CN-122004150-A - Intelligent pet collar based on unconstrained inertial sensing and physiological signal processing method thereof

CN122004150ACN 122004150 ACN122004150 ACN 122004150ACN-122004150-A

Abstract

The invention provides an intelligent pet collar based on unconstrained inertial sensing and a physiological signal processing method thereof, which comprise the steps of being worn on the neck of a pet in an unconstrained mode, allowing a gap of 0-30 mm to change between the collar and skin, and allowing the collar to rotate randomly relative to the body. The method comprises the steps of acquiring multi-axis acceleration data, converting the multi-axis acceleration data into scalar signals irrelevant to wearing directions through space irrelevant fusion, performing self-adaptive frequency domain adjustment according to physiological characteristics of a target pet to restrain motion noise and respiratory interference, separating cardiovascular micro-vibration components from mixed signals based on signal statistics independence characteristics through a blind source separation algorithm, and finally extracting cardiovascular activity parameters including heart rate, heart rate variability, heart beat interval sequences and the like. The invention only depends on a general inertial sensor and an advanced signal processing algorithm, and realizes cardiovascular micro-vibration signal extraction and health monitoring with high robustness under the conditions of random wearing posture and uncertain direction.

Inventors

LI YUNLONG
LIN ZIJIAN
LV GUOHAO
XIE YUAN
LIANG ZHIHENG
YU YUAN

Assignees

李云龙

Dates

Publication Date: 20260512
Application Date: 20260203

Claims (10)

1. A smart pet collar based on unconstrained inertial sensing, comprising: the intelligent pet necklace body surrounds the neck of the pet in a non-constraint mode, and an inertia measurement unit is arranged in the intelligent pet necklace body; a signal processing unit in communication with the smart pet collar body and configured to perform: (a) Receiving multi-axis acceleration time series data from an inertia measurement unit; (b) Performing space independent fusion, namely performing direction-invariant transformation on the multi-axis acceleration data to generate a composite energy signal independent of the wearing direction of the intelligent pet collar, wherein the direction-invariant transformation comprises at least one of Euclidean norms, weighted square sums or time change rates of vector magnitudes of the multi-axis acceleration; (c) The self-adaptive frequency domain regulation comprises the steps of configuring band-pass filtering parameters according to physiological characteristics of a target pet, and inhibiting motion noise and respiratory interference of a non-cardiovascular source, wherein the physiological characteristics comprise at least one of species, body type, heart rate priori range and respiratory frequency priori range; (d) Blind source separation, namely inputting the filtered signals into a calculation reasoning model, and separating cardiovascular micro-vibration components from the mixed signals based on statistical independence characteristics or non-Gaussian characteristics of the signals; (e) And (3) parameter extraction, namely determining and outputting cardiovascular activity parameters of the pet based on the separated cardiovascular differential vibration quantity.
2. The intelligent pet collar based on unconstrained inertial sensing according to claim 1, wherein the unconstrained manner allows the distance between the intelligent pet collar body and the skin of a pet to be varied within a range of 0-30 mm, and the unconstrained manner allows the rotation angle of the intelligent pet collar body relative to the body of the pet to be in any direction.
3. The intelligent pet collar based on unconstrained inertial sensing according to claim 1, wherein the signal processing unit adopts any one of the following structural forms: an embedded processor integrated within the smart pet collar body; a mobile terminal device in wireless communication with the smart pet collar body; Cloud computing server connected through internet.
4. The intelligent pet collar based on unconstrained inertial sensing according to claim 1, wherein the computational inference model comprises at least one of an independent component analysis, a principal component analysis, a singular value decomposition, a convolutional neural network, a recurrent neural network or a long-short term memory network, and a converter architecture based on an attention mechanism.
5. The smart pet collar based on unconstrained inertial sensing of claim 1, further comprising a motion state discrimination module configured to: Judging whether the pet is in a resting state or not based on the statistical characteristics of the multi-axis acceleration data; the blind source separation step is performed only in the resting state.
6. The smart pet collar based on unconstrained inertial sensing of claim 1, wherein the method of extracting heart rate from cardiovascular differential motion components comprises: The method comprises the steps of performing autocorrelation analysis on a separation signal, taking a time delay corresponding to a first significant peak value as a heart beat period, performing FFT conversion on the separation signal, taking peak frequency in a heart beat frequency band as a heart rate, or detecting a characteristic peak value by using a peak detection algorithm, and calculating the interval between adjacent peak values.
7. A physiological signal processing method of a smart pet collar based on unconstrained inertial sensing is characterized by comprising the following steps: S1, acquiring inertial sensor data from a smart pet collar body worn on the neck of a pet in an unconstrained mode, wherein the inertial sensor data comprises a multi-axis acceleration time sequence, and the unconstrained mode allows the collar to be in a range of 0-30mm from the skin and allows the collar to rotate randomly relative to the body; S2, performing space-independent fusion operation on the multi-axis acceleration data, and generating scalar energy signals irrelevant to the wearing direction by combining multi-axis components; S3, self-adaptive band-pass filtering based on target pet physiological characteristics is applied to the scalar energy signal, and motion noise and respiratory interference are restrained; S4, inputting the filtered signals into a blind source separation model, and separating cardiovascular micro-vibration components from the mixed signals based on the statistical independence or non-Gaussian characteristic of the signals; S5, quantifying cardiovascular activity parameters based on the cardiovascular differential vibration quantity; s6, outputting the cardiovascular activity parameters.
8. The physiological signal processing method of the intelligent pet collar based on unconstrained inertial sensing of claim 7, the method is characterized in that the space-independent fusion operation comprises at least one of the following operations: calculating Euclidean norms of the triaxial acceleration; calculating the weight vector amplitude; And calculating the dynamic acceleration amplitude after the gravity component is eliminated by performing coordinate transformation.
9. The method for processing physiological signals of a smart pet collar based on unconstrained inertial sensing according to claim 8, wherein the blind source separation model is trained based on any of the following ways: Performing supervised learning by using the labeled cardiovascular signal data set; performing unsupervised learning based on signal statistics using unlabeled data; Adaptive parameter configuration is performed for different species.
10. The method for processing physiological signals of a smart pet collar based on unconstrained inertial sensing of claim 9, further comprising: performing quality assessment on the extracted cardiovascular activity parameters; And when the quality is lower than a preset threshold value, marking the time period data as low confidence.

Description

Intelligent pet collar based on unconstrained inertial sensing and physiological signal processing method thereof Technical Field The invention relates to the technical field of wearable intelligent equipment for pets, in particular to an intelligent pet collar based on unconstrained inertial sensing and a physiological signal processing method thereof. Background As pet economies develop, there is an increasing need for health monitoring of companion animals. The intelligent pet collar is used as the most common pet wearable equipment and carries multiple functions such as activity monitoring, health tracking and position positioning. The ballistocardiogram is a physiological signal reflecting the micro-vibration of the body caused by the mechanical pulsation of the heart and the blood jet flow, has the advantages of non-invasiveness and no need of electrodes, and is suitable for long-term health monitoring of animals. In the prior art, various intelligent pet necklace schemes exist, but certain technical limitations exist: In the prior art 1, a mode of arranging elastic bulges on a sensor shell is adopted, so that the bulge parts of the sensor are required to face the skin of the neck of an animal. The proposal has the problems that the elastic bulge needs to be in direct contact with the skin to conduct vibration signals, strict requirement is made on wearing tightness, the sensor is fixed in direction and can not adapt to the rotation of the necklace during animal activities, and the bulge structure can cause discomfort to the animal skin. Prior art 2 Integrated Speech recognition translation units attempt to translate animal calls into human language. The scheme has the following problems of high system complexity, high power consumption, lack of scientific basis for voice translation function, doubtful practicality, and difficulty in optimizing core functions by stacking multiple functions in a single device. In the prior art 3, a double-speed comparison scheme of matching the collar with the leg wearing assembly is adopted, and the movement state of the animal is judged by comparing the movement speed of the collar and the movement speed of the leg. Such a solution has the problems of requiring additional leg wear components, increasing system complexity and animal wear burden, increased failure points due to the cooperative operation of the two components, and inability to fit animals that are not wearing leg components. In the prior art 4, the self-adaptive adjustment of the tightness of the necklace is realized by adopting a shape memory alloy. Such solutions have the problems of high cost of shape memory alloy materials and the need for accurate temperature control, mechanical adjustment mechanisms add complexity and weight to the system, and frequent adjustments can affect animal wear comfort. Therefore, there is a need in the art for a smart pet collar and a physiological signal processing method thereof that has a simple structure, is comfortable to wear, has comprehensive functions, and does not require special sensor orientations or additional wearing components. Disclosure of Invention In order to solve at least one technical problem in the background art, the invention provides the intelligent pet collar based on the unconstrained inertial sensing and the physiological signal processing method thereof, which can realize multiple functions of cardiovascular micro-vibration signal extraction, activity monitoring, sleep detection, gesture recognition, health early warning and the like only by relying on a universal inertial sensor under the unconstrained wearing condition of random equipment gesture and uncertain direction. The invention solves the technical problems by adopting the following technical scheme: A smart pet collar based on unconstrained inertial sensing, comprising: the intelligent pet necklace body surrounds the neck of the pet in a non-constraint mode, and an inertia measurement unit is arranged in the intelligent pet necklace body; a signal processing unit in communication with the smart pet collar body and configured to perform: (a) Receiving multi-axis acceleration time series data from an inertia measurement unit; (b) Performing space independent fusion, namely performing direction-invariant transformation on the multi-axis acceleration data to generate a composite energy signal independent of the wearing direction of the intelligent pet collar, wherein the direction-invariant transformation comprises at least one of Euclidean norms, weighted square sums or time change rates of vector magnitudes of the multi-axis acceleration; (c) The self-adaptive frequency domain regulation comprises the steps of configuring band-pass filtering parameters according to physiological characteristics of a target pet, and inhibiting motion noise and respiratory interference of a non-cardiovascular source, wherein the physiological characteristics comprise at least one of species, body type,