Animal vital sign measuring method based on millimeter wave radar, applied device and pet collar

Abstract

The invention discloses an animal vital sign measuring method based on a millimeter wave radar, an applied device and a pet necklace, and the method comprises the steps: collecting radar data of a pet target area through the millimeter wave radar, and dynamically locking an echo section of the target area according to the body type of a pet in the collection process, a certain distance is formed between the target area and the millimeter-wave radar; eliminating environmental noise and motion artifacts in radar data by adopting a band-pass filter, and reserving effective signals in a range of 0.2-5Hz; then, performing sliding window segmentation on the radar data; extracting a heart rate signal and a respiration signal from the segmented radar data; and inputting the extracted heart rate signal and respiration signal into the trained measurement and calculation model, and outputting heart rate and respiration frequency predicted values in real time. The millimeter wave radar is used for collecting signals and removing environmental noise and motion artifacts in the signals, the problems of low measurement precision and the like caused by long hair shielding and high-frequency physiological rhythm of the pet are solved, and the heart rate and the respiratory rate of the pet can be accurately measured in real time.

Description

Technical Field

[0001] This invention relates to the field of pet health monitoring technology, and in particular to a method and device for measuring animal vital signs based on millimeter-wave radar, and a pet collar. Background Technology

[0002] Currently, wearable products such as pet collars and vests on the market all have functions for detecting pet heart rate and respiratory rate. For example, the PetPace collar-type pet health monitoring system in the United States and the Inupathy heart rate collar in Japan measure the vital signs of pets, such as heart rate and respiratory rate. These products mainly monitor heart rate indicators by capturing the sound of the pet's heartbeat, which is easily affected by external noise and has relatively low accuracy.

[0003] Current methods for detecting animal heart rate and respiratory rate include: 1. The detection equipment integrates PPG sensors or ECG electrodes for detection. For example, SeaverCEEFITPulse & ECG is used on saddles, FeedHeatMonitor is used on cows, and PetPace is used on dogs. 2. Infrared thermal imaging / visible light video detection method: that is, extracting the micro-movement of the chest or the rhythmic blood flow of the face from the camera image.

[0004] Among them, PPG sensors (Photoplethysmography) typically consist of light-emitting diodes (LEDs) and photodetectors (such as photodiodes). It is a non-invasive detection technology based on optical principles to detect changes in blood volume. It acquires physiological signals related to the cardiovascular system by emitting light into the skin and measuring changes in reflected or transmitted light. ECG (Electrocardiogram) electrodes are sensors used to collect signals of cardiac electrical activity. By contacting the skin surface, they transmit the weak electrical signals generated during the depolarization and repolarization of myocardial cells to the electrocardiograph, forming an electrocardiogram waveform.

[0005] For long-haired pets (such as most cats and long-haired dogs (Alaskan Malamutes, Afghan Hounds, etc.)), the above detection methods are not ideal. On the one hand, due to the strong absorption and scattering of visible / infrared light by long hair, the signal-to-noise ratio of PPG / RGB / IR measurement signals is extremely low. On the other hand, the static electricity generated by pet sebum and hair can also affect the electrodes. For example, cats and dogs have thick sebum and a lot of static electricity, which makes the contact impedance of ECG electrodes large and unstable. In addition, the rapid movement of the pet's tail and limbs interferes with the extraction of micro-motion signals, resulting in low measurement accuracy.

[0006] Existing wearable pet devices typically use Bluetooth to upload data to the client terminal, which is not suitable for long-distance transmission.

[0007] Millimeter-wave radar (such as OMRON millimeter-wave vital signs radar) is generally used to measure micro-movements in the human chest cavity. It utilizes the reflection characteristics of electromagnetic waves to achieve target detection, ranging, velocity measurement, and imaging. Combined with detection algorithms for human heart rate / respiratory rate, more accurate human vital signs data can be obtained. However, because pets' physiological characteristics (including physical features, heart rate, and respiratory rate) differ from those of humans—for example, the heart rate and respiratory rate of pets such as cats and dogs (cats have a heart rate of 140-240 bpm and a respiratory rate of 20-30 rpm, while dogs have a heart rate of 60-160 bpm and a respiratory rate of 15-40 rpm) are significantly higher than those of humans—there are currently no cases of millimeter-wave radar being used for measuring the vital signs of pets.

[0008] In view of this, the inventors have conducted in-depth research in the field of animal health monitoring technology and proposed a method for measuring animal vital signs based on millimeter-wave radar, as well as a device and pet collar. Summary of the Invention

[0009] The purpose of this invention is to provide a method and device for measuring animal vital signs based on millimeter-wave radar, as well as a pet collar. By using millimeter-wave radar to collect signals and removing environmental noise and motion artifacts from the signals, the invention overcomes the problems of low measurement accuracy caused by long fur and high-frequency physiological rhythms in pets, and can measure the heart rate and respiratory rate of pets in real time and accurately.

[0010] To achieve the above-mentioned technical objectives, the present invention adopts the following technical solution: On one hand, the present invention provides a method for measuring the vital signs of pets based on millimeter-wave radar, comprising: S1. Data Acquisition: Millimeter-wave radar is used to acquire radar data of the target area of ​​the pet. During the acquisition process, the echo segment of the target area is dynamically locked according to the size of the pet, and the target area is at a certain distance from the millimeter-wave radar. S2. Data preprocessing: A bandpass filter is used to eliminate environmental noise and motion artifacts in the radar data, retaining the effective signal in the 0.2-5Hz range; then, the radar data is divided into sliding windows. S3. Data Extraction: Extract heart rate and respiratory signals from the segmented radar data; S4. Data Calculation: Input the extracted heart rate and respiratory signals into the trained calculation model, and output the predicted values ​​of heart rate and respiratory rate in real time.

[0011] Furthermore, the data collection process is as follows: S11. The millimeter-wave radar should be directed towards the target area of ​​the pet's carotid artery or chest, and the distance between the radar antenna and the target area should be controlled between 0.3-2m. S12: Millimeter-wave radar transmits continuous wave or frequency-modulated continuous wave signals and collects reflected echo data; S13. The reflected echo data is converted into a time-domain data stream after analog-to-digital conversion.

[0012] Furthermore, a Butterworth bandpass filter is used, and the data preprocessing specifically includes: S21. Polarization Differential Hair Suppression: The vertical or horizontal polarization echo difference method is used to remove the hair scattering baseline in radar data. S22. Dual-channel differential motion elimination: The method of synthesizing a reference channel by combining dual-channel differential and virtual phase center is used to eliminate large-amplitude limb movements; S23. Set a fixed window and use a sliding window to divide the data.

[0013] Furthermore, data extraction specifically includes: S31. Heart rate signal extraction: The heart rate signal is extracted from the segmented radar data using continuous wavelet transform analysis. S32. Respiratory signal extraction: The respiratory signal is extracted from the segmented radar data using fast Fourier transform spectral analysis.

[0014] Furthermore, the measurement model adopts a CNN-based neural network model, the structure of which includes, in sequence: Input layer: Used to input preprocessed and extracted timing signals; Hidden layer: used for radar signal analysis; Output layer: A fully connected layer is used to output predicted values ​​of heart rate and respiratory rate; Loss Functions and Optimizers: Loss function: Mean squared error; Optimizer: Adam optimizer is used with a learning rate of 0.001; The training process of the measurement model includes: Data labeling: Tag data for pet heart rate and respiratory rate were collected using standard heart rate and respiratory rate monitoring equipment; Dataset partitioning: 70% of the labeled data is divided into a training set, 15% into a validation set, and 15% into a test set; Model training: Set batch size=64, epochs=100, input the training set into the model for training, and use an early stopping mechanism during training to avoid overfitting.

[0015] Furthermore, the data calculation specifically includes: S41. Model Loading: Deploying the model in the computing unit; S42. Real-time signal input: Input the extracted real-time heart rate and respiratory signals of the pet into the model; S43. Frequency Prediction: The model outputs predicted values ​​of heart rate and respiratory rate in real time.

[0016] On the other hand, the present invention provides a pet vital signs measurement device based on millimeter-wave radar, which can implement the aforementioned pet vital signs measurement method based on millimeter-wave radar. The device includes: Data acquisition module: Used to acquire radar data of the target area of ​​the pet using millimeter-wave radar. During the acquisition process, the echo segment of the target area is dynamically locked according to the size of the pet, and the target area is at a certain distance from the millimeter-wave radar. Data processing unit: Used to eliminate environmental noise and motion artifacts in radar data using a bandpass filter, retaining the effective signal in the 0.2-5Hz range; then, the radar data is divided into sliding windows; Data extraction unit: used to extract heart rate and respiratory signals from the segmented radar data; Data calculation unit: used to input the extracted heart rate and respiratory signals into the trained calculation model and output the predicted values ​​of heart rate and respiratory rate in real time.

[0017] Furthermore, the data acquisition module uses the IWR6843 millimeter-wave radar chip, which operates in the 60-64GHz frequency band, and the radar antenna is directed towards the pet's carotid artery area or chest.

[0018] A pet collar includes a control unit and a communication unit, a power management unit, and the aforementioned measuring device connected thereto.

[0019] Furthermore, the control unit integrates a microcontroller or edge computing unit, and the communication unit adopts Bluetooth, WiFi module and / or cellular module; The pet collar also includes a six-degree-of-freedom inertial measurement unit and a temperature and humidity sensor connected to the control unit.

[0020] By adopting the above solution, the present invention has the following advantages compared with the prior art: This invention uses millimeter-wave radar to acquire signals and combines it with a dual-polarization differential long hair suppression algorithm and dual-channel differential motion elimination to remove environmental noise and motion artifacts from the signals. This overcomes the problems of low measurement accuracy caused by long hair occlusion and high-frequency physiological rhythms in pets, and enables real-time and accurate measurement of pet heart rate and respiratory rate within a distance of 0.3-2m, improving adaptability and measurement accuracy for long-haired pets.

[0021] Adaptive gating and long-haired target recognition: During the acquisition process, the echo segment of the target area is dynamically locked according to the pet's body shape. That is, the measurement requirements of pets with different body shapes and fur are met by dynamically adjusting parameters. At the same time, it also improves posture and multi-scene robustness. The target can be re-locked within 2 seconds for postures such as lying down, lying on the side, standing, and licking fur. Enhanced safety and comfort: The millimeter-wave radar is completely non-contact with pets and has a low-power transmission of <10dBm, meeting the requirements of FCC / CESAR. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other modifications can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of a pet vital signs measurement method based on millimeter-wave radar, as described in an embodiment of the present invention. Figure 2 A block diagram of a measuring device and a pet collar for pet vital sign measurement methods based on millimeter-wave radar. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] like Figure 1 As shown in the figure, an embodiment of the present invention discloses a method for measuring the vital signs of pets based on millimeter-wave radar, including the following steps: S1. Data Acquisition: Millimeter-wave radar is used to acquire radar data of the target area of ​​the pet. During the acquisition process, the echo segment of the target area is dynamically locked according to the size of the pet, and the target area is at a certain distance from the millimeter-wave radar. S2. Data preprocessing: A bandpass filter is used to eliminate environmental noise and motion artifacts in the radar data, retaining the effective signal in the 0.2-5Hz range; then, the radar data is divided into sliding windows. S3. Data Extraction: Extract heart rate and respiratory signals from the segmented radar data; S4. Data Calculation: Input the extracted heart rate and respiratory signals into the trained calculation model, and output the predicted values ​​of heart rate and respiratory rate in real time.

[0026] Preferably, the data acquisition process is as follows: S11. The millimeter-wave radar should be directed towards the target area of ​​the pet's carotid artery or chest, and the distance between the radar antenna and the target area should be controlled between 0.3-2m. S12. The millimeter-wave radar transmits continuous wave (CW) or frequency modulated continuous wave (FMCW) signals and collects reflected echo data. S13. The reflected echo data is converted into a time-domain data stream by analog-to-digital conversion (ADC).

[0027] Furthermore, a Butterworth bandpass filter is used, and the data preprocessing specifically includes: S21. Polarization Differential Hair Suppression: This technique removes hair scattering baselines from radar data using vertical or horizontal polarization echo difference. Polarization differential hair suppression is a technique that eliminates hair scattering baselines in radar data through the difference between vertical (V) and horizontal (H) polarization echoes. The technical principle and core mechanism are as follows: ① Differences in polarization characteristics: Hair typically exhibits volume scattering, and its polarized echo has the following characteristics: The vertically polarized (VV) and horizontally polarized (HH) echoes are relatively similar, resulting in a low polarization ratio (e.g., VH / VV). Cross-polarized (VH or HV) echoes are relatively strong, reflecting multiple scattering from the hair crown. In contrast, cat and dog skin hair primarily exhibits surface scattering or secondary scattering, resulting in greater polarization differences and weaker cross-polarized echoes. By calculating the differences between polarization channels (e.g., VV-HH, VH / VV), the scattering characteristics of the hair can be highlighted and a suppression model can be constructed.

[0028] ② Polarization Difference Method Polarization difference method: Directly calculate the difference between vertical and horizontal polarization echoes (e.g., Δσ=σ_VV-σ_HH). Δσ in hair regions is usually small and can be suppressed by threshold filtering.

[0029] Polarization ratio method: Construct polarization ratios (e.g., R = σ_VH / σ_VV). Hair has a higher R value (due to volume scattering contribution) and can be used for mask extraction.

[0030] Dual-frequency polarization difference ratio factor (R): Combining the polarization difference between two frequency bands to eliminate background interference such as surface temperature, as shown in the formula: R=δT(f)δT(f)=T(f)−T(f)T(f)−T(f); where δT is the polarization brightness temperature difference value, and f1 and f2 are two frequency bands (such as 6GHz and 36GHz). By constructing the R factor, the baseline of hair scattering can be stripped away.

[0031] S22. Dual-channel differential motion elimination: The method of synthesizing a reference channel by combining dual-channel differential and virtual phase center is used to eliminate large-amplitude limb movements; Dual-channel differential acquisition involves synchronously acquiring signals through two independent channels (such as a measurement channel and a reference channel) and using differential operations to eliminate common-mode interference.

[0032] Virtual array construction involves recombining a single-channel phase signal into a multi-channel virtual array by dividing it into time slices. For example, in vital sign detection, the single-channel phase signal of a continuous wave radar is decomposed into m virtual array elements, which is equivalent to the multi-channel data of a single-input multiple-output (SIMO) radar.

[0033] Dual-channel differential motion cancellation technology achieves efficient suppression of large-amplitude limb motion interference through algorithmic synthesis of the physical differences between the two-channel signals and a virtual phase center. Its core advantage lies in generating a high-precision reference signal without requiring additional hardware.

[0034] S23. Set a fixed window (e.g., 30 seconds) and perform sliding window segmentation on the data. This can effectively convert long-term time series data into segmented processing units, which is convenient for subsequent feature extraction, pattern recognition or real-time monitoring.

[0035] Preferably, the data extraction process includes: S31. Heart rate signal extraction: The heart rate signal is extracted from the segmented radar data using continuous wavelet transform (CWT) analysis. Continuous wavelet transform (CWT) enables multi-resolution analysis of signals in the time and frequency domains by translating and scaling mother wavelet functions (such as the Morlet wavelet), making it particularly suitable for processing non-stationary, weak heart rate signals. Its mathematical expression is:

[0036] in, Ψ For Morlet wavelet functions, x(t) This is the original radar signal; a: Scale parameter, controls the scaling of the wavelet function (inversely proportional to frequency); b: Translation parameter, which controls the position of the wavelet function on the time axis.

[0037] Continuous wavelet transform (CWT), through time-frequency multiresolution analysis, has the advantage of effectively handling non-stationary signals, distinguishing between heart rate and respiratory interference, and adapting to dynamic changes in heart rate. By rationally selecting the mother wavelet, optimizing the scaling parameters, and combining preprocessing techniques, CWT can achieve high-precision and robust heart rate monitoring.

[0038] S32. Respiratory Signal Extraction: Respiratory signals are extracted from the segmented radar data using Fast Fourier Transform (FFT) spectral analysis. FFT is an efficient algorithm of Discrete Fourier Transform (DFT), converting time-domain signals into frequency-domain representations, making it suitable for extracting the frequency components of periodic signals. The stationarity of respiratory signals makes them particularly suitable for FFT analysis. Model formula:

[0039] Where x(t) is the original time-domain signal, typically in volts (V) or meters (m), or any time-varying integrable signal, such as radar echoes, ECG / respiratory waveforms, etc. t is a free and continuous time variable in seconds (s), and the integral iterates over the entire time axis; f is the frequency variable Hertz (Hz) continuous frequency; f>0 corresponds to a positive frequency component, and f<0 corresponds to a conjugate negative frequency component; j is the imaginary unit; 2πft is the product of angular frequency and time in radians. Convert the linear frequency ff to the angular frequency 2πf. e −j2πft Complex exponential basis functions—Fourier inner product kernels—are used to decompose signals in the frequency domain; in essence, they are rotating phase factors.

[0040] This invention utilizes reasonable radar data preprocessing, windowing, and spectrum analysis to enable FFT to accurately identify respiratory rate, meeting the needs of most pet medical and health monitoring scenarios.

[0041] Preferably, the measurement model adopts a CNN-based neural network model, the structure of which includes, in sequence: Input layer: Used to input preprocessed and extracted timing signals; Hidden layer: Proprietary architecture for radar signal analysis; Output layer: A fully connected layer is used to output predicted values ​​of heart rate and respiratory rate; Loss Functions and Optimizers: Loss function: Mean Squared Error (MSE) Optimizer: Adam optimizer is used with a learning rate of 0.001; The training process of the measurement model includes: Data labeling: Pets wear standard heart rate and respiratory rate monitoring devices to collect labeled data on their heart rate and respiratory rate; Dataset partitioning: 70% of the labeled data is divided into a training set, 15% into a validation set, and 15% into a test set; Model training: Set batch size=64, epochs=100, input the training set into the model for training, and use an early stopping mechanism during training to avoid overfitting.

[0042] Preferably, the data calculation specifically includes: S41. Model Loading: Deploying the model in the computing unit; S42. Real-time signal input: Input the extracted real-time heart rate and respiratory signals of the pet into the model; S43. Frequency Prediction: The model outputs predicted values ​​of heart rate and respiratory rate in real time.

[0043] The model was compared and validated with standard medical monitoring equipment (such as veterinary diagnostic equipment). When the accuracy of the model's predictions reached the clinically practical standard (error less than ±3 bpm), the model training was deemed effective.

[0044] The pet vital signs measurement method based on millimeter-wave radar of this invention has the following advantages. 1) This invention uses millimeter-wave radar to acquire signals and combines it with a dual-polarization differential long-hair suppression algorithm and dual-channel differential motion cancellation to remove environmental noise and motion artifacts from the signal. This overcomes the problems of low measurement accuracy caused by long fur occlusion and high-frequency physiological rhythms in pets, enabling real-time and accurate measurement of pet heart rate and respiratory rate within a distance of 0.3-2m. This improves the adaptability and measurement accuracy for long-haired pets. For example, in tests on long-haired breeds such as Maine Coons and Afghan Hounds with hair length of 5-15mm, the signal-to-noise ratio is improved by 8dB, and the reliability is improved by 70% compared to optical methods. Measurement accuracy: Compared with ECG / breathing bag benchmarks, the average error of heart rate is ≤±3bpm (cats) and ±2bpm (dogs); the error of respiratory rate is ≤±2rpm (cats and dogs).

[0045] 2) Adaptive gating and long-haired target recognition: During the acquisition process, the echo segment of the target area is dynamically locked according to the pet's body shape. That is, the measurement requirements of pets with different body shapes and fur are met by dynamically adjusting parameters. At the same time, it also improves posture and multi-scene robustness. The target can be re-locked within 2 seconds for postures such as lying down, lying on the side, standing, and licking fur. 3) Enhanced safety and comfort: The millimeter-wave radar is completely non-contact with pets and has a low-power transmission of <10dBm, meeting the requirements of FCC / CE SAR.

[0046] 4) Multi-stage high-frequency signal separation: For example, actual tests show that even under high coupling conditions of a pet (cat) heart rate of 200 bpm and respiratory rate of 30 rpm, heart rate and respiratory rate data can still be accurately distinguished; like Figure 2As shown, this embodiment also provides a pet vital signs measurement device based on millimeter-wave radar, which can implement the aforementioned pet vital signs measurement method based on millimeter-wave radar. The device includes: Data acquisition module: Used to acquire radar data of the target area of ​​the pet using millimeter-wave radar. During the acquisition process, the echo segment of the target area is dynamically locked according to the size of the pet, and the target area is at a certain distance from the millimeter-wave radar. Data processing unit: Used to eliminate environmental noise and motion artifacts in radar data using a bandpass filter, retaining the effective signal in the 0.2-5Hz range; then, the radar data is divided into sliding windows; Data extraction unit: used to extract heart rate and respiratory signals from the segmented radar data; Data calculation unit: used to input the extracted heart rate and respiratory signals into the trained calculation model and output the predicted values ​​of heart rate and respiratory rate in real time.

[0047] Preferably, the millimeter-wave radar used in the above data acquisition module is the IWR6843 radar chip, and the radar antenna is oriented towards the pet's carotid artery area or chest. In practical applications, other models of millimeter-wave radar can also be selected. This invention takes the IWR6843 radar chip as an example.

[0048] The IWR6843 is a single-chip smart millimeter-wave sensor from 60GHz to 64GHz with integrated processing capabilities, launched by Texas Instruments (TI). It features high integration, wide frequency range, beamforming capability, and powerful processing capabilities, and is used in industrial inspection, security monitoring, intelligent transportation, healthcare and other scenarios.

[0049] like Figure 2 As shown, the present invention provides a pet collar, including a control unit and a communication unit, a power management unit, and the aforementioned measuring device connected thereto.

[0050] Preferably, the control unit integrates a microcontroller or edge computing unit, and the communication unit adopts a Bluetooth, WiFi module and / or cellular module. The communication unit can upload the pet's heart rate and respiratory rate data monitored by the measuring device to the cloud, user terminal or other related smart devices in real time. The control unit can control the transmission rhythm of the communication unit and push it in time when there is an abnormal heart rate or respiratory arrest. The controller or edge computing unit can be used for the model loading and calculation process of the measuring device. The pet collar also includes a six-degree-of-freedom inertial measurement unit (6-DoF IMU) and a temperature and humidity sensor connected to the control unit. When calculating the predicted values ​​of heart rate and respiratory rate, the pet's movement data measured by the six-DoF IMU and the environmental temperature and humidity data detected by the temperature and humidity sensor can be used as posture and environmental compensation to more accurately calculate the required predicted values.

[0051] In the description of this specification, references to terms such as "an embodiment," "some embodiments," "example," "specific example," or "optional embodiment," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0052] The embodiments described above do not constitute a limitation on the scope of protection of this technical solution. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the above embodiments should be included within the scope of protection of this technical solution.

Claims

1. A method for measuring pet vital signs based on millimeter-wave radar, characterized in that, include: S1. Data Acquisition: Millimeter-wave radar is used to acquire radar data of the target area of ​​the pet. During the acquisition process, the echo segment of the target area is dynamically locked according to the size of the pet, and the target area is at a certain distance from the millimeter-wave radar. S2. Data preprocessing: A bandpass filter is used to eliminate environmental noise and motion artifacts in the radar data, retaining the effective signal in the range of 0.2-5Hz; then, the radar data is divided into sliding windows. S3. Data Extraction: Extract heart rate and respiratory signals from the segmented radar data; S4. Data Calculation: Input the extracted heart rate and respiratory signals into the trained calculation model, and output the predicted values ​​of heart rate and respiratory rate in real time.

2. The pet vital signs measurement method based on millimeter-wave radar as described in claim 1, characterized in that... The data collection process is as follows: S11. The millimeter-wave radar should be directed towards the target area of ​​the pet's carotid artery or chest, and the distance between the radar antenna and the target area should be controlled between 0.3-2m. S12: Millimeter-wave radar transmits continuous wave or frequency-modulated continuous wave signals and collects reflected echo data; S13. The reflected echo data is converted into a time-domain data stream after analog-to-digital conversion.

3. The pet vital signs measurement method based on millimeter-wave radar as described in claim 1, characterized in that, The bandpass filter used is a Butterworth bandpass filter, and the data preprocessing specifically includes: S21. Polarization Differential Hair Suppression: The vertical or horizontal polarization echo difference method is used to remove the hair scattering baseline in radar data. S22. Dual-channel differential motion elimination: The method of synthesizing a reference channel by combining dual-channel differential and virtual phase center is used to eliminate large-amplitude limb movements; S23. Set a fixed window and use a sliding window to divide the data.

4. The pet vital signs measurement method based on millimeter-wave radar as described in claim 1, characterized in that, Data extraction specifically includes: S31. Heart rate signal extraction: The heart rate signal is extracted from the segmented radar data using continuous wavelet transform analysis. S32. Respiratory signal extraction: The respiratory signal is extracted from the segmented radar data using fast Fourier transform spectral analysis.

5. The pet vital signs measurement method based on millimeter-wave radar as described in claim 1, characterized in that, The measurement model adopts a CNN-based neural network model, the structure of which includes the following in sequence: Input layer: Used to input preprocessed and extracted timing signals; Hidden layer: used for radar signal analysis; Output layer: A fully connected layer is used to output predicted values ​​of heart rate and respiratory rate; Loss Functions and Optimizers: Loss function: Mean squared error; Optimizer: Adam optimizer is used with a learning rate of 0.001; The training process of the measurement model includes: Data labeling: Tag data for pet heart rate and respiratory rate were collected using standard heart rate and respiratory rate monitoring equipment; Dataset partitioning: 70% of the labeled data is divided into a training set, 15% into a validation set, and 15% into a test set; Model training: Set batch size=64, epochs=100, input the training set into the model for training, and use an early stopping mechanism during training to avoid overfitting.

6. The method for measuring pet vital signs based on millimeter-wave radar as described in claim 1, characterized in that, The data calculation specifically includes: S41. Model Loading: Deploy the trained model to the computing unit; S42. Real-time signal input: Input the extracted real-time heart rate and respiratory signals of the pet into the model; S43. Frequency Prediction: The model outputs predicted values ​​of heart rate and respiratory rate in real time.

7. A pet vital signs measurement device based on millimeter-wave radar, capable of implementing the pet vital signs measurement method based on millimeter-wave radar as described in any one of claims 1-6, characterized in that... The measuring device includes: Data acquisition module: Used to acquire radar data of the target area of ​​the pet using millimeter-wave radar. During the acquisition process, the echo segment of the target area is dynamically locked according to the size of the pet, and the target area is at a certain distance from the millimeter-wave radar. Data processing unit: Used to eliminate environmental noise and motion artifacts in radar data using a bandpass filter, retaining the effective signal in the 0.2-5Hz range; then, the radar data is divided into sliding windows; Data extraction unit: used to extract heart rate and respiratory signals from the segmented radar data; Data calculation unit: used to input the extracted heart rate and respiratory signals into the trained calculation model and output the predicted values ​​of heart rate and respiratory rate in real time.

8. The pet vital signs measurement device based on millimeter-wave radar as described in claim 7, characterized in that, The data acquisition module uses the IWR6843 millimeter-wave radar chip, which operates in the 60-64GHz frequency band, and the radar antenna is directed towards the pet's carotid artery area or chest.

9. A pet collar, characterized in that: It includes a control unit and its connected communication unit, power management unit, and the measuring device as described in claim 7 or 8.

10. The pet collar as described in claim 9, characterized in that: The control unit integrates a microcontroller or edge computing unit, and the communication unit adopts Bluetooth, WiFi module and / or cellular module; The pet collar also includes a six-degree-of-freedom inertial measurement unit and a temperature and humidity sensor connected to the control unit.

Images