Self-powered wearable monitoring system and method based on lightweight machine learning

Abstract

The invention discloses a self-powered wearable monitoring system and method based on lightweight machine learning, and the monitoring system organically integrates a multi-type distributed sensing architecture, lightweight machine learning perception, and self-powered energy supply and demand balance into a flexibly integrated wearable system without depending on cloud computing resources. The real-time performance of intelligent sensing of the motion state is improved, tasks and energy consumption of wireless data transmission are reduced, and meanwhile, a traditional battery with limited capacity is replaced through efficient electric energy conversion of the human body kinetic energy collector and a working mode dynamic switching mechanism based on event triggering. And long-time stable and sustainable real-time intelligent monitoring of the motion state is realized on the wearable side. The monitoring system has high reconfigurability in the aspects of sensing architecture, intelligent sensing and the like, and provides a feasible idea and possibility of intelligent wearing for constructing a remote active medical blueprint for applications of sports injury rehabilitation, chronic disease monitoring and the like.

Description

Technical Field

[0001] The present invention belongs to the technical field of health monitoring, and in particular relates to a self-powered wearable monitoring system and method based on lightweight machine learning. Background Art

[0002] Technological progress has promoted the development and changes of social life. While people's pursuit of sports and health is constantly improving, they are also facing new problems such as sub-health in life. The demand for wearable technology in chronic disease monitoring, sports rehabilitation and competitive status analysis is developing rapidly. It plays an irreplaceable role in improving the efficiency of individual health management and promoting the health level of the whole people. It is an indispensable and important tool for modern health management.

[0003] Wearable technology has great application potential in the fields of sports health monitoring, which helps improve the efficiency and convenience of universal health services, helps users understand their own health status, and timely warns of potential problems, promoting active health management. Today, wearable technology has entered a diversified development stage, which is inseparable from the cross-cooperation of cutting-edge technologies such as flexible electronics, Internet of Things communications, artificial intelligence, and ubiquitous energy collection to solve the many challenges and problems currently faced, and move towards intelligence, lightness, and personalization.

[0004] As the main means of achieving long-term monitoring of human motion status, smart wearable technology still has problems such as single sensor parameters, low intelligence, long algorithm delay, high power consumption and power supply limitations that need to be solved urgently. Many wearable monitoring systems lack collaborative optimization in sensor design, algorithm deployment, energy management, etc., and the system integration is insufficient. For example, the patent "An Intelligent Wearable Fatigue Monitoring and Early Warning System" combines ECG signal monitoring with human fatigue status recognition, transmits the user's vital sign parameters to the remote platform through wireless transmission, and completes the feature extraction and recognition calculation of the ECG signal. However, the continuous collection and transmission of the original vital sign parameters brings higher working energy consumption and redundant operation. There is a certain delay in the data interaction process between the wearable device and the remote platform, which limits the sustainability and real-time response of autonomous intelligent monitoring in actual application scenarios.

[0005] Traditional wearable technology faces challenges in terms of rich vital sign sensing, autonomous intelligent perception, and energy supply and demand balance. Starting from sensor deployment, state perception, energy balance and other aspects, building a reconfigurable flexible wearable self-powered intelligent perception system with strong autonomy, good reliability, high accuracy, flexibility and effectiveness is the mainstream trend of the current development of intelligent wearable technology. Summary of the invention

[0006] The present invention provides a self-powered wearable monitoring system and method based on lightweight machine learning. The wearable detection system can locally execute the machine learning model through a microprocessor, can instantly perform intelligent analysis and feedback of multiple types of sensor data collected in real time, and organically combine the efficient collection of human kinetic energy with the low-power consumption optimization design of the system, thereby realizing sustainable intelligent monitoring.

[0007] In order to achieve the above object, the present invention adopts the following specific technical solutions: The present invention provides a self-powered wearable monitoring system based on lightweight machine learning, which includes an integrated multi-type human body sensor, a signal processing circuit, a microprocessor, a wireless communication module, a human body kinetic energy collector, and an electric energy management module; The multiple types of human body sensors are used to collect vital sign parameters of the human body; The signal processing circuit is used to process the sensing signals collected by the multi-type human body sensors, convert the sensing signals into electrical signals, and convert them into a voltage range compatible with the microprocessor; The microprocessor is connected to the signal processing circuit signal, and is used to process the electrical signal from the signal processing circuit through a locally deployed lightweight machine learning model to obtain vital sign information and intelligent perception results of motion status, realize real-time monitoring of human motion status, and dynamically adjust the working mode according to external trigger events to reduce the average power consumption of the system; The wireless communication module is used to wirelessly transmit the vital sign information and perception results obtained by the microprocessor, and has a data transceiver interface; The human body kinetic energy collector is used to collect the low-frequency random kinetic energy of the human body and convert the kinetic energy into electrical energy and store it in the electrical energy management module; The power management module is used to store the energy captured by the human kinetic energy collector and output a stable power supply voltage to the multiple types of human sensors, the signal processing circuit, the microprocessor and the wireless communication module to achieve long-term stable and sustainable intelligent monitoring.

[0008] Furthermore, it also includes Flexible Printed Circuit (FPC); The multi-type human body sensors are connected to the flexible circuit board through FPC cables and FPC connectors; The signal processing circuit, the microprocessor and the wireless communication module are all integrated on the flexible circuit board.

[0009] Furthermore, the multi-type human body sensor includes at least two types of vital sign parameter sensing modules, wherein at least one of the vital sign parameter sensing modules adopts a flexible multi-channel distributed sensing design.

[0010] Furthermore, the signal processing circuit includes a measurement circuit connected to the multi-type human body sensor signals and a time-division multiplexing module based on a multi-way switch; the multi-way switch is used to realize the time-division multiplexing of the measurement circuit.

[0011] Furthermore, the multi-type human body sensors include a multi-channel flexible film plantar pressure sensor and an inertial meter; The multi-channel flexible thin film plantar pressure sensor is made into a shoe insole shape and placed on the top surface of the sole; The inertial meter is integrated into the flexible circuit board to fit the curved surface of the foot to reduce the burden of wearing; The signal processing circuit includes a multiplexed scanning foot pressure sensing channel connected to the multi-channel flexible film plantar pressure sensor signal; the multiplexed scanning foot pressure sensing channel is used for time-sharing capture of the measurement circuit; The flexible circuit board, the human kinetic energy collector and the electric energy management circuit are placed on the side interlayer of the shoe body.

[0012] Furthermore, the wireless communication module is a low-power Bluetooth wireless communication module and has a normal working mode and a sleep mode.

[0013] Furthermore, the human kinetic energy collector converts low-frequency random linear reciprocating motion into rotational motion based on a secondary pendulum nonlinear mechanism to generate electrical energy output.

[0014] Furthermore, the lightweight machine learning model is a convolutional neural network model, a binary neural network model or a pulse neural network model.

[0015] Furthermore, the microprocessor adopts a wearable scenario embedded chip.

[0016] In addition, the present invention also provides a monitoring method based on the above-mentioned wearable monitoring system, and the monitoring method comprises the following steps: Step 1: Multi-type sensor data collection and data set establishment: Under the set sampling frequency, all sensor data of different individuals in each motion state are collected and saved, the time series data are aligned and divided into segments, and the motion state perception data set for building machine learning models is established; Step 2: Construction and lightweight deployment of machine learning models. Design machine learning models for intelligent perception monitoring of human motion status, use motion status perception data sets to train, verify and test the models; quantize, compress and optimize the trained model parameter files, and convert them into a format and capacity executable by microprocessors, so as to deploy them lightweight on microprocessors and realize local real-time processing of intelligent perception algorithms. Step 3: Design a low-power event trigger algorithm. Set an external event under a predetermined state. When and only when the event is triggered, the monitoring system enters the normal working mode. After intelligent perception and analysis of the human body's motion state, it immediately enters the low-power sleep mode, achieving efficient event capture and dynamic intelligent monitoring while ensuring that the power consumption of the monitoring system is maintained below 20mW, maintaining the energy supply and demand balance with the human kinetic energy harvester. Step 4: Acquisition and display of motion state perception results. The monitoring system locally executes the machine learning model to process the sensor data collected in real time, and instantly obtains the perception results of the current motion state, which can be directly displayed on the wearable side or transmitted to a remote display through the wireless communication module.

[0017] Compared with the prior art, the technical solution of the present invention has the following beneficial effects: 1. The wearable monitoring system of the present invention integrates multi-type distributed sensing architecture, lightweight machine learning perception and self-powered energy supply and demand balance. Multi-type vital sign parameter sensing and low-power intelligent perception of motion status are realized on the wearable side, and long-term stable and sustainable intelligent monitoring is realized through the coordinated cooperation of human kinetic energy collector and low-power event trigger algorithm, providing feasible ideas and possibilities for intelligent wearables to build a remote active medical blueprint for applications such as sports injury rehabilitation and chronic disease monitoring.

[0018] 2. The present invention constructs a distributed sensing architecture of multiple vital sign parameters based on human body characteristics. The flexible, multi-channel sensing design can efficiently capture the dynamic changes of multiple vital sign parameters, ensure the consistency of the sensing signal with the human body state characteristics, and provide data guarantee for the expansion of intelligent perception and continuous monitoring of complex human body states.

[0019] 3. The present invention realizes the deployment and application of machine learning on the wearable microprocessor through lightweight algorithm. The lightweight route is to use the microprocessor to execute the machine learning method locally for intelligent perception, which has the advantages of reducing the workload of wireless data transmission, reducing the power consumption of data sensing, and shortening the system response time. It can realize the real-time and localized lightweight perception processing of sensor data, which meets the development needs of personalized health services.

[0020] 4. The present invention uses human kinetic energy collectors to replace traditional batteries, and designs a dynamic switching mechanism for working modes based on event triggering to reduce power consumption. The two work together to ensure the energy supply and demand balance of the system, converting and storing the low-frequency random kinetic energy of the human body into electrical energy, and realizing an efficient and low-power monitoring mode. By rationally utilizing effective sensor information and dynamically adjusting the working mode according to the characteristics of human movement, the average power consumption of the system can be significantly reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 Schematic diagram of the side structure of the wearable system according to an embodiment of the present invention.

[0022] Figure 2 The figure is a flow chart of a monitoring method according to an embodiment of the present invention.

[0023] Figure numerals: 1-shoe body, 2-multi-channel flexible thin film plantar pressure sensor, 3-shoe sole, 4-inertial meter, 5-flexible circuit board, 6-human kinetic energy collector. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] Embodiment 1 As an important support and driving force for human movement, the foot's physical parameters can directly reflect gait characteristics, force distribution and movement patterns, and have important monitoring value. It has a wide range of potential applications in injury risk, rehabilitation training, etc. The side view structure of the wearable monitoring system is shown in Figure 1. Figure 1 As shown, the wearable device in this embodiment is illustrated by taking sports shoes as an example. Based on the foot pressure distribution and acceleration, this embodiment provides a self-powered wearable monitoring system based on lightweight machine learning. The wearable monitoring system is integrated between the shoe body 1 and the sole 3 of the sports shoe, and includes an integrated multi-type human body sensor, a signal processing circuit, a microprocessor, a wireless communication module, a human kinetic energy collector 6, an electric energy management module and a flexible circuit board 5; the signal processing circuit, the microprocessor and the wireless communication module are all integrated in the flexible circuit board 5.

[0026] The multi-type human body sensors are connected to the flexible circuit board 5 via FPC cables and FPC connectors, and are used to collect the vital sign parameters of the human body; the multi-type human body sensors include at least two vital sign parameter sensing modules, at least one of which adopts a flexible multi-channel distributed sensing design. The vital sign parameter sensing module adopts a flexible and multi-channel array design, which can better fit the human skin and adapt to individual differences, accurately obtain the distributed information of vital sign parameters, and is the source of vital sign parameter sensing for the signal processing circuit. Since the foot is an important support and power part of the human body movement, its vital sign parameters can directly reflect the gait characteristics, force distribution and movement pattern, and have important monitoring value, and have potential for wide application in injury risk, rehabilitation training and other aspects. Figure 1As shown in the structure, the multi-type human body sensors installed in the sports shoes include a multi-channel flexible film plantar pressure sensor 2 and an inertial meter 4; the multi-channel flexible film plantar pressure sensor 2 is made into an insole shape and placed on the top surface of the sole 3; the inertial meter 4 is integrated in the flexible circuit board 5, which is used to fit the curved surface of the foot to reduce the burden of wearing.

[0027] The signal processing circuit is used to process the sensing signals collected by various types of human body sensors, convert the sensing signals into electrical signals, and convert them into a voltage range compatible with the microprocessor; the signal processing circuit includes a measurement circuit connected to the signals of various types of human body sensors and a time-division multiplexing module based on a multi-way switch; the multi-way switch is used to realize the time-division multiplexing of the measurement circuit.

[0028] The microprocessor is connected to the signal processing circuit signal, and is used to process the electrical signal from the signal processing circuit through the locally deployed lightweight machine learning model to obtain the vital sign information and intelligent perception results of the motion state, realize the real-time monitoring of the human motion state, and dynamically adjust the working mode according to the external trigger event to reduce the average power consumption of the system; the microprocessor is the computing core of the system to realize the intelligent monitoring of the human motion state, and adopts the wearable scene embedded chip with limited computing power but low power consumption. The lightweight machine learning model is a convolutional neural network (CNN) model, a binary neural network (BNN) model or a pulse neural network (SNN) model. The microprocessor can realize the low-power instant intelligent perception of real-time sensor data through the locally deployed lightweight machine learning model, without transmitting a large amount of raw sensor data to the remote platform, and can directly obtain the perception results of the motion state on the wearable side.

[0029] The wireless communication module is used to wirelessly transmit the vital sign information and perception results obtained through the microprocessor, and has a data transceiver interface; the wireless communication module is a low-power Bluetooth (BLE) wireless communication module, and has a normal working mode and a sleep mode.

[0030] The human kinetic energy collector 6 is used to collect the low-frequency random kinetic energy of the human body, and convert the kinetic energy into electrical energy and store it in the power management module; the human kinetic energy collector 6 converts the low-frequency random linear reciprocating motion into rotational motion based on the secondary pendulum nonlinear mechanism to generate electrical energy output. The power management module is used to store the energy captured by the human kinetic energy collector 6, and output a stable power supply voltage to multiple types of human sensors, signal processing circuits, microprocessors and wireless communication modules, so as to achieve long-term stable and sustainable intelligent monitoring.

[0031] When the wearable device is a sports shoe, the multi-type human body sensor includes a multi-channel flexible film plantar pressure sensor 2 and an inertial meter 4; the multi-channel flexible film plantar pressure sensor 2 is made into an insole shape and placed on the top surface of the sole 3; the inertial meter 4 is integrated in the flexible circuit board 5, which is used to fit the curved surface of the foot to reduce the wearing burden; the signal processing circuit includes a multiplexed scanning foot pressure sensing channel connected to the signal of the multi-channel flexible film plantar pressure sensor 2; the multiplexed scanning foot pressure sensing channel is used for time-sharing capture of the measurement circuit; the flexible circuit board 5, the human kinetic energy collector 6 and the power management circuit are placed in the side interlayer of the shoe body 1, which can improve the kinetic energy collection efficiency as much as possible, make full use of the kinetic energy generated by the leg swing process, and maintain the energy supply of the system for long-term stable intelligent monitoring. The signal processing circuit includes a multiplexed scanning foot pressure sensing channel for time-sharing capture of its measurement circuit.

[0032] The above-mentioned wearable monitoring system integrates multiple types of human sensors, lightweight machine learning perception models, and self-powered energy supply and demand balance. On the wearable side, it realizes multiple types of vital sign parameter sensing and low-power intelligent perception of motion status, and realizes long-term stable and sustainable intelligent monitoring through the coordinated cooperation of human kinetic energy collector and low-power event trigger algorithm, providing feasible ideas and possibilities for intelligent wearables to build a remote active medical blueprint for applications such as sports injury rehabilitation and chronic disease monitoring.

[0033] A distributed sensing architecture with multiple vital sign parameters is constructed based on human characteristics. The flexible, multi-channel sensing design can efficiently capture the dynamic changes of multiple vital sign parameters, ensure the consistency of sensing signals with human state characteristics, and provide data guarantee for the expansion of intelligent perception and continuous monitoring of complex human states.

[0034] By lightweighting the algorithm, machine learning can be deployed and applied in the wearable microprocessor. The lightweight route is to locally execute machine learning methods on the microprocessor for intelligent perception. This has the advantages of reducing the amount of wireless data transmission tasks, reducing data sensing power consumption, and shortening system response time. It can realize real-time, localized, lightweight perception processing of sensor data, which meets the development needs of personalized health services.

[0035] The wearable monitoring system uses human kinetic energy harvesters to replace traditional batteries, and designs a dynamic switching mechanism for working modes based on event triggering to reduce power consumption. The two work together to ensure the energy supply and demand balance of the system, converting and storing the low-frequency random kinetic energy of the human body into electrical energy, and realizing an efficient and low-power monitoring mode. By rationally utilizing effective sensor information and dynamically adjusting the working mode according to the characteristics of human movement, the average power consumption of the system can be significantly reduced.

[0036] Embodiment 2 This embodiment provides a monitoring method based on the above wearable monitoring system, and the monitoring method includes the following steps: Step 1: Multi-type sensor data collection and data set establishment: Under the set sampling frequency, all sensor data of different individuals in each motion state are collected and saved, the time series data are aligned and divided into segments, and the motion state perception data set for building machine learning models is established; Step 2: Construction and lightweight deployment of machine learning models. Design machine learning models for intelligent perception monitoring of human motion status, use motion status perception data sets to train, verify and test the models; quantize, compress and optimize the trained model parameter files, and convert them into a format and capacity executable by microprocessors, so as to deploy them lightweight on microprocessors and realize local real-time processing of intelligent perception algorithms. Step three, design a low-power event trigger algorithm, set an external event under a predetermined state, and the monitoring system enters a normal working mode when and only when the event is triggered. After intelligent perception and analysis of the human body's movement state, it immediately enters a low-power sleep mode, achieving efficient event capture and dynamic intelligent monitoring while ensuring that the power consumption of the monitoring system is maintained below 20mW, maintaining the energy supply and demand balance with the human kinetic energy collector 6; the predetermined state can be a sudden change in a certain sensor signal when the human body changes state or the body becomes abnormal, such as: acceleration signal, plantar pressure signal, etc.

[0037] Step 4: Acquisition and display of motion state perception results. The monitoring system locally executes the machine learning model to process the sensor data collected in real time, and instantly obtains the perception results of the current motion state, which can be directly displayed on the wearable side or transmitted to the remote display through the wireless communication module.

[0038] Embodiment 3 According to the above-mentioned second embodiment, a specific method for intelligent monitoring of human motion state is provided, which is carried out in the following steps: collecting foot pressure and acceleration sensor data and establishing a data set, constructing and lightweight deployment of a convolutional neural network model, designing a low-power event triggering algorithm, acquiring and displaying motion state perception results, and realizing intelligent monitoring of various daily motion states such as standing, sitting, walking, running, squatting, etc., specifically: Step 1: Use the multi-channel distributed foot pressure and three-axis acceleration sensor acquisition unit to provide more comprehensive and accurate motion information. In order to obtain sufficient motion characteristics, the sampling frequency of the physical sign parameters is set to 100Hz. With the help of the low-power Bluetooth (BLE) wireless communication module and the host computer serial port, the original sensor data of all testers in different motion states are stored, the time series data is aligned and divided into segments, and a data set is established.

[0039] Step 2: Build a convolutional neural network (CNN) model and deploy lightweight perception computing on a low-power microcontroller. The low-complexity lightweight motion state convolutional neural network recognition algorithm comprehensively utilizes two types of sensor data and is deployed on the wearable side to intelligently monitor the user's motion state in real time. Without relying on cloud computing resources, low-latency response is achieved, reducing dependence on network connections and enhancing system stability and reliability.

[0040] Step 3: According to the characteristics of human motion, a trigger mechanism based on acceleration changes and low-power timer events was established, which significantly reduced the average power consumption of the system. The system will enter the normal working mode for intelligent perception only after detecting motion or sleeping for a set time, otherwise it will be in low-power sleep mode. The trigger time interval is determined by the human motion state. The system optimizes the redundant energy consumption process and cooperates with the human body's low-frequency random kinetic energy collector to achieve energy supply and demand balance.

[0041] Step 4: Use the Bluetooth Low Energy (BLE) wireless communication module to realize data transmission between the system and the host computer, and update the intelligent perception results of human motion status in real time on the host computer interface, with an accuracy rate of more than 95%.

[0042] Embodiment 4 The technical solutions of the above-mentioned embodiments 1 and 2 can also be implemented on the wrist of the human body. As an important hub of human upper limb movement, the wrist contains important signs such as pulse fluctuations, arm swing characteristics and movement patterns, and has high application value in cardiovascular health assessment, sports training monitoring and other aspects. In the daily state of the human body, the wrist also contains rich mechanical energy, thereby providing an energy source for the human kinetic energy collector 6 to realize system self-power supply. This embodiment provides a self-powered intelligent monitoring system flexibly integrated in the wrist, using a flexible pulse sensor and a six-axis inertial sensor distributed in a 3×3 matrix, combined with a human kinetic energy collector 6, to convert and store the low-frequency random kinetic energy changes generated by the arm swing into electrical energy, based on the low-power, low-latency intelligent perception of the human body state by the monitoring method of embodiment 2, and integrates multi-type sensing, local lightweight intelligence and self-powered operation on the wrist flexible wearable system.

[0043] In summary, the embodiments of the present invention realize lightweight self-powered intelligent perception of various daily motion states on the wearable side, integrate multiple types of vital sign parameter sensing and low-power intelligent perception of motion states into a wearable device integrated in the foot, power the system through the human kinetic energy collector 6, and visualize the monitoring results of sensor perception on the host computer.

[0044] Obviously, those skilled in the art can make various changes and modifications to the embodiments of the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A self-powered wearable monitoring system based on lightweight machine learning, characterized in that: It includes integrated multi-type human body sensors, signal processing circuits, microprocessors, wireless communication modules, human body kinetic energy collectors and power management modules; The multiple types of human body sensors are used to collect vital sign parameters of the human body; The signal processing circuit is used to process the sensing signals collected by the multi-type human body sensors, convert the sensing signals into electrical signals, and convert them into a voltage range compatible with the microprocessor; The microprocessor is connected to the signal processing circuit signal, and is used to process the electrical signal from the signal processing circuit through a locally deployed lightweight machine learning model to obtain vital sign information and intelligent perception results of motion status, realize real-time monitoring of human motion status, and dynamically adjust the working mode according to external trigger events to reduce the average power consumption of the system; The wireless communication module is used to wirelessly transmit the vital sign information and perception results obtained by the microprocessor, and has a data transceiver interface; The human body kinetic energy collector is used to collect the low-frequency random kinetic energy of the human body and convert the kinetic energy into electrical energy and store it in the electrical energy management module; The power management module is used to store the energy captured by the human kinetic energy collector and output a stable power supply voltage to the multiple types of human sensors, the signal processing circuit, the microprocessor and the wireless communication module to achieve long-term stable and sustainable intelligent monitoring.

2. The wearable monitoring system according to claim 1, characterized in that: Also included are flexible circuit boards; The multi-type human body sensors are connected to the flexible circuit board through FPC cables and FPC connectors; The signal processing circuit, the microprocessor and the wireless communication module are all integrated on the flexible circuit board.

3. The wearable monitoring system according to claim 2, characterized in that: The multi-type human body sensor includes at least two types of vital sign parameter sensing modules, wherein at least one of the vital sign parameter sensing modules adopts a flexible multi-channel distributed sensing design.

4. The wearable monitoring system according to claim 3, characterized in that: The signal processing circuit comprises a measurement circuit connected to the multi-type human body sensor signals and a time-division multiplexing module based on a multi-way switch; the multi-way switch is used to realize the time-division multiplexing of the measurement circuit.

5. The wearable monitoring system according to claim 4, characterized in that: The multi-type human body sensors include a multi-channel flexible film plantar pressure sensor and an inertial meter; The multi-channel flexible thin film plantar pressure sensor is made into a shoe insole shape and placed on the top surface of the sole; The inertial meter is integrated into the flexible circuit board to fit the curved surface of the foot to reduce the burden of wearing; The signal processing circuit includes a multiplexed scanning foot pressure sensing channel connected to the multi-channel flexible film plantar pressure sensor signal; the multiplexed scanning foot pressure sensing channel is used for time-sharing capture of the measurement circuit; The flexible circuit board, the human kinetic energy collector and the electric energy management circuit are placed on the side interlayer of the shoe body.

6. The wearable monitoring system according to any one of claims 1 to 5, characterized in that: The wireless communication module is a low-power Bluetooth wireless communication module and has a normal operation mode and a sleep mode.

7. A wearable monitoring system as claimed in any one of claims 1 to 5, characterized in that: The human body kinetic energy collector converts low-frequency random linear reciprocating motion into rotational motion based on the secondary pendulum nonlinear mechanism to generate electrical energy output.

8. The wearable monitoring system according to any one of claims 1 to 5, characterized in that: The lightweight machine learning model is a convolutional neural network model, a binary neural network model or a pulse neural network model.

9. The wearable monitoring system according to any one of claims 1 to 5, characterized in that: The microprocessor adopts a wearable scenario embedded chip.

10. A monitoring method based on the wearable monitoring system according to any one of claims 1 to 9, characterized in that: The following steps are involved: Step 1: Multi-type sensor data collection and data set establishment: Under the set sampling frequency, all sensor data of different individuals in each motion state are collected and saved, the time series data are aligned and divided into segments, and the motion state perception data set for building machine learning models is established; Step 2: Construction and lightweight deployment of machine learning models. Design machine learning models for intelligent perception monitoring of human motion status, use motion status perception data sets to train, verify and test the models; quantize, compress and optimize the trained model parameter files, and convert them into a format and capacity executable by microprocessors, so as to deploy them lightweight on microprocessors and realize local real-time processing of intelligent perception algorithms. Step 3: Design a low-power event trigger algorithm. Set an external event under a predetermined state. When and only when the event is triggered, the monitoring system enters the normal working mode. After intelligent perception and analysis of the human body's motion state, it immediately enters the low-power sleep mode, achieving efficient event capture and dynamic intelligent monitoring while ensuring that the power consumption of the monitoring system is maintained below 20mW, maintaining the energy supply and demand balance with the human kinetic energy harvester. Step 4: Acquisition and display of motion state perception results. The monitoring system locally executes the machine learning model to process the sensor data collected in real time, and instantly obtains the perception results of the current motion state, which can be directly displayed on the wearable side or transmitted to a remote display through the wireless communication module.

Images