Heart rate detection method and system based on millimeter wave radar and storage medium

Abstract

The invention discloses a heart rate detection method and system based on a millimeter wave radar and a storage medium, and the method comprises the steps: carrying out the signal preprocessing of a vital sign signal collected by the millimeter wave radar, and obtaining a target phase signal; extracting a heartbeat signal from the target phase signal through an SSA-VME algorithm; performing time-domain heart rate estimation and frequency-domain heart rate estimation on the heartbeat signals in each window to obtain a time-domain heart rate estimation value and a frequency-domain heart rate estimation value of each window; fusing the time domain heart rate estimation value and the frequency domain heart rate estimation value of each window through Kalman filtering to obtain a heartbeat estimation value of each window; and realizing heart rate detection according to the heartbeat estimation value of each window. According to the method, time domain data and frequency domain data are fused by using Kalman filtering on the basis of the SSA-VME algorithm, so that the accuracy and the stability of heart rate estimation are improved.

Description

Technical Field

[0001] This invention relates to the field of signal processing technology, and in particular to a heart rate detection method, system and storage medium based on millimeter-wave radar. Background Technology

[0002] With the continuous improvement of public health awareness, people's health management concepts are gradually shifting towards "prevention first," and more and more people are choosing regular physical examinations to detect potential health risks in advance. Heart rate is an important basis for judging whether cardiopulmonary function is normal in a physical examination report. In sudden emergencies, rapid and accurate vital sign detection can provide reliable data support for medical staff to race against time in their rescue efforts, thereby saving lives.

[0003] Traditional heart rate monitoring devices are mostly contact-based, such as electrocardiogram (ECG) monitors, phonocardiographs, wristband pulse oximeters, and EEG caps. Later, to accommodate the need for mobile users, many manufacturers integrated heart rate monitoring into smart bracelets. While these contact-based devices can monitor heart rate in real time, they still have limitations: they are not well-suited for situations involving damaged skin or long-term monitoring. Therefore, there is an urgent need for a non-contact detection technology that can accurately track heart rate signals in real time.

[0004] Non-contact heartbeat detection devices include millimeter-wave radar. However, current millimeter-wave radar estimates target distance by detecting the time delay of reflected echoes, but it cannot effectively distinguish between complex respiratory and heartbeat signals, and it is also difficult to cope with clutter interference, limiting its application in high-precision vital sign monitoring. Furthermore, due to its wide operating frequency band, it may be subject to spectrum resource limitations and interference issues, making system implementation complex. Therefore, how to achieve high-precision vital sign detection in complex and dynamic real-world application scenarios has become an urgent problem to be solved.

[0005] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0006] The main objective of this invention is to provide a heart rate detection method, system, and storage medium based on millimeter-wave radar, aiming to address the technical problem of achieving high-precision vital sign detection in complex and dynamic real-world application scenarios.

[0007] To achieve the above objectives, the present invention provides a heart rate detection method based on millimeter-wave radar, the heart rate detection method based on millimeter-wave radar comprising:

[0008] The vital signs signals collected by millimeter-wave radar are preprocessed to obtain the target phase signal;

[0009] Heartbeat signals are extracted from the target phase signal based on the SSA-VME algorithm;

[0010] Time-domain and frequency-domain heart rate estimations are performed on the heartbeat signals in each window to obtain the time-domain and frequency-domain heart rate estimates for each window.

[0011] The time-domain and frequency-domain heart rate estimates for each window are adaptively fused using Kalman filtering to obtain the heart rate estimate for each window.

[0012] Heart rate detection is achieved based on the estimated heart rate values ​​from each window.

[0013] Optionally, the step of preprocessing the vital sign signals acquired by the millimeter-wave radar to obtain the target phase signal includes:

[0014] Range-dimensional Fourier transform is performed on the vital sign signals collected by millimeter-wave radar to obtain the range-frequency signal;

[0015] Static clutter filtering is performed on the distance-frequency signal to obtain echo signals at different distances;

[0016] Extract the echo signal of the target distance from the echo signals at different distances;

[0017] The echo signal from the target distance is subjected to phase unwrapping processing to obtain the target phase signal.

[0018] Optionally, the step of statically filtering out clutter from the distance frequency signal to obtain echo signals at different distances includes:

[0019] The distance frequency signal is averaged to obtain a reference received pulse;

[0020] Based on the reference received pulse, static clutter filtering is performed on the distance-frequency signal using the average cancellation method to obtain echo signals at different distances.

[0021] Optionally, the step of performing phase unwrapping processing on the echo signal of the target distance to obtain the target phase signal includes:

[0022] The target phase at each moment is calculated using the tangent formula based on the echo signal from the target distance.

[0023] Based on the echo signal of the target distance, the target phase at each time moment is de-wrapped to obtain the target phase signal.

[0024] Optionally, the phase unwrapping processing of the target phase at each time step based on the echo signal of the target distance includes:

[0025] Based on the target distance, the difference between adjacent phases is calculated according to the target phase at each time point from the echo signal.

[0026] Based on the adjacent phase difference, the jump phase is extracted from the target phase at each moment;

[0027] The phase-unwrapping process of the phase transition is performed by a preset transition elimination algorithm.

[0028] Optionally, the extraction of the heartbeat signal from the target phase signal based on the SSA-VME algorithm includes:

[0029] The parameters of the VME algorithm are optimized using the SSA algorithm based on the target phase signal;

[0030] The heartbeat signal is extracted from the target phase signal based on the optimized VME algorithm.

[0031] Optionally, time-domain heart rate estimation is performed on the heartbeat signals in each window to obtain the time-domain heart rate estimate for each window, including:

[0032] The invalid peak and valley signals in the heartbeat signal of each window are removed by the peak and valley difference denoising method to obtain the peak and valley difference denoised signal in each window;

[0033] Peak finding operation is performed on the peak-valley difference denoised signal within each window to determine multiple signal peak values;

[0034] Based on a preset peak-to-peak time interval, select multiple valid signal peaks within each window from multiple signal peaks;

[0035] The time-domain heart rate estimate for each window is calculated based on the peak values ​​of multiple valid signals within each window.

[0036] Optionally, frequency domain heart rate estimation is performed on the intracardiac signal in each window to obtain the frequency domain heart rate estimate for each window, including:

[0037] The highest frequency value of each window is determined using the Double-CZT algorithm based on the internal rap signal of each window.

[0038] Calculate the frequency domain heart rate estimate for each window based on the highest frequency value of each window.

[0039] Furthermore, to achieve the above objectives, the present invention also proposes a heart rate detection system based on millimeter-wave radar, the heart rate detection system based on millimeter-wave radar comprising:

[0040] The processing module is used to preprocess the vital signs signals collected by the millimeter-wave radar to obtain the target phase signal;

[0041] The extraction module is used to extract the heartbeat signal from the target phase signal based on the SSA-VME algorithm;

[0042] The calculation module is used to perform time-domain heart rate estimation and frequency-domain heart rate estimation on the heartbeat signals of each window, and obtain the time-domain heart rate estimate and frequency-domain heart rate estimate of each window;

[0043] The fusion module is used to adaptively fuse the time-domain heart rate estimates and frequency-domain heart rate estimates of each window through Kalman filtering to obtain the heart rate estimate of each window.

[0044] The visual module is used to detect heart rate based on the heart rate estimates of each window.

[0045] Furthermore, to achieve the above objectives, the present invention also proposes a heart rate detection device based on millimeter-wave radar, the device comprising: a memory, a processor, and a heart rate detection program based on millimeter-wave radar stored in the memory and executable on the processor, the heart rate detection program based on millimeter-wave radar being configured to implement the steps of the heart rate detection method based on millimeter-wave radar as described above.

[0046] Furthermore, to achieve the above objectives, the present invention also proposes a storage medium storing a heart rate detection program based on millimeter-wave radar, wherein when the millimeter-wave radar-based heart rate detection program is executed by a processor, it implements the steps of the heart rate detection method based on millimeter-wave radar as described above.

[0047] This invention first preprocesses the vital sign signals acquired by millimeter-wave radar to obtain the target phase signal. Then, based on the SSA-VME algorithm, it extracts the heartbeat signal from the target phase signal. Next, it performs time-domain and frequency-domain heart rate estimation on the heartbeat signals in each window, obtaining the time-domain and frequency-domain heart rate estimates for each window. Kalman filtering is then used to adaptively fuse these estimates to obtain the heartbeat estimate for each window. Finally, heart rate detection is achieved based on the heartbeat estimates for each window. This invention preprocesses the acquired raw signal to obtain a precise phase signal. The heartbeat signal is then extracted and separated using a VME algorithm optimized by the SSA algorithm. The introduction of the SSA algorithm to optimize key parameters in the VME significantly improves the separation effect of the heartbeat signal. The fusion of time-domain and frequency-domain data through Kalman filtering significantly improves the accuracy and stability of heart rate estimation, enabling high-precision vital sign detection in complex and dynamic real-world application scenarios. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the structure of a heart rate detection device based on millimeter-wave radar in the hardware operating environment of the embodiment of the present invention;

[0049] Figure 2This is a flowchart illustrating the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention.

[0050] Figure 3 This is a schematic diagram of the radar signal preprocessing process of the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention;

[0051] Figure 4 This is a schematic diagram of target distance unit selection in the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention;

[0052] Figure 5 This is a schematic diagram of the signals before and after phase unwinding in the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention;

[0053] Figure 6 This is a flowchart illustrating the optimization of VME algorithm parameters using the SSA algorithm in the first embodiment of the heart rate detection method based on millimeter-wave radar according to the present invention.

[0054] Figure 7 This is a schematic diagram of the peak-valley difference denoising result of the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention;

[0055] Figure 8 This is a schematic diagram of peak-to-peak time interval denoising in the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention;

[0056] Figure 9 This is a flowchart of the Double-CZT algorithm in the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention;

[0057] Figure 10 This is a schematic diagram of the time-frequency domain signal fusion method based on Kalman filtering in the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention.

[0058] Figure 11 This is a graph showing the estimated heart rate curve of the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention.

[0059] Figure 12 This is a flowchart illustrating the overall process of the millimeter-wave radar heart rate detection method according to the first embodiment of the present invention.

[0060] Figure 13 This is a structural block diagram of the first embodiment of the heart rate detection system based on millimeter-wave radar of the present invention.

[0061] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0062] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0063] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of a heart rate detection device based on millimeter-wave radar, which is part of the hardware operating environment of the embodiment of the present invention.

[0064] like Figure 1 As shown, the millimeter-wave radar-based heart rate detection device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen and an input unit such as a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wireless-Fidelity (Wi-Fi) interface). The memory 1005 may be high-speed random access memory (RAM) or stable non-volatile memory (NVM), such as a disk storage device. The memory 1005 may also optionally be a storage system independent of the aforementioned processor 1001.

[0065] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on millimeter-wave radar-based heart rate detection devices, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0066] like Figure 1 As shown, the memory 1005, which serves as a storage medium, may include an operating system, a network communication module, a user interface module, and a heart rate detection program based on millimeter-wave radar.

[0067] exist Figure 1In the millimeter-wave radar-based heart rate detection device shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and memory 1005 in the millimeter-wave radar-based heart rate detection device of the present invention can be set in the millimeter-wave radar-based heart rate detection device. The millimeter-wave radar-based heart rate detection device calls the millimeter-wave radar-based heart rate detection program stored in the memory 1005 through the processor 1001 and executes the millimeter-wave radar-based heart rate detection method provided in the embodiment of the present invention.

[0068] This invention provides a heart rate detection method based on millimeter-wave radar, referring to... Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention.

[0069] In this embodiment, the heart rate detection method based on millimeter-wave radar includes the following steps:

[0070] Step S10: Perform signal preprocessing on the vital signs signals acquired by the millimeter-wave radar to obtain the target phase signal.

[0071] It is easy to understand that the executing entity of this embodiment can be a heart rate detection system based on millimeter-wave radar with functions such as data processing, network communication and program execution, or other computer devices with similar functions. This embodiment does not limit it.

[0072] Millimeter-wave radar can be a frequency-modulated continuous wave (FM CW) radar.

[0073] It should also be noted that the vital signs signals acquired by millimeter-wave radar often contain more complex noise. To successfully extract clean and accurate heartbeat signals, preprocessing of the vital signs signals is necessary to improve the quality of useful signals, filter out unwanted clutter, and ensure the accuracy of subsequent analysis. (Reference) Figure 3 , Figure 3 This is a schematic diagram of the radar signal preprocessing process of the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention. It mainly includes range-fourth-FT, static clutter filtering, target range cell selection, phase extraction and phase unwinding.

[0074] Furthermore, the signal preprocessing method for obtaining the target phase signal from the vital signs signals acquired by the millimeter-wave radar is as follows: performing a range-dimensional Fourier transform on the vital signs signals acquired by the millimeter-wave radar to obtain the range-frequency signal (i.e., the range-frequency signal); performing static clutter filtering on the range-frequency signal to obtain echo signals at different distances; extracting the target range echo signal from the echo signals at different distances; and performing phase unwrapping processing on the target range echo signal to obtain the target phase signal.

[0075] Furthermore, the method for statically filtering out clutter from the range-frequency signal to obtain echo signals at different distances is as follows: the range-frequency signal is averaged to obtain a reference received pulse; based on the reference received pulse, the range-frequency signal is statically filtered out using the average cancellation method to obtain echo signals at different distances.

[0076] In practical implementation, average subtraction is used to filter out static noise. Unlike dynamic targets, the distance between a static target and the radar antenna remains constant. Therefore, when the radar's transmitted signal contacts the same stationary object, the time delay of its echo signal should remain consistent, and neither the frequency nor the phase should change. Signals with this characteristic are also called DC signals. The reference received pulse C[M] can be obtained by averaging all received DC signals (i.e., range-frequency signals).

[0077]

[0078] Where M represents the sampling points in the fast time dimension (i.e., the distance dimension), and N represents the sampling points in the slow time dimension.

[0079] Subsequently, each received DC signal is subtracted from the reference received pulse to obtain echo signals at different distances. Figure 4 , Figure 4 This is a schematic diagram of target range cell selection in the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention. The expression for the average cancellation method is as follows:

[0080] R[M,N]=R[M,N]-C[M]

[0081] Figure 4 The target distance cell selection diagram is a two-dimensional distance-time graph. The distance cell with the highest energy is selected as the target cell, and the echo signal of the target distance is determined based on the target cell.

[0082] Furthermore, the target phase signal is obtained by performing phase unwrapping processing on the echo signal of the target distance. The processing method is as follows: the target phase at each time moment is calculated based on the echo signal of the target distance using the tangent formula; the target phase at each time moment is then unwrapped based on the echo signal of the target distance to obtain the target phase signal, as referenced. Figure 5 , Figure 5 This is a schematic diagram of the signals before and after phase unwinding in the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention.

[0083] In this embodiment, phase extraction can be performed on the echo signal of the target distance to obtain key information such as distance and speed contained in the signal.

[0084] The echo signal of the target distance is a complex signal z(t), consisting of a real part I(t) and an imaginary part Q(t), and its expression is:

[0085] z(t) = I(t) + jQ(t)

[0086] By performing the arctangent operation on the complex signal z(t), the phase φ(t) of the target can be obtained, expressed as:

[0087]

[0088] The phase φ(t) ranges from (-π, π).

[0089] Furthermore, the processing method for unwinding the target phase at each moment based on the echo signal of the target distance is as follows: the difference between adjacent phases is calculated based on the echo signal of the target distance at each moment; the jump phase is extracted from the target phase at each moment based on the difference between adjacent phases; and the jump phase is unwound by a preset jump elimination algorithm.

[0090] Because the phase is periodic, there is a jump from π to -π during the phase extraction process. This phenomenon is called phase entanglement. To address this, a phase unentanglement operation is needed to restore the unentangled continuous phase, so that the phase information can change smoothly, which is more conducive to the extraction of heartbeat information.

[0091] The steps for phase unwinding are as follows:

[0092] Detecting phase jumps: Take the difference between adjacent phases. If the absolute value of the difference exceeds π, a phase jump is considered to have occurred.

[0093] Phase correction: Corrects the phase that jumps, using a preset jump elimination algorithm (i.e., adding or subtracting 2π) to eliminate the jump;

[0094] Cumulative correction: Gradually perform phase-to-phase unwinding on all phases until all transitions are eliminated.

[0095] like Figure 5 The image shows a comparison of the measured signal phase before and after unwinding. Observation shows that before unwinding, the waveform is periodic and the signal fluctuates within the range of (-π, π], making the signal appear discontinuous and unstable; after unwinding, the waveform is relatively smooth and continuous, eliminating the jump problem across π to -π.

[0096] Step S20: Extract the heartbeat signal from the target phase signal based on the SSA-VME algorithm.

[0097] Furthermore, the parameters of the VME algorithm are optimized using the SSA algorithm based on the target phase signal; the heartbeat signal is then extracted from the target phase signal based on the optimized VME algorithm.

[0098] In this embodiment, reference Figure 6 , Figure 6 The flowchart below shows the optimization of VME algorithm parameters using the SSA algorithm in the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention:

[0099] To address the issue of VME parameter selection, the SSA algorithm is used to optimize the VME algorithm and return the most suitable parameters.

[0100] In the process of extracting heartbeat signals, the goal of signal denoising is to preserve the signal in the target frequency band (0.8-2.5Hz) while minimizing noise interference. The fitness function (FE) can sensitively reflect changes in signal complexity and distinguish high-quality signals from noise. Therefore, an FE-based fitness function can more accurately guide the optimization algorithm towards the optimal solution, thereby improving the accuracy and robustness of heart rate estimation.

[0101] FE utilizes exponential functions The FE algorithm uses fitness functions to measure the similarity between two vectors. The computation steps are as follows:

[0102] 1) Constructing the embedding matrix: Given an N-dimensional time series {x(j), 1 ≤ j ≤ N}, construct an embedding matrix of dimension k:

[0103] Y i =[x(1),x(2),…,x(N-k+1)]-x0(i)

[0104] i = 1, 2, ..., N-k+1

[0105] Where x0(i) is the mean of k consecutive x(i), and its expression is:

[0106]

[0107] 2) Calculate the similarity distance: For any two k-dimensional vectors Y in the matrix i and Y j Calculate their distance The maximum distance is usually chosen:

[0108]

[0109] Where i,j=1,2,…,Nk and i≠j.

[0110] 3) Introduce fuzzy membership function Determine the vector and Membership degree between and the average membership degree of all members other than itself

[0111]

[0112] 4) The final definition of fuzzy entropy FE is:

[0113]

[0114] Fuzzy entropy (FE) is primarily used to quantify the complexity of time series at a specific scale. A smaller FE value indicates a more regular internal structure of the time series, exhibiting higher self-similarity and suggesting stronger predictability. In this case, the corresponding alpha value is optimal, which helps improve the accuracy and stability of subsequent signal processing.

[0115] Optimization of penalty factor α:

[0116] This embodiment uses the SSA algorithm to optimize the penalty factor α in the VME parameters and selects FE as the fitness function of the SSA algorithm. This variational mode extraction based on the sparrow search algorithm is named the SSA-VME algorithm in this paper. The specific steps of SSA-VME are as follows:

[0117] 1) Set the initial parameters of the VME algorithm: Input the initial center frequency omega_int = 0.06 and the range of values ​​for the penalty factor alpha;

[0118] 2) SSA algorithm initialization: Randomly set the population location, the number of sparrows in the population, and the number of iterations.

[0119] 3) Sparrow population classification: The entire sparrow population is divided into two parts: foragers and followers;

[0120] 4) Perform VME decomposition on the preprocessed phase-form vital sign signals and calculate the fuzzy entropy;

[0121] 5) Fitness calculation: Randomly select parameter alpha, extract the corresponding signal through VME, and calculate the corresponding fitness;

[0122] 6) Update the positions of population members according to the principles and implementation rules of the SSA algorithm;

[0123] Convergence and stopping conditions: a) When the maximum fitness value is reached, the optimal solution is considered to have been found, and the iteration stops; b) When the average position change of the sparrow population is less than a set threshold, the algorithm is considered to have reached a stable state, and the iteration stops; c) When the algorithm reaches the preset maximum number of iterations, the parameters of the individual with the best fitness in the current population are output as the final solution.

[0124] Optimized VME algorithm:

[0125] Input: Initialization n ← 0. Where The center frequency of the desired mode.

[0126] Repeat: n←n+1

[0127] 1) For all cases where ω≥0, update

[0128]

[0129] 2) Update ω d :

[0130]

[0131] 3) For all cases where ω≥0, update the Lagrange multipliers.

[0132]

[0133] Until: The following convergence conditions must be met:

[0134]

[0135] Output:

[0136] in, The target modal component (i.e., the heartbeat signal).

[0137] Step S30: Perform time-domain heart rate estimation and frequency-domain heart rate estimation on the heartbeat signal of each window to obtain the time-domain heart rate estimate and frequency-domain heart rate estimate of each window.

[0138] It should also be noted that heartbeat signals typically exhibit a periodic waveform. A complete heartbeat cycle contains a peak (the high point during cardiac contraction) and a trough (the low point during cardiac relaxation). The peak-trough interval of the heartbeat signal reflects the heart's pulsation cycle. By detecting these peak-trough features, the signal's periodicity can be captured, thus enabling heart rate estimation. The basic idea of ​​peak-trough difference denoising is to extract the peaks and troughs from the signal and calculate their difference to reflect the signal's periodic variation characteristics.

[0139] It should also be noted that a sliding window is used to extract the heartbeat signal from the SSA-VME algorithm. The initial sliding window size is set to 12s and the sliding step size is 1s. The heart rate is estimated from both the time domain and the frequency domain for the heartbeat signal in each window.

[0140] Furthermore, the temporal heart rate estimation of the heartbeat signal in each window is performed. The processing method for obtaining the temporal heart rate estimate of each window is as follows: invalid peak and valley signals in the heartbeat signal of each window are removed by peak-valley difference denoising method (i.e., peak-valley difference denoising method) to obtain peak-valley difference denoised signal in each window; peak finding operation is performed on the peak-valley difference denoised signal in each window to determine multiple signal peaks; multiple effective signal peaks in each window are selected from the multiple signal peaks based on the preset peak-peak time interval; and the temporal heart rate estimate of each window is calculated according to the multiple effective signal peaks in each window using the heart rate conversion formula.

[0141] Heart rate conversion formula:

[0142] Heart rate = 60 / t

[0143] t is the time interval between effective signal peaks.

[0144] The implementation steps of the peak-valley difference denoising method are as follows:

[0145] 1) Finding peaks and troughs: such as Figure 7 As shown, Figure 7 This is a schematic diagram of the peak-valley difference denoising result of the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention. The signal is traversed to find the local maximum value, which satisfies formula (1) and is the peak value; the local minimum value is found to satisfy formula (2) and is the valley value.

[0146] x(t)>x(t-1); x(t)>x(t+1) (1)

[0147] x(t) <x(t-1);x(t)<x(t+1) (2)

[0148] 2) Calculate the peak-valley difference: Calculate the difference for each pair of peak and valley values, as shown in equation (3):

[0149] Δx=x peak-x valley (3)

[0150] Where, x peak The peak value in the time-domain waveform of the heartbeat; x valley This represents the valley value in the heartbeat time-domain waveform.

[0151] 3) Removing invalid peaks and troughs: Real heartbeat signals usually have significant amplitude variations, while noise is typically high-frequency random fluctuation, resulting in smaller false peaks and troughs. This can be mitigated by setting a reasonable minimum amplitude threshold Δx. min After eliminating invalid peaks and valleys, the Δx set in this scheme min It is 0.3.

[0152] Figure 7 In the diagram, green and yellow dots are used to mark the effective peak and valley values ​​with amplitude changes greater than 0.3, respectively.

[0153] The implementation steps of the peak-to-peak time interval denoising method are as follows:

[0154] 1) Finding peak values: such as Figure 8 As shown, Figure 8 This is a schematic diagram of peak-to-peak time interval denoising in the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention. Peak-finding operation is performed on the signal after the peak-valley difference method to find all the peak values ​​in the signal.

[0155] 2) Calculate the peak-to-peak time interval Δd: Set the expressions for the minimum and maximum time intervals as shown in equation (4). If Δd is within the interval, retain the first peak; otherwise, discard the peak.

[0156]

[0157] like Figure 8 As shown, peak points with a peak-to-valley difference greater than 0.3, selected by the peak-to-valley difference denoising method, are marked with green dots. These peak points are then subjected to peak-to-peak time interval denoising, and discarded peak points are marked with red. Peak-to-valley difference denoising effectively removes short-term noise, making the heartbeat signal smoother and reducing the impact of instantaneous fluctuations on signal quality. Peak-to-peak time interval denoising accurately extracts the heartbeat peaks of the heartbeat cycle, improving the reliability of heart rate estimation. Combining these two methods not only improves the accuracy and stability of heart rate estimation but also effectively reduces noise interference and minimizes the impact of outlier data on the results. The combination of peak-to-valley difference denoising and peak-to-peak time interval denoising enhances signal robustness and improves overall accuracy, providing a guarantee for more accurate vital sign detection.

[0158] Furthermore, the frequency domain heart rate estimation of the heartbeat signal in each window is performed. The processing method to obtain the frequency domain heart rate estimate of each window is as follows: the highest frequency value of each window is determined by the Double-CZT algorithm based on the heartbeat signal of each window; the frequency domain heart rate estimate of each window is calculated based on the highest frequency value of each window.

[0159] In this embodiment, reference Figure 9 , Figure 9 This is a flowchart of the Double-CZT algorithm in the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention. Double-CZT introduces a second CZT process on top of the traditional CZT. By performing two CZT processes with continuously narrowing frequency ranges, a more refined scan of the frequency range of interest can be performed based on the initial spectral analysis, thereby achieving more accurate heart rate localization. Compared to a single CZT, Double-CZT can effectively reduce the impact of noise interference and significantly reduce the estimation error caused by insufficient frequency resolution, thereby further improving the accuracy and robustness of heart rate estimation.

[0160] The core idea of ​​this algorithm is to employ a two-stage CZT transform strategy, progressing from coarse to fine, to gradually improve frequency resolution. First, a preliminary CZT transform is performed to obtain spectral information for the entire target frequency band and determine the locations of major frequency peaks. Then, based on the preliminary spectral results, a finer CZT transform is performed near the frequency peaks of interest, further narrowing the frequency scanning range and improving the accuracy of frequency localization, ensuring that the extracted heartbeat signal is more accurate and reliable.

[0161] Let the original signal (i.e., the heartbeat signal) be x[n], and its length be N. The specific steps for performing Double-CZT on this signal are as follows:

[0162] 1) First CZT Transform: Set a relatively wide frequency range, such as [f min ,f max Calculate the spectrum of M1 sampling points:

[0163]

[0164] M1 is the number of sampling points in the first CZT, covering the entire initial frequency range.

[0165] 2) Determine the fine frequency band range, and select the peak value f based on the first CZT transform. peak The narrow frequency range in which it is located, such as pf peak -Δf,f peak +Δf], and determine the new starting sampling point position A2 and the spiral extension W2:

[0166]

[0167] M2 represents the number of sampling points in the second CZT, with a higher resolution than M1, covering the frequency range of the second CZT.

[0168] Double-CZT accurately pinpoints the peak position through two CZT transformations, obtaining the highest frequency value of the maximum amplitude for each window. Then, the heart rate (i.e., the frequency domain heart rate estimate) is calculated by multiplying the frequency corresponding to the maximum amplitude by 60. Simultaneously, it effectively suppresses noise interference and reduces sensitivity to non-target frequency band noise. In the process of refining the heartbeat signal spectrum, this method not only improves the accuracy of frequency estimation but also demonstrates strong robustness, enabling stable extraction of the heartbeat frequency even in complex environments.

[0169] Step S40: Adaptively fuse the time-domain heart rate estimates and frequency-domain heart rate estimates of each window using Kalman filtering to obtain the heart rate estimate for each window.

[0170] refer to Figure 10 , Figure 10 This diagram illustrates the time-frequency domain signal fusion method based on Kalman filtering in the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention. The time-domain heart rate estimate determined by the peak-valley difference denoising method and the peak-peak time interval denoising method is set as the predicted value of Kalman filtering, and the frequency-domain heart rate estimate determined by the Double-CZT method is set as the observed value of Kalman filtering. Finally, the predicted value is adjusted by the observed value and the Kalman gain to obtain the final state value.

[0171] In this embodiment, the heart rate prediction value is set to... Heart rate observation value z k The Kalman fusion estimate of heart rate is expressed as:

[0172]

[0173] Where K is the Kalman gain.

[0174] According to the property of variance, if the predicted value and the observed value are independent, then Variance can be obtained expression:

[0175]

[0176] right Taking the derivative with respect to K and setting it to zero, we obtain the point of minimum variance:

[0177]

[0178] Solving for:

[0179]

[0180] Based on the above derivation, the formula for heart rate estimation using the Kalman filter method can be summarized as follows:

[0181] Algorithm: Heart rate estimation using the Kalman filter method:

[0182] Input: Initial estimate P t-1

[0183] Repeat:

[0184] 1) The time-domain heart rate estimate determined by the peak-valley difference denoising method and the peak-peak time interval denoising method is used as the predicted value of the Kalman filter:

[0185] P k|k-1 =Predicted covariance

[0186] 2) The frequency domain heart rate estimate determined by the Double-CZT method is used as the observation value for Kalman filtering:

[0187] z k =Observed value

[0188] 3) Update steps:

[0189] i. Calculate the Kalman gain:

[0190] ii. Updated state estimation:

[0191] iii. Updated covariance: P k|k = (1-K) k )P k|k-1 ;

[0192] Iteration count: k = k + 1.

[0193] It should also be noted that the collected vital signs signals were all 60 seconds long. The initial size of the sliding window was set to 12 seconds, with a step size of 1 second. The window was slid 48 times in sequence, and a heart rate estimate was calculated at each slide. Finally, a total of 48 heart rate data points were output.

[0194] Step S50: Detect heart rate based on the heart rate estimates of each window.

[0195] Connecting the data points obtained after sliding the window in chronological order allows for the creation of an estimated heart rate signal curve. This curve not only clearly displays the dynamic trend of heart rate changes over time but also helps in observing the fluctuation trend of the heartbeat signal in different time periods. Comparing this curve with the actual heartbeat signal measured by the electrocardiogram monitor yields the following results: Figure 11 As shown, Figure 11 This is a graph showing the estimated heart rate curve of the first embodiment of the heart rate detection method based on millimeter-wave radar of the present invention:

[0196] like Figure 11 As shown, the black curve represents the actual heart rate signal measured by the heart rate monitor, while the red curve represents the estimated heart rate signal calculated by the Kalman filter-based heart rate estimation method. Overall, the two curves move in tandem, indicating that the estimated heart rate signal is numerically close to the actual heart rate signal and maintains a relatively synchronized trend in both the rise and fall of heart rate. The comparative results demonstrate that the heart rate estimation method proposed in this paper possesses high accuracy and reliability in dynamically tracking heart rate changes, accurately reflecting the true heart rate fluctuations of the subjects, and validating its effectiveness in heart rate detection experiments.

[0197] In this embodiment, reference Figure 12 , Figure 12This is a flowchart illustrating the overall process of the millimeter-wave radar heart rate detection method according to the first embodiment of the present invention. Signal Acquisition: Vital signs signals are acquired using a combination of an IWR6843AOP millimeter-wave radar and a DCA1000 acquisition card. The radar is responsible for transmitting and receiving millimeter-wave signals, while the DCA1000 acquisition card converts the echo signals into digital data and stores them. The acquired raw data is transmitted to the computer in .bin format to provide data support for further signal processing. Radar Signal Preprocessing: Preprocessing the acquired data is a key step in improving target signal quality and reducing environmental noise interference. Range-dimensional FFT is used to calculate the frequency domain information of the target at different range cells, providing a foundation for subsequent signal analysis. Static clutter filtering removes fixed echo interference from the background environment, ensuring effective extraction of dynamic target signals. Appropriate range cells are selected to focus on the area of ​​interest and improve data effectiveness. Phase extraction and phase dewinding correct errors caused by phase abrupt changes, making the signal more continuous. After these preprocessing steps, the radar signal quality is significantly improved, and the heartbeat signal extraction is more accurate. Heartbeat Signal Extraction and Separation: Signal separation processing is performed on the specific frequency band containing the heartbeat frequency, retaining only the effective signal within the target frequency band. This reduces interference from irrelevant frequency components and lowers the computational complexity of subsequent calculations, allowing the algorithm to more efficiently focus on changes in the target signal. The Sparrow Search Algorithm (SSA) is introduced to optimize key parameters of the VME algorithm, constructing the SSA-VME algorithm. This algorithm is then used to extract the heartbeat signal from radar data, making vital sign information more accurate. Heart Rate Estimation: After acquiring the heartbeat signal, a sliding window method is used to segment the data, and heart rate estimation is performed in both the time and frequency domains. To further improve estimation accuracy, Kalman filtering is used to fuse time and frequency domain information to obtain the final heartbeat value. Based on the estimation results, a heartbeat waveform is plotted to visually present the trend of heart rate changes.

[0198] In this embodiment, the vital sign signals acquired by millimeter-wave radar are first preprocessed to obtain the target phase signal. Then, the heartbeat signal is extracted from the target phase signal based on the SSA-VME algorithm. Next, time-domain and frequency-domain heart rate estimations are performed on the heartbeat signals in each window to obtain the estimated heart rate values ​​for each window. Kalman filtering is then used to adaptively fuse the time-domain and frequency-domain heart rate estimates for each window to obtain the estimated heartbeat value for each window. Finally, heart rate detection is achieved based on the estimated heartbeat values ​​for each window. This embodiment preprocesses the acquired raw signal to obtain an accurate phase signal. The heartbeat signal is then extracted and separated using the VME algorithm optimized by the SSA algorithm. The introduction of the SSA algorithm to optimize key parameters in the VME significantly improves the separation effect of the heartbeat signal. The fusion of time-domain and frequency-domain data through Kalman filtering significantly improves the accuracy and stability of heart rate estimation, enabling high-precision vital sign detection in complex and dynamic real-world application scenarios.

[0199] Reference Figure 13 , Figure 13 This is a structural block diagram of the first embodiment of the heart rate detection system based on millimeter-wave radar of the present invention.

[0200] like Figure 13 As shown, the heart rate detection system based on millimeter-wave radar proposed in this embodiment of the invention includes:

[0201] Processing module 1301 is used to preprocess the vital signs signals collected by millimeter-wave radar to obtain the target phase signal;

[0202] Extraction module 1302 is used to extract heartbeat signal from the target phase signal based on the SSA-VME algorithm;

[0203] The calculation module 1303 is used to perform time-domain heart rate estimation and frequency-domain heart rate estimation on the heartbeat signal of each window, and obtain the time-domain heart rate estimate and frequency-domain heart rate estimate of each window;

[0204] The fusion module 1304 is used to adaptively fuse the time-domain heart rate estimates and frequency-domain heart rate estimates of each window through Kalman filtering to obtain the heart rate estimates of each window.

[0205] The visual module 1305 is used to detect heart rate based on the heart rate estimates of each window.

[0206] Other embodiments or specific implementations of the heart rate detection system based on millimeter-wave radar of the present invention can be referred to the above-described method embodiments, and will not be repeated here.

[0207] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0208] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0209] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0210] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A heart rate detection method based on millimeter-wave radar, characterized in that, The method includes the following steps: The vital signs signals collected by millimeter-wave radar are preprocessed to obtain the target phase signal; Heartbeat signals are extracted from the target phase signal based on the SSA-VME algorithm; Time-domain and frequency-domain heart rate estimations are performed on the heartbeat signals in each window to obtain the time-domain and frequency-domain heart rate estimates for each window. The time-domain and frequency-domain heart rate estimates for each window are adaptively fused using Kalman filtering to obtain the heart rate estimate for each window. Heart rate detection is achieved based on the estimated heart rate values ​​from each window.

2. The method as described in claim 1, characterized in that, The process of preprocessing the vital signs signals acquired by the millimeter-wave radar to obtain the target phase signal includes: Range-dimensional Fourier transform is performed on the vital sign signals collected by millimeter-wave radar to obtain the range-frequency signal; Static clutter filtering is performed on the distance-frequency signal to obtain echo signals at different distances; Extract the echo signal of the target distance from the echo signals at different distances; The echo signal from the target distance is subjected to phase unwrapping processing to obtain the target phase signal.

3. The method as described in claim 2, characterized in that, The static clutter filtering of the distance frequency signal to obtain echo signals at different distances includes: The distance frequency signal is averaged to obtain a reference received pulse; Based on the reference received pulse, the distance-frequency signal is statically filtered out using the average cancellation method to obtain echo signals at different distances.

4. The method as described in claim 2, characterized in that, The step of performing phase unwrapping processing on the echo signal at the target distance to obtain the target phase signal includes: The target phase at each moment is calculated using the tangent formula based on the echo signal from the target distance. Based on the echo signal of the target distance, the target phase at each time moment is de-wrapped to obtain the target phase signal.

5. The method as described in claim 4, characterized in that, The phase unwrapping processing of the target phase at each time moment based on the echo signal of the target distance includes: Based on the target distance, the difference between adjacent phases is calculated according to the target phase at each time point from the echo signal. Based on the adjacent phase difference, the jump phase is extracted from the target phase at each moment; The phase-unwrapping process of the phase transition is performed by a preset transition elimination algorithm.

6. The method according to any one of claims 1-5, characterized in that, The extraction of the heartbeat signal from the target phase signal based on the SSA-VME algorithm includes: The parameters of the VME algorithm are optimized using the SSA algorithm based on the target phase signal; The heartbeat signal is extracted from the target phase signal based on the optimized VME algorithm.

7. The method as described in claim 6, characterized in that, Time-domain heart rate estimation is performed on the intracardiac signals in each window to obtain the estimated time-domain heart rate values ​​for each window, including: The invalid peak and valley signals in the heartbeat signal of each window are removed by the peak and valley difference denoising method to obtain the peak and valley difference denoised signal in each window; Peak finding operation is performed on the peak-valley difference denoised signal within each window to determine multiple signal peak values; Based on a preset peak-to-peak time interval, select multiple valid signal peaks within each window from multiple signal peaks; The time-domain heart rate estimate for each window is calculated based on the peak values ​​of multiple valid signals within each window.

8. The method as described in claim 6, characterized in that, Frequency domain heart rate estimation is performed on the intracardiac signals in each window to obtain the frequency domain heart rate estimate for each window, including: The highest frequency value of each window is determined using the Double-CZT algorithm based on the internal rap signal of each window. Calculate the frequency domain heart rate estimate for each window based on the highest frequency value of each window.

9. A heart rate detection system based on millimeter-wave radar, characterized in that, The system includes: The processing module is used to preprocess the vital signs signals collected by the millimeter-wave radar to obtain the target phase signal; The extraction module is used to extract the heartbeat signal from the target phase signal based on the SSA-VME algorithm; The calculation module is used to perform time-domain heart rate estimation and frequency-domain heart rate estimation on the heartbeat signals of each window, and obtain the time-domain heart rate estimate and frequency-domain heart rate estimate of each window; The fusion module is used to adaptively fuse the time-domain heart rate estimates and frequency-domain heart rate estimates of each window through Kalman filtering to obtain the heart rate estimate of each window. The visual module is used to detect heart rate based on the heart rate estimates of each window.

10. A storage medium, characterized in that, The storage medium stores a heart rate detection program based on millimeter-wave radar, which, when executed by a processor, implements the steps of the heart rate detection method based on millimeter-wave radar as described in any one of claims 1 to 8.

Images