chip 221: responsible for executing the inference task of the AI ​​model, supporting network speeds up to 20 Gb / s and 132 GB / s memory bandwidth; built-in lightweight AI model 222: preset AI model for local data processing and reasoning; privacy desensitization module 223: desensitizes sensitive data to ensure user privacy security; storage device 224: used to store locally processed data and models; and other related components 225: edge connector: used to connect external devices and sensors, smart gateway: to achieve communication and data exchange between devices; sensor interface: receives data collected by sensors and supports multiple sensor types; its control process is: 1. Data acquisition: the sensor collects environmental data (such as temperature, humidity, light, etc.) through the sensor interface 225, and the DSP digital signal processor 213 receives the sensor data and performs preliminary signal processing, 2. Data transmission: the processed data is transmitted to the edge computing module 22 through the main control circuit board card 21, and the data enters the NPU localized model reasoning chip 221 through the edge connector 225; 3. Data reasoning: The NPU chip 221 uses the built-in lightweight AI model 222 to infer the data and generate control instructions; the privacy desensitization module 223 desensitizes sensitive data to ensure privacy security; data storage: the processed data and control instructions are stored in the storage device 224 to ensure data traceability and system stability; 4. Control instruction execution: the control module 217 sends control signals to smart home devices through the smart gateway 225 according to the generated instructions; the adaptive communication module 219 ensures the stable transmission of control signals and supports multiple communication protocols; 5. System monitoring and feedback: the main control circuit board card 21 monitors the operating status of each module in real time to ensure the stability and reliability of the system, and The system can communicate with the external network to realize remote monitoring and management; the edge computing module 22 realizes localized data processing and control of smart home devices through highly integrated hardware design and optimized control process. Its core components NPU localized model inference chip 221 and built-in lightweight AI model 222 ensure efficient data inference capability, while the privacy desensitization module 223 and storage device 224 ensure data security and stability. By working in collaboration with the DSP digital signal processor 213 and the main control circuit board card 21, the edge computing module 22 can efficiently process sensor data and generate control instructions, and finally realize precise control of smart home devices through the intelligent gateway and adaptive communication module.

[0047] like Figure 3As shown, the sensor module 218 is the core component of the smart home system, responsible for the collection, processing and transmission of environmental data, including: MCU controller: as the core of the module, the MCU is responsible for collecting data from various sensors and transmitting the data to the main control system through the communication interface; signal processing converter: responsible for converting the analog signal collected by the sensor into a digital signal for processing by the MCU. The signal processing converter usually includes an analog-to-digital converter (ADC) and necessary filtering circuits to ensure the accuracy and stability of the signal; sensor array: including various types of sensors, such as thermistors, thermocouples, humidity sensors, infrared sensors, gas sensors, etc., responsible for monitoring parameters such as temperature, humidity, motion, and gas concentration in the environment; communication interface: used to transmit the processed data to the central control system. The communication interface includes I2C, SPI, UART, etc., and supports connection with the adaptive communication module 219; power management module 214: responsible for providing a stable power supply for the entire sensor module to ensure that the module can still work normally in low power consumption mode; edge computing module 22: used to perform simple data processing and logic operations locally, reduce dependence on the central control system, and improve the response speed and reliability of the system.

[0048] Its control process is: 1. Data acquisition: The sensor array collects environmental data in real time, such as temperature, humidity, motion status, gas concentration, etc., and these data are output in the form of analog signals; 2. Signal conversion: The signal processing converter converts the analog signal output by the sensor into a digital signal, and performs necessary filtering and amplification processing to ensure the accuracy of the data; 3. Data processing: After receiving the digital signal, the MCU controller performs preliminary data processing, such as data calibration, outlier filtering, etc.; the edge computing module 22 can perform simple logical operations locally, such as determining whether the temperature exceeds the threshold, detecting whether there is movement, etc.; 4. Data transmission: The processed data is transmitted to the central control system through the communication interface; the MCU controller is responsible for data packaging and transmission to ensure the integrity and real-time nature of the data; 5. System feedback: The central control system performs corresponding automated operations based on the received data, such as adjusting the air-conditioning temperature, starting the ventilation system, triggering security alarms, etc.

[0049] like Figure 3As shown, the adaptive communication module 219 includes an MCU microcontroller: a core processing unit responsible for executing communication protocols, data processing and device control; a wireless communication chip: supporting multiple wireless protocols (such as Wi-Fi, Zigbee, Z-Wave, Bluetooth, etc.) to achieve communication between devices; an antenna: used to send and receive wireless signals; a power management unit: providing a stable power supply and managing energy consumption; a sensor interface: connecting to sensors such as temperature, humidity, and light to collect environmental data; a memory: storing firmware, configuration data, and temporary data; an I / O interface: connecting to other devices or modules, supporting wireless communication with smart home devices and cloud servers (dynamically switching local / cloud decision logic), and achieving communication and coordination between smart home devices, such as: protocol adaptation: the module can automatically identify and switch Different communication protocols can be changed to ensure the compatibility of devices in different network environments; data acquisition and processing: environmental data is collected through the sensor interface, processed by the microcontroller, and sent to other devices or the cloud through the wireless communication chip; device control: receiving instructions from users or other devices, and controlling the operating status of home appliances through the I / O interface; energy consumption management: the power management unit optimizes energy consumption and extends device life, especially in battery-powered devices; network ad hoc: supports ad hoc network technology, automatically establishes and maintains network topology, and ensures stable communication between devices; secure communication: adopts encryption and authentication mechanisms to ensure the security of data transmission and prevent unauthorized access; the adaptive communication module integrates multiple communication protocols and intelligent algorithms to achieve interconnection and intelligent control between devices, thereby improving the user experience and energy efficiency of the smart home system.

[0050] like Figure 3 As shown, the privacy desensitizing module 223 includes a storage device, which is responsible for local data processing, for temporarily storing original data and desensitized data, and is connected to the adaptive communication module 23, supporting Wi-Fi, Bluetooth, Zigbee, etc., for communication between devices and with the cloud; and is connected to the DSP digital signal processor 213: obtaining original data from sensors and cameras: identifying sensitive information in the data, such as faces, license plates, etc.; desensitizing sensitive information, such as blurring, encrypting, etc., storing desensitized data and transmitting it to the cloud or other devices.

[0051] like Figure 4 As shown, a method for applying a portable modular edge AI in a smart home system is characterized in that the system management, artificial intelligence algorithm processing including storage of local AI models and data and the operating system AIOS of the smart home application running on the main control circuit board (21) and the edge computing module (22) are as follows:

[0052] 1. The user interaction module (215) receives user input and sends instructions through the control module (217);

[0053] 2. The sensor module (218) collects environmental data and sends it to the MCU processor (211) for processing;

[0054] 3. The MCU processor (211) processes data and communicates with smart home devices through the adaptive communication module (219);

[0055] 4. The NPU localized model inference chip (221) in the edge computing module (22) performs localized AI inference and processes complex tasks;

[0056] 5. The GPU graphics processor (212) and the DSP digital signal processor (213) process graphics and signal data respectively;

[0057] 6. The power management module (214) manages the power supply of the entire system to ensure low power consumption operation;

[0058] 7. The storage module (216) and the storage device (224) store the processed data and AI model;

[0059] 8. The expansion circuit board system module (23) provides additional functional expansion;

[0060] And follow these steps:

[0061] S1: The operating system connects the cloud and the device end, connects the home devices through Wifi access, Bluetooth gateway, infrared access and Zigbee, and applies for cloud authorization from the cloud through configuration requests, so that the modular edge AI can achieve interconnection and resource configuration in multiple scenarios and update authentication information. If the cloud authentication fails, the device authentication access management is directly provided to the edge computing module 22; the basic model sent from the cloud is retrained using local data, and the smart home detection model and feature extraction model are loaded into the AI ​​chip of the edge computing module 22 to identify the input information;

[0062] S2: local data preprocessing (noise reduction, feature extraction). The edge processing module supports dynamic model switching. When it detects that the network bandwidth is higher than the threshold, it automatically downloads a high-precision AI model to replace the local lightweight model.

[0063] S3: Edge model reasoning (dynamic loading of lightweight AI models), multi-device collaborative decision-making (direct communication between devices to avoid cloud transit), the edge computing module performs AI model reasoning locally, reducing dependence on cloud servers;

[0064] S4: Optimize the control strategy of smart home devices through machine learning algorithms, and implement the method steps of applying modular edge AI in a smart home system as described in any one of claims 1 to 7 when the operating system AIOS of the smart home application is executed by the processor.

[0065] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0066] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general hardware. Based on this understanding, the above technical solution can be essentially or the part that contributes to the prior art can be embodied in the form of a hardware software product, and the software product can be stored in a hardware computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0067] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features can be replaced by equivalents. However, these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A portable modular edge AI device (100) for use in a smart home system, characterized in that: include: A portable outer shell (10) having a PCB multi-layer stacked modular board (20) installed inside; The PCB multi-layer stacked modular board (20) comprises: a main control circuit board (21), an edge computing module (22), and one or more expansion circuit board system modules (23); the main control circuit board (21) comprises an MCU processor (211), a PU graphics processor (212), a DSP digital signal processor (213), a power management module (214), a user interaction module (215), a storage module (216), a control module (217), a sensor module (218), and an adaptive communication module (219); the edge computing module (22) comprises an NPU localized model inference chip (221), a built-in lightweight AI model (222), a privacy desensitization module (223), and a storage device (224); the above-mentioned devices are connected and integrated into a complete system for realizing localized data processing and device control in a smart home system, capable of performing complex tasks under low power consumption conditions, dispersing heat, reducing size and cost, improving the intelligence and automation of the system, and suitable for portable mobile or Internet of Things nodes or other resource-constrained scenarios.

2. The portable modular edge AI device for use in a smart home system according to claim 1, characterized in that: The portable outer shell (10) comprises: an upper shell (11) and a lower shell (12), the upper and lower shells being movably connected by a clamp to form a hollow accommodating cavity, the upper shell (11) being provided with a carrying handle (13), the lower shell being provided with a heat dissipation vent (14), an antenna (15) and other related device interfaces (16) (including a power switch, a power interface, a network connection interface, a serial port, a USB port, and other interfaces).

3. The portable modular edge AI device for use in a smart home system according to claim 1, characterized in that: The PCB multi-layer stacked modular board (20) is formed by stacking a plurality of single-layer or double-layer boards, the layers are isolated by insulating materials (such as FR-4), and electrical connections are achieved through pre-drilling and electroplating processes, circuits with different functions are modularized, each module is independently designed and integrated into the multi-layer PCB, and the modules are interconnected through internal wiring or connectors; the MCU processor (211) on the main control circuit board (21) is responsible for executing computer instructions, processing data in computer software, and controlling the computer's calculation, storage and input and output devices; the GPU graphics processor (212) focuses on graphics processing and parallel computing, and supports visually intensive tasks such as three-dimensional animation, graphic design and video editing; The DSP digital signal processor (213) supports high-speed, real-time transformation and extraction of information, can extract information from various noise and interference environments, and transform it into a programmable form that is easy for people or machines to use; a power management module (214), which supports battery power supply and external power supply; a user interaction module (215), including a touch screen and a voice input and output interface, supporting a voice assistant function, and being able to interact with users through natural language processing technology; a storage module (216), used to store local AI models and data; a control module (217), used to coordinate the operation of each module and execute the AI ​​algorithm, a sensor module (218), wherein the sensor module (218) includes at least one of a temperature sensor, a humidity sensor, a light sensor, and a motion sensor; an adaptive communication module (219), wherein the adaptive communication module (219) supports multiple wireless communication protocols, including Wi-Fi, Bluetooth, Zigbee, and 5G; and multiple expansion circuit boards (suitable for application scenarios requiring more resources), wherein the main control circuit board is connected with the above-mentioned module system functions to integrate a complete system.

4. The portable modular edge AI device for use in a smart home system according to claim 1, characterized in that: The edge computing module (22) is a highly integrated hardware system that processes data from smart home devices locally and generates control instructions, including: NPU localized model inference chip (221): responsible for executing the inference task of the AI ​​model, supporting a network speed of up to 20 Gb / s and 132 GB / s memory bandwidth; built-in lightweight AI model (222): a preset AI model for local data processing and reasoning; privacy desensitization module (223): desensitizes sensitive data to ensure user privacy security; storage device (224): used to store locally processed data and models; and other related components (225): edge connector: used to connect external devices and sensors, smart gateway: to achieve communication and data exchange between devices; sensor interface: receives data collected by the sensor and supports multiple sensor types; its control process is as follows:

1. Data acquisition: the sensor collects environmental data (such as temperature, humidity, light, etc.) through the sensor interface (225), the DSP digital signal processor (213) receives the sensor data and performs preliminary signal processing, 2. Data transmission: the processed data is transmitted to the edge computing module (22) through the main control circuit board card (21), and the data enters the NPU localized model reasoning chip (221) through the edge connector (225); 3. Data reasoning: The NPU chip (221) uses the built-in lightweight AI model (222) to infer the data and generate control instructions; the privacy desensitization module (223) desensitizes the sensitive data to ensure privacy security; data storage: the processed data and control instructions are stored in the storage device (224) to ensure data traceability and system stability; 4. Control instruction execution: the control module (217) sends the control signal to the smart home device through the smart gateway (225) according to the generated instructions; the adaptive communication module (219) ensures the stable transmission of the control signal and supports multiple communication protocols; 5. System monitoring and feedback: the main control circuit board (21) monitors the operating status of each module in real time to ensure the stability and reliability of the system, and ), the system can communicate with the external network to achieve remote monitoring and management; the edge computing module (22) realizes localized data processing and control of smart home devices through highly integrated hardware design and optimized control process, and its core components NPU localized model inference chip (221) and built-in lightweight AI model (222) ensure efficient data inference capability, while the privacy desensitization module (223) and storage device (224) ensure data security and stability. By working in coordination with the DSP digital signal processor (213) and the main control circuit board (21), the edge computing module (22) can efficiently process sensor data and generate control instructions, and finally realize precise control of smart home devices through the intelligent gateway and adaptive communication module.

5. The device for applying portable modular edge AI in a smart home system according to claim 1 or 3, characterized in that the sensor module (218) is a core component of the smart home system, responsible for collecting, processing and transmitting environmental data, including: MCU controller: as the core of the module, the MCU is responsible for collecting data from various sensors and transmitting the data to the main control system through the communication interface; signal processing converter: responsible for converting the analog signal collected by the sensor into a digital signal for processing by the MCU, the signal processing converter usually includes an analog-to-digital converter (ADC) and necessary filtering circuits to ensure the accuracy and stability of the signal; sensor array: including various types of sensors, such as thermistors, thermocouples, humidity sensors, infrared sensors, gas sensors, etc., responsible for monitoring parameters such as temperature, humidity, motion, gas concentration, etc. in the environment; communication interface: used to transmit the processed data to the central control system. The communication interface includes I2C, SPI, UART, etc., and supports connection with the adaptive communication module (219); power management module (214) : Responsible for providing a stable power supply for the entire sensor module to ensure that the module can still work normally in low power mode; Edge computing module (22): used to perform simple data processing and logical operations locally, reduce dependence on the central control system, and improve the response speed and reliability of the system; its control process is:

1. Data acquisition: The sensor array collects environmental data in real time, such as temperature, humidity, motion status, gas concentration, etc., and these data are output in the form of analog signals; 2. Signal conversion: The signal processing converter converts the analog signal output by the sensor into a digital signal, and performs necessary filtering and amplification processing to ensure the accuracy of the data; 3. Data processing: After receiving the digital signal, the MCU controller performs preliminary data processing, such as data calibration, outlier filtering, etc.; The edge computing module (22) can perform simple logical operations locally, such as determining whether the temperature exceeds the threshold, detecting whether there is movement, etc.; 4. Data transmission: The processed data is transmitted to the central control system through the communication interface; The MCU controller is responsible for data packaging and transmission to ensure the integrity and real-time nature of the data; 5. System feedback: The central control system performs corresponding automated operations based on the data received, such as adjusting the air-conditioning temperature, starting the ventilation system, triggering the security alarm, etc.

6. The portable modular edge AI device for use in a smart home system according to claim 3 or 4, characterized in that: The adaptive communication module (219) comprises an MCU microcontroller: a core processing unit responsible for executing communication protocols, data processing and device control; a wireless communication chip: supporting multiple wireless protocols (such as Wi-Fi, Zigbee, Z-Wave, Bluetooth, etc.) to achieve communication between devices; Antenna: used to send and receive wireless signals; Power management unit: provides stable power supply and manages energy consumption; Sensor interface: connects temperature, humidity, light and other sensors to collect environmental data; Memory: stores firmware, configuration data and temporary data; I / O interface: connects to other devices or modules, supports wireless communication with smart home devices and cloud servers (dynamically switches local / cloud decision logic), and realizes communication and coordination between smart home devices, such as: Protocol adaptation: the module can automatically identify and switch different communication protocols to ensure the compatibility of devices in different network environments; Data acquisition and processing: collect environmental data through the sensor interface, and after processing by the microcontroller, send it to other devices or the cloud through the wireless communication chip; Device control: receives instructions from users or other devices, and controls the operating status of home appliances through the I / O interface; Energy consumption management: the power management unit optimizes energy consumption and extends device life, especially in battery-powered devices; Network ad hoc: supports ad hoc networking technology, automatically establishes and maintains network topology, and ensures stable communication between devices; Secure communication: uses encryption and authentication mechanisms to ensure the security of data transmission and prevent unauthorized access; The adaptive communication module integrates multiple communication protocols and intelligent algorithms to achieve interconnection and intelligent control between devices, improving the user experience and energy efficiency of the smart home system.

7. The portable modular edge AI device for use in a smart home system according to claim 1 or 4, characterized in that: The privacy desensitization module (223) includes a storage device, which is responsible for local data processing, for temporarily storing original data and desensitized data, and is connected to the adaptive communication module (23), supporting Wi-Fi, Bluetooth, Zigbee, etc., for communication between devices and with the cloud; and is connected to the DSP digital signal processor (213): obtaining original data from sensors and cameras: identifying sensitive information in the data, such as faces, license plates, etc.; Desensitize sensitive information, such as blurring and encrypting it, store the desensitized data and transmit it to the cloud or other devices.

8. A method for applying portable modular edge AI in a smart home system, characterized in that: The system management, artificial intelligence algorithm processing, storage of local AI models and data, and the operating system AIOS process of smart home applications running on the main control circuit board (21) and the edge computing module (22) are as follows: (1) The user interaction module (215) receives user input and sends instructions through the control module (217); (2) The sensor module (218) collects environmental data and sends it to the MCU processor (211) for processing; (3). The MCU processor (211) processes data and communicates with the smart home device through the adaptive communication module (219); (4) The NPU localized model inference chip (221) in the edge computing module (22) performs localized AI inference and processes complex tasks; (5) The GPU graphics processor (212) and the DSP digital signal processor (213) process graphics and signal data respectively; (6) The power management module (214) manages the power supply of the entire system to ensure low power consumption operation; (7) The storage module (216) and the storage device (224) store the processed data and the AI ​​model; (8) The expansion circuit board system module (23) provides additional functional expansion; And follow these steps: S1: The operating system connects the cloud and the device end, connects the home device through Wifi access, Bluetooth gateway, infrared access and Zigbee, and applies for cloud authorization to the cloud through configuration request, so that the modular edge AI can achieve interconnection and resource configuration in multiple scenarios and update authentication information. If the cloud authentication fails, the device authentication access management is directly provided to the edge computing module (22); the basic model sent from the cloud is retrained using local data, and the smart home detection model and feature extraction model are loaded into the AI ​​chip of the edge computing module (22) to identify the input information; S2: local data preprocessing (noise reduction, feature extraction). The edge processing module supports dynamic model switching. When it detects that the network bandwidth is higher than the threshold, it automatically downloads a high-precision AI model to replace the local lightweight model. S3: Edge model reasoning (dynamic loading of lightweight AI models), multi-device collaborative decision-making (direct communication between devices to avoid cloud transfer), the edge computing module performs AI model reasoning locally, reducing dependence on cloud servers; S4: Optimize the control strategy of smart home devices through machine learning algorithms, and implement the method steps of applying modular edge AI in a smart home system as described in any one of claims 1 to 7 when the operating system AIOS of the smart home application is executed by the processor.

Images