An ai human body sensor and human body state monitoring system

Abstract

The application discloses an AI human body sensor and a human body state monitoring system, which comprises a millimeter wave radar module for emitting millimeter wave signals and receiving reflected signals; a signal processing module connected with the millimeter wave radar module for processing the reflected signals received by the millimeter wave radar module to obtain processing signals; an AI algorithm processing module connected with the signal processing module for human body existence detection, movement detection, fall detection and position detection based on the processing signals to obtain detection results; a communication module connected with the AI algorithm processing module for sending the detection results to external equipment; and a power module for supplying power to the millimeter wave radar module, the signal processing module, the AI algorithm processing module and the communication module. The application aims to solve the problems of single function, privacy leakage risk, low data processing efficiency and lack of linkage mechanism of the human body sensor in the prior art.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sensors, in particular to an AI human body sensor and a human body state monitoring system. BACKGROUND

[0002] At present, under the background of rapid development of Internet of Things and intelligent monitoring technology, human body sensors are widely used in smart home, smart care for the elderly, security monitoring and other fields. Traditional human body sensors, such as infrared sensors, can only detect human body movement signals and cannot accurately determine the presence state, position information and posture change of human body. When the human body is stationary, it is easy to miss detection, the monitoring function is single, and it is difficult to meet the use requirements in complex scenes.

[0003] Part of the human body monitoring scheme based on image recognition technology can obtain rich human body state information, but there is a serious risk of privacy leakage, especially in private scenes such as home and bedroom, the user's acceptance of image collection is low. In addition, the existing human body monitoring system often relies on cloud data processing, data transmission is delayed, and cannot work normally when the network is not good, and there are security risks in the data transmission process. In addition, different function sensors are independent of each other, lack of linkage mechanism, cannot form a complementary function ecosystem, resource utilization efficiency is low, and it is difficult to realize intelligent comprehensive management. SUMMARY

[0004] The present application aims to at least one of the above technical problems in the technical field. To this end, the purpose of the present application is to propose the AI human body sensor and the human body state monitoring system of the present application, which aims to solve the problems of single function of human body sensor, privacy leakage risk, low data processing efficiency and lack of linkage mechanism in the prior art.

[0005] To achieve the above purpose, the present application embodiment proposes an AI human body sensor, comprising:

[0006] A millimeter wave radar module for transmitting millimeter wave signals and receiving reflected signals;

[0007] A signal processing module connected with the millimeter wave radar module for processing the reflected signals received by the millimeter wave radar module to obtain a processing signal;

[0008] An AI algorithm processing module connected with the signal processing module for human body presence detection, movement detection, fall detection and position detection based on the processing signal to obtain a detection result;

[0009] A communication module connected with the AI algorithm processing module for sending the detection result to an external device;

[0010] A power module is configured to supply power to the millimeter wave radar module, the signal processing module, the AI algorithm processing module, and the communication module.

[0011] According to some embodiments of the present application, the signal processing module comprises:

[0012] A filtering module is configured to perform filtering processing on the reflected signal to obtain a filtered signal.

[0013] An amplification module is configured to perform amplification processing on the filtered signal to obtain an amplified signal.

[0014] An analog-to-digital conversion module is configured to perform analog-to-digital conversion processing on the amplified signal to obtain a digital signal as a processed signal.

[0015] According to some embodiments of the present application, the filtering module comprises:

[0016] A first generation module is configured to split the reflected signal into a plurality of sub-signals, determine the spectral amplitude of each sub-signal, calculate the average spectral amplitude of the plurality of sub-signals, calculate the absolute value of the difference between the spectral amplitude of each sub-signal and the average spectral amplitude, and filter out the sub-signals corresponding to the spectral amplitudes with absolute values greater than a preset difference value to generate a first set.

[0017] A second generation module is configured to analyze the reflected signal to determine all local maximum and minimum values of the reflected signal, fit the maximum value points by cubic spline interpolation to obtain an upper envelope line, fit the minimum value points by cubic spline interpolation to obtain a lower envelope line, calculate the mean value of the upper and lower envelope lines as a trend item, and subtract the trend item from the reflected signal to obtain an IMF component, and generate a second set based on the sub-signals corresponding to the IMF component.

[0018] A first determination module is configured to determine a target sub-signal based on the first set and the second set, analyze the type of the target sub-signal, perform filtering processing according to the type of the target sub-signal, perform signal reconstruction based on the processed target sub-signal, and obtain a filtered signal.

[0019] According to some embodiments of the present application, the first generation module splits the reflected signal into a plurality of sub-signals, comprising:

[0020] The width of the sliding window is determined based on the sampling rate of the reflected signal and the expected width of the peak, and the window step is set.

[0021] The width constraint condition is set, and the peak detection is performed in the sliding window based on the width constraint condition to obtain a plurality of peaks.

[0022] The peaks spanning multiple sliding windows are filtered out as to-be-merged peaks, and the to-be-merged peaks are merged to obtain a plurality of target peaks.

[0023] Determine the time point corresponding to each target wave crest, segment the reflection signal according to the time points corresponding to adjacent target wave crests to obtain a plurality of sub-signals.

[0024] According to some embodiments of the application, the AI algorithm processing module comprises:

[0025] The 3D-FFT-based processing signal is processed to obtain distance-Doppler-angle three-dimensional imaging information;

[0026] According to the distance-Doppler-angle three-dimensional imaging information and the LSTM prediction model, the respiratory signal existence is determined; according to the distance-Doppler-angle three-dimensional imaging information and the CNN classification module, the micro-motion feature is determined; and based on the respiratory signal existence and the micro-motion feature, the human existence detection is performed.

[0027] According to some embodiments of the application, the AI algorithm processing module further comprises a movement detection module configured to:

[0028] Based on the distance-Doppler-angle three-dimensional imaging information, trajectory analysis is performed; the trajectory analysis includes Kalman filter tracking and velocity mutation detection;

[0029] Based on the trajectory analysis result, when the processing signal appears continuous change in time sequence and meets the human movement feature, it is determined that the human is in a moving state, and the movement-related information is recorded.

[0030] According to some embodiments of the application, the AI algorithm processing module further comprises a fall detection module configured to:

[0031] Based on the distance-Doppler-angle three-dimensional imaging information, feature analysis is performed to determine feature information; the feature information includes vertical velocity mutation, attitude angle change and impact duration;

[0032] The feature information is input into a preset fall posture model for judgment; when the matching degree of the feature information and the fall posture model is greater than a preset threshold, it is determined that the human has a fall action.

[0033] According to some embodiments of the application, the AI algorithm processing module further comprises a position detection module configured to:

[0034] Based on the distance-Doppler-angle three-dimensional imaging information, time delay and angle information are determined;

[0035] According to the time delay and angle information, the Doppler positioning algorithm is used to determine the position information of the human in the region.

[0036] According to some embodiments of the present application, the system further comprises a guard alarm module configured to select a corresponding alarm mode based on a preset alarm rule according to the detection result and transmit the alarm information to a monitoring terminal through a wired or wireless mode.

[0037] According to some embodiments of the present application, a human body state monitoring system comprises a control module, an execution device, a server and an AI human body sensor as described above.

[0038] The control module is connected with the AI human body sensor and the server respectively, and is configured to:

[0039] receive the monitoring information of the human body state transmitted by the AI human body sensor and transmit the monitoring information to the server;

[0040] generate a corresponding control strategy according to the monitoring information, determine a corresponding execution device based on the control strategy, and transmit an execution instruction to the execution device;

[0041] The server is configured to acquire the monitoring information and the execution information of the execution device, generate a monitoring data table, and transmit the monitoring data table to a mobile terminal.

[0042] The present application provides an AI human body sensor and a human body state monitoring system, which realizes accurate detection of multi-dimensional information such as human body existence, position and posture by combining a millimeter wave radar module with a multi-in-one AI algorithm, and makes up for the defects of traditional sensors that cannot detect comprehensively; while ensuring the care function, the user privacy is strictly protected; with the linkage design of the communication module and the external device, based on the multi-dimensional detection results such as posture, area and direction, a rich intelligent linkage scene is triggered, and the intelligent level and practicability of the monitoring system are improved.

[0043] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application will be realized and achieved by means of the structures particularly pointed out in the written description and claims.

[0044] The technical solutions of the present application will be further described in detail below with reference to the drawings and embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0045] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, and are used to explain the present application together with embodiments of the present application, and do not constitute a limitation on the present application. In the drawings:

[0046] Figure 1 is a block diagram of an AI human body sensor according to an embodiment of the present application;

[0047] Figure 2is a block diagram of a human state monitoring system according to an embodiment of the present application. DETAILED DESCRIPTION

[0048] The preferred embodiments of the present application will be described below in conjunction with the accompanying drawings, it should be understood that the preferred embodiments described herein are only used to explain and illustrate the present application, and are not used to limit the present application.

[0049] As shown in Figure 1 An AI human sensor is provided according to an embodiment of the present application, comprising:

[0050] a millimeter wave radar module for transmitting millimeter wave signals and receiving reflected signals;

[0051] a signal processing module connected with the millimeter wave radar module, for processing the reflected signals received by the millimeter wave radar module to obtain a processed signal;

[0052] an AI algorithm processing module connected with the signal processing module, for performing human existence detection, movement detection, fall detection and position detection based on the processed signal to obtain a detection result;

[0053] a communication module connected with the AI algorithm processing module, for sending the detection result to an external device;

[0054] a power module for supplying power to the millimeter wave radar module, the signal processing module, the AI algorithm processing module and the communication module.

[0055] The working principle of the above technical solution is as follows: the millimeter wave radar module transmits a linear frequency modulation signal through FMCW technology, and obtains an intermediate frequency signal after mixing the received signal with the transmitted signal, and calculates the target distance and speed through FFT. Through the MIMO antenna array, the target angle is calculated by using the beam forming technology (such as the MUSIC algorithm). The signal processing module processes the reflected signals received by the millimeter wave radar module, including filtering, amplification, analog-to-digital conversion, etc. The AI algorithm processing module performs human existence detection, movement detection, fall detection and position detection based on the processed signal to obtain a detection result; the detection result is sent to an external device based on the communication module, which is convenient for triggering a rich intelligent linkage scene.

[0056] The above technical solution has the following beneficial effects: through the millimeter wave radar module combined with the all-in-one AI algorithm, the precise detection of multi-dimensional information such as human existence, position, posture, etc. is realized, which makes up for the defects of the traditional sensor detection being not comprehensive; while ensuring the care function, the user privacy is strictly protected; with the linkage design of the communication module and the external device, based on the multi-dimensional detection results such as posture, area, direction, etc., a rich intelligent linkage scene is triggered, which improves the intelligent level and practicality of the monitoring system.

[0057] According to some embodiments of the present application, the signal processing module comprises:

[0058] a filtering module configured to perform filtering processing on the reflected signal to obtain a filtered signal;

[0059] an amplification module configured to perform amplification processing on the filtered signal to obtain an amplified signal;

[0060] an analog-to-digital conversion module configured to perform analog-to-digital conversion processing on the amplified signal to obtain a digital signal as a processed signal.

[0061] The working principle of the above technical solution is that the filtering module is configured to suppress noise and interference and improve the accuracy of the signal. The amplification module is configured to increase the amplitude of the signal to ensure that the dynamic range of subsequent analog-to-digital conversion covers the effective signal. The analog-to-digital conversion module is configured to convert the analog signal into a digital signal for facilitating signal processing.

[0062] The beneficial effects of the above technical solution are that the signal processing module converts the reflected signal of the millimeter wave radar into a high-quality digital signal through filtering, amplification, and analog-to-digital conversion.

[0063] According to some embodiments of the present application, the filtering module comprises:

[0064] a first generation module configured to split the reflected signal into a plurality of sub-signals, determine the spectral amplitude of each sub-signal, calculate the average spectral amplitude of the plurality of sub-signals, calculate the absolute value of the difference between the spectral amplitude of each sub-signal and the average spectral amplitude, and filter out the sub-signal corresponding to the spectral amplitude with an absolute value of the difference greater than a preset difference to generate a first set;

[0065] a second generation module configured to analyze the reflected signal to determine all local maximum and minimum values of the reflected signal, fit the maximum value points using cubic spline interpolation to obtain an upper envelope line, fit the minimum value points using cubic spline interpolation to obtain a lower envelope line, calculate the mean value of the upper and lower envelope lines as a trend item, subtract the trend item from the reflected signal to obtain an IMF component, and generate a second set based on the sub-signal corresponding to the IMF component;

[0066] a first determination module configured to determine a target sub-signal based on the first set and the second set, analyze the type of the target sub-signal, perform filtering processing according to the type of the target sub-signal, perform signal reconstruction based on the processed target sub-signal, and obtain a filtered signal.

[0067] The working principle of the technical solution is as follows: the first generation module filters abnormal sub-signals through spectrum analysis to generate a first set. The reflected signal is divided into a plurality of sub-signals, and the FFT of each sub-signal is calculated to calculate the spectrum amplitude. The average spectrum amplitude of all sub-signals is calculated, the absolute value of the difference between the spectrum amplitude of each sub-signal and the average spectrum amplitude is calculated, and the sub-signal corresponding to the spectrum amplitude with an absolute value greater than a preset difference value is filtered out to generate the first set; the preset difference value is determined according to the noise statistical characteristics, and an example is 2σ, and σ is the standard deviation of the spectrum amplitude. The second generation module decomposes the reflected signal through EMD to extract the IMF component to generate a second set. All local maximum and minimum values of the reflected signal are detected, the maximum value points are fitted by cubic spline interpolation to obtain the upper envelope line, and the minimum value points are fitted by cubic spline interpolation to obtain the lower envelope line. The mean of the upper and lower envelope lines is taken as the trend term; the reflected signal is subtracted from the trend term to obtain the IMF component, and the second set is generated based on the sub-signal corresponding to the IMF component; the intersection of the first set and the second set is processed to determine the corresponding sub-signal as the target sub-signal, which is the to-be-filtered sub-signal. The type of the target sub-signal is analyzed, including a detail signal, a transition signal and a trend signal. The detail signal contains noise, a sudden signal or high-frequency details (such as edges and transient characteristics). The threshold-based denoising: a wavelet packet coefficient threshold (such as a soft threshold or a hard threshold) is set to filter out noise components below the threshold. The transition signal contains the main structure or periodic characteristics of the signal, and is between high-frequency noise and low-frequency trend. The spectrum analysis-based: the frequency components are analyzed by short-time Fourier transform (STFT) to remove interference frequencies. The trend signal reflects the overall trend or background noise (such as a slowly varying drift) of the signal. The low-pass filtering-based: low-frequency noise (such as baseline drift) is filtered out. The signal is reconstructed based on the processed target sub-signal to obtain a filtered signal.

[0068] The beneficial effects of the technical solution are as follows: the reflected signal is filtered and decomposed by EMD, the target sub-signal is accurately determined, the filtering processing is performed according to the type of the target sub-signal, the accuracy of the filtering is improved, and the signal is reconstructed based on the processed target sub-signal to obtain an accurate filtered signal.

[0069] According to some embodiments of the application, the sliding window width is determined based on the sampling rate of the reflected signal and the expected width of the peak, and the window step is set;

[0070] The width constraint condition is set, the peak detection is performed in the sliding window based on the width constraint condition, and a plurality of peaks are obtained;

[0071] The peaks across a plurality of sliding windows are filtered out as to-be-merged peaks, the to-be-merged peaks are merged, and a plurality of target peaks are obtained;

[0072] Determine the time point corresponding to each target wave crest, and segment the reflection signal according to the time points corresponding to adjacent target wave crests to obtain a plurality of sub-signals.

[0073] The working principle of the technical solution is as follows: the sliding window width is determined based on the sampling rate of the reflection signal and the expected width of the wave crest, that is, the product of the sampling rate and the expected width of the wave crest is the sliding window width. The window step is 1 / 3 of the sliding window width. A width constraint condition is set, and wave crest detection is performed in the sliding window based on the width constraint condition to obtain a plurality of wave crests. The width constraint condition is that there are at least N points on both sides of the wave crest that are less than 50% of the wave crest value. The wave crests that span multiple sliding windows are screened out as to-be-merged wave crests, and the to-be-merged wave crests are merged to obtain a plurality of target wave crests. The reflection signal is segmented according to the time points of the target wave crests to generate sub-signals. The midpoint of the time points of adjacent wave crests is used as the segmentation boundary.

[0074] The technical solution has the beneficial effects that the dynamic window width adapts to different wave crest widths, the wave crest merging strategy avoids repeated detection and missed detection, the combination of the sliding window and the wave crest merging balances real-time performance and robustness, and efficient positioning and segmentation of the wave crests in the reflection signal are realized.

[0075] According to some embodiments of the present application, the AI algorithm processing module comprises:

[0076] The 3D-FFT is used to process the processing signal to obtain distance-Doppler-angle three-dimensional imaging information.

[0077] The presence of the respiratory signal is determined based on the distance-Doppler-angle three-dimensional imaging information and the LSTM prediction model, the micro-motion features are determined based on the distance-Doppler-angle three-dimensional imaging information and the CNN classification module, and the human presence detection is performed based on the presence of the respiratory signal and the micro-motion features.

[0078] The working principle of the technical solution is as follows: the 3D-FFT is used to process the processing signal to obtain distance-Doppler-angle three-dimensional imaging information. The LSTM prediction model is used to analyze the distance-Doppler-angle three-dimensional imaging information, the Doppler time sequence signal of the chest region (fixed distance-angle range) is extracted from the distance-Doppler-angle three-dimensional imaging information, the Doppler time sequence signal is input into the LSTM prediction model, and the presence of the respiratory signal or the absence of the respiratory signal is output. The distance-Doppler slice is determined by feature extraction on the distance-Doppler-angle three-dimensional imaging information, the distance-Doppler slice is input into the CNN classification module, the micro-motion features are determined, and the micro-motion features include limb swinging and chest fluctuation. The human presence detection is performed based on the presence of the respiratory signal and the micro-motion features.

[0079] The beneficial effects of the above technical solutions are: through the joint method of 3D-FFT and deep learning, combining distance, Doppler, and angle information, the LSTM captures the respiratory rhythm, and the CNN extracts the micro-motion features, so that the human existence detection with high robustness is realized.

[0080] According to some embodiments of the present application, the AI algorithm processing module further comprises a movement detection module for:

[0081] Trajectory analysis is performed based on distance-Doppler-angle three-dimensional imaging information; the trajectory analysis includes Kalman filter tracking and velocity mutation detection;

[0082] When the processed signal appears continuous change in time sequence and meets the human body movement characteristics based on the trajectory analysis result, it is determined that the human body is in a moving state, and the movement related information is recorded.

[0083] The working principle of the above technical solutions is: based on distance-Doppler-angle (RDA) three-dimensional imaging information, the target trajectory is tracked through Kalman filter, the velocity mutation is detected, and the movement behavior is identified. When the processed signal appears continuous change in time sequence and meets the human body movement characteristics based on the trajectory analysis result, it is determined that the human body is in a moving state, and the movement related information is recorded.

[0084] The beneficial effects of the above technical solutions are: through the joint method of Kalman filter and velocity mutation detection, the human body movement state detection with high robustness is realized.

[0085] According to some embodiments of the present application, the AI algorithm processing module further comprises a fall detection module for:

[0086] Feature analysis is performed based on distance-Doppler-angle three-dimensional imaging information to determine feature information; the feature information includes vertical velocity mutation, attitude angle change, and impact duration;

[0087] The feature information is input into a preset fall posture model for judgment; when the matching degree of the feature information and the fall posture model is greater than a preset threshold, it is determined that the human body has a fall action.

[0088] The working principle of the above technical solutions is: based on distance-Doppler-angle three-dimensional imaging information, feature analysis is performed, human target points are extracted, vertical velocity mutation, attitude angle change (pitch angle θ and roll angle φ), and impact duration are calculated. The feature information is input into a preset fall posture model for judgment; when the matching degree of the feature information and the fall posture model is greater than a preset threshold, it is determined that the human body has a fall action.

[0089] The beneficial effects of the above technical solutions are: the accuracy of fall detection is improved.

[0090] According to some embodiments of the present invention, the AI ​​algorithm processing module further includes: a position detection module, used for:

[0091] Determine time delay and angle information based on distance-Doppler-angle 3D imaging information;

[0092] The location of a human body within a region is determined using a Doppler positioning algorithm based on time delay and angle information.

[0093] The working principle and beneficial effects of the above technical solution are as follows: Based on distance-Doppler-angle (RDA) three-dimensional imaging information, the time delay and angle information of the target are determined. Using a Doppler positioning algorithm based on the time delay and angle information, the position information of the human body within the area is determined, thus improving positioning accuracy.

[0094] According to some embodiments of the present invention, it further includes: a guardian alarm module, which selects the corresponding alarm method to issue an alarm based on the detection result and a preset alarm rule, and transmits the alarm information to the monitoring terminal via wired or wireless means.

[0095] The working principle of the above technical solution is as follows: It receives detection results (such as fall events and abnormal locations) from fall detection modules, location detection modules, etc. It selects an alarm method according to preset alarm rules (such as priority and time threshold) and transmits the alarm information to the monitoring terminal via wired or wireless means.

[0096] The beneficial effects of the above technical solution are: efficient and reliable intelligent alarms are achieved through dynamic rule matching and multi-channel transmission, which facilitates timely detection and handling of abnormal information.

[0097] like Figure 2 As shown, according to some embodiments of the present invention, a human body status monitoring system includes a control module, an execution device, a server, and an AI human body sensor as described above; wherein;

[0098] The control module is connected to both the AI ​​human sensor and the server, and is used for:

[0099] It receives monitoring information about the human body status transmitted by AI human body sensors and transmits it to the server;

[0100] Based on the monitoring information, a corresponding control strategy is generated, the corresponding execution device is determined based on the control strategy, and the execution command is transmitted to the execution device.

[0101] The server is used to acquire monitoring information and execution information of the execution device, generate a monitoring data table, and transmit it to the mobile terminal.

[0102] The working principle and beneficial effects of the above technical solution are as follows: the AI human body sensor integrates modules such as millimeter wave radar, Doppler positioning, and fall detection, and collects human state data (such as position, posture, and motion state) in real time. The control module receives the monitoring information of the human state transmitted by the AI human body sensor and transmits it to the server; a corresponding control strategy is generated according to the monitoring information, the corresponding execution device is determined based on the control strategy, and the execution instruction is transmitted to the execution device; the server acquires the monitoring information and the execution information of the execution device, generates a monitoring data table, and transmits it to the mobile terminal. Through the edge-cloud collaborative architecture, high-precision, low-latency, and privacy-protected human state monitoring is realized, and the accuracy of human state monitoring is improved.

[0103] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. An AI human sensor, characterized by, The application relates to a human body detection device based on millimeter wave radar. The device comprises: a millimeter wave radar module for transmitting millimeter wave signals and receiving reflected signals; a signal processing module connected with the millimeter wave radar module, for processing the reflected signals received by the millimeter wave radar module to obtain processed signals; an AI algorithm processing module connected with the signal processing module, for performing human body existence detection, movement detection, fall detection and position detection based on the processed signals to obtain detection results; a communication module connected with the AI algorithm processing module, for sending the detection results to external equipment; a power module for supplying power to the millimeter wave radar module, the signal processing module, the AI algorithm processing module and the communication module. The signal processing module comprises: a filtering module for filtering the reflected signals to obtain filtered signals; an amplification module for amplifying the filtered signals to obtain amplified signals; an analog-to-digital conversion module for analog-to-digital conversion of the amplified signals to obtain digital signals as the processed signals. The filtering module comprises: a first generation module for dividing the reflected signals into a plurality of sub-signals, determining the spectral amplitude of each sub-signal, calculating the average spectral amplitude of the plurality of sub-signals, calculating the absolute value of the difference between the spectral amplitude of each sub-signal and the average spectral amplitude, and screening out the sub-signals corresponding to the spectral amplitudes with absolute values greater than a preset difference value to generate a first set; a second generation module for analyzing the reflected signals to determine all local maximum and minimum values of the reflected signals, fitting the maximum value points by cubic spline interpolation to obtain an upper envelope line, fitting the minimum value points by cubic spline interpolation to obtain a lower envelope line, calculating the mean value of the upper and lower envelope lines as a trend item, and subtracting the trend item from the reflected signals to obtain an IMF component, and generating a second set based on the sub-signals corresponding to the IMF component; 2. The AI human sensor of claim 1, wherein, a first determination module for determining a target sub-signal according to the first set and the second set, analyzing the type of the target sub-signal, performing filtering processing according to the type of the target sub-signal, reconstructing the signal based on the processed target sub-signal, and obtaining the filtered signals. The first generation module divides the reflected signals into a plurality of sub-signals, comprising: determining the sliding window width based on the sampling rate of the reflected signals and the expected width of the wave peak, and setting the window step length; setting a width constraint condition, performing wave peak detection in the sliding window based on the width constraint condition to obtain a plurality of wave peaks; screening out wave peaks that span multiple sliding windows as to-be-merged wave peaks, performing merging processing on the to-be-merged wave peaks to obtain a plurality of target wave peaks; 3.The AI human sensor of claim 1, wherein determining the time point corresponding to each target wave peak, and dividing the reflected signals according to the time points corresponding to adjacent target wave peaks to obtain a plurality of sub-signals. The AI algorithm processing module comprises an existence detection module for: processing the processed signals based on 3D-FFT to obtain distance-doppler-angle three-dimensional imaging information; determining the existence of a breathing signal based on the distance-doppler-angle three-dimensional imaging information and an LSTM prediction model, determining micro-motion features based on the distance-doppler-angle three-dimensional imaging information and a CNN classification module, and performing human body existence detection based on the existence of the breathing signal and the micro-motion features.

4. The AI human sensor of claim 3, wherein, The AI algorithm processing module further comprises a movement detection module configured to: perform trajectory analysis based on the distance-Doppler-angle three-dimensional imaging information; the trajectory analysis comprises Kalman filter tracking and velocity mutation detection; determine that the human body is in a moving state and record movement-related information when the trajectory analysis result indicates that the processing signal has a continuous change in the time sequence and meets the human body movement characteristics.

5. The AI human sensor of claim 3, wherein, The AI algorithm processing module further comprises a fall detection module configured to: perform feature analysis based on the distance-Doppler-angle three-dimensional imaging information to determine feature information; the feature information comprises vertical velocity mutation, posture angle change, and impact duration; input the feature information into a preset fall posture model for judgment; when the feature information and the fall posture model have a matching degree greater than a preset threshold, determine that the human body has a fall action.

6. The AI human sensor of claim 3, wherein, The AI algorithm processing module further comprises a position detection module configured to: determine time delay and angle information based on the distance-Doppler-angle three-dimensional imaging information; determine the position information of the human body in the area by a Doppler positioning algorithm based on the time delay and angle information. 7.The AI human sensor of claim 1, wherein Further comprising: a guard alarm module configured to select a corresponding alarm mode based on a preset alarm rule according to the detection result, and transmit the alarm information to a monitoring terminal through a wired or wireless mode.

8. A human state monitoring system, characterized by comprise a control module, an execution device, a server, and the AI human body sensor of any one of claims 1-7; wherein; the control module is connected with the AI human body sensor and the server respectively, and is configured to: receive the monitoring information of the human body state transmitted by the AI human body sensor and transmit the monitoring information to the server; generate a corresponding control strategy according to the monitoring information, determine a corresponding execution device based on the control strategy, and transmit an execution instruction to the execution device; the server is configured to acquire the monitoring information and the execution information of the execution device, generate a monitoring data table, and transmit the monitoring data table to a mobile terminal.

Images