Intelligent dual-band microwave vital sign monitoring method and device

Abstract

The application discloses a kind of intelligent dual-band microwave vital sign monitoring method and device, to solve the insufficient microwave monitoring of existing anti-interference, multi-target resolution etc..Dog walks according to planning route, rotates head and emits 2.36GHz and 2.45GHz+1310nm laser auxiliary dual-band microwave acquisition signal, when signal is poor, adjust route and head direction, pre-process, FastICA algorithm separates heartbeat and breathing signal;Combined with SVM classification motion state, establish reference signal library and adopt Kalman filtering, GRU network etc. mode optimization signal according to SNR grading;Through joint time-frequency analysis, Hilbert transform is calculated respectively heart rate and respiratory rate, mark monitoring position, can also judge vital sign anomaly and set data transmission priority according to three levels of emergency, warning, normal, realize high-precision, real-time multi-target vital sign monitoring in harsh environment.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and specifically to an intelligent dual-band microwave vital sign monitoring method. Background Technology

[0002] In recent years, non-contact vital sign monitoring technology has made significant progress. CN119344705A discloses a vital sign detection method and device based on microwave radar, and CN118749912A discloses a night patrol and nursing system and method based on microwave monitoring. Microwave radar technology has achieved high-precision detection of human heartbeat (0.8-3Hz) and respiration (0.1-0.5Hz) by emitting frequency-modulated continuous wave (FMCW) and analyzing the reflected signal.

[0003] The current mainstream solution uses the 2.36GHz and 2.45GHz frequency bands (with attenuation <3dB when penetrating clothing) combined with MIMO antenna arrays, enabling multi-target tracking within a 3-meter range. In terms of signal processing, the application of FastICA blind source separation and adaptive Kalman filtering significantly improves the signal-to-noise ratio (SNR>30dB) in dynamic environments. The introduction of artificial intelligence technologies (such as SVM motion classification and GRU temporal filtering) further controls the heart rate detection error to within ±2bpm (static) and ±5bpm (dynamic).

[0004] However, the current mainstream solutions have shortcomings:

[0005] 1. Low spatial resolution in detection and tracking. Microwaves have relatively long wavelengths (typically in the millimeter to meter range), making it difficult to accurately identify the shape, height, and details of objects. The generated point cloud data is relatively sparse and blurry, affecting the accuracy of target classification and modeling, and limiting the ability to identify small or stationary targets.

[0006] Second, it is susceptible to electromagnetic interference and clutter. Microwave radar operates in the radio frequency band, making it easily affected by interference from other electronic devices, wireless communication signals, or strong electromagnetic sources. It is highly sensitive to reflected signals and clutter from stationary objects such as walls, guardrails, and road signs, and has difficulty distinguishing moving targets from background clutter.

[0007] Third, the cost and size are relatively high. Although conventional microwave radar is cheaper than lidar, the research, development, production and maintenance costs of high-performance microwave radar systems and imaging radar are still relatively high. The size of their antennas and transmitter / receiver units is usually larger than that of infrared or ultrasonic sensors, which affects portability and integration flexibility in space-constrained scenarios.

[0008] IV. High Environmental Sensitivity and Risk of False Alarms. Although microwave radar can penetrate clouds and fog, certain environmental factors can still affect its performance. For example, in security applications, microwaves may penetrate non-metallic walls, causing false alarms due to outdoor moving targets (such as pedestrians and vehicles). Electromagnetic modulation signals generated by commonly used gas discharge light sources (such as fluorescent lamps and mercury lamps) may also be mistaken for moving targets, resulting in false alarms. The detection range is usually limited to within 100 meters, and the angular resolution is insufficient (especially in the height direction), making it prone to producing false targets or missed detections. Summary of the Invention

[0009] To address the aforementioned technical problems, this invention proposes an intelligent dual-band microwave vital sign monitoring method and device, the method comprising the following steps: S1. The intelligent dual-band microwave vital sign monitoring robot dog rotates its head during walking to emit dual-band microwaves to collect surrounding heartbeat and respiratory signals. If the collected signals are poor, it adjusts its walking route and head direction to search for a better location to collect signals. The signal processing unit sets two different passband range filter parameters for the two microwave signals according to the different frequency domain characteristics of the respiratory and heartbeat signals, filtering and separating the respiratory and heartbeat signals of four channels: long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, and short-band microwave heartbeat signal. Preferably, the passband filter sets the frequency range of the respiratory signal to 0.1-0.5Hz and the frequency range of the heartbeat signal to 0.8-3.0Hz, and the amplitude of the respiratory signal is much higher than that of the heartbeat signal. The S1.1 intelligent dual-band microwave vital signs monitoring robot dog walks along the planned route and rotates its head in a cycle to emit dual-band microwaves in different directions to collect heartbeat and breathing signals from the surrounding area. If the collected signals are poor or the strength of the suspected heartbeat and breathing signals is related to the robot dog's walking route and head direction, the walking route and head direction are adjusted to search for a better location to collect signals. S1.1.1 The user assigns a task planning route to the intelligent dual-band microwave vital signs monitoring robot dog, and the intelligent dual-band microwave vital signs monitoring robot dog follows the route. S1.1.2 The user feeds the intelligent dual-band microwave vital signs monitoring robot dog with a cyclical rotation of its head to emit dual-band microwaves in different directions around it, and continuously collects microwave signals reflected from the surroundings during the walking process; The intelligent dual-band microwave vital signs monitoring robot dog uses 2.36GHz and 2.45GHz industrial, scientific, and medical frequency bands, plus a 1310nm laser-assisted dual-band microwave linear frequency modulated continuous wave (FMCW) for detection. Distance measurement is achieved through linear frequency modulation. ; Where A is the microwave transmission power; B is the microwave center frequency; B is the microwave sweep bandwidth; T is the microwave sweep period; The S1.2 intelligent dual-band microwave vital signs monitoring robot dog preprocesses and denoises microwave signals. The signal processing unit sets two different denoising parameters for the two microwave signals with different passband ranges based on the different frequency domain characteristics of the breathing and heartbeat signals. S1.2.1 The intelligent dual-band microwave vital signs monitoring robot dog receives reflected microwave signals, i.e., echo signals. The intelligent dual-band microwave vital signs monitoring robot dog receives and processes echo signals, and uses down-conversion processing to convert the signal into an intermediate frequency signal. The expression is: ; in, The amplitude of the microwave intermediate frequency signal, in V, is proportional to the target's reflection coefficient. Microwave round-trip time, in seconds (s). d is the target distance, and c is the speed of light; A represents the microwave transmission power; B is the microwave center frequency; B is the microwave sweep bandwidth; T is the microwave sweep period; S1.2.2 The intelligent dual-band microwave vital signs monitoring robot dog denoises the reflected microwave signals; the signal processing unit sets two different passband range microwave signal denoising parameters for the two microwave signals according to the different frequency domain characteristics of the breathing and heartbeat signals; preferably, the intelligent dual-band microwave vital signs monitoring robot dog uses wavelet denoising: ; in, These are wavelet basis functions; These are wavelet coefficients, reflecting the energy distribution of the signal at different scales; The number of decomposition layers is determined by the sampling rate. choose; S1.2.3 If the microwave acquisition signal is poor, a weak suspected heartbeat and breathing signal will appear after filtering in S1.2.2. The signal strength is related to the robot dog's walking route and head direction. If the signal is fluctuating, it is determined that there is obstruction or a fast-moving target. Furthermore, the intelligent dual-band microwave vital signs monitoring robot dog will stop walking or adjust its walking route and head direction to try to bypass the obstruction and search for a better position to acquire the signal, i.e., the position with the strongest signal. The intelligent dual-band microwave vital signs monitoring robot dog sets a threshold for poor microwave acquisition signals to determine whether the walking route and head direction need to be adjusted. The intelligent dual-band microwave vital signs monitoring robot dog is set to have a microwave signal dynamic range of 80dB after noise reduction, i.e., -40dBm to +40dBm. If this range is insufficient, it is judged that the microwave acquisition signal is poor, for example, due to obstruction. Furthermore, the intelligent dual-band microwave vital signs monitoring robot dog experiences signal fluctuations during its movement, and the signal strength is related to the robot dog's walking path and head direction. The robot dog adjusts its walking path and head direction around the microwave detection target, and stops walking when it finds the position with the strongest signal. The intelligent dual-band microwave vital signs monitoring robot dog is set to have a microwave signal response time of <10ms after noise reduction to adapt to target movement. If this is insufficient, it is judged that the microwave acquisition signal is poor. For example, if a fast-moving target appears, the intelligent dual-band microwave vital signs monitoring robot dog will further pause its movement and adjust its head direction to align with the fast-moving target to ensure that the microwave accurately illuminates the fast-moving target.

[0010] The S1.3 intelligent dual-band microwave vital signs monitoring robot dog separates the heartbeat and respiration microwave signals from two microwave signal bands. The signal processing unit filters and separates the respiration and heartbeat signals into four channels: long-band microwave respiration signal, long-band microwave heartbeat signal, short-band microwave respiration signal, and short-band microwave heartbeat signal. Preferably, the passband filter sets the frequency range of the respiration signal to 0.1-0.5Hz and the frequency range of the heartbeat signal to 0.8-3.0Hz, with the amplitude of the respiration signal being much higher than that of the heartbeat signal. Preferably, the respiration signal uses a 4th-order Butterworth bandpass filter with a passband range of 0.1-0.5Hz, and the heartbeat signal uses an 8th-order Butterworth bandpass filter with a passband range of 0.8-3.0Hz. S1.3.1 The intelligent dual-band microwave vital signs monitoring robot dog filters the microwave signal, separating the breathing and heartbeat signals into four channels: long-wavelength microwave breathing signal, long-wavelength microwave heartbeat signal, short-wavelength microwave breathing signal, and short-wavelength microwave heartbeat signal. Preferably, the passband filter sets the frequency range of the breathing signal to 0.1-0.5Hz and the frequency range of the heartbeat signal to 0.8-3.0Hz, with the amplitude of the breathing signal being much higher than that of the heartbeat signal. Preferably, a 4th-order Butterworth bandpass filter with a passband range of 0.1-0.5Hz is used for the breathing signal, and an 8th-order Butterworth bandpass filter with a passband range of 0.8-3.0Hz is used for the heartbeat signal. The intelligent dual-band microwave vital signs monitoring robot dog filters the denoised microwave signal and then uses the Fast Independent Component Analysis (FastICA) algorithm to separate and process the filtered microwave signal. This is because although the heartbeat and respiration signals have overlapping frequency bands, they are statistically independent. FastICA can effectively separate independent source signals with non-Gaussian properties and finds the separation matrix by maximizing negative entropy, thus preserving the complete characteristics of the microwave signal.

[0011] The preprocessed four-channel microwave signal matrix is ​​as follows: ; This is a 4-channel microwave receiving signal matrix, with units in volts (V). Let i be the microwave signal vector of the i-th channel, i=1,2,3,4; The number of microwave sampling points is dimensionless. It is a space of 4 rows and N columns of real numbers; S1.3.2 The intelligent dual-band microwave vital signs monitoring robot dog amplifies and corrects the filtered four-channel signals, performs group delay and phase linearity correction, and obtains the filtered and corrected long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, and short-band microwave heartbeat signal. The intelligent dual-band microwave vital signs monitoring robot dog sets the input signal quality to SNR>15dB to ensure separation effect; the intelligent dual-band microwave vital signs monitoring robot dog sets the motion state information provided by preprocessing to guide the FastICA parameter adjustment, and outputs the separated heartbeat and respiratory signals to lay the foundation for subsequent feature extraction.

[0012] S1.3.2.1 The intelligent dual-band microwave vital signs monitoring robot amplifies and corrects the amplitude of the four-channel signal after it has been denoised and down-frequency preprocessed: ; in, This is the microwave signal matrix after signal amplification and amplitude correction. Microwave signal matrix for signal amplification and amplitude correction (unit: ),pass calculate; The diagonal matrix of microwave signal eigenvalues ​​with covariance matrix (unit: ); Solving for the separation matrix of the eigenvector matrix of the microwave signal: ; Where w is the microwave signal separation vector; z is the column vector of the microwave signal for signal amplification and amplitude correction; It is a nonlinear function; It is the derivative; E is the expectation operator; S1.3.2.2 The intelligent dual-band microwave vital signs monitoring robot dog performs group delay and phase linearity correction on the four-channel signal after signal amplification and amplitude correction; if the group delay is nonlinear, a digital equalizer or predistortion circuit can be used for compensation. The intelligent dual-band microwave vital signs monitoring robot dog separates the results based on the following characteristics. Group delay and phase linearity correction are performed in d11 to identify heartbeat signals: ; in, Calculate the group delay and phase linearity correction of the signal in the 0.8-3Hz frequency band, corresponding to the normal heart rate range of 48-180 bpm; The signal is calculated as energy across the entire frequency band; the ratio with the largest value is selected. As a heartbeat signal; The separated heartbeat signal is expressed in V. For Fast Fourier Transform; Frequency, in Hz; The sampling frequency is expressed in Hz. Furthermore, the intelligent dual-band microwave vital sign monitoring robot dog identifies respiratory signals from the remaining signals: ; in, Calculate the group delay and phase linearity correction for the 0.1-0.5Hz frequency band, corresponding to a respiratory rate of 6-30 rpm; Exclude identified heartbeat signals Choose later; The isolated respiratory signal is expressed in V. For Fast Fourier Transform; Frequency, in Hz; The sampling frequency is expressed in Hz.

[0013] The S1.3.2.3 intelligent dual-band microwave vital signs monitoring robot dog inputs the four-channel signal after signal amplification and amplitude correction transformation into the analog-to-digital converter (ADC) for analog-to-digital conversion, and checks whether its bandwidth meets the Nyquist sampling theorem. Considering the high frequency of microwave signals, if the filter does not completely suppress high-frequency components, an anti-aliasing low-pass filter needs to be added before the ADC for further filtering. S2. The intelligent dual-band microwave vital signs monitoring robot dog uses artificial intelligence to learn and enhance weak signals based on dual-band microwave signals from different angles during walking. The signal processing unit compares and analyzes the respiratory and heartbeat signal characteristics of four channels separated by filtering in both bands, identifying common and different features of the four channels in both bands. If a valid respiratory and heartbeat signal is separated in one channel, the artificial intelligence can enhance the respiratory and heartbeat signals at the corresponding positions in other channels based on the common and different features of the two band microwave signals, generating enhanced respiratory and heartbeat signals for the four channels, namely, enhanced long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, and short-band microwave heartbeat signal, improving the detection capability of weak signals and anti-interference capability.

[0014] The S2.1 intelligent dual-band microwave vital signs monitoring robot dog learns and filters the four channels of respiratory and heartbeat signals, namely long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, and short-band microwave heartbeat signal. The intelligent dual-band microwave vital signs monitoring robot dog classifies motion states into static or stationary, dynamic or moving. For different states, the robot dog employs a multi-feature fusion strategy to learn the motion state from the four channels of respiratory and heartbeat signals separated by filtering. A high-precision mode is activated when the robot is static, and an anti-interference mode is activated when the robot is dynamic. Support vector machine (SVM) is used for motion state classification. ; in, The motion state feature vectors of the four channels of respiratory and heartbeat signals separated by filtering [variance, kurtosis, spectral entropy, zero-crossing rate, sample entropy]; The labels represent the motion status of the four channels of respiratory and heartbeat signals separated by filtering, with -1 indicating static and 1 indicating dynamic. The RBF kernel function is used to filter and separate the four channels of respiratory and heartbeat signals for motion state. ; The weights of the motion state support vectors for the four channels of respiratory and heartbeat signals separated by filtering; The motion state decision bias is applied to the four channels of respiratory and heartbeat signals separated by filtering. The S2.2 intelligent dual-band microwave vital signs monitoring robot dog compares and analyzes the respiratory and heartbeat signal characteristics of four channels through filtering and separation of two bands, and identifies the common and different characteristics of the microwave signals of the four channels in the two bands. The S2.2.1 intelligent dual-band microwave vital signs monitoring robot dog filters and separates the breathing and heartbeat signals from four channels into two bands, inputting them into a neural network to learn and recognize their respective characteristics.

[0015] The heartbeat reference signal includes three types of feature templates: sinus rhythm, motion disturbance, and abnormal rhythm. The sinus rhythm feature template is modeled using a Gaussian-modulated cosine function. ; in, This refers to the amplitude of the R-wave. For R-wave timestamps; To control the pulse width; This refers to the instantaneous heart rate.

[0016] The respiratory reference signals are divided into three categories: steady breathing, deep breathing, and apnea characteristics. The steady breathing characteristic template is represented as follows: ; in, Baseline amplitude; It is the modulation depth; Respiratory rate; It is a dual-band microwave carrier frequency with laser assistance of 2.36 GHz and 2.45 GHz + 1310nm in the industrial, scientific and medical frequency bands, namely the dual microwave carrier frequency.

[0017] The deep breathing template requires an amplitude increase of more than 1.5 times, and the apnea template detects a decrease in amplitude of more than 90% within 10 seconds.

[0018] S2.2.2 Intelligent dual-band microwave vital signs monitoring robot dog identifies the common and different characteristics of microwave signals from two bands and four channels; Two bands and four channels of microwave signals were set up and correlated. Matching is prioritized if the match value is below this threshold.

[0019] S2.2.3 The intelligent dual-band microwave vital signs monitoring robot dog performs template fusion on the respiratory and heartbeat signal features of four channels separated by filtering in two bands; The intelligent dual-band microwave vital sign monitoring robot dog adopts a dynamic weighting mechanism. Template fusion is performed, and the respiratory and heartbeat signal feature weighting coefficients of the four channels are separated by two-band filtering. Signal quality index Decision. When the quality of the heartbeat signal... Prioritize matching heartbeat templates and respiratory signal quality. During exercise, the breathing template is prioritized, while during exercise, the exercise interference and deep breathing template are mixed in a 6:4 ratio.

[0020] The reference template is dynamically updated via a sliding window, which displays the latest 100 cycles. The update condition is... and adopt The exponentially weighted forgetting mechanism.

[0021] The S2.3 intelligent dual-band microwave vital signs monitoring robot dog enhances weak signals. If effective breathing and heartbeat signals are separated in one channel, the artificial intelligence can enhance the breathing and heartbeat signals at the corresponding positions in other channels based on the common and different characteristics of the two band microwave signals, generating enhanced breathing and heartbeat signals for four channels, namely, enhanced long-band microwave breathing signal, long-band microwave heartbeat signal, short-band microwave breathing signal, and short-band microwave heartbeat signal. S2.3.1 Intelligent dual-band microwave vital signs monitoring robot dog establishes signal strength classification and enhancement strategies; Based on the signal-to-noise ratio (SNR) and motion state, signal strength is divided into three levels: Level 1, high-quality signal (SNR≥20dB), using adaptive Kalman filtering: ; in, for The state estimate at time 1, i.e., the filtered signal; This is the state transition matrix, which describes the model of how the signal changes over time. The Kalman gain is used to weigh the predictions against the observations. for The observed value at time, i.e., the original signal; The observation matrix maps the state to the observation space; The prediction error covariance matrix; To observe the noise covariance; Level 2, mild interference (10dB≤SNR<20dB): use gated cyclic unit (GRU) network filtering; Network structure: Input layer (16-dimensional features) → 2-layer GRU (32 units / layer) → Fully connected output Loss function: MSE + heart rate smoothing constraint term, the expression is:

[0022] in, To control the smoothing intensity; This represents the total loss value. For the first The true and predicted values ​​of each sample; The difference between heart rate estimates at adjacent time points; Level 3, severe interference (SNR<10dB), enable reference signal replacement mechanism, preferably using S2.3.2 for weak signal enhancement; The S2.3.2 intelligent dual-band microwave vital signs monitoring robot dog uses an optimal template for signal strength matching to enhance weak signals. If effective respiratory and heartbeat signals are separated on one channel, the artificial intelligence can enhance the respiratory and heartbeat signals at the corresponding locations on the other three channels based on the common and different characteristics of the two microwave signal bands. ; Threshold for enhancing or replacing weak signals: correlation coefficient < 0.6 or amplitude deviation > 30%; in, The output signal is the result of amplifying a weak signal; For reference templates in the template library A template for breathing and heartbeat signals; for Real-time respiratory and heart rate signal estimates at any given moment; For the first Reference respiratory and heart rate signal values ​​for each template; The penalty coefficient for differences in respiratory and heartbeat signals; L2 norm, Euclidean distance; S2.3.3 The intelligent dual-band microwave vital signs monitoring robot dog generates enhanced respiratory and heartbeat signals from four channels, namely, enhanced long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, and short-band microwave heartbeat signal, and extracts multimodal features. The multimodal features include: Temporal characteristics: Peak interval variation (SDNN), signal amplitude area (SMA); Frequency domain characteristics: wavelet packet energy entropy (frequency band 0.1-5Hz, 5-level decomposition); Nonlinear characteristics: approximate entropy (ApEn, m=2, r=0.2).

[0023] The S3 intelligent dual-band microwave vital signs monitoring robot dog performs coherent fusion processing on the four channels of respiratory and heartbeat signals, which are enhanced by signal enhancement. The data processing unit calculates a wider equivalent bandwidth microwave signal and separates the equivalent bandwidth microwave respiratory signal and the equivalent bandwidth microwave heartbeat signal. Based on the enhanced long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, short-band microwave heartbeat signal generated by S2, the equivalent bandwidth microwave respiratory signal generated by S3, and the equivalent bandwidth microwave heartbeat signal, it intelligently calculates the heart rate and respiratory rate. It analyzes the time delay, frequency shift characteristics, phase offset, and equivalent bandwidth microwave signal between the transmitted waveform and the echo signal of the dual-band microwave signal, calculates the spatial position, movement speed, and azimuth angle of the respiratory and heartbeat signals, and marks the location on the display control unit.

[0024] The S3.1 intelligent dual-band microwave vital signs monitoring robot dog uses a full-pole model to coherently fuse the heartbeat signals from the four channels that have been enhanced, and calculates a wider equivalent bandwidth microwave signal. Based on the enhanced long-band microwave heartbeat signal, short-band microwave heartbeat signal, and equivalent bandwidth microwave heartbeat signal generated by S2, the heart rate is estimated. Because the spectrum becomes blurred during exercise, the heart rate is estimated using joint time-frequency analysis of FFT and time-domain R-wave detection for enhanced long-wavelength microwave heart rate signals, short-wavelength microwave heart rate signals, and equivalent bandwidth microwave heart rate signals. The formula is as follows: ; in, The estimated heart rate frequency is in Hz. To convert it to bpm, multiply by 60. For frequency search variables; The envelope of the heartbeat signal is derived from the filtered output of step S2, and the amplitude is normalized. For Fast Fourier Transform, the input is a time-domain signal, and the output is a frequency-domain energy spectrum. To calculate the spectral energy by taking the square of the modulus; The parameter that maximizes the function value is the frequency with the highest energy. This heart rate estimation algorithm, by constraining the frequency domain search range to 0.8-3Hz and combining it with Blackman-Harris window FFT, effectively suppresses motion artifacts and spectral leakage while ensuring high accuracy. The computational complexity of this heart rate estimation algorithm is O(n log n). It meets real-time requirements and improves clinical reliability through energy peak validation, making it suitable for robust monitoring in both static and dynamic scenarios.

[0025] The S3.2 intelligent dual-band microwave vital signs monitoring robot dog estimates the respiratory rate based on the enhanced long-band microwave respiratory signal generated by S2, the short-band microwave respiratory signal, and the equivalent bandwidth microwave respiratory signal generated by S3.1. Because low-frequency respiratory signals are susceptible to baseline drift, Hilbert transform combined with envelope extraction and peak detection is used for long-wavelength microwave respiratory signals, short-wavelength microwave respiratory signals, and equivalent bandwidth microwave respiratory signals. The formula is as follows: ; in, The estimated respiratory rate is in Hz. To convert it to rpm, multiply by 60. The envelope of the respiratory signal is derived from the output of the Hilbert transform. For Fast Fourier Transform, the input is the time-domain envelope, and the output is the frequency-domain energy spectrum; The frequency point that returns the maximum spectral energy, in Hz; For frequency search variables; The denominator 60 is the unit conversion factor, converting Hz to rpm; Hilbert transform accurately extracts the main frequency of the respiratory envelope signal from 0.1 to 0.5 Hz through frequency domain peak detection, achieving clinical-grade accuracy while suppressing cardiac interference and noise, with an error of ≤1 rpm, and meeting real-time processing requirements.

[0026] The S3.3 intelligent dual-band microwave vital signs monitoring robot dog analyzes the time delay, frequency shift characteristics, phase offset, and equivalent bandwidth microwave signal between the transmitted and echoed waveforms of the dual-band microwave signal, calculates the spatial location, speed, and azimuth of the breathing and heartbeat signals, and marks the location and corresponding vital signs monitoring data on the display control unit. S3.3.1 The intelligent dual-band microwave vital signs monitoring robot dog calculates the time delay, frequency shift characteristics, phase offset, and equivalent bandwidth microwave signal between the transmitted waveform and the echo signal of the dual-band microwave signal, and obtains the spatial location, movement speed and azimuth angle of the breathing and heartbeat signals. The intelligent dual-band microwave vital signs monitoring robot dog calculates that when the linear frequency modulated microwave signal transmitted in the dual band encounters the target object in the propagation path, it is reflected. The front-end antenna and transceiver module capture the echo signal and compare it with the transmitted signal to extract the frequency difference or time delay information and calculate the target distance and speed parameters. S3.3.2 The intelligent dual-band microwave vital sign monitoring robot dog determines the normal or abnormal state of vital sign monitoring data, including bradycardia and apnea; preferably, the bradycardia threshold can be set with reference to the adult's resting heart rate being continuously below 60 beats per minute, and the apnea threshold can be set with reference to the apnea interval of less than 10 seconds in an adult or the apnea interval of less than 5 times per hour in an elderly person. One of the abnormal conditions is bradycardia: when If bradycardia persists for 5 consecutive cycles, it is considered a warning sign; if it persists for more than 15 cycles, it is considered an emergency sign of bradycardia. The second abnormal condition is apnea: when If it lasts for 60 seconds, it is considered a warning of respiratory arrest; if it lasts for 180 seconds, it is considered an emergency of respiratory arrest. The S3.3.3 intelligent dual-band microwave vital sign monitoring robot dog marks its location and corresponding vital sign monitoring data on the display control unit, and dynamically controls the transmission of vital sign monitoring data. The intelligent dual-band microwave vital sign monitoring robot dog sets the data transmission priority based on the severity of abnormal vital sign monitoring data in step S3.3.1 and the normal vital sign monitoring data in step S3.3.2. The highest priority is emergency, including bradycardia emergency and apnea emergency; the second priority is warning, including bradycardia warning and apnea warning; the lowest priority is normal, i.e., the normal vital sign monitoring data in step S3.3.2. Furthermore, higher priority vital sign monitoring data is transmitted first, with shorter data packet intervals. The method for determining the data packet interval is as follows: .

[0027] Based on the above method, an intelligent dual-band microwave vital signs monitoring robot dog is provided. Its structure is a long-band microwave radar including a front-end antenna, a transceiver module, a signal processing unit, a data processing unit, and a display control unit. Its front-end antenna and transceiver module are installed at a high position, enabling long-distance wide-area search and all-weather detection of heartbeat and breathing signals, and sharing the location of suspected heartbeat and breathing signals with the short-band microwave radar. The shortwave microwave radar includes a front-end antenna, a transceiver module, a signal processing unit, a data processing unit, and a display control unit; its front-end antenna and transceiver module are installed in a low position, enabling precise tracking and imaging of suspected heartbeat and breathing signal locations shared by the longwave microwave radar. The dual-band microwave radar adopts a shared back-end design, meaning that the two microwave radars have their own independent front-end antennas and transceiver modules for transmitting and reading microwave signals, but share back-end equipment such as signal processing units, data processing units, and display control units to process and calculate heartbeat and respiratory signals. A front-end antenna is used to convert electrical signals into microwave signals of a specific frequency for transmission, or to convert received echo signals into electrical signals. The transceiver module is used to load the transmitted signal onto the front-end antenna, or to read out the echo signal received by the front-end antenna. The signal processing unit is used to amplify and modulate the transmitted signal, or to demodulate and filter the echo signal; The data processing unit runs artificial intelligence algorithms to analyze and calculate heartbeat and respiratory signals; The display control unit is used for human-computer interaction, displaying the detected heartbeat and breathing signals according to the corresponding location.

[0028] Compared with the prior art, the beneficial effects of the present invention include: I. High Spatial Resolution for Detection and Tracking. It combines the different characteristics of microwave radars in different frequency bands. For example, long-wavelength microwave radar has a longer wavelength and lower propagation loss, making it suitable for long-range, wide-area searches and all-weather detection; while short-wavelength microwave radar has a shorter wavelength and narrower beam, providing higher range and angular resolution, suitable for precise tracking and imaging. The dual-band microwave radar system combines the advantages of both, achieving a "detect first, lock later" closed loop, significantly improving the spatial identification accuracy and tracking stability of targets.

[0029] II. Strong resistance to electromagnetic interference and clutter. This invention, by comparing and analyzing the characteristics of microwave signals received in two bands, can effectively identify and filter out interference signals or clutter, distinguishing moving targets from background clutter. Its identification accuracy can exceed that of single-band microwave radar. Furthermore, some electromagnetic interference and clutter resistance is primarily effective against a specific frequency band (such as the long-wave band), while its radar reflection characteristics are significantly enhanced in another frequency band (such as the short-wave band). Dual-band collaborative operation can effectively weaken the electromagnetic interference and clutter resistance, increasing the probability of target detection.

[0030] Third, lower cost and smaller size. Dual-band microwave radars typically employ a shared back-end design, meaning that the front-end antennas and transceiver modules for the two bands are independent, but the back-end equipment such as signal processing, data processing, and display control are shared. This avoids the complexity of information fusion via a data bus required by traditional multiple independent microwave radars, making it easier to form a single, robust target trajectory, reducing system size, cost, weight, and power consumption, and improving overall efficiency and reliability.

[0031] IV. Low Environmental Sensitivity and False Alarm Risk. By coherently fusing microwave echo signals from two bands, a wider microwave bandwidth signal can be obtained, significantly improving the one-dimensional range profile (HRRP) resolution of the target and enabling more refined imaging and feature recognition. Microwave signals have high phase resolution; by accumulating multiple pulses with a fixed phase relationship, the signal-to-noise ratio is effectively improved. Modern microwave radar phase noise can be as low as 0.1° RMS. When a displacement detection accuracy of 0.1 mm is required, the phase noise only needs to meet 2.39°, improving the efficiency of automatic target identification and classification based on breathing and heartbeat. Attached Figure Description

[0032] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0033] Figure 1 This is a schematic diagram of the structure of the method of the present invention.

[0034] Figure 2 This is a flowchart of the process of the method of the present invention.

[0035] In the diagram: 101 is the robot dog, 102 is the front-end antenna and transceiver module of the long-band microwave radar, 103 is the front-end antenna and transceiver module of the short-band microwave radar, 104 is the signal processing unit, 105 is the data processing unit, and 106 is the display control unit. Detailed Implementation

[0036] Example 1 This invention provides an intelligent dual-band microwave vital sign monitoring robot dog 101, such as... Figure 1 The diagram shows the structure of the method of the present invention. Its structure includes a robot dog, a long-wavelength microwave radar, and a short-wavelength microwave radar; The long-wavelength microwave radar includes a front-end antenna 102, a transceiver module, a signal processing unit, a data processing unit, and a display control unit. Its front-end antenna and transceiver module are installed at a high position, enabling long-distance wide-area search and all-weather detection of heartbeat and breathing signals, and sharing the location of suspected heartbeat and breathing signals with the short-wavelength microwave radar. The shortwave microwave radar includes a front-end antenna 103, a transceiver module, a signal processing unit, a data processing unit, and a display control unit; its front-end antenna and transceiver module are installed in a low position, enabling precise tracking and imaging of the suspected heartbeat and breathing signal locations shared by the longwave microwave radar. The dual-band microwave radar adopts a shared back-end design, meaning that the two microwave radars have their own independent front-end antennas and transceiver modules for transmitting and reading microwave signals, but share back-end equipment such as signal processing units, data processing units, and display control units to process and calculate heartbeat and respiratory signals. The front-end antenna is used to convert electrical signals into microwave signals of a specific frequency and transmit them, or to convert received echo signals into electrical signals. The transceiver module is used to load the transmitted signal onto the front-end antenna, or to read out the echo signal received by the front-end antenna. The signal processing unit 104 is used to amplify and modulate the transmitted signal, or to demodulate and filter the echo signal; Data processing unit 105 runs artificial intelligence algorithms to analyze and calculate heartbeat and respiratory signals; The display control unit 106 is used for human-computer interaction to display the detected heartbeat and breathing signals, corresponding to the location; Example 2 like Figure 2 As shown, this invention provides an intelligent dual-band microwave vital sign monitoring method, comprising the following steps: S1. The intelligent dual-band microwave vital sign monitoring robot dog rotates its head during walking to emit dual-band microwaves to collect surrounding heartbeat and respiratory signals. If the collected signals are poor, it adjusts its walking route and head direction to search for a better location to collect signals. The signal processing unit sets two different passband range filter parameters for the two microwave signals according to the different frequency domain characteristics of the respiratory and heartbeat signals, filtering and separating the respiratory and heartbeat signals of four channels: long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, and short-band microwave heartbeat signal. Preferably, the passband filter sets the frequency range of the respiratory signal to 0.1-0.5Hz and the frequency range of the heartbeat signal to 0.8-3.0Hz, and the amplitude of the respiratory signal is much higher than that of the heartbeat signal. The S1.1 intelligent dual-band microwave vital signs monitoring robot dog walks along the planned route and rotates its head in a cycle to emit dual-band microwaves in different directions to collect heartbeat and breathing signals from the surrounding area. If the collected signals are poor or the strength of the suspected heartbeat and breathing signals is related to the robot dog's walking route and head direction, the walking route and head direction are adjusted to search for a better location to collect signals. S1.1.1 The user assigns a task planning route to the intelligent dual-band microwave vital signs monitoring robot dog, and the intelligent dual-band microwave vital signs monitoring robot dog follows the route. S1.1.2 The user feeds the intelligent dual-band microwave vital signs monitoring robot dog with a cyclical rotation of its head to emit dual-band microwaves in different directions around it, and continuously collects microwave signals reflected from the surroundings during the walking process; The intelligent dual-band microwave vital sign monitoring robot dog uses 2.36GHz and 2.45GHz industrial, scientific, and medical frequency bands, plus a 1310nm laser-assisted dual-band microwave linear frequency modulated continuous wave (FMCW) for detection. These two microwave frequency bands are widely used in non-contact vital sign monitoring devices, achieving a good balance between tissue penetration and signal processing. Preferably, the long-band microwave is 2.36GHz, and the short-band microwave is 2.45GHz. Distance measurement is achieved through linear frequency modulation. ; Wherein, A: microwave emission power, which complies with FCC Part 15 radiation safety standards, preferably 10mW; The microwave center frequency has strong penetrating power through clothing and is harmless to the human body. Preferably, it is a dual-band microwave with 2.36 GHz and 2.45 GHz + 1310 nm laser-assisted microwave in the industrial, scientific and medical frequency bands, i.e., dual microwave center frequencies. B: Microwave sweep bandwidth, providing a distance resolution of 0.3m, preferably 500MHz; T: Microwave sweep period, preferably 1ms, corresponding to a maximum detection distance of 3m; Preferably, the intelligent dual-band microwave vital signs monitoring robot dog uses a GaAs power amplifier with an efficiency >40%, a 4×4 MIMO phased array antenna, a beam electronic scanning range of ±30°, an integrated temperature compensation circuit, and a frequency stability of ±10ppm. The S1.2 intelligent dual-band microwave vital signs monitoring robot dog preprocesses and denoises microwave signals. The signal processing unit sets two different denoising parameters for the two microwave signals with different passband ranges based on the different frequency domain characteristics of the breathing and heartbeat signals. S1.2.1 The intelligent dual-band microwave vital signs monitoring robot dog receives reflected microwave signals, i.e., echo signals. The intelligent dual-band microwave vital signs monitoring robot dog receives and processes echo signals, and uses down-conversion processing to convert the signal into an intermediate frequency signal. The expression is: ; in, The amplitude of the microwave intermediate frequency signal, in units of V, is proportional to the target's reflection coefficient. Microwave round-trip time, in seconds (s). d is the target distance, and c is the speed of light; A: Microwave emission power, in compliance with FCC Part 15 radiation safety standards, preferably 10mW; The microwave center frequency has strong penetrating power through clothing and is harmless to the human body. Preferably, it is a dual-band microwave with 2.36GHz and 2.45GHz + 1310nm laser-assisted microwave in the industrial, scientific and medical frequency bands, i.e., dual microwave center frequencies. B: Microwave sweep bandwidth, providing a distance resolution of 0.3m, preferably 500MHz; T: Microwave sweep period, preferably 1ms, corresponding to a maximum detection distance of 3m; S1.2.2 The intelligent dual-band microwave vital signs monitoring robot dog denoises the reflected microwave signals; the signal processing unit sets two different passband range microwave signal denoising parameters for the two microwave signals respectively, based on the different frequency domain characteristics of the breathing and heartbeat signals; preferably, The intelligent dual-band microwave vital sign monitoring robot dog uses wavelet denoising: ; in, The wavelet basis function is a sym4 wavelet with a support length of 7. These are wavelet coefficients, reflecting the energy distribution of the signal at different scales; The number of decomposition layers is determined by the sampling rate. choose; S1.2.3 If the microwave acquisition signal is poor, a weak suspected heartbeat and breathing signal will appear after filtering in S1.2.2. The signal strength is related to the robot dog's walking route and head direction. If the signal is fluctuating, it is determined that there is obstruction or a fast-moving target. Furthermore, the intelligent dual-band microwave vital signs monitoring robot dog will stop walking or adjust its walking route and head direction to try to bypass the obstruction and search for a better position to acquire the signal, i.e., the position with the strongest signal. The intelligent dual-band microwave vital signs monitoring robot dog sets a threshold for poor microwave acquisition signals to determine whether the walking route and head direction need to be adjusted. The intelligent dual-band microwave vital signs monitoring robot dog is set to have a microwave signal dynamic range of 80dB after noise reduction, i.e., -40dBm to +40dBm. If this range is insufficient, it is judged that the microwave acquisition signal is poor, for example, due to obstruction. Furthermore, the intelligent dual-band microwave vital signs monitoring robot dog experiences signal fluctuations during its movement, and the signal strength is related to the robot dog's walking path and head direction. The robot dog adjusts its walking path and head direction around the microwave detection target, and stops walking when it finds the position with the strongest signal. The intelligent dual-band microwave vital signs monitoring robot dog is set to have a microwave signal response time of <10ms after noise reduction to adapt to target movement. If this is insufficient, it is judged that the microwave acquisition signal is poor. For example, if a fast-moving target appears, the intelligent dual-band microwave vital signs monitoring robot dog will further pause its movement and adjust its head direction to align with the fast-moving target to ensure that the microwave accurately illuminates the fast-moving target.

[0037] The S1.3 intelligent dual-band microwave vital signs monitoring robot dog separates the heartbeat and respiration microwave signals from two microwave signal bands. The signal processing unit filters and separates the respiration and heartbeat signals into four channels: long-band microwave respiration signal, long-band microwave heartbeat signal, short-band microwave respiration signal, and short-band microwave heartbeat signal. Preferably, the passband filter sets the frequency range of the respiration signal to 0.1-0.5Hz and the frequency range of the heartbeat signal to 0.8-3.0Hz, with the amplitude of the respiration signal being much higher than that of the heartbeat signal. Preferably, the respiration signal uses a 4th-order Butterworth bandpass filter with a passband range of 0.1-0.5Hz, and the heartbeat signal uses an 8th-order Butterworth bandpass filter with a passband range of 0.8-3.0Hz. S1.3.1 The intelligent dual-band microwave vital signs monitoring robot dog filters the microwave signal, separating the breathing and heartbeat signals into four channels: long-wavelength microwave breathing signal, long-wavelength microwave heartbeat signal, short-wavelength microwave breathing signal, and short-wavelength microwave heartbeat signal. Preferably, the passband filter sets the frequency range of the breathing signal to 0.1-0.5Hz and the frequency range of the heartbeat signal to 0.8-3.0Hz, with the amplitude of the breathing signal being much higher than that of the heartbeat signal. Preferably, a 4th-order Butterworth bandpass filter with a passband range of 0.1-0.5Hz is used for the breathing signal, and an 8th-order Butterworth bandpass filter with a passband range of 0.8-3.0Hz is used for the heartbeat signal. The intelligent dual-band microwave vital signs monitoring robot dog filters the denoised microwave signal and then uses the Fast Independent Component Analysis (FastICA) algorithm to separate and process the filtered microwave signal. This is because although the heartbeat and respiration signals have overlapping frequency bands, they are statistically independent. FastICA can effectively separate independent source signals with non-Gaussian properties and finds the separation matrix by maximizing negative entropy, thus preserving the complete characteristics of the microwave signal.

[0038] The preprocessed four-channel microwave signal matrix is ​​as follows: ; This is a 4-channel microwave receiving signal matrix, with units in volts (V). Let i be the microwave signal vector of the i-th channel, i=1,2,3,4; The number of microwave sampling points is dimensionless. It is a space of 4 rows and N columns of real numbers; S1.3.2 The intelligent dual-band microwave vital signs monitoring robot dog amplifies and corrects the filtered four-channel signals, performs group delay and phase linearity correction, and obtains the filtered and corrected long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, and short-band microwave heartbeat signal. The intelligent dual-band microwave vital signs monitoring robot dog sets the input signal quality to SNR>15dB to ensure separation effect; the intelligent dual-band microwave vital signs monitoring robot dog sets the motion state information provided by preprocessing to guide the FastICA parameter adjustment, and outputs the separated heartbeat and respiratory signals to lay the foundation for subsequent feature extraction.

[0039] S1.3.2.1 The intelligent dual-band microwave vital signs monitoring robot amplifies and corrects the amplitude of the four-channel signal after it has been denoised and down-frequency preprocessed: ; in, This is the microwave signal matrix after signal amplification and amplitude correction. Microwave signal matrix for signal amplification and amplitude correction (unit: ),pass calculate; The diagonal matrix of microwave signal eigenvalues ​​with covariance matrix (unit: ); Solving for the separation matrix of the eigenvector matrix of the microwave signal: ; Where w is the microwave signal separation vector; z is the column vector of the microwave signal for signal amplification and amplitude correction; It is a nonlinear function; It is the derivative; E is the expectation operator; S1.3.2.2 The intelligent dual-band microwave vital signs monitoring robot dog performs group delay and phase linearity correction on the four-channel signal after signal amplification and amplitude correction; if the group delay is nonlinear, a digital equalizer or predistortion circuit can be used for compensation. The intelligent dual-band microwave vital signs monitoring robot dog separates the results based on the following characteristics. Group delay and phase linearity correction in d11 for identifying heartbeat signals: ; in, Calculate the group delay and phase linearity correction of the signal in the 0.8-3Hz frequency band, corresponding to the normal heart rate range of 48-180 bpm; The signal is calculated as energy across the entire frequency band; the ratio with the largest value is selected. As a heartbeat signal; The separated heartbeat signal is expressed in V. For Fast Fourier Transform; Frequency, in Hz; The sampling frequency is expressed in Hz. Furthermore, the intelligent dual-band microwave vital sign monitoring robot dog identifies respiratory signals from the remaining signals: ; in, Calculate the group delay and phase linearity correction for the 0.1-0.5Hz frequency band, corresponding to a respiratory rate of 6-30 rpm; Exclude identified heartbeat signals Choose later; The isolated respiratory signal is expressed in V. For Fast Fourier Transform; Frequency, in Hz; The sampling frequency is expressed in Hz.

[0040] The S1.3.2.3 intelligent dual-band microwave vital signs monitoring robot dog inputs the four-channel signal after signal amplification and amplitude correction transformation into the analog-to-digital converter (ADC) for analog-to-digital conversion, and checks whether its bandwidth meets the Nyquist sampling theorem. Considering the high frequency of microwave signals, if the filter does not completely suppress high-frequency components, an anti-aliasing low-pass filter needs to be added before the ADC for further filtering. S2. The intelligent dual-band microwave vital signs monitoring robot dog uses artificial intelligence to learn and enhance weak signals based on dual-band microwave signals from different angles during walking. The signal processing unit compares and analyzes the respiratory and heartbeat signal characteristics of four channels separated by filtering in both bands, identifying common and different features of the four channels in both bands. If a valid respiratory and heartbeat signal is separated in one channel, the artificial intelligence can enhance the respiratory and heartbeat signals at the corresponding positions in other channels based on the common and different features of the two band microwave signals, generating enhanced respiratory and heartbeat signals for the four channels, namely, enhanced long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, and short-band microwave heartbeat signal, improving the detection capability of weak signals and anti-interference capability.

[0041] The S2.1 intelligent dual-band microwave vital signs monitoring robot dog learns and filters the four channels of respiratory and heartbeat signals, namely long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, and short-band microwave heartbeat signal. The intelligent dual-band microwave vital signs monitoring robot dog classifies motion states into static or stationary, dynamic or moving. For different states, the robot dog employs a multi-feature fusion strategy to learn the motion state from the four channels of respiratory and heartbeat signals separated by filtering. A high-precision mode is activated when the robot is static, and an anti-interference mode is activated when the robot is dynamic. Support vector machine (SVM) is used for motion state classification. ; in, The motion state feature vectors of the four channels of respiratory and heartbeat signals separated by filtering [variance, kurtosis, spectral entropy, zero-crossing rate, sample entropy]; The labels represent the motion status of the four channels of respiratory and heartbeat signals separated by filtering, with -1 indicating static and 1 indicating dynamic. The RBF kernel function is used to filter and separate the four channels of respiratory and heartbeat signals for motion state. ; The weights of the motion state support vectors for the four channels of respiratory and heartbeat signals separated by filtering; The motion state decision bias is applied to the four channels of respiratory and heartbeat signals separated by filtering. The S2.2 intelligent dual-band microwave vital signs monitoring robot dog compares and analyzes the respiratory and heartbeat signal characteristics of four channels through filtering and separation of two bands, and identifies the common and different characteristics of the microwave signals of the four channels in the two bands. The S2.2.1 intelligent dual-band microwave vital signs monitoring robot dog filters and separates the breathing and heartbeat signals from four channels into two bands, inputting them into a neural network to learn and recognize their respective characteristics.

[0042] The heartbeat reference signal includes three types of feature templates: sinus rhythm, motion disturbance, and abnormal rhythm. The sinus rhythm feature template is modeled using a Gaussian-modulated cosine function. ; in, The amplitude of the R wave is 0.5-2mV; For R-wave timestamps; To control the pulse width, it is set to 0.1-0.3 seconds; This refers to the instantaneous heart rate.

[0043] The respiratory reference signal is divided into three categories: steady breathing, deep breathing, and apnea. The steady breathing feature template is characterized as follows: ; in, The baseline amplitude is 0.2-1V; It is the modulation depth, which is 0.3-0.8. Respiratory rate; It is a dual-band microwave carrier frequency with laser assistance of 2.36 GHz and 2.45 GHz + 1310nm in the industrial, scientific and medical frequency bands, namely the dual microwave carrier frequency.

[0044] The deep breathing template requires an amplitude increase of more than 1.5 times, and the apnea template detects a decrease in amplitude of more than 90% within 10 seconds.

[0045] S2.2.2 Intelligent dual-band microwave vital signs monitoring robot dog identifies the common and different characteristics of microwave signals from two bands and four channels; Two bands and four channels of microwave signals were set up and correlated. Matching is prioritized if the match value is below this threshold.

[0046] S2.2.3 The intelligent dual-band microwave vital signs monitoring robot dog performs template fusion on the respiratory and heartbeat signal features of four channels separated by filtering in two bands; The intelligent dual-band microwave vital sign monitoring robot dog adopts a dynamic weighting mechanism. Template fusion is performed, and the respiratory and heartbeat signal feature weighting coefficients of the four channels are separated by two-band filtering. Signal quality index Decision. When the quality of the heartbeat signal... Prioritize matching heartbeat templates and respiratory signal quality. During exercise, the breathing template is prioritized, while during exercise, the exercise interference and deep breathing template are mixed in a 6:4 ratio.

[0047] The reference template is dynamically updated via a sliding window, which displays the latest 100 cycles. The update condition is... and adopt The exponentially weighted forgetting mechanism.

[0048] The S2.3 intelligent dual-band microwave vital signs monitoring robot dog enhances weak signals. If effective breathing and heartbeat signals are separated in one channel, the artificial intelligence can enhance the breathing and heartbeat signals at the corresponding positions in other channels based on the common and different characteristics of the two band microwave signals, generating enhanced breathing and heartbeat signals for four channels, namely, enhanced long-band microwave breathing signal, long-band microwave heartbeat signal, short-band microwave breathing signal, and short-band microwave heartbeat signal. S2.3.1 Intelligent dual-band microwave vital signs monitoring robot dog establishes signal strength classification and enhancement strategies; Based on the signal-to-noise ratio (SNR) and motion state, signal strength is divided into three levels: Level 1, high-quality signal (SNR≥20dB), using adaptive Kalman filtering: ; in, for The state estimate at time 1, i.e., the filtered signal; This is the state transition matrix, which describes the model of how the signal changes over time. The Kalman gain is used to weigh the predictions against the observations. for The observed value at time, i.e., the original signal; The observation matrix maps the state to the observation space; The prediction error covariance matrix; To observe the noise covariance; Level 2, mild interference (10dB≤SNR<20dB): use gated cyclic unit (GRU) network filtering; Network structure: Input layer (16-dimensional features) → 2-layer GRU (32 units / layer) → Fully connected output Loss function: MSE + heart rate smoothing constraint term, the expression is:

[0049] in, To control the smoothing intensity; This represents the total loss value. For the first The true and predicted values ​​of each sample; The difference between heart rate estimates at adjacent time points; Level 3, severe interference (SNR<10dB), enable reference signal replacement mechanism, preferably using S2.3.2 for weak signal enhancement; The S2.3.2 intelligent dual-band microwave vital signs monitoring robot dog uses an optimal template for signal strength matching to enhance weak signals. If effective respiratory and heartbeat signals are separated on one channel, the artificial intelligence can enhance the respiratory and heartbeat signals at the corresponding locations on the other three channels based on the common and different characteristics of the two microwave signal bands. ; Threshold for enhancing or replacing weak signals: correlation coefficient < 0.6 or amplitude deviation > 30%; in, The output signal is the result of amplifying a weak signal; For reference templates in the template library A template for breathing and heartbeat signals; for Real-time respiratory and heart rate signal estimates at any given moment; For the first Reference respiratory and heart rate signal values ​​for each template; This is the penalty coefficient for differences in breathing and heartbeat signals, with a default value of 0.5. L2 norm, Euclidean distance; S2.3.3 The intelligent dual-band microwave vital signs monitoring robot dog generates enhanced respiratory and heartbeat signals from four channels, namely, enhanced long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, and short-band microwave heartbeat signal, and extracts multimodal features. The multimodal features include: Temporal characteristics: Peak interval variation (SDNN), signal amplitude area (SMA); Frequency domain characteristics: wavelet packet energy entropy (frequency band 0.1-5Hz, 5-level decomposition); Nonlinear characteristics: approximate entropy (ApEn, m=2, r=0.2).

[0050] The S3 intelligent dual-band microwave vital signs monitoring robot dog performs coherent fusion processing on the four channels of respiratory and heartbeat signals, which are enhanced by signal enhancement. The data processing unit calculates a wider equivalent bandwidth microwave signal and separates the equivalent bandwidth microwave respiratory signal and the equivalent bandwidth microwave heartbeat signal. Based on the enhanced long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, short-band microwave heartbeat signal generated by S2, the equivalent bandwidth microwave respiratory signal generated by S3, and the equivalent bandwidth microwave heartbeat signal, it intelligently calculates the heart rate and respiratory rate. It analyzes the time delay, frequency shift characteristics, phase offset, and equivalent bandwidth microwave signal between the transmitted waveform and the echo signal of the dual-band microwave signal, calculates the spatial position, movement speed, and azimuth angle of the respiratory and heartbeat signals, and marks the location on the display control unit.

[0051] The S3.1 intelligent dual-band microwave vital signs monitoring robot dog uses a full-pole model to coherently fuse the heartbeat signals from the four channels that have been enhanced, and calculates a wider equivalent bandwidth microwave signal. Based on the enhanced long-band microwave heartbeat signal, short-band microwave heartbeat signal, and equivalent bandwidth microwave heartbeat signal generated by S2, the heart rate is estimated. The intelligent dual-band microwave vital signs monitoring robot performs differential operations on the phase of the unfolded heartbeat signal to enhance the heartbeat component in the mixed signal, suppress interference from respiratory signals and their harmonics, and eliminate phase drift caused by the hardware receiver. The robot uses a Fast Fourier Transform-based spectral estimation (FFT-SE) algorithm combined with an autocorrelation function and peak detection algorithm to jointly estimate the heartbeat frequency for long-band, short-band, and equivalent bandwidth microwave heartbeat signals. Because the spectrum is blurred during movement, the robot employs joint time-frequency analysis of FFT and time-domain R-wave detection to estimate the heart rate for the enhanced long-band, short-band, and equivalent bandwidth microwave heartbeat signals. The formula is as follows: ; in, The estimated heart rate frequency is in Hz. To convert it to bpm, multiply by 60. The frequency search variable is 0.8-3Hz, corresponding to 48-180bpm; The envelope of the heartbeat signal is derived from the filtered output of step S2, and the amplitude is normalized. For Fast Fourier Transform, the input is a time-domain signal, and the output is a frequency-domain energy spectrum. To calculate the spectral energy by taking the square of the modulus; The parameter that maximizes the function value is the frequency with the highest energy. This heart rate estimation algorithm, by constraining the frequency domain search range to 0.8-3Hz and combining it with Blackman-Harris window FFT, effectively suppresses motion artifacts and spectral leakage while ensuring high accuracy. The computational complexity of this heart rate estimation algorithm is O(n log n). It meets real-time requirements and improves clinical reliability through energy peak validation, making it suitable for robust monitoring in both static and dynamic scenarios.

[0052] The S3.2 intelligent dual-band microwave vital signs monitoring robot dog estimates the respiratory rate based on the enhanced long-band microwave respiratory signal generated by S2, the short-band microwave respiratory signal, and the equivalent bandwidth microwave respiratory signal generated by S3.1. The intelligent dual-band microwave vital sign monitoring robot dog uses a Fast Fourier Transform-based spectral estimation (FFT-SE) algorithm combined with an autocorrelation function and peak detection algorithm to jointly estimate the respiratory frequency for long-band microwave respiratory signals, short-band microwave respiratory signals, and equivalent bandwidth microwave respiratory signals. Because low-frequency respiratory signals are easily affected by baseline drift, Hilbert transform is used to extract the envelope and then perform peak detection for long-band, short-band, and equivalent bandwidth microwave respiratory signals. The formula is as follows: ; in, The estimated respiratory rate is in Hz. To convert it to rpm, multiply by 60. The envelope of the respiratory signal is derived from the output of the Hilbert transform. For Fast Fourier Transform, the input is the time-domain envelope, and the output is the frequency-domain energy spectrum; The frequency point that returns the maximum spectral energy, in Hz; The frequency search variable is 0.1-0.5Hz, corresponding to 6-30rpm; The denominator 60 is the unit conversion factor, converting Hz to rpm; Hilbert transform accurately extracts the main frequency of the respiratory envelope signal from 0.1 to 0.5 Hz through frequency domain peak detection, achieving clinical-grade accuracy while suppressing cardiac interference and noise, with an error of ≤1 rpm, and meeting real-time processing requirements.

[0053] The S3.3 intelligent dual-band microwave vital signs monitoring robot dog analyzes the time delay, frequency shift characteristics, phase offset, and equivalent bandwidth microwave signal between the transmitted and echoed waveforms of the dual-band microwave signal, calculates the spatial location, speed, and azimuth of the breathing and heartbeat signals, and marks the location and corresponding vital signs monitoring data on the display control unit. S3.3.1 The intelligent dual-band microwave vital signs monitoring robot dog calculates the time delay, frequency shift characteristics, phase offset, and equivalent bandwidth microwave signal between the transmitted waveform and the echo signal of the dual-band microwave signal, and obtains the spatial location, movement speed and azimuth angle of the breathing and heartbeat signals. S3.3.2 The intelligent dual-band microwave vital signs monitoring robot dog determines the normal or abnormal state of vital signs monitoring data, including bradycardia and apnea. One of the abnormal conditions is bradycardia: when If bradycardia persists for 5 consecutive cycles, it is considered a warning sign; if it persists for more than 15 cycles, it is considered an emergency sign of bradycardia. The second abnormal condition is apnea: when If it lasts for 60 seconds, it is considered a warning of respiratory arrest; if it lasts for 180 seconds, it is considered an emergency of respiratory arrest. The S3.3.3 intelligent dual-band microwave vital sign monitoring robot dog marks its location and corresponding vital sign monitoring data on the display control unit, and dynamically controls the transmission of vital sign monitoring data. The intelligent dual-band microwave vital sign monitoring robot dog sets the data transmission priority based on the severity of abnormal vital sign monitoring data in step S3.3.1 and the normal vital sign monitoring data in step S3.3.2. The highest priority is emergency, including bradycardia emergency and apnea emergency; the second priority is warning, including bradycardia warning and apnea warning; the lowest priority is normal, i.e., the normal vital sign monitoring data in step S3.3.2. Furthermore, higher priority vital sign monitoring data is transmitted first, with shorter data packet intervals. The method for determining the data packet interval is as follows: .

[0054] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent dual-band microwave vital sign monitoring, characterized in that, Includes the following steps: S1. The intelligent dual-band microwave vital signs monitoring robot dog rotates its head to emit dual-band microwaves to collect surrounding heartbeat and respiratory signals while walking. If the collected signal is not good, it adjusts its walking route and head direction to search for a better position to collect the signal. Based on the different frequency domain characteristics of the breathing and heartbeat signals, the signal processing unit sets two filter parameters with different passband ranges for the two microwave signals of the two bands, and filters and separates the breathing and heartbeat signals of four channels, namely long-wavelength microwave breathing signal, long-wavelength microwave heartbeat signal, short-wavelength microwave breathing signal, and short-wavelength microwave heartbeat signal. S2. The intelligent dual-band microwave vital signs monitoring robot dog performs artificial intelligence learning and weak signal enhancement based on dual-band microwave signals from different angles during walking. The signal processing unit compares and analyzes the respiratory and heartbeat signal characteristics of the four channels filtered and separated by the two bands, and identifies the common and different characteristics of the microwave signals of the four channels in the two bands. If effective breathing and heartbeat signals are separated in one channel, artificial intelligence can enhance the breathing and heartbeat signals at the corresponding positions in other channels based on the common and different characteristics of the two microwave signals, generating enhanced breathing and heartbeat signals for four channels, namely, enhanced long-wavelength microwave breathing signal, long-wavelength microwave heartbeat signal, short-wavelength microwave breathing signal, and short-wavelength microwave heartbeat signal, thereby improving the ability to detect weak signals and resist interference. The S3 intelligent dual-band microwave vital signs monitoring robot dog performs coherent fusion processing on the four channels of respiratory and heartbeat signals, which are enhanced by signal enhancement. The data processing unit calculates a wider equivalent bandwidth microwave signal and separates the equivalent bandwidth microwave respiratory signal and the equivalent bandwidth microwave heartbeat signal. Based on the enhanced long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, short-band microwave heartbeat signal generated by S2, the equivalent bandwidth microwave respiratory signal generated by S3, and the equivalent bandwidth microwave heartbeat signal, the heart rate and respiratory rate are calculated. The time delay, frequency shift characteristics, phase offset, and equivalent bandwidth microwave signal between the transmitted waveform and the echo signal of the dual-band microwave signal are analyzed to calculate the spatial position, movement speed, and azimuth angle of the respiratory and heartbeat signals, and the location is marked on the display control unit.

2. The intelligent dual-band microwave vital sign monitoring method according to claim 1, characterized in that, Step S1 includes the following steps: The S1.1 intelligent dual-band microwave vital signs monitoring robot dog rotates its head during walking to emit dual-band microwaves to collect surrounding heartbeat and breathing signals. If the collected signals are poor or the strength of the suspected heartbeat and breathing signals is related to the robot dog's walking route and head direction, the walking route and head direction will be adjusted to search for a better location to collect signals. The S1.2 intelligent dual-band microwave vital signs monitoring robot dog preprocesses and denoises microwave signals. The signal processing unit sets two different passband range denoising parameters for the two microwave signals according to the different frequency domain characteristics of the breathing and heartbeat signals. The S1.3 intelligent dual-band microwave vital signs monitoring robot dog separates the heartbeat and respiration microwave signals from two bands of microwave signals. The signal processing unit filters and separates the respiration and heartbeat signals from four channels, namely long-band microwave respiration signal, long-band microwave heartbeat signal, short-band microwave respiration signal, and short-band microwave heartbeat signal. The passband filter sets the frequency range of the breathing and heartbeat signals, with the amplitude of the breathing signal being higher than that of the heartbeat signal.

3. The intelligent dual-band microwave vital sign monitoring method according to claim 2, characterized in that, Step S1.1 includes the following steps: S1.1.1 The user assigns a task planning route to the intelligent dual-band microwave vital signs monitoring robot dog, and the intelligent dual-band microwave vital signs monitoring robot dog follows the route. S1.1.2 The user feeds the intelligent dual-band microwave vital signs monitoring robot dog with a cyclical rotation of its head to emit dual-band microwaves in different directions around it, and continuously collects microwave signals reflected from the surroundings during the walking process; The monitoring robot dog cable achieves distance measurement through linear frequency modulation: ； Where A is the microwave transmission power; B is the microwave center frequency; B is the microwave sweep bandwidth; T is the microwave sweep period.

4. The intelligent dual-band microwave vital sign monitoring method according to claim 2, characterized in that, Step S1.2 includes the following steps: S1.2.1 The intelligent dual-band microwave vital signs monitoring robot dog receives reflected microwave signals, i.e., echo signals. S1.2.2 The intelligent dual-band microwave vital signs monitoring robot dog denoises the reflected microwave signals; the signal processing unit sets two different passband range microwave signal denoising parameters for the two microwave signals according to the different frequency domain characteristics of the breathing and heartbeat signals; preferably, the intelligent dual-band microwave vital signs monitoring robot dog uses wavelet denoising. S1.2.3 If the microwave acquisition signal is poor, weak suspected heartbeat and breathing signals will appear after filtering in S1.2.

2. The signal strength is related to the robot dog's walking route and head direction. If the signal strength fluctuates, it is determined that there is obstruction or a fast-moving target. Furthermore, the intelligent dual-band microwave vital signs monitoring robot dog will stop walking or adjust its walking route and head direction to try to bypass the obstruction and search for a better position to acquire the signal, that is, the position with the strongest signal.

5. The intelligent dual-band microwave vital sign monitoring method according to claim 2, characterized in that, Step S1.3 further includes the intelligent dual-band microwave vital signs monitoring robot dog amplifying and correcting the filtered four-channel signals, and performing group delay and phase linear correction to obtain filtered and corrected long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, and short-band microwave heartbeat signal.

6. The intelligent dual-band microwave vital sign monitoring method according to claim 2, characterized in that, Step S1.3.2 includes the following steps: S1.3.2.1 The intelligent dual-band microwave vital signs monitoring robot dog amplifies and corrects the amplitude of the four-channel signal after it has been denoised and down-frequency preprocessed; S1.3.2.2 The intelligent dual-band microwave vital signs monitoring robot dog performs group delay and phase linearity correction on the four-channel signal after signal amplification and amplitude correction; if the group delay is nonlinear, a digital equalizer or predistortion circuit can be used for compensation. The S1.3.2.3 intelligent dual-band microwave vital signs monitoring robot dog inputs the four-channel signal after signal amplification and amplitude correction transformation into the analog-to-digital converter for analog-to-digital conversion, and checks whether its bandwidth meets the Nyquist sampling theorem. Considering the high frequency of microwave signals, if the filter does not completely suppress high-frequency components, an anti-aliasing low-pass filter needs to be added before the analog-to-digital converter for further filtering.

7. The intelligent dual-band microwave vital sign monitoring method according to claim 1, characterized in that, Step S2 includes the following steps: The S2.1 intelligent dual-band microwave vital signs monitoring robot dog learns and filters the four channels of respiratory and heartbeat signals, namely long-band microwave respiratory signal, long-band microwave heartbeat signal, short-band microwave respiratory signal, and short-band microwave heartbeat signal. The S2.2 intelligent dual-band microwave vital signs monitoring robot dog compares and analyzes the respiratory and heartbeat signal characteristics of four channels through filtering and separation of two bands, and identifies the common and different characteristics of the microwave signals of the four channels in the two bands. The S2.3 intelligent dual-band microwave vital signs monitoring robot dog enhances weak signals. If effective breathing and heartbeat signals are separated on one channel, the artificial intelligence can enhance the breathing and heartbeat signals at the corresponding positions in other channels based on the common and different characteristics of the two band microwave signals, generating enhanced breathing and heartbeat signals for four channels, namely, enhanced long-band microwave breathing signal, long-band microwave heartbeat signal, short-band microwave breathing signal, and short-band microwave heartbeat signal.

8. The intelligent dual-band microwave vital sign monitoring method according to claim 1, characterized in that, Step S3 includes the following steps: The S3.1 intelligent dual-band microwave vital signs monitoring robot dog uses a full-pole model to coherently fuse the heartbeat signals from the four channels that have been enhanced, and calculates a wider equivalent bandwidth microwave signal. Based on the enhanced long-band microwave heartbeat signal, short-band microwave heartbeat signal, and equivalent bandwidth microwave heartbeat signal generated by S2, the heart rate is estimated. The S3.2 intelligent dual-band microwave vital signs monitoring robot dog estimates the respiratory rate based on the enhanced long-band microwave respiratory signal generated by S2, the short-band microwave respiratory signal, and the equivalent bandwidth microwave respiratory signal generated by S3.

1. The S3.3 intelligent dual-band microwave vital signs monitoring robot dog analyzes the time delay, frequency shift characteristics, phase offset, and equivalent bandwidth microwave signal between the transmitted and echoed waveforms of the dual-band microwave signal, calculates the spatial location, speed, and azimuth of the breathing and heartbeat signals, and marks the location and corresponding vital signs monitoring data on the display control unit. Step S3.3 specifically includes the following steps: S3.3.1 The intelligent dual-band microwave vital signs monitoring robot dog calculates the time delay, frequency shift characteristics, phase offset, and equivalent bandwidth microwave signal between the transmitted waveform and the echo signal of the dual-band microwave signal, and obtains the spatial location, movement speed and azimuth angle of the breathing and heartbeat signals. S3.3.2 The intelligent dual-band microwave vital signs monitoring robot dog determines the normal or abnormal state of vital signs monitoring data, including bradycardia and apnea. The S3.3.3 intelligent dual-band microwave vital sign monitoring robot dog marks its location and corresponding vital sign monitoring data on the display and control unit, and dynamically controls the transmission of vital sign monitoring data.

9. The intelligent dual-band microwave vital sign monitoring method according to claim 8, characterized in that, The data dynamic transmission control in step S3.3.3 specifically involves the intelligent dual-band microwave vital sign monitoring robot dog setting the priority of vital sign monitoring data transmission based on the severity of the abnormal vital sign monitoring data in step S3.3.1 and the normal vital sign monitoring data in step S3.3.

2. The highest priority is emergency, including bradycardia emergency and respiratory arrest emergency. The second level is a warning, including bradycardia warning and apnea warning; The lowest level is normal, which is the vital signs monitoring data in normal state in step S3.3.2; High-priority vital sign monitoring data are transmitted first, with shorter data packet intervals. The method for determining the data packet interval is as follows: .

10. A robot dog based on the intelligent dual-band microwave vital sign monitoring method of any one of claims 1-9, characterized in that, Including robot dogs, long-band microwave radar, and short-band microwave radar; Long-band microwave radar includes a front-end antenna, a transceiver module, a signal processing unit, a data processing unit, and a display control unit; Its front-end antenna and transceiver module are installed high up, enabling long-distance wide-area search and all-weather detection of heartbeat and breathing signals, and sharing the location of suspected heartbeat and breathing signals with short-band microwave radar; The shortwave microwave radar includes a front-end antenna, a transceiver module, a signal processing unit, a data processing unit, and a display control unit; Its front-end antenna and transceiver module are installed in a low position, enabling precise tracking and imaging of the suspected heartbeat and breathing signal locations shared by long-band microwave radar; The dual-band microwave radar adopts a shared back-end design, meaning that the two microwave radars have their own independent front-end antennas and transceiver modules for transmitting and reading microwave signals, but share back-end equipment such as signal processing units, data processing units, and display control units to process and calculate heartbeat and respiratory signals. A front-end antenna is used to convert electrical signals into microwave signals of a specific frequency for transmission, or to convert received echo signals into electrical signals. The transceiver module is used to load the transmitted signal onto the front-end antenna, or to read out the echo signal received by the front-end antenna. The signal processing unit is used to amplify and modulate the transmitted signal, or to demodulate and filter the echo signal; The data processing unit runs artificial intelligence algorithms to analyze and calculate heartbeat and respiratory signals; The display control unit is used for human-computer interaction, displaying the detected heartbeat and breathing signals according to the corresponding location.

Images