Old people falling detection system and method based on reserve pool calculation transistor

Abstract

The invention discloses an old people tumble detection system and method based on a reservoir calculation transistor. The system comprises a human body motion state monitoring sensing module and a physical reservoir calculation module. The human body motion state monitoring sensing module comprises a three-axis acceleration sensor and a main control chip; the three-axis acceleration sensor is integrated with the main control chip to serve as sensing end equipment and is used for collecting and sending inertial acceleration data in the movement process in real time; the physical reservoir calculation module comprises an electric double layer transistor and is used for calculating the ion relaxation process of the reservoir. The building method comprises the following steps: S1, building a physical reserve pool calculation module based on an electric double-layer transistor; s2, integrating a three-axis acceleration sensor with a main control chip, and building a human body motion state monitoring sensing module; and S3, integrating a human body motion state monitoring and sensing module and a physical reserve pool calculation module. The equipment can continuously work for a long time under the condition of low energy consumption, and is suitable for scenes needing long-time monitoring, such as a home nursing environment and a pension institution.

Description

Technical Field

[0001] The present invention belongs to the technical field of fall detection, and in particular relates to a system and method for detecting falls of the elderly based on a reserve pool computing transistor. Background Art

[0002] As people age, their sense of balance, muscle strength and bone density gradually weaken, and the risk of falling increases significantly. Falls can lead to serious consequences such as fractures, head injuries and soft tissue injuries, and the associated medical expenses and care costs are also a heavy social burden. Therefore, convenient and efficient fall detection technology is particularly important.

[0003] Edge computing is a technology that transfers computing power from the cloud to edge devices (such as sensors, cameras, smartphones, etc.) close to the data source. This method processes and analyzes data locally, improves real-time performance, reduces latency, and reduces bandwidth requirements and privacy risks for data transmission. Edge computing uses the computing resources of edge devices to process data, avoiding the process of transmitting large amounts of data to the cloud, thereby saving bandwidth and processing time. It is suitable for application scenarios that require fast response, such as real-time video analysis, smart homes, and industrial automation, and is also the data processing method for most fall detection. For example, Chinese patent CN114067390A (publication date February 18, 2022) adopts edge computing to process face and behavior recognition data with an independent system.

[0004] Among the many edge computing algorithms, reservoir computing is a computational method for processing time series data and dynamic systems. It has a simplified training process, powerful time-dependent processing capabilities, and good robustness. Reservoir computing uses a fixed, randomly connected recursive neural network (reservoir) to map input signals to a high-dimensional state space and predict outputs through a simple read layer. The connection weights inside the reservoir remain unchanged, and model training can be completed by training only the weights of the read layer. This method simplifies computational complexity and training time and reduces energy consumption. At the same time, its recursive structure retains historical information in the time series and is suitable for real-time processing tasks such as dynamic system modeling and pattern recognition.

[0005] The double-layer coupled transistor is a new type of electronic device that uses the double-layer effect to achieve efficient electrical signal regulation. Through the movement of ions at the interface between the electrolyte and the semiconductor, the DLCT forms a double layer with high capacitance, thereby regulating the conductivity of the transistor. By utilizing the high nonlinearity and time correlation provided by its ion relaxation process and constructing a hardware reserve pool virtual node, the hardware-based reserve pool calculation algorithm can be implemented.

[0006] The existing technology suffers from high computing load and energy consumption, mainly due to the limited computing resources of edge devices and the need to process complex data tasks; secondly, the system complexity and maintenance cost are high, and it is necessary to manage diverse sensor data and complex application requirements, which increases the challenges of system design and maintenance. Summary of the invention

[0007] By constructing electronic devices adapted to machine learning algorithms and building dedicated processing modules, the present application is expected to solve the above problems. The present invention proposes a system and method for detecting falls of the elderly based on a reserve pool computing transistor, wherein the system includes a human motion state sensing module and a physical reserve pool computing module;

[0008] The human body motion state sensing module includes a three-axis acceleration sensor and a main control chip;

[0009] The three-axis acceleration sensor is integrated with the main control chip as a sensor terminal device, which is used to collect and send inertial acceleration data during movement in real time;

[0010] The physical reserve pool calculation module includes a double-layer transistor, which is used for calculating the ion relaxation process of the reserve pool.

[0011] Furthermore, the method for constructing a fall detection system for the elderly based on a reserve pool computing transistor specifically includes:

[0012] Step S1, building a physical reserve pool calculation module based on double-layer transistors;

[0013] Step S2, integrating the three-axis acceleration sensor with the main control chip to build a human motion state sensing module;

[0014] Step S3, integrating the human body motion state sensing module and the physical reserve pool calculation module.

[0015] Furthermore, the step S1 specifically includes:

[0016] Step S101, preparing a patterned gate electrode, including metal aluminum Al and indium tin oxide ITO;

[0017] Step S102, preparing a patterned gate dielectric layer, including inorganic solid electrolyte silicon dioxide and ion gel chitosan;

[0018] Step S103, preparing a patterned semiconductor channel layer, including indium gallium zinc oxide IGZO and indium tin oxide ITO;

[0019] Step S104, preparing patterned source and drain electrodes, including metal aluminum Al and silver Ag;

[0020] Step S105: Use a wire bonding machine to wire the double-layer transistor array electrodes on the silicon wafer to a PCB board for packaging, so as to facilitate integration with a main control chip board.

[0021] Furthermore, in step S101, a patterned Al electrode is prepared by thermal evaporation: the mask plate and the nitrogen-doped silicon layer are clamped together flatly and tightly, placed in a sample holder, and evacuated to a vacuum degree of less than 10 -4 After Pa, Al is evaporated as the gate electrode.

[0022] Furthermore, in step S102, a patterned silicon dioxide film is deposited on the gate electrode using a plasma enhanced chemical vapor method: the mask and the gate electrode are clamped together flatly and tightly and placed in a sample holder. During the deposition process, 40 sccm SiH4 / N2 mixture and 120 sccm N2O gas are used as reaction gases to react under certain temperature conditions.

[0023] Furthermore, in step S103, a patterned IGZO channel layer is grown using a magnetron sputtering method: the mask and the silicon dioxide layer are clamped together flatly and tightly and placed in a sample holder. During the deposition process, the chamber pressure is set to 0.3 Pa, the power is 100 W, the argon flow rate is 30 sccm, and the sputtering is performed for 3 minutes.

[0024] Furthermore, in step S104, a patterned Al electrode is prepared by thermal evaporation: the mask plate and the IGZO layer are clamped together flatly and tightly, placed in a sample holder, and evacuated to a vacuum degree of less than 10 -4 After Pa, Al was evaporated as source and drain electrodes.

[0025] Furthermore, in step S1, the data collection process of the physical reserve pool calculation module includes:

[0026] Step a: The human motion state sensing module collects the signal of the three-axis acceleration sensor and multiplies it by a mask matrix to perform a time division multiplexing process; the mask matrix is ​​a randomly generated multi-value matrix ranging from -1 to 1;

[0027] Step b: The signal of the triaxial acceleration sensor is used to generate a pulse sequence with the average value as the threshold, a pulse greater than the threshold is defined as a pulse, and a pulse less than the threshold is defined as a resting state;

[0028] Step c: The pulse sequence is applied to the double-layer transistor, and the current at each time step is collected as the reservoir state.

[0029] Furthermore, in step S2, the human motion state sensing module uses a gyroscope to collect signals from a three-axis acceleration sensor; the human motion state sensing module transmits the signals to a main control chip through an I2C communication interface; and the main control chip uses a digital filtering algorithm to process the collected data, including the following steps:

[0030] Step 1: First use Kalman filtering to remove high-frequency noise and retain the valid acceleration signal;

[0031] Step 2: Secondly, use high-pass filtering to remove the DC component in the acceleration data and obtain dynamic acceleration information, including vibration signals;

[0032] Step 3: The processed signals finally form a three-axis acceleration data set.

[0033] Furthermore, the method of using the system includes:

[0034] Step A: The data set of the human motion state sensing module integrated with the three-axis acceleration sensor and the main control chip is collected, including the front-to-back fall, the left-side fall, the right-side fall, the stretching movement, the squatting movement, and the sitting movement, and each movement is collected no less than 200 times;

[0035] Step B: 90% of the acquired data set is used as a training set to generate reservoirs, and the training weights are obtained through one-step linear regression;

[0036] Step C: 10% of the collected data set is used as the test set to generate a reservoir of the test set, and the action label is output by multiplying the trained weights.

[0037] Compared with the prior art, the advantages and effects of this application are as follows:

[0038] 1. The module integrates the human motion state monitoring sensor module and the physical reserve pool calculation module. The MPU6050 and ESP32 chips are integrated as sensor devices to collect and send inertial acceleration data during movement in real time. Secondly, a physical reserve pool system constructed with double-layer-based IGZO artificial synaptic transistors is used for edge calculations. Thanks to the electrostatic coupling effect, double-layer transistors can show synaptic plasticity under low operating voltage drive. The ion relaxation process it exhibits can provide rich dynamic response currents as virtual nodes in the reserve pool, thereby significantly reducing computing energy consumption and reducing the working energy consumption of traditional edge devices to the nanojoule level.

[0039] 2. This application significantly reduces computing load and energy consumption through reserve pool computing, making the device more suitable for long-term monitoring and use. The power consumption level is at the nanowatt level, close to the energy consumption of biological neurons. The reserve pool computing device is used to achieve energy-efficient processing of sensor data and provide support for the long-term and continuous operation of edge devices. The equipment in this application can work continuously for a long time with low energy consumption. It is not only suitable for home care environments, but also for scenarios such as nursing homes that require long-term monitoring.

[0040] 3. This application simplifies the system design by integrating the human motion state monitoring sensor module with the physical reserve pool calculation module, reduces the multiple separate components and complex wiring required in the traditional method, and reduces the system complexity and maintenance costs. At the same time, this integrated design not only reduces the size of the device, making it more portable and easy to install, but also reduces production and assembly costs. In addition, since the system architecture is more concise, maintenance and upgrades also become more convenient. The time and human resources required for troubleshooting and parts replacement are reduced, further improving the reliability and service life of the equipment.

[0041] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application so that it can be implemented in accordance with the contents of the specification, and to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the following is a detailed description of the preferred embodiments of the present application in conjunction with the accompanying drawings as follows.

[0042] Based on the detailed description of the specific embodiments of the present application in combination with the accompanying drawings below, those skilled in the art will become more aware of the above and other objects, advantages and features of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings without creative work. In all drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, each element or part is not necessarily drawn according to the actual scale.

[0044] in:

[0045] Figure 1 It is the overall calculation framework diagram of the present invention;

[0046] Figure 2 The structure of the double-electric-layer-based IGZO artificial synaptic transistor and the nonlinear short relaxation process diagram provided by the present invention;

[0047] Figure 3The circuit structure of the human motion state sensing module provided by the present invention and the demonstration diagram of the collected acceleration signal;

[0048] Figure 4 Demonstration of reservoir state during calculation of the physical reserve pool provided by the present invention;

[0049] Figure 5 The test results for the feasibility verification of the present invention;

[0050] Figure 6 It is the overall flow chart of the present invention;

[0051] Figure 7 A diagram of a method for constructing a fall detection system for the elderly based on a reserve pool computing transistor according to the present invention. Specific embodiments

[0052] To make the purpose, technical scheme and advantages of the embodiment of the present application clearer, the technical scheme in the embodiment of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiment of the present application. Obviously, the described embodiment is a part of the embodiment of the present application, rather than all of the embodiments. In the following description, specific details such as specific configuration and components are provided only to help fully understand the embodiments of the present application. Therefore, it should be clear to those skilled in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. In addition, for clarity and brevity, the description of known functions and structures is omitted in the embodiment.

[0053] It should be understood that the references to "one embodiment" or "this embodiment" throughout the specification mean that the specific features, structures, or characteristics associated with the embodiment are included in at least one embodiment of the present application. Therefore, the references to "one embodiment" or "this embodiment" appearing throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0054] In addition, the present application may repeat reference numerals and / or letters in different examples. This repetition is for the purpose of simplicity and clarity, and does not in itself indicate the relationship between the various embodiments and / or settings discussed.

[0055] The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, B exists alone, and A and B exist at the same time. The term " / and" in this article describes another type of association object relationship, indicating that there can be two relationships. For example, A / and B can mean: A exists alone, and A and B exist alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0056] The term "at least one" in this article is merely a description of the association relationship of associated objects, indicating that there may be three relationships. For example, at least one of A and B can mean: A exists alone, A and B exist at the same time, and B exists alone.

[0057] It should also be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusions.

[0058] Example 1

[0059] This embodiment introduces a fall detection system for the elderly based on a reserve pool computing transistor. Please refer to Figure 1 As shown, a fall detection system for the elderly based on a reservoir transistor includes a human motion state monitoring sensor module;

[0060] Furthermore, the human motion state monitoring sensor module includes an MPU6050 three-axis acceleration sensor and an ESP32 main control chip;

[0061] For further information, please refer to Figure 2 As shown, a fall detection system for the elderly based on a reservoir transistor includes a physical reservoir computing module;

[0062] The physical reserve pool calculation module is composed of N double-layer transistors, and its structure and corresponding nonlinear short-term relaxation process are used for calculating the ion relaxation process of the reserve pool.

[0063] The technical effect of this embodiment is as follows: by integrating the human motion state monitoring sensor module and the physical reserve pool calculation module, the system design is simplified, the multiple separate components and complex wiring required in the traditional method are reduced, and the system complexity and maintenance cost are reduced.

[0064] Example 2

[0065] This embodiment introduces the operation process of an elderly fall detection system based on a reserve pool computing transistor. Please refer to Figure 3 As shown, further introduction is given based on the above embodiments.

[0066] The complete data collection process of the human motion state monitoring sensor module is as follows:

[0067] The sensor uses a gyroscope to collect three-axis acceleration signals and transmits them to the main control chip through the I2C communication interface. The digital filtering algorithm of the main control chip processes the collected data:

[0068] Step 1: First use Kalman filtering to remove high-frequency noise and retain the valid acceleration signal;

[0069] Step 2: Next, use high-pass filtering to remove the DC component in the acceleration data and obtain dynamic acceleration information, such as vibration signals.

[0070] Step 3: The processed signals finally form a three-axis acceleration data set.

[0071] For further information, please refer to Figure 2 As shown, the complete calculation process of the physical reserve pool calculation module is as follows:

[0072] Step a: First, the three-axis acceleration signal collected by the sensor end is multiplied by the mask matrix for time division multiplexing, wherein the mask matrix is ​​a randomly generated multi-value matrix ranging from -1 to 1.

[0073] Step b: Secondly, the signal generated by the previous process is used to generate a pulse sequence with the average value as the threshold, and the signal greater than the threshold is defined as a pulse (time step is 50ms, pulse duration is 40ms, pulse interval is 10ms), and the signal less than the threshold is defined as a resting state (time step is 10ms, pulse interval is 10ms).

[0074] Step c: Finally, the generated pulse train is applied to the transistor and the current at each time step is collected as the reservoir state.

[0075] For a demonstration of reservoir status, please refer to Figure 4 As shown, in the whole process, N different mask matrices are used to generate different pulse sequences, and the N pulse sequences are applied to N transistors, so the current states of the N transistors are collected to serve as a reservoir.

[0076] For further information, please refer to Figure 5 As shown, the feasibility of the fall detection system of the present invention is verified:

[0077] Firstly, a data set is collected through the human motion state sensing module integrated by MPU6050 and ESP32, which includes six actions (three falling actions: falling from front to back, falling to the left and falling to the right; three daily actions: stretching, squatting and sitting down), and each action is collected 200 times.

[0078] Secondly, 90% of the data set is used as a training set to generate the reservoir, and the training weights are obtained through one-step linear regression

[0079] Finally, 10% of the dataset is used as a test set to generate a reservoir for the test set, and the action label is output by multiplying the trained weights.

[0080] The average recognition rate of falling actions and normal actions reached 94.89%.

[0081] The single-reservoir power consumption and energy consumption of the system to achieve this recognition rate are shown in Table 1. The extremely low power consumption and energy consumption are precisely the advantages of the system.

[0082] Table 1 Energy consumption and power consumption data of single reservoir

[0083] Average current Read and write voltage Pulse Width Average power consumption per reservoir Average energy consumption per reservoir 9.8nA 1V 40ms 9.8nW 0.39nJ

[0084] Please refer to Figure 6 As shown, the entire operation process includes: after the human motion state sensing module detects the human motion acceleration signal, it is converted into a pulse sequence after time division multiplexing, the pulse sequence is applied to the double-layer transistor, and then the current state is obtained to perform physical reserve pool calculations, and finally output and determine whether it is a fall behavior.

[0085] The technical effect of this embodiment is: the computing load and energy consumption are significantly reduced through the reserve pool calculation, making the device more suitable for long-term monitoring and use, and the power consumption level is at the nanowatt level, close to the energy consumption of biological neurons. The reserve pool computing device is used to achieve energy-efficient processing of sensor data, providing support for the long-term and continuous operation of edge devices.

[0086] Example 3

[0087] This embodiment introduces a method for building a fall detection system for the elderly based on a reserve pool computing transistor. Please refer to Figure 7 As shown, based on the above embodiment, the method is further described, and specifically includes:

[0088] Step S1, building a physical reserve pool calculation module based on double-layer transistors;

[0089] Step S2, integrating the three-axis acceleration sensor with the main control chip to build a human motion state monitoring sensor module;

[0090] Step S3, integrating a human motion state monitoring sensor module and a physical reserve pool calculation module;

[0091] Furthermore, the step S1 specifically includes:

[0092] Step S101, preparing a patterned gate electrode (such as metal aluminum Al, indium tin oxide ITO, etc.);

[0093] Step S102, preparing a patterned gate dielectric layer (such as inorganic solid electrolyte silicon dioxide, ion gel chitosan, etc.);

[0094] Step S103, preparing a patterned semiconductor channel layer (such as indium gallium zinc oxide IGZO, indium tin oxide ITO, etc.);

[0095] Step S104, preparing patterned source and drain electrodes (such as metal aluminum Al, silver Ag, etc.);

[0096] Step S105: Use a wire bonding machine to wire the double-layer transistor array electrodes on the silicon wafer to a PCB board for packaging, so as to facilitate integration with a main control chip board.

[0097] Furthermore, in step S101, a patterned Al electrode is prepared by thermal evaporation: the mask plate and the nitrogen-doped silicon layer are clamped together flatly and tightly, placed in a sample holder, and evacuated to a vacuum degree of less than 10 -4 After Pa, Al is evaporated as the gate electrode.

[0098] Furthermore, in step S102, a patterned silicon dioxide film is deposited on the gate electrode using a plasma enhanced chemical vapor deposition (PECVD) method: the mask and the gate electrode are clamped together flatly and tightly and placed in a sample holder. During the deposition process, 40 sccm SiH4 / N2 mixture (5% SiH4 / 95% N2) and 120 sccm N2O gas are used as reaction gases, and the reaction is carried out at a temperature of 140°C for 10 minutes.

[0099] Furthermore, in step S103, a patterned IGZO channel layer is grown using a magnetron sputtering method: the mask and the silicon dioxide layer are clamped together flatly and tightly and placed in a sample holder. During the deposition process, the chamber pressure is set to 0.3 Pa, the power is 100 W, the argon flow rate is 30 sccm, and the sputtering is performed for 3 minutes.

[0100] Furthermore, in step S104, a patterned Al electrode is prepared by thermal evaporation: the mask plate and the IGZO layer are clamped together flatly and tightly, placed in a sample holder, and evacuated to a vacuum degree of less than 10 -4 After Pa, Al was evaporated as source and drain electrodes.

[0101] Furthermore, in step S2, the three-axis acceleration sensor includes an MPU6050 model, and the main control chip includes an ESP32 model.

[0102] The technical effect of this embodiment is that the integrated design of the human motion state monitoring sensor module and the physical reserve pool calculation module not only reduces the size of the device, making it more portable and easy to install, but also reduces the production and assembly costs. In addition, since the system architecture is more concise, maintenance and upgrading are also more convenient. The time and human resources required for troubleshooting and component replacement are reduced, further improving the reliability and service life of the device.

[0103] The above description is only the preferred embodiment of the present invention, which does not limit the protection scope of the present invention. For those skilled in the art, the present invention can be modified and varied in various ways. Within the spirit and principle of the present invention, any change, modification, replacement, integration and parameter change of these embodiments by conventional substitution or capable of achieving the same function without departing from the principle and spirit of the present invention shall fall within the protection scope of the present invention.

Claims

1. A fall detection system for the elderly based on a reserve pool computing transistor, characterized in that: The system includes a human body motion state sensing module and a physical reserve pool calculation module; The human body motion state sensing module includes a three-axis acceleration sensor and a main control chip; The three-axis acceleration sensor is integrated with the main control chip as a sensor terminal device, which is used to collect and send inertial acceleration data during movement in real time; The physical reserve cell calculation module includes a double-layer transistor, which is used for calculating the ion relaxation process of the reserve cell.

2. The method for constructing an elderly fall detection system based on a reserve pool computing transistor according to claim 1, characterized in that: The method comprises: Step S1, building a physical reserve pool calculation module based on double-layer transistors; Step S2, integrating the three-axis acceleration sensor with the main control chip to build a human motion state sensing module; Step S3, integrating the human body motion state sensing module and the physical reserve pool calculation module.

3. The method for constructing an elderly fall detection system based on a reserve pool computing transistor according to claim 2, characterized in that: The step S1 specifically includes: Step S101, preparing a patterned gate electrode, including metal aluminum Al and indium tin oxide ITO; Step S102, preparing a patterned gate dielectric layer, including inorganic solid electrolyte silicon dioxide and ion gel chitosan; Step S103, preparing a patterned semiconductor channel layer, including indium gallium zinc oxide IGZO and indium tin oxide ITO; Step S104, preparing patterned source and drain electrodes, including metal aluminum Al and silver Ag; Step S105: Use a wire bonding machine to wire the double-layer transistor array electrodes on the silicon wafer to a PCB board for packaging, so as to facilitate integration with a main control chip board.

4. The method for constructing an elderly fall detection system based on a reserve pool computing transistor according to claim 3 is characterized in that: In step S101, a patterned Al electrode is prepared by thermal evaporation: the mask plate and the nitrogen-doped silicon layer are clamped together flatly and tightly, placed in a sample holder, and evacuated to a vacuum degree of less than 10 -4 After Pa, Al is evaporated as the gate electrode.

5. The method for constructing an elderly fall detection system based on a reserve pool computing transistor according to claim 3, characterized in that: In step S102, a patterned silicon dioxide film is deposited on the gate electrode using a plasma enhanced chemical vapor method: the mask and the gate electrode are clamped together flatly and tightly and placed in a sample holder. During the deposition process, 40 sccm SiH4 / N2 mixture and 120 sccm N2O gas are used as reaction gases to react under certain temperature conditions.

6. The method for constructing an elderly fall detection system based on a reserve pool computing transistor according to claim 3, characterized in that: In step S103, a patterned IGZO channel layer is grown by magnetron sputtering: the mask and the silicon dioxide layer are clamped together flatly and tightly and placed in a sample holder. During the deposition process, the chamber pressure is set to 0.3 Pa, the power is 100 W, the argon flow rate is 30 sccm, and the sputtering is performed for 3 minutes.

7. The method for constructing an elderly fall detection system based on a reserve pool computing transistor according to claim 3, characterized in that: In step S104, a patterned Al electrode is prepared by thermal evaporation: the mask plate and the IGZO layer are clamped together flatly and tightly, placed in a sample holder, and evacuated to a vacuum degree of less than 10 -4 After Pa, Al was evaporated as source and drain electrodes.

8. The method for constructing an elderly fall detection system based on a reserve pool computing transistor according to claim 2, characterized in that: In step S1, the data collection process of the physical reserve pool calculation module includes: Step a: The human motion state sensing module collects the signal of the three-axis acceleration sensor and multiplies it by a mask matrix to perform a time division multiplexing process; the mask matrix is ​​a randomly generated multi-value matrix ranging from -1 to 1; Step b: The signal of the triaxial acceleration sensor is used to generate a pulse sequence with the average value as the threshold, a pulse greater than the threshold is defined as a pulse, and a pulse less than the threshold is defined as a resting state; Step c: The pulse sequence is applied to the double-layer transistor, and the current at each time step is collected as the reservoir state.

9. The method for constructing an elderly fall detection system based on a reserve pool computing transistor according to claim 2, characterized in that: In step S2, the human body motion state sensing module uses a gyroscope to collect signals from a three-axis acceleration sensor; The human body motion state sensor module transmits the data to the main control chip through the I2C communication interface; the main control chip uses a digital filtering algorithm to process the collected data, including: Step 1: First use Kalman filtering to remove high-frequency noise and retain the valid acceleration signal; Step 2: Secondly, use high-pass filtering to remove the DC component in the acceleration data and obtain dynamic acceleration information, including vibration signals; Step 3: The processed signals finally form a three-axis acceleration data set.

10. The method for constructing an elderly fall detection system based on a reserve pool computing transistor according to claim 2, characterized in that: The method of using the system includes: Step A: The data set of the human motion state sensing module integrated with the three-axis acceleration sensor and the main control chip is collected, including the front-to-back fall, the left-side fall, the right-side fall, the stretching movement, the squatting movement, and the sitting movement, and each movement is collected no less than 200 times; Step B: 90% of the acquired data set is used as a training set to generate reservoirs, and the training weights are obtained through one-step linear regression; Step C: 10% of the collected data set is used as the test set to generate a reservoir of the test set, and the action label is output by multiplying the trained weights.

Images