A method and system for identifying the health status of newborn lambs based on flexible piezoelectric sensors

Abstract

The present invention belongs to the field of flexible sensing and biosignal processing technology and provides a method and system for identifying the health status of newborn lambs based on flexible piezoelectric sensors. The method comprises: acquiring a subset of newborn lamb biometric features and identifying a final newborn lamb status identification model. The final newborn lamb status identification model is constructed through the following steps: preparing the flexible piezoelectric sensor, acquiring and extracting superimposed biosignals, determining a health index calculation formula and health index intervals, adjusting a first variable parameter, optimizing the input matrix, adjusting a second variable parameter, and selecting the model. The present invention utilizes flexible piezoelectric sensors to non-invasively acquire newborn lamb biosignals, reducing the difficulty of data collection. By constructing the final newborn lamb status identification model, the method accurately identifies the health status of unknown newborn lambs, contributing to improved animal welfare and farm economic efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of flexible sensing and biological signal processing, and in particular to a method and system for identifying the health status of newborn lambs based on a flexible piezoelectric sensor. Background Art

[0002] The health of newborn lambs significantly impacts their growth, development, immunity, and subsequent production performance, significantly impacting farming efficiency and economic returns. Common biological signals such as heartbeat, respiration, vocalizations, and behavioral signals are key indicators of health. Poor health can lead to decreased immunity, slowed growth, and even death. Therefore, real-time health monitoring is crucial for improving animal welfare, ensuring food safety, and increasing production efficiency.

[0003] Currently, monitoring the health of newborn lambs relies primarily on two methods: manual observation and biological indicator measurement. Manual observation is highly subjective, easily influenced by the observer's experience and environmental factors, and has low accuracy. Biological indicator measurement requires the collection of samples such as blood and saliva, which is complex and invasive, making it unsuitable for long-term monitoring. Furthermore, each indicator must be collected and tested separately. Summary of the Invention

[0004] In order to overcome the shortcomings of the existing technology, the purpose of the present invention is to provide a method and system for identifying the health status of newborn lambs based on flexible piezoelectric sensors. Through the flexible piezoelectric sensors, non-invasive acquisition of newborn lamb biological signals can be achieved, reducing the difficulty of data collection; by constructing the final newborn lamb status recognition model, accurate identification of the health status of unknown newborn lambs can be achieved, thereby improving animal welfare and farm economic benefits.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for identifying the health status of a newborn lamb based on a flexible piezoelectric sensor, comprising:

[0007] Obtaining a subset of biological characteristics of a newborn lamb of unknown health status; the health status includes: normal state, listless state, stress state, suspended animation state, and trembling state;

[0008] Inputting the newborn lamb biological feature subset into the final newborn lamb state recognition model to perform state recognition to obtain a target state recognition result;

[0009] The process of constructing the final newborn lamb status recognition model includes:

[0010] A flexible piezoelectric sensor was prepared; the flexible piezoelectric sensor comprised a sandwich-structured substrate layer, a sensitive layer, and an electrode layer; the electrode layer was used to transmit vibration signals from the surface of a target newborn lamb; the sensitive layer was used to convert the vibration signals into voltage signals; the substrate layer and the electrode layer were prepared using a one-step molding process combining a Posilicone flexible substrate and laser-induced graphene electrodes; the sensitive layer was composed of a 1 mm thick foam-structured composite piezoelectric material;

[0011] Acquiring superimposed biological signals using the flexible piezoelectric sensor, and performing signal feature extraction on the superimposed biological signals to obtain an eigenvalue input matrix;

[0012] Determining a newborn lamb health index calculation formula and a health index division interval, and establishing a mapping relationship between a calculation result of the newborn lamb health index calculation formula and the health status according to the health index division interval, to obtain an original newborn lamb status recognition model;

[0013] Inputting the eigenvalue input matrix into the original newborn lamb state recognition model to adjust variable parameters to obtain a first optimized recognition model;

[0014] Optimizing the eigenvalue input matrix using a correlation index to obtain a first optimized input matrix;

[0015] Inputting the first optimization input matrix into the first optimization recognition model to adjust variable parameters to obtain a second optimization recognition model;

[0016] The recognition accuracy and recognition speed of the first optimized recognition model and the second optimized recognition model are extracted to obtain model operation data, and the first optimized recognition model and the second optimized recognition model are screened according to the model operation data to obtain the final newborn lamb state recognition model.

[0017] Preferably, the preparation process of the electrode layer includes:

[0018] A 5 cm x 5 cm glass slide was cleaned with deionized water, 5 mL of Posilicone silica gel was placed on the glass slide, and the glass slide was placed in a spin coater and spin-coated at a speed of 1000 rpm. A 5 cm x 5 cm PI film was attached to the surface of the glass slide coated with Posilicone silica gel and heated at 80°C for 10 minutes to obtain a substrate; the thickness of the PI film was 0.075 mm.

[0019] Laser-induced graphene operation was performed on the substrate according to a preset electrode pattern using a CO2 near-infrared laser machine, 15 mL of Posilicone silica gel was placed on the PI film, the glass sheet was placed in a spin coater and spin-coated at a speed of 800 r / min for 20 s, the glass sheet was placed in a vacuum defoaming machine for defoaming treatment for 10 min, the glass sheet was placed on a heating table and heated for 1 h for curing, and the PI film coated with Posilicone silica gel was kneaded for 10 min to obtain the electrode layer.

[0020] Preferably, the preparation process of the sensitive layer includes:

[0021] Posilicone silica gel, polyvinylidene fluoride and zinc oxide powder are mixed to obtain a mixed silica gel; the mass fraction of polyvinylidene fluoride and the mass fraction of zinc oxide powder in the mixed silica gel are both 10%;

[0022] The prepared foam is immersed in the mixed silica gel and heated and cured at 80° C. for 10 minutes to obtain the sensitive layer; the diameter of the prepared foam is 10 mm; and the thickness of the prepared foam is less than 1 mm.

[0023] Preferably, the flexible piezoelectric sensor is used to obtain superimposed biological signals, and signal features are extracted from the superimposed biological signals to obtain an eigenvalue input matrix, including:

[0024] Performing signal deconstruction and entropy calculation on the superimposed biological signal to obtain intrinsic piezoelectric component and total energy data;

[0025] Dividing the total energy data into heartbeat frequency band energy and respiratory frequency band energy according to a preset frequency interval;

[0026] An index calculation is performed based on the heartbeat frequency band energy and the respiratory frequency band energy to obtain a mode index; the calculation formula of the mode index is: ;in, 、 、 as well as are the first index, the second index, the third index, and the fourth index respectively; The energy in the heart rate band from 1.0Hz to 3.0Hz; is the total energy of the intrinsic piezoelectric component; The energy in the respiratory frequency band from 0.1Hz to 0.5Hz; Calculate the total energy in the frequency domain for the target; is the threshold; is the piezoelectric reconstruction quantity;

[0027] Reconstructing and filtering the intrinsic piezoelectric component using the pattern index to obtain filtered data;

[0028] The time domain features, frequency domain features and disturbance features of the filtered data are extracted to obtain the eigenvalue input matrix; the calculation formula of the disturbance feature is: ;in, ; is the disturbance characteristic; is the signal voltage value at the i-th moment; is the average disturbance; is the basic disturbance; is the piezoelectric value at the i-th moment; is the average piezoelectric value within 180s.

[0029] Preferably, the newborn lamb health index calculation formula is:

[0030]

[0031] in, The calculation result of the newborn lamb health index calculation formula; The i-th eigenvalue of the eigenvalue input matrix; is the i-th index parameter; is the i-th weight coefficient; the number of eigenvalues ​​of the eigenvalue input matrix; is the j-th shape parameter; is the jth core parameter; is the bias term; is the random error term.

[0032] Preferably, the parameter update formula of the first optimized recognition model includes:

[0033]

[0034]

[0035] in, ; is the updated i-th weight coefficient; is the correction constant; is the updated i-th index parameter; is the sample size; The first sample eigenvalues; The health index of the samples; The average health index.

[0036] Preferably, the calculation formula of the correlation index is:

[0037]

[0038] in, is the correlation index; For all samples The mean of the features; is the mean of the health index of all samples; All are adjustment parameters; For the Weight coefficients; for Second The health index of the samples; for Second The health index of the samples; for Second The first sample eigenvalues; for Second The first sample eigenvalues; is the adjustment factor.

[0039] Preferably, the eigenvalue input matrix is ​​optimized using a correlation index to obtain a first optimized input matrix, including:

[0040] Calculating the correlation indices of all the eigenvalue input matrices to obtain an original index set;

[0041] Arrange the original index set in descending order to obtain a sequential index set;

[0042] Inputting the hth indicator in the sequential indicator set into the first optimized recognition model, and calculating the accuracy of the first optimized recognition model to obtain the hth model accuracy; h is any indicator in the sequential indicator set except the last one;

[0043] Inputting the hth indicator and the h+1th indicator in the sequential indicator set into the first optimization recognition model, and calculating the accuracy of the first optimization recognition model to obtain the h+1th model accuracy;

[0044] When the hth model accuracy is less than the h+1th model accuracy, the h+1th indicator is retained; otherwise, the h+1th indicator is eliminated;

[0045] The sequential indicator set is updated each time an indicator is eliminated, and the sequential indicator set is iterated to obtain the first optimized input matrix after iteration.

[0046] Preferably, a newborn lamb health status identification system based on a flexible piezoelectric sensor comprises: a flexible piezoelectric sensing unit, a flexible circuit unit, a data analysis unit, and an elastic band;

[0047] The flexible piezoelectric sensing unit and the flexible circuit unit are fixed on the surface of the target newborn lamb through the elastic band; the data analysis unit is connected to the flexible circuit unit; the flexible piezoelectric sensing unit is connected to the flexible circuit unit;

[0048] The flexible piezoelectric sensing unit is used to collect superimposed biological signals; the flexible circuit unit is used to amplify, store and transmit the superimposed biological signals; and the data analysis unit is used to identify the health status of the target newborn lamb.

[0049] The present invention discloses the following technical effects:

[0050] The present invention provides a method and system for identifying the health status of newborn lambs based on flexible piezoelectric sensors. Through the flexible piezoelectric sensor, the method solves the defects of the conventional technology of complex and invasive operation when collecting biological signals of newborn lambs, and realizes the function of directly obtaining biological signals from the surface of newborn lambs; by constructing the final newborn lamb status recognition model, the method solves the disadvantage of low accuracy caused by relying on manual judgment of health status in conventional methods, and realizes accurate recognition of the health status of unknown newborn lambs. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 A schematic diagram of a process for identifying the health status of a newborn lamb provided by an embodiment of the present invention;

[0053] Figure 2 A schematic diagram of the process for preparing a flexible piezoelectric sensor according to an embodiment of the present invention;

[0054] Figure 3 A flow chart for preparing a flexible piezoelectric sensor according to an embodiment of the present invention;

[0055] Figure 4 This is an effect diagram of the electrode layer provided by an embodiment of the present invention;

[0056] Figure 5 An overall effect diagram provided for an embodiment of the present invention;

[0057] Description of reference numerals:

[0058] 201 - base layer, 202 - electrode layer, 203 - sensitive layer, 401 - signal receiving domain, 402 - signal transmission domain. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0060] The purpose of the present invention is to provide a method and system for identifying the health status of newborn lambs based on flexible piezoelectric sensors. Through the flexible piezoelectric sensors, non-invasive acquisition of biological signals of newborn lambs can be achieved, reducing the difficulty of data collection; by constructing a final newborn lamb status recognition model, accurate identification of the health status of unknown newborn lambs can be achieved, thereby improving animal welfare and farm economic benefits.

[0061] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0062] Figure 1 A schematic diagram of a process for identifying the health status of newborn lambs provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the present invention provides a method for identifying the health status of a newborn lamb based on a flexible piezoelectric sensor, comprising:

[0063] Step 10: Obtain a subset of biological characteristics of newborn lambs of unknown health status; health status includes: normal state, listless state, stress state, suspended animation state, and trembling state;

[0064] Step 20: Inputting the newborn lamb biometric subset into the final newborn lamb state recognition model to perform state recognition and obtain a target state recognition result;

[0065] The final process of constructing the newborn lamb status recognition model includes:

[0066] Step 21: Prepare a flexible piezoelectric sensor. The flexible piezoelectric sensor includes a sandwich-structured substrate layer 201, a sensitive layer 203, and an electrode layer 202. The electrode layer 202 is used to transmit vibration signals from the surface of the target newborn lamb. The sensitive layer 203 is used to convert the vibration signals into voltage signals. The substrate layer 201 and the electrode layer 202 are prepared using a one-step molding process using a Posilicone flexible substrate and laser-induced graphene electrodes. The sensitive layer 203 is composed of a 1 mm thick foam structure composite piezoelectric material.

[0067] Step 22: Acquire the superimposed biological signal using the flexible piezoelectric sensor, and perform signal feature extraction on the superimposed biological signal to obtain an eigenvalue input matrix;

[0068] Step 23: determining a newborn lamb health index calculation formula and health index division intervals, and establishing a mapping relationship between the calculation result of the newborn lamb health index calculation formula and the health status according to the health index division intervals, to obtain an original newborn lamb status recognition model;

[0069] Step 24: inputting the eigenvalue input matrix into the original newborn lamb state recognition model to adjust the variable parameters to obtain a first optimized recognition model;

[0070] Step 25: Optimize the eigenvalue input matrix using the correlation index to obtain a first optimized input matrix;

[0071] Step 26: Input the first optimized input matrix into the first optimized recognition model to adjust the variable parameters, thereby obtaining a second optimized recognition model;

[0072] Step 27: Extract the recognition accuracy and recognition speed of the first optimized recognition model and the second optimized recognition model to obtain model operation data, and screen the first optimized recognition model and the second optimized recognition model according to the model operation data to obtain the final newborn lamb state recognition model.

[0073] Further explanation of the health status of newborn lambs: the normal state is characterized by full vitality, the ability to stand up and find the ewe quickly, suckling normally, normal body temperature and heart rate, good physical condition, dry and shiny hair, and piezoelectric signal fluctuations within ±30mV; the listless state shows obvious weakness and drowsiness, poor activity, usually unable to stand up or find the ewe immediately, which may be accompanied by a decrease in body temperature, weak sucking response, and piezoelectric signal fluctuations within ±15mV; the stress state is usually caused by severe stress (such as excessive feeding, abnormal delivery, etc.) and manifests as restlessness, rapid breathing, and accelerated heart rate, with piezoelectric signal fluctuations greater than ±30mV and less than ±50mV; lambs in the suspended animation state appear to be dead, with low body temperature, weak heartbeat, and lack of response, and the piezoelectric signal fluctuates within ±10mV; the trembling state is usually characterized by whole body trembling, which may be caused by factors such as cold, hypothermia or fright, and may struggle or shout violently, with the piezoelectric signal fluctuating beyond ±50mV.

[0074] refer to Figure 2 and Figure 3 The preparation process of the electrode layer 202 includes:

[0075] Clean a 5cm*5cm glass slide with deionized water, place 5mL of Posilicone on the slide, and spin coat the slide at 1000 rpm in a spin coater. Attach a 0.075mm thick 5cm*5cm PI film to the surface of the Posilicone-coated glass slide and heat at 80°C for 10 minutes to obtain a substrate.

[0076] A CO2 near-infrared laser machine was used to perform laser-induced graphene operation on the substrate according to a preset electrode pattern. 15 mL of Posilicone silica gel was placed on the PI film. The glass sheet was placed in a spin coater and spin-coated at a speed of 800 r / min for 20 seconds. The glass sheet was placed in a vacuum defoaming machine for defoaming treatment for 10 minutes. The glass sheet was placed on a heating table and heated for 1 hour for curing. The PI film coated with Posilicone silica gel was kneaded for 10 minutes to obtain an electrode layer 202.

[0077] refer to Figure 2 and Figure 3 The preparation process of the sensitive layer 203 includes:

[0078] Posilicone silica gel, polyvinylidene fluoride and zinc oxide powder are mixed to obtain a mixed silica gel; the mass fraction of polyvinylidene fluoride and the mass fraction of zinc oxide powder in the mixed silica gel are both 10%;

[0079] A foam with a diameter of 10 mm and a thickness of less than 1 mm is immersed in mixed silica gel and heated and cured at 80° C. for 10 minutes to obtain a sensitive layer 203 .

[0080] Furthermore, the flexible piezoelectric sensor is used to obtain the superimposed biological signal, and the signal feature extraction of the superimposed biological signal is performed to obtain the eigenvalue input matrix, including:

[0081] Perform signal deconstruction and entropy calculation on the superimposed biological signals to obtain intrinsic piezoelectric components and total energy data;

[0082] The total energy data is divided into heartbeat frequency band energy and respiratory frequency band energy according to the preset frequency interval;

[0083] The index is calculated based on the energy of the heartbeat frequency band and the energy of the respiratory frequency band to obtain the mode index. The calculation formula of the mode index is: ;in, 、 、 as well as are the first index, the second index, the third index, and the fourth index respectively; The energy in the heart rate band from 1.0Hz to 3.0Hz; is the total energy of the intrinsic piezoelectric component; The energy in the respiratory frequency band from 0.1Hz to 0.5Hz; Calculate the total energy in the frequency domain for the target; is the threshold; is the piezoelectric reconstruction quantity;

[0084] Reconstructing and filtering the intrinsic piezoelectric component using the pattern index to obtain filtered data;

[0085] The time domain features, frequency domain features and disturbance features of the filtered data are extracted to obtain the eigenvalue input matrix; the calculation formula of the disturbance feature is: ;in, ; is the disturbance characteristic; is the signal voltage value at the i-th moment; is the average disturbance; is the basic disturbance; is the piezoelectric value at the i-th moment; is the average piezoelectric value within 180s.

[0086] Specifically, the formula for calculating the health index of newborn lambs is:

[0087]

[0088] in, The calculation result of the formula for calculating the health index of newborn lambs; The i-th eigenvalue of the eigenvalue input matrix; is the i-th index parameter; is the i-th weight coefficient; The number of eigenvalues ​​of the matrix for Eigenvalue input; is the j-th shape parameter; is the jth core parameter; is the bias term; is the random error term.

[0089] Furthermore, the parameter update formula of the first optimized recognition model includes:

[0090]

[0091]

[0092] in, ; is the updated i-th weight coefficient; is the correction constant; is the updated i-th index parameter; is the sample size; The first sample eigenvalues; The health index of the samples; The average health index.

[0093] Specifically, the calculation formula of the correlation index is:

[0094]

[0095] in, is the correlation index; For all samples The mean of the features; is the mean of the health index of all samples; All are adjustment parameters; For the Weight coefficients; for Second The health index of the samples; for Second The health index of the samples; for Second The first sample eigenvalues; for Second The first sample eigenvalues; is the adjustment factor.

[0096] Furthermore, the eigenvalue input matrix is ​​optimized using the correlation index to obtain a first optimized input matrix, including:

[0097] Calculating the correlation indices of all the eigenvalue input matrices to obtain an original index set;

[0098] Arrange the original indicator set in descending order to obtain a sequential indicator set;

[0099] Inputting the hth indicator in the sequential indicator set into the first optimized recognition model, and calculating the accuracy of the first optimized recognition model to obtain the hth model accuracy; h is any indicator in the sequential indicator set except the last one;

[0100] Inputting the hth indicator and the h+1th indicator in the sequential indicator set into the first optimized recognition model, and calculating the accuracy of the first optimized recognition model to obtain the h+1th model accuracy;

[0101] When the hth model accuracy is less than the h+1th model accuracy, the h+1th indicator is retained; otherwise, the h+1th indicator is eliminated;

[0102] The sequential indicator set is updated each time an indicator is eliminated, and the sequential indicator set is iterated to obtain the first optimized input matrix after iteration.

[0103] Specifically, a newborn lamb health status identification system based on a flexible piezoelectric sensor includes: a flexible piezoelectric sensing unit, a flexible circuit unit, a data analysis unit, and an elastic band;

[0104] The flexible piezoelectric sensing unit and the flexible circuit unit are fixed on the surface of the target newborn lamb through an elastic band; the data analysis unit is connected to the flexible circuit unit; and the flexible piezoelectric sensing unit is connected to the flexible circuit unit.

[0105] The flexible piezoelectric sensing unit is used to collect superimposed biological signals; the flexible circuit unit is used to amplify, store and transmit the superimposed biological signals; and the data analysis unit is used to identify the health status of the target newborn lamb.

[0106] Preferably, a flexible piezoelectric sensor comprises a substrate layer 201, an electrode layer 202, and a sensitive layer 203. The substrate layer 201 and electrode layer 202 of the flexible piezoelectric sensor are fabricated using a one-step molding process of a Posilicone flexible substrate and laser-induced graphene electrodes. The sensitive layer 203 of the flexible piezoelectric sensor is composed of a loose, porous foam-structured composite piezoelectric material. The structure is a sandwich, with the sensitive layer 203 sandwiched between the electrode layer 202 and the substrate layer 201. The sensitive layer 203 and electrode layer 202 should be in close contact to ensure that mechanical vibrations can be effectively transmitted and captured by the electrodes.

[0107] refer to Figure 3 Posilicone silicone is a synthetic polymer material composed of epoxy resin (Component A) and hardener (Component B). It offers advantages such as low bubble formation, non-stickiness, and corrosion resistance. Due to its excellent biocompatibility, it was selected as the base material for the preparation of a flexible piezoelectric sensor used to detect and identify the health status of newborn lambs. Posilicone silicone is prepared by mixing epoxy resin (Component A) and hardener (Component B) in a 1:1 mass ratio using a stirring rod for 10 minutes. The mixture is then set aside for subsequent preparation of the flexible substrate (base layer 201) and sensitive layer 203.

[0108] refer to Figure 3 The one-step formation of the Posilicone flexible substrate and the laser-induced graphene electrode is as follows: a 5cm*5cm glass sheet is cleaned with deionized water, 1mL of pre-prepared 30# hardness Posilicone silica gel is placed on the spin coater and spin-coated at a speed of 1000r / min. Then, a 5cm*5cm PI film pre-cut with 0.075mm is attached to the surface of the glass sheet coated with Posilicone silica gel, and heated at 80°C for 10 minutes to completely adhere the PI film to the surface of the glass sheet. Figure 4 As shown, the PI film attached to the glass sheet was placed in a CO2 near-infrared laser machine for laser-induced graphene operation. The electrode pattern pre-drawn in CAD (Autodesk Computer Aided Design) was imported into the dedicated software supporting the machine. The parameters were adjusted to a speed of 120 mm / min, a power of 12%, a scanning interval of 0.02 mm, and the height of the laser emitter was adjusted to 7 mm. Figure 4As shown, a glass sheet is designed with four electrode patterns, and the electrode pattern includes a signal receiving domain 401 and a signal transmission domain 402. PI film is a polyimide film that can be converted into graphene material under high-temperature activation of CO2. Because it has good conductivity and the prepared material is uniform, it is selected as the raw material for preparing the electrode layer 202. Place 1mL of pre-prepared 30-hardness Posilicone silica gel and put it into a spin coater and spin-coat it at a speed of 1000r / min. The role of Posilicone silica gel in this step is to act as an adhesive for the subsequent attachment of the PI film to the glass sheet and preparation for the adhesion of the two; heat it at 80°C for 10 minutes to make the PI film completely attached to the surface of the glass sheet. This process is to prepare for the use of glass sheets to accelerate heat dissipation in the subsequent laser-induced graphene process to prevent the PI film from curling.

[0109] refer to Figure 4 The signal receiving domain 401 is used to receive the voltage signal of the sensitive layer 203, and the positive and negative charges generated on the upper and lower surfaces of the sensitive layer 203 when receiving the biological signal of the newborn lamb; the signal transmission domain 402 is used to transmit the signal of the signal receiving domain, and plays the role of connecting the signal receiving domain and the flexible circuit unit.

[0110] refer to Figure 3 , the one-time molding of the Posilicone flexible substrate and the laser-induced graphene electrode, the laser-induced PI film is placed in a spin coater, 15mL of pre-prepared Posilicone silica gel is added and spin-coated at 800r / min for 20s, the laser-induced graphene PI film with the surface spin-coated with Posilicone silica gel is placed in a vacuum defoamer for defoaming treatment for 10 minutes, heated on a heating table for 1 hour for curing, the laser-induced graphene PI film with the surface spin-coated with Posilicone silica gel is separated from the glass sheet, and the laser-induced graphene PI film coated with Posilicone silica gel is repeatedly rubbed for 10 minutes, and finally separated to complete the transfer operation, realizing the one-time molding of the flexible substrate and the electrode layer 202. The obtained transferred Posilicone silica gel surface contains four electrodes, which are cut into four 2.5cm*2.5cm electrodes using a CO2 near-infrared laser with a power parameter of 90% and a speed of 80mm / min.

[0111] Specifically, the laser-induced graphene PI film coated with Posilicone silicone is rubbed repeatedly for 10 minutes. The operation is to pinch the PI film coated with Posilicone silicone with the thumb and index finger, and rub it repeatedly with both hands. Pay special attention to repeatedly rubbing the edges and corners of the laser-induced graphene pattern. This is the key in the one-time molding technology and directly affects the one-time molding of the substrate and the electrode layer 202.

[0112] Furthermore, the sensitive layer 203 of the flexible piezoelectric sensor is composed of a loose, porous foam-structured composite piezoelectric material. Posilicone silica gel is mixed with PVDF and ZnO to create a mixed silica gel with a mass fraction of 10% PVDF and 10% ZnO. A pre-cut, 10mm-diameter ultrathin foam is placed in the mixed silica gel solution to fully immerse it. The mixture is then heated and cured on a heating plate at 80°C for 30 minutes to form the sensitive layer 203. PVDF and ZnO, respectively, are organic polymer piezoelectric materials, while ZnO is an inorganic piezoelectric material.

[0113] Specifically, the loose and porous ultra-thin foam is a polyurethane foam with a thickness of 1 mm, which is cut into polyurethane foam discs with a diameter of 1 cm; the full impregnation process is to first spread a mixed silica gel with a mass fraction of 10% PVDF and 10% ZnO on the bottom of a small plastic cup, and then put the cut polyurethane foam disc into the small plastic cup, and then pour the mixed silica gel into it for the second time. The purpose of this step is to allow the polyurethane foam disc to fully absorb the mixed silica gel solution.

[0114] Furthermore, the flexible substrate and the electrode layer 202 are formed in one piece, and the sensitive layer 203 and the flexible substrate are made of the same Posilicone silicone material, so the three components of the sensor can be adhered together without glue or packaging. Figure 5 .

[0115] Specifically, a method for identifying the health status of newborn lambs based on flexible piezoelectric sensors has the following process:

[0116] S1: Acquisition and processing of biological signals of newborn lambs: Acquisition of biological signals of newborn lambs, processing of superimposed biological signals, extraction of signal features, and construction of input matrix of characteristic values ​​of biological signals of newborn lambs;

[0117] S2: Preset newborn lamb health status identification model: propose a calculation method for the newborn lamb health index Y based on a mathematical model, divide the health index Y into different intervals, establish a mapping relationship between the health index Y and the health status, and preset the newborn lamb health status identification model;

[0118] S3: Determination of the newborn lamb health status identification model: Input the newborn lamb biological signal characteristic value input matrix into the preset newborn lamb health status identification model, adjust the variable parameters; optimize the characteristic input matrix to find the newborn lamb key biological characteristic subset , continue to adjust the variable parameters; compare the model recognition accuracy and recognition speed before and after the feature input matrix optimization, and finally obtain the newborn lamb health status recognition model;

[0119] S4: Validation of the model for identifying the health status of newborn lambs: Acquiring a subset of key biological characteristics of newborn lambs with unknown health status , and input it into the newborn lamb health status recognition model, output the corresponding health status recognition result, and complete the newborn lamb health status recognition.

[0120] Furthermore, in S1, biological signals of newborn lambs are acquired. Signal data is collected from newborn lambs, including lambs in normal, depressed, stressed, suspended animation, and trembling states, selected by professionals. The corresponding health states are represented by numbers 1, 2, 3, 4, and 5. Flexible piezoelectric sensors are worn on the lambs' chests to collect biological and behavioral signals.

[0121] Preferably, the superimposed biological signals are processed, including signal decoupling and reconstruction, and a signal decoupling method is constructed to process the superposition and modal aliasing of nonlinear and non-stationary piezoelectric signals, and the breathing, heart rate, behavior, and call signals in the piezoelectric signals are reconstructed based on the total energy.

[0122] Furthermore, for the signal decoupling part, according to the signal collected by the sensor , which includes four signals: breathing signal , heartbeat signal , call signal , behavioral signals Among them, the breathing signal With heartbeat signal Classified as a periodic signal , the call signal and behavioral signals Classified as non-periodic signal .

[0123] in,

[0124]

[0125]

[0126] in represents the amplitude of the periodic signal, represents the frequency of the sine wave, represents the initial phase of the sine wave, The amplitude of the breathing signal, represents the frequency of the respiratory wave, represents the initial phase of the respiratory wave, Indicates the frequency of the heartbeat wave, Indicates the initial phase of the heartbeat wave.

[0127] The signal to be decomposed is recorded as x(t), where t represents time. The upper and lower envelopes of the signal are obtained by linear interpolation, and the upper envelope u(t) is:

[0128] u(t)=max(x(t),x(t-1),x(t+1))

[0129] Lower envelope L(t):

[0130] L(t)=min(x(t),x(t-1),x(t+1))

[0131] Define the local mean value of the signal as m(t), the local mean value of the signal:

[0132] m(t)=(u(t)+l(t)) / 2

[0133] That is, the average of the upper envelope and the lower envelope.

[0134] Subtract the local mean m(t) from the signal x(t) to obtain the signal residual r(t)=x(t)-m(t).

[0135] If the residual r(t) is a monotonic function or the number of extreme points is less than 2, then r(t) is regarded as an intrinsic piezoelectric component and the decomposition is stopped; otherwise, r(t) is regarded as a new signal for the next round of decomposition.

[0136] Repeat until all ingredients are obtained.

[0137] Furthermore, the Fourier transform is performed on each intrinsic piezoelectric component and the total energy E(j) in the frequency domain is calculated. The calculated total energy is , the energy in the heart rate band from 1.0Hz to 3.0Hz is ; The energy of the respiratory frequency band from 0.1Hz to 0.5Hz is , =1, 2, ..., n. Then they are:

[0138]

[0139] In the formula, δ is the set threshold, is the total energy of a certain intrinsic piezoelectric component in the entire frequency domain. Taking δ as 0.5 gives the best reconstruction effect.

[0140] Signal reconstruction, assuming periodic signal is decomposed into These n intrinsic piezoelectric components have pattern indices n1, n2, n3, and n4. The heartbeat signal sh(t) is reconstructed by (n2-n1+1) intrinsic piezoelectric components, and the respiratory signal sr(t) is reconstructed by (n4-n3+1) intrinsic piezoelectric components:

[0141]

[0142] The reconstructed signal is filtered and the threshold formula is improved to perform wavelet transform on the signal: the wavelet transform is used to decompose the signal into components of different frequencies and scales.

[0143]

[0144] in, is the input signal, is the wavelet function, and Represent the scale and translation parameters, respectively. Select an appropriate threshold: Threshold the coefficients after wavelet transformation, setting smaller coefficients to 0 and retaining larger coefficients. This can filter out noise or unwanted components.

[0145] Thresholding:

[0146]

[0147] Furthermore, signal features are extracted from S1, and characteristic values ​​of breathing, heart rate and behavioral signals are extracted. The biological signal is D, which contains L newborn lambs. Each newborn lamb contains M biological features. In addition to common time domain and frequency domain features such as average value, maximum and minimum values, and number of zero points, a basic disturbance (dd) is proposed to reflect the degree of disturbance of a signal. The specific method is as follows: the 30-minute time domain data is divided into a time window of 10 seconds for calculation, and a total of 180 groups of disturbance dd data are obtained. Then the average disturbance da of this group of data is calculated, and finally the standard disturbance difference dsd of this group of data is calculated. Represents the voltage signal at a certain moment, It represents the average voltage within these 10s. The specific calculation formula is as follows:

[0148]

[0149]

[0150]

[0151] Furthermore, a method for calculating the health index Y of newborn lambs based on a mathematical model in S2 is as follows:

[0152]

[0153] Among them, the biological characteristics of each newborn lamb can be expressed as , is an exponential parameter that can adjust the nonlinearity of each biological characteristic; A nonlinear term was introduced to better capture the complex relationship between biological characteristics and health index, and then the health status of newborn lambs was inferred from the health index.

[0154] Specifically, the variable parameters are adjusted in S3 as follows:

[0155] Determination of parameters in the model : Randomly initialize parameters and calculate the gradient of parameters. :

[0156]

[0157] for :

[0158]

[0159] Similarly, we can calculate gradient. It can usually be regarded as a random noise term, and its value is difficult to determine directly, but it can be indirectly influenced by optimizing other parameters.

[0160] The final adjusted parameters are:

[0161]

[0162] Furthermore, in S3, the feature input matrix is ​​optimized as follows:

[0163] In order to reduce the difficulty of calculating the health index and the number of features, a feature importance ranking method was proposed to find the key biological feature subset of newborn lambs. , , let the correlation index be

[0164]

[0165] in, Different values ​​can be assigned according to actual conditions to adjust the nonlinearity of characteristics and health status. Controls the influence of the exponential function. is the weight coefficient, which indicates the influence of other features on the correlation between the current feature and health status. is the total number of features, and finally re-sorted from small to large according to the correlation index .

[0166] The most important feature in the input feature subset ,Will Bring in the health index , judged only by Can the health status of newborn lambs be identified? And calculate the accuracy rate k1;

[0167] Among them, when there is only one feature subset Based on this, input the second feature , and form a feature set To jointly predict the health status of newborn lambs And calculate the accuracy rate k2; if k1 < k2, then retain ; if k1 > k2, then eliminate .

[0168] Continue to add features according to the above steps , when the feature Is retained, form a component feature set ; when the feature Is eliminated, form a component feature set , and at the same time calculate the prediction accuracy rate k3 of the newly formed feature set. If k2 < k3, then retain ; if k2 > k3, then eliminate .

[0169] Repeat the above steps until the last feature set , calculate its prediction accuracy rate km, determine whether to retain or eliminate it, and the finally formed new feature set is the key biological feature subset of newborn lambs .

[0170] Optionally, in S4, identify newborn lambs that may have potential non - health problems in the healthy state, extract the key feature subsets of healthy lambs and non - healthy lambs , denoted as And Calculate the difference of the corresponding items respectively and denote it as Among them:

[0171]

[0172]

[0173]

[0174] Calculate the change rate of each item , among which , this represents the change rate of the key biological feature subset of newborn lambs from the healthy state to the abnormal state. A threshold t of the change rate can be specified. When the change rate of any key feature subset Exceeds the threshold t, it is determined that the newborn lamb may or will soon have an abnormal health condition.

[0175] Specifically, a newborn lamb health status identification device based on a flexible piezoelectric sensor consists of a flexible piezoelectric sensing unit, a flexible circuit unit, a data analysis unit and an elastic band suitable for newborn lambs. The flexible piezoelectric sensor, flexible circuit unit and data analysis unit are located at the center of the elastic band suitable for newborn lambs.

[0176] The flexible piezoelectric sensor collects biological signals from newborn lambs; the flexible circuit unit amplifies, stores, transmits, and supplies power; and the data analysis unit integrates newborn lamb health status identification methods. The elastic band, designed for newborn lambs, houses the flexible piezoelectric sensor, flexible circuit unit, and data analysis unit, and also fits the chest circumference of newborn lambs. The elastic band is made of 40-gauge Posilicone silicone. Component A, epoxy resin, and component B, hardener, are mixed in a 1:1 ratio by mass and poured into a pre-designed mold to create a flexible elastic band measuring 20 cm long, 5 cm wide, and 1 cm thick.

[0177] Furthermore, the flexible circuit unit has the functions of signal amplification, signal storage, signal transmission and power supply, including the following four circuits: signal conditioning circuit (preamplifier circuit, signal filtering circuit, trap circuit), signal processing circuit (AD conversion circuit, main core chip), signal storage / transmission circuit (SD card storage module / Bluetooth transmission module), power supply circuit (supplying power to the above three circuits), and then connected with the flexible composite piezoelectric sensor to form a flexible composite piezoelectric sensing unit.

[0178] Preferably, an elastic band is used to install the above-mentioned flexible piezoelectric sensor and flexible circuit unit and is adapted to the chest circumference of a newborn lamb. Since the flexible piezoelectric sensor, flexible circuit unit, data analysis unit and elastic band adapted to newborn lambs are all made of flexible materials and have a smooth surface, they can be self-adsorbed and assembled without the need for screws to fix them, ultimately completing a newborn lamb health status identification device based on a flexible piezoelectric sensor.

[0179] The beneficial effects of the present invention are as follows:

[0180] The present invention uses flexible piezoelectric sensors to achieve non-invasive acquisition of biological signals of newborn lambs, reducing the difficulty of data collection; by constructing a final newborn lamb status recognition model, it achieves accurate identification of the health status of unknown newborn lambs, which helps to improve animal welfare and farm economic benefits.

[0181] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0182] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for identifying the health status of newborn lambs based on flexible piezoelectric sensors, characterized in that: include: Obtain a subset of biometric characteristics of newborn lambs of unknown health status; The health status includes: normal state, depressed state, stress state, suspended animation state and trembling state; Inputting the newborn lamb biological feature subset into the final newborn lamb state recognition model to perform state recognition to obtain a target state recognition result; The process of constructing the final newborn lamb status recognition model includes: A flexible piezoelectric sensor was prepared; the flexible piezoelectric sensor comprised a sandwich-structured substrate layer, a sensitive layer, and an electrode layer; the electrode layer was used to transmit vibration signals from the surface of a target newborn lamb; the sensitive layer was used to convert the vibration signals into voltage signals; the substrate layer and the electrode layer were prepared using a one-step molding process combining a Posilicone flexible substrate and laser-induced graphene electrodes; the sensitive layer was composed of a 1 mm thick foam-structured composite piezoelectric material; Acquiring superimposed biological signals using the flexible piezoelectric sensor, and performing signal feature extraction on the superimposed biological signals to obtain an eigenvalue input matrix; Determining a newborn lamb health index calculation formula and a health index division interval, and establishing a mapping relationship between a calculation result of the newborn lamb health index calculation formula and the health status according to the health index division interval, to obtain an original newborn lamb status recognition model; Inputting the eigenvalue input matrix into the original newborn lamb state recognition model to adjust variable parameters to obtain a first optimized recognition model; Optimizing the eigenvalue input matrix using a correlation index to obtain a first optimized input matrix; Inputting the first optimization input matrix into the first optimization recognition model to adjust variable parameters to obtain a second optimization recognition model; The recognition accuracy and recognition speed of the first optimized recognition model and the second optimized recognition model are extracted to obtain model operation data, and the first optimized recognition model and the second optimized recognition model are screened according to the model operation data to obtain the final newborn lamb state recognition model.

2. A method for identifying the health status of newborn lambs based on a flexible piezoelectric sensor according to claim 1, characterized in that: The preparation process of the electrode layer includes: A 5 cm x 5 cm glass slide was cleaned with deionized water, 5 mL of Posilicone silica gel was placed on the glass slide, and the glass slide was placed in a spin coater and spin-coated at a speed of 1000 rpm. A 5 cm x 5 cm PI film was attached to the surface of the glass slide coated with Posilicone silica gel and heated at 80°C for 10 minutes to obtain a substrate; the thickness of the PI film was 0.075 mm. Laser-induced graphene operation was performed on the substrate according to a preset electrode pattern using a CO2 near-infrared laser machine, 15 mL of Posilicone silica gel was placed on the PI film, the glass sheet was placed in a spin coater and spin-coated at a speed of 800 r / min for 20 s, the glass sheet was placed in a vacuum defoaming machine for defoaming treatment for 10 min, the glass sheet was placed on a heating table and heated for 1 h for curing, and the PI film coated with Posilicone silica gel was kneaded for 10 min to obtain the electrode layer.

3. The method for identifying the health status of newborn lambs based on a flexible piezoelectric sensor according to claim 1, characterized in that: The preparation process of the sensitive layer includes: Posilicone silica gel, polyvinylidene fluoride and zinc oxide powder are mixed to obtain a mixed silica gel; the mass fraction of polyvinylidene fluoride and the mass fraction of zinc oxide powder in the mixed silica gel are both 10%; The prepared foam is immersed in the mixed silica gel and heated and cured at 80° C. for 10 minutes to obtain the sensitive layer; the diameter of the prepared foam is 10 mm; and the thickness of the prepared foam is less than 1 mm.

4. The method for identifying the health status of newborn lambs based on a flexible piezoelectric sensor according to claim 1, characterized in that: The flexible piezoelectric sensor is used to obtain superimposed biological signals, and signal features are extracted from the superimposed biological signals to obtain an eigenvalue input matrix, including: Performing signal deconstruction and entropy calculation on the superimposed biological signal to obtain intrinsic piezoelectric component and total energy data; Dividing the total energy data into heartbeat frequency band energy and respiratory frequency band energy according to a preset frequency interval; An index calculation is performed based on the heartbeat frequency band energy and the respiratory frequency band energy to obtain a mode index; the calculation formula of the mode index is: ;in, 、 、 as well as are the first index, the second index, the third index, and the fourth index respectively; The energy in the heart rate band from 1.0Hz to 3.0Hz; is the total energy of the intrinsic piezoelectric component; The energy in the respiratory frequency band from 0.1Hz to 0.5Hz; Calculate the total energy in the frequency domain for the target; is the threshold; is the piezoelectric reconstruction quantity; Reconstructing and filtering the intrinsic piezoelectric component using the pattern index to obtain filtered data; The time domain features, frequency domain features and disturbance features of the filtered data are extracted to obtain the eigenvalue input matrix; the calculation formula of the disturbance feature is: ;in, ; is the disturbance characteristic; is the signal voltage value at the i-th moment; is the average disturbance; is the basic disturbance; is the piezoelectric value at the i-th moment; is the average piezoelectric value within 180s.

5. The method for identifying the health status of newborn lambs based on a flexible piezoelectric sensor according to claim 1, characterized in that: The calculation formula for the newborn lamb health index is: ； in, The calculation result of the newborn lamb health index calculation formula; The i-th eigenvalue of the eigenvalue input matrix; is the i-th index parameter; is the i-th weight coefficient; the number of eigenvalues ​​of the eigenvalue input matrix; is the j-th shape parameter; is the jth core parameter; is the bias term; is the random error term.

6. A method for identifying the health status of newborn lambs based on a flexible piezoelectric sensor according to claim 5, characterized in that: The parameter update formula of the first optimization recognition model includes: ； ； in, ; is the updated i-th weight coefficient; is the correction constant; is the updated i-th index parameter; is the sample size; The first sample eigenvalues; The health index of the samples; The average health index.

7. A method for identifying the health status of newborn lambs based on a flexible piezoelectric sensor according to claim 6, characterized in that: The calculation formula of the correlation index is: ； in, is the correlation index; For all samples The mean of the features; is the mean of the health index of all samples; All are adjustment parameters; For the Weight coefficients; for Second The health index of the samples; for Second The health index of the samples; for Second The first sample eigenvalues; for Second The first sample eigenvalues; is the adjustment factor.

8. The method for identifying the health status of newborn lambs based on flexible piezoelectric sensors according to claim 7, characterized in that: The eigenvalue input matrix is ​​optimized using a correlation index to obtain a first optimized input matrix, including: Calculating the correlation indices of all the eigenvalue input matrices to obtain an original index set; Arrange the original index set in descending order to obtain a sequential index set; Inputting the hth indicator in the sequential indicator set into the first optimized recognition model, and calculating the accuracy of the first optimized recognition model to obtain the hth model accuracy; h is any indicator in the sequential indicator set except the last one; Inputting the hth indicator and the h+1th indicator in the sequential indicator set into the first optimization recognition model, and calculating the accuracy of the first optimization recognition model to obtain the h+1th model accuracy; When the hth model accuracy is less than the h+1th model accuracy, the h+1th indicator is retained; otherwise, the h+1th indicator is eliminated; The sequential indicator set is updated each time an indicator is eliminated, and the sequential indicator set is iterated to obtain the first optimized input matrix after iteration.

9. A newborn lamb health status identification system based on flexible piezoelectric sensors, characterized in that: A method for identifying the health status of a newborn lamb based on a flexible piezoelectric sensor as described in claim 1, comprising: a flexible piezoelectric sensing unit, a flexible circuit unit, a data analysis unit, and an elastic band; The flexible piezoelectric sensing unit and the flexible circuit unit are fixed on the surface of the target newborn lamb through the elastic band; the data analysis unit is connected to the flexible circuit unit; the flexible piezoelectric sensing unit is connected to the flexible circuit unit; The flexible piezoelectric sensing unit is used to collect superimposed biological signals; the flexible circuit unit is used to amplify, store and transmit the superimposed biological signals; and the data analysis unit is used to identify the health status of the target newborn lamb.

Images