Replacement gilt oestrus monitoring intelligent ear tag based on estradiol sensor

Abstract

The invention relates to the field of electrochemical biosensing intelligent detection, in particular to a replacement gilt oestrus monitoring intelligent ear tag based on an estradiol sensor. The replacement gilt oestrus monitoring intelligent ear tag comprises an estradiol detection element, a Bluetooth communication module, an electromagnetic interference detection module, a power supply module, a micro-control unit and a cloud data processing module. The local module rapidly screens suspected oestrus data, 90% of invalid data transmission is reduced, and the network bandwidth requirement of a small-scale farm is lowered; the cloud module performs secondary accurate analysis on suspected data in combination with the historical estrus cycle data and the group estrus characteristics of the sows, so that the judgment precision is ensured, and the cost requirement of a small-scale breeding scene is met.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of electrochemical biosensing and intelligent detection, and particularly relates to a backup sow estrus monitoring intelligent ear tag based on an estradiol sensor. BACKGROUND

[0002] In the context of backup sow estrus monitoring, the prior art has involved single technical elements such as estradiol electrochemical aptamer sensors, PDMS gas-permeable membranes, and conventional data transmission links. However, the prior art does not teach how to solve the following complex technical problems, which directly result in the inability of existing solutions to meet the requirements of precise, stable, and efficient monitoring.

[0003] Firstly, the contradiction between "detection stability - response speed" in the dynamic physiological environment of sow ears. The composition of backup sow ear sweat fluctuates dynamically (such as a sudden increase in salt concentration after exercise, and random changes in sebum secretion), and there is a problem of non-uniform contact (sows shaking their heads and rubbing against walls, causing intermittent detachment between the sensor and the skin). The prior art only uses a PDMS gas-permeable membrane to achieve static impurity filtration, without considering the optimization of detection stability in a dynamic environment: when the composition of sweat fluctuates, traditional sensors are prone to "false positive signals", such as abnormal adsorption of electroactive labels caused by high salt concentration; when the sensor and the skin are intermittently detached, the traditional detection mode will be interrupted due to "data supply interruption" in trend analysis. In the prior art, the application of the PDMS gas-permeable membrane only stays at the basic function of "filtering macromolecular impurities", without involving how to balance "impurity filtration efficiency" and "signal response speed" through membrane structure optimization, nor providing a data completion solution for "intermittent contact", resulting in the inability to balance the reliability and continuity of detection data.

[0004] Secondly, the conflict between "estradiol specific recognition - environmental anti-interference" under the superposition of multiple interference factors. In the breeding environment of backup sows, there are multiple non-target interfering substances (such as plant estrogens left in feed, and alkaline substances formed by ammonia gas in the air of the pig house dissolving in sweat) and physical interference (such as electromagnetic radiation generated by pig house equipment, and temperature differences in the ears of different sows). The existing electrochemical aptamer sensor only relies on the specific binding of aptamer and estradiol to achieve recognition, but does not solve the superposition problem of "non-target substances competing for aptamer binding sites" and "electromagnetic / temperature interference causing signal baseline drift": plant estrogens will compete with estradiol for aptamer binding sites, resulting in a high detection value of traditional sensors (the error can be more than 20%); electromagnetic radiation will cause the baseline of electrochemical signals to fluctuate, further confusing the trend of estradiol concentration changes. In the prior art, conventional temperature compensation only addresses "overall changes in environmental temperature", without involving the coordinated correction of "local ear temperature differences" and "electromagnetic interference", nor providing a specific shielding solution for non-target substances, resulting in a significant decline in the specificity and accuracy of estradiol concentration detection.

[0005] Again, in the small-scale breeding scenario, the "data link cost-coverage rate" is unbalanced. Although the existing data transmission scheme (such as "ear tag-gateway-cloud") can realize data transmission, it has significant limitations in small-scale breeding farms (50 or fewer replacement sows). On the one hand, the cost of traditional gateway equipment is high (more than 800 yuan per unit), and it needs to be deployed and debugged by professional personnel, which is difficult for small-scale breeding farms to bear. On the other hand, if the ear tag is directly connected to the cloud (such as NB-IoT), data loss rate will exceed 15% due to signal blind area in the breeding area (such as weak 4G / NB signal in remote areas). In the prior art, the combination of conventional communication technologies only considers a single dimension of "transmission distance" or "power consumption", and does not provide a lightweight data link solution with low cost and high coverage for the dual needs of "cost sensitivity in small-scale scenarios" and "signal blind area", resulting in the existing solution being unable to adapt to the differentiated needs of different scale breeding farms. SUMMARY

[0006] The purpose of the present application is to provide a replacement sow estrus monitoring intelligent ear tag based on an estradiol sensor.

[0007] The replacement sow estrus monitoring intelligent ear tag based on an estradiol sensor according to the present application comprises an estradiol detection element, a Bluetooth communication module, an electromagnetic interference detection module, a power supply module, a micro control unit (MCU), and a cloud data processing module, wherein, The estradiol detection element comprises an electrochemical aptamer sensor and a three-electrode system, wherein a temperature compensation resistor is arranged around the three-electrode system, and a polydimethylsiloxane gas-permeable membrane is coated on the surface of the electrochemical aptamer sensor, with a thickness of 50 μm, for filtering impurities and macromolecular proteins in sow ear sweat to avoid interference with detection. The polydimethylsiloxane gas-permeable membrane has a gradient pore structure, wherein the upper layer has a pore size of 5 μm, and the lower layer has a pore size of 0.2 μm, and an estradiol guiding channel is constructed on the surface of the lower layer membrane; The power supply module supplies power to the micro control unit, the electrochemical aptamer sensor of the estradiol detection element, the Bluetooth communication module, and the electromagnetic interference detection module, respectively; The electromagnetic interference detection module comprises an EMI sensor, which is connected to the micro control unit. The EMI sensor collects electromagnetic interference intensity in the pig house environment in real time and transmits the data to the micro control unit. The micro control unit includes a local storage module, a lightweight LSTM module, a probability determination module, a real-time completion module based on an LSTM-attention mechanism, and a real-time clock module. The micro control unit transmits the electromagnetic interference strength value collected by the EMI sensor of the electromagnetic interference detection module, the estradiol concentration original value detected by the estradiol detection element, the ear temperature value, the ear tag ID, and the detection time to the gateway through the Bluetooth communication module, and then sends them to the cloud data processing module. The ear tag ID is stored in the local storage module of the micro control unit (MCU) and is actively called and included in the data frame by the MCU. The detection time is generated by the real-time clock module (RTC) built-in the MCU and is synchronously triggered with each detection period to ensure that the timestamp accurately corresponds to the detection data. Both of them are integrated with other detection data by the MCU and are transmitted through the Bluetooth communication module, The ear temperature is measured by a temperature compensation resistor (thermistor) arranged around the three-electrode system in the estradiol detection element. The temperature compensation resistor has an accuracy of ±0.1℃ and a measurement range of 0-40℃. It collects the local temperature of the sow's ear in real time, rather than the environmental temperature. The data is directly transmitted to the micro control unit MCU for subsequent double-factor collaborative correction. The micro control unit includes a lightweight LSTM module, a probability determination module, and a real-time completion module based on an LSTM-attention mechanism. The lightweight LSTM module outputs the estrus probability by extracting the short-term estradiol concentration change trend. The probability determination module compares the estrus probability output by the lightweight LSTM module with a preset threshold value. The real-time completion module based on the LSTM-attention mechanism is used to complete the estradiol concentration value and output the actual concentration of estradiol. The cloud data processing module includes a data preprocessing unit and a double-factor collaborative correction unit. The data preprocessing unit filters the collected estradiol concentration data, eliminates abnormal data, and eliminates interference from valid data. The preprocessed estradiol concentration data is associated with the ear tag ID and the sow's basic information. The double-factor collaborative correction unit is used to correct the calculation of the temperature-aptamer binding constant and the calculation of the estradiol concentration. When the micro control unit continuously samples data missing, the double-factor collaborative correction unit is triggered. The temperature-aptamer binding constant is corrected by the following formula: , Where, K 25KT is the standard binding constant of estradiol aptamer and estradiol at 25℃, in the formula, T is the local real-time temperature of the sow's ear collected by the temperature compensation resistance of the estradiol detection element (unit: ℃), the measurement range is 0-40℃, and the accuracy is ±0.1℃, which is directly used to correct the temperature-aptamer binding constant KT, If T <25℃, then K T , K 25 , indicating that low temperature reduces the binding capacity of the aptamer, and the concentration value needs to be corrected; if T >25℃, then K T , K 25 , indicating that high temperature enhances the binding capacity of the aptamer, and the concentration value also needs to be corrected, Based on the corrected binding constant K T , the accurate estradiol concentration value is recalculated C corrected , and the calculation formula is as follows: , Among them, I is the original electrochemical signal strength uploaded by the MCU, unit: nA, I 0 is the blank signal strength, which is calibrated in the absence of estradiol environment in advance, and the value is 10nA, S is the sensor sensitivity, The corrected estradiol concentration value C corrected The data preprocessing unit of the cloud data processing module is further fused to eliminate electromagnetic interference, and finally the estradiol concentration data used for estrus determination is obtained.

[0008] According to the technical scheme of the present application, the estradiol guiding channel is constructed on the surface of the lower layer film by the following method: The prepared double-layer gradient hole PDMS film is placed in an isopropanol solution and ultrasonically cleaned, and then dried after cleaning; The pretreated PDMS film is etched by an inductively coupled plasma etching device, and the estradiol guiding channel on the surface of the lower layer film is formed by ion directional bombardment, and a large number of hydroxyl groups are generated on the inner wall of the channel.

[0009] According to the technical scheme of the present application, a nano-gold particle modification layer is constructed on the surface of the working electrode of the electrochemical aptamer sensor, the nano-gold particle has a diameter of 10nm, and is connected to the end of the aptamer through a mercapto group.

[0010] According to the technical scheme of the application, the nano gold particles are spherical nano gold particles prepared by a trisodium citrate reduction method.

[0011] The method for intelligently monitoring estrus conditions of gilt hogs based on an estradiol sensor according to the application comprises the following steps: An estradiol detection element is worn on the ear of the gilt hog, and estradiol in ear sweat is detected, the estradiol detection element comprising an electrochemical aptamer sensor, a three-electrode system, a polydimethylsiloxane (PDMS) breathable film covering the surface of the sensor, the thickness of the breathable film being 50 microns, the PDMS film being a double-layer gradient pore structure, wherein the upper layer has a pore diameter of 5 microns, and the lower layer has a pore diameter of 0.2 microns, and an estradiol guide channel is constructed on the surface of the lower layer film, and a temperature compensation resistor is arranged around the three-electrode system to correct the influence of the ambient temperature on the detection signal in real time, the three-electrode system comprising a working electrode, a reference electrode and an auxiliary electrode, wherein the working electrode is a gold electrode surface co-assembled with mercaptopolyethylene glycol and a capture chain, the recognition element is an estradiol aptamer, and the capture chain is complementary to a signal chain; when a sample containing estradiol to be detected is detected, a complex formed by the estradiol aptamer and the signal chain is mixed with the sample to be detected, estradiol binds with the estradiol aptamer to release the signal chain, the released signal chain is hybridized with the capture chain to form a DNA double-stranded structure, an electrically active label is introduced and combined with the DNA double strand, and the electrochemical signal is measured by an electrochemical workstation to realize the detection of the sex hormone estradiol; The electrochemical signal generated when the estradiol detection element detects estradiol is output to the micro control unit by the electrochemical aptamer sensor; The electromagnetic interference intensity in the pig house environment is collected in real time by an EMI sensor, and the data is transmitted to the micro control unit; The ear temperature value, the estradiol detection element ID and the detection time data are transmitted to the micro control unit; The lightweight LSTM module of the micro control unit extracts the short-term estradiol concentration change trend, outputs the estrus probability (range 0-100%), compares the output estrus probability with the preset threshold value, wherein if the probability is greater than or equal to 70%, it is marked as “suspected estrus data”, and the electromagnetic interference intensity value, the original estradiol concentration value, the ear temperature value, the estradiol detection element ID and the detection time are uploaded to the cloud, and at the same time, “ear tag ID + estradiol concentration value + detection time + estrus probability” is stored synchronously; if the probability is less than 70%, it is marked as “non-suspected data”, and only the electromagnetic interference intensity value, the original estradiol concentration value, the ear temperature value, the estradiol detection element ID and the detection time are stored; When the micro control unit detects that the continuous sampling data is missing, the real-time completion module of the LSTM-attention mechanism is automatically triggered to complete the data consistent with the actual trend; In the cloud server, the collected estradiol concentration data is screened to eliminate abnormal data, and the effective data is eliminated interference, the preprocessed estradiol concentration data is associated with ear tag ID and sow basic information, When the micro control unit is continuously sampled data missing, the temperature-aptamer binding constant is corrected and calculated, and the estradiol concentration is corrected and calculated, wherein, The temperature-aptamer binding constant is corrected and calculated by the following formula: , Wherein, K 25 The standard binding constant of estradiol aptamer and estradiol at 25℃, T in the formula is the temperature compensation resistance of estradiol detection element, which is the real-time temperature of sow ear part collected (unit: ℃), the measurement range is 0-40℃, the accuracy is ±0.1℃, and it is directly used to correct the temperature-aptamer binding constant KT, If T <25℃, then K T , K 25 , which indicates that low temperature reduces the binding capacity of aptamer, and the concentration value needs to be corrected; if T >25℃, then K T , K 25 , which indicates that high temperature enhances the binding capacity of aptamer, and the concentration value also needs to be corrected, Based on the corrected binding constant K T , the accurate estradiol concentration value is recalculated C corrected , and the calculation formula is as follows: , Wherein, I The original electrochemical signal strength uploaded by MCU, unit: nA, I 0 The blank signal strength is 10nA, which is calibrated in the environment without estradiol, S The sensor sensitivity is The corrected estradiol concentration value C corrected Is transmitted to the cloud for data preprocessing, further fusion with electromagnetic interference elimination, and finally the estradiol concentration data for estrus determination is obtained.

[0012] Beneficial technical effects: In view of the problems not solved by the prior art, the following technical improvements are made in the present application: 1. PDMS gas-permeable membrane-electrode collaborative optimization design to solve the contradiction between "detection stability-response speed".

[0013] In the existing electrochemical aptamer sensor for detecting the sex hormone estradiol, the PDMS membrane has a "uniform pore size". In the present application, the PDMS gas-permeable membrane adopts a double-layer gradient pore structure, the upper layer has a pore size of 5 μm for filtering macromolecular impurities such as hair and dandruff, and the lower layer has a pore size of 0.2 μm for selectively permeating estradiol molecules while blocking interfering substances such as salts and sebum. By means of plasma etching process, an "estradiol guiding channel" is constructed on the surface of the lower layer membrane (the inner wall of the channel is modified with hydroxyl groups, which form weak hydrogen bonds with estradiol molecules to accelerate their migration to the electrode surface), which increases the permeation efficiency of estradiol molecules by 40% and reduces the permeation rate of non-target ions (such as Na⁺ and Cl⁻) by 60%. This improvement not only solves the problem of "high filtering efficiency and slow response" of traditional PDMS membranes, but also reduces the false positive signals caused by salt fluctuations, so that the detection error in dynamic environment is reduced from 15% in the prior art to ≤3%.

[0014] 2. Intermittent contact data intelligent completion algorithm: To solve the problem of data interruption caused by intermittent separation between sensor and skin, a real-time completion model based on LSTM-attention mechanism is integrated in the MCU. When it is detected that data is missing for 2 consecutive samplings (interval of 2 hours), the model automatically calls the sow's historical estrus cycle data of the previous 7 days (such as estradiol concentration fluctuation rules in the same period), real-time activity data (obtained through the three-axis acceleration sensor built-in the ear tag, estradiol concentration fluctuation is small when activity is low), and generates completion data through attention mechanism weight distribution (historical cycle data weight 0.6, activity data weight 0.4). Through testing, the deviation between the completed data and the actual detection data is ≤5% when the data missing rate is ≤30%, which is much better than the traditional "linear interpolation" (deviation ≥15%), ensuring the continuity of trend analysis.

[0015] 3. Multi-dimensional anti-interference collaborative correction to solve the conflict between "specific recognition-environmental anti-interference".

[0016] Firstly, a "aptamer-nanogold particle" composite modification layer is constructed on the surface of the working electrode of the original electrochemical aptamer sensor, and the nanogold particles have a diameter of 10 nm and are connected to the end of the aptamer through thiol groups. The surface plasmon effect of nanogold particles can enhance the electrochemical signal intensity of the electroactive label (methylene blue), and the signal amplitude is increased by 2 times. At the same time, its positive charge characteristics can repel negative interference substances such as sebacate ions in sweat, reducing the adsorption of non-target substances on the electrode surface.

[0017] Secondly, the aptamer fixation density is adjusted from 1x10^12 / cm^2 to 5x10^11 / cm^2, so as to reduce the competitive binding probability of non-target substances such as phytoestrogens. It is verified through experiments that the modified layer can improve the specific recognition rate of the sensor to estradiol from 80% of the traditional sensor to 98%, and reduce the cross-reaction rate of phytoestrogens to below 2%.

[0018] 4. Electromagnetic-temperature dual-factor collaborative correction algorithm: In view of the superimposed influence of electromagnetic interference and local temperature difference, a dual-factor collaborative correction model is introduced in the cloud data preprocessing stage.

[0019] Firstly, the real-time interference intensity is collected through the electromagnetic interference detection module (EMI sensor, detection range 10kHz-1GHz) built in the ear tag, and a “interference intensity-signal baseline offset” mapping relationship is established to correct the baseline of the original signal.

[0020] Secondly, the ear local temperature (not the environmental temperature) collected by the temperature compensation resistor is combined to construct a “temperature-aptamer binding constant” correction formula: K_{T}=K_{25}×(1 + 0.02×(T - 25)) (where K_{T} is the binding constant at temperature T, and K_{25} is the standard binding constant at 25°C), and the estradiol concentration calculation result is adjusted by correcting the binding constant. The algorithm reduces the signal error caused by electromagnetic interference from 12% to 3%, and the error caused by local temperature difference from 8% to 2%, and the comprehensive anti-interference ability is significantly better than the existing single temperature compensation scheme.

[0021] 5. In order to reduce the cloud data transmission pressure and the network dependence of small-scale breeding farms, after the mobile gateway receives the ear tag data in real time, it calculates the estrus probability through the local model (the calculation time is ≤0.5 seconds), and only uploads the “suspected estrus data” (the probability is ≥70%) to the cloud, reducing 90% of the invalid data transmission; the cloud carries out secondary accurate analysis on the “suspected estrus data”, combines historical data and group characteristics, and finally generates an early warning, which not only ensures the analysis accuracy, but also reduces the network bandwidth demand of small-scale breeding farms. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 The flowchart of the method for intelligently monitoring the estrus condition of replacement sows based on the estradiol sensor of the application; Figure 2 The structure schematic diagram of the replacement sow estrus monitoring intelligent ear tag based on the estradiol sensor of the application is shown. DETAILED DESCRIPTION

[0023] As Figure 1 and Figure 2As shown, the smart ear tag according to the present application comprises: an estradiol detection element, a Bluetooth (BLE) communication module, an electromagnetic interference detection module, a power supply module, a micro control unit (MCU), and a cloud data processing module. I. Estradiol detection element

[0024] The estradiol detection element includes an electrochemical aptamer sensor and a three-electrode system, and a polydimethylsiloxane (PDMS) breathable film is covered on the surface of the sensor, with a thickness of 50μm, for filtering impurities and macromolecular proteins in the ear sweat of sows to avoid interference with detection. The gradient pore structure of the PDMS breathable film is improved: breaking through the design limitation of "uniform pore size" in the prior art, a double-layer gradient pore structure is adopted, wherein the upper layer has a pore size of 5μm for filtering macromolecular impurities such as hair and dandruff; the lower layer has a pore size of 0.2μm for selectively permeating estradiol molecules while blocking interfering substances such as salt and sebum. By means of plasma etching process, an "estradiol guiding channel" is constructed on the surface of the lower layer film (the inner wall of the channel is modified with hydroxyl groups, which form weak hydrogen bonds with estradiol molecules to accelerate their migration to the electrode surface), which improves the permeation efficiency of estradiol molecules by 40% and reduces the permeation rate of non-target ions (such as Na⁺ and Cl⁻) by 60%. This improvement not only solves the problem of "high filtering efficiency and slow response" of traditional PDMS films, but also reduces false positive signals caused by salt fluctuations, reducing the detection error in dynamic environment from 15% in the prior art to ≤3%.

[0025] The process of constructing the "estradiol guiding channel" on the surface of the lower layer film by plasma etching process is as follows: place the prepared double-layer gradient pore PDMS film (lower layer pore size 0.2μm) in an isopropanol solution, and clean it with 40kHz ultrasonic wave for 15 minutes to completely remove the residual release agent, dust and other impurities on the film surface; after cleaning, transfer the film to a vacuum drying oven, set the temperature to 60℃ and the vacuum degree to -0.09MPa, and dry for 2 hours to ensure that there is no water or organic solvent residue on the film surface, so as to avoid affecting the subsequent etching effect.

[0026] Using an inductively coupled plasma (ICP) etching device, fix the pretreated PDMS film on the sample table to ensure that the lower layer film is facing up and completely flat; set the etching gas to "oxygen-water vapor mixed gas" (volume ratio 8:2), control the gas flow to 20sccm, set the etching power to 150W, and etch for 30 seconds; apply a 50V radio frequency bias during etching to construct a "estradiol guiding channel" with a diameter of 50-80nm and a depth of 200nm on the surface of the lower layer film by directional ion bombardment, and at the same time generate a large number of hydroxyl (-OH) groups on the inner wall of the channel to realize the specific molecule guiding function.

[0027] After etching, immediately transfer the PDMS film to a nitrogen protection box (nitrogen purity 99.99%), cool to room temperature (25±2℃), prevent impurities in the air from adhering; then immerse the film in deionized water, clean for 5 minutes with 20 kHz ultrasonic waves, remove the PDMS debris generated during etching; finally, place it in a vacuum drying oven again, set the temperature to 40℃, the vacuum degree to-0.09MPa, dry for 1 hour, complete the hydroxyl modification process, and obtain the PDMS breathable film that can be used for sensors.

[0028] At the same time, a temperature compensation resistor (accuracy ±0.1℃) is arranged around the electrode to correct the influence of the ambient temperature on the detection signal in real time (signal error ≤5% within the temperature range of 0-40℃), and a three-electrode system includes a working electrode, a reference electrode and an auxiliary electrode, wherein the working electrode is a gold electrode surface co-assembled with thiol polyethylene glycol and a capture chain, the recognition element is an estradiol aptamer, and the capture chain is complementary to the signal chain; when detecting a sample containing estradiol, the complex formed by the estradiol aptamer and the signal chain is mixed with the sample, estradiol binds with the estradiol aptamer to release the signal chain, the released signal chain is hybridized with the capture chain to form a DNA double-stranded structure, an electrically active label is introduced and combined with the DNA double strand, and the detection of the sex hormone estradiol is realized by measuring the electrochemical signal through an electrochemical workstation.

[0029] An "aptamer-nanogold particle" composite modification layer is constructed on the surface of the working electrode of the electrochemical aptamer sensor, and the nanogold particle has a diameter of 10nm and is connected to the end of the aptamer through thiol.

[0030] The signal output end of the electrochemical aptamer sensor is connected to the MCU through the SPI interface, and the MCU controls the detection cycle of the sensor, which is 1 detection per 2 hours by default and can be adjusted remotely through the cloud.

[0031] The nanogold particle is a spherical nanogold particle prepared by the trisodium citrate reduction method, and has the following specific properties: morphology: regular spherical shape, no agglomeration (observed by transmission electron microscope TEM, particle size variation coefficient ≤10%); Surface charge: zeta potential -25~-30mV (detected by Malvern particle size analyzer), which can be repelled by negative charge to avoid agglomeration; Purity: gold element purity ≥99.9% (inductively coupled plasma emission spectrometry ICP-OES detection), no impurity ions (such as Cl⁻, Na⁺ residual ≤1ppm); Particle size: 10±1nm (TEM statistics of 500 particles diameter average), to ensure that the surface plasmon effect is optimal and the electrochemical signal is enhanced.

[0032] The preparation method of the nanogold particle is as follows: Reagent preparation: chloroauric acid solution: weigh 0.1 g of chloroauric acid (HAuCl4·4H2O, analytical pure), dissolve with deionized water (resistivity ≥ 18.2 MΩ·cm) and dilute to 10 mL to obtain 1% (w / v) chloroauric acid stock solution; trisodium citrate solution: weigh 0.1 g of trisodium citrate (Na3C6H5O7·2H2O, analytical pure), dissolve with deionized water and dilute to 10 mL to obtain 1% (w / v) trisodium citrate stock solution.

[0033] Reduction reaction: add 200 mL of deionized water to a 500 mL three-necked flask, place it on a constant temperature magnetic stirrer, set the stirring speed to 500 rpm, heat to boiling (100°C, keep refluxing); quickly add 5 mL of 1% chloroauric acid stock solution, stir for 30 seconds, the solution turns light yellow; then slowly add 8 mL of 1% trisodium citrate stock solution (addition speed 1 mL / min), continue stirring for 15 minutes, the solution color gradually changes from light yellow → light red → wine red, indicating the formation of gold nanoparticles; after the addition is completed, continue stirring for 30 minutes to ensure complete reaction, control the particle size by adjusting the amount of trisodium citrate (for every 1 mL increase in trisodium citrate, the particle size decreases by about 1 nm, 8 mL in this scheme corresponds to a particle size of 10 nm).

[0034] After the sensor comes into contact with ear sweat, the estradiol in the sweat specifically binds to the aptamer, releasing a signal chain, which hybridizes with the capture chain to form a DNA double strand. The electroactive label (methylene blue, concentration 50 μM) is embedded in the double strand, generating an electrochemical signal. The micro control unit (MCU) collects this signal and converts it into an estradiol concentration value. II. Bluetooth (BLE) communication module

[0035] The BLE module receives ear tag data, connects with the micro control unit (MCU) through the UART interface, and uploads the data to the cloud server through the 4G module after the MCU verifies the data integrity. The BLE communication module is only awakened during data transmission and enters sleep mode during non-transmission.

[0036] The core function of the BLE module is ear tag data transmission: receiving the "electromagnetic interference intensity value, estradiol concentration raw value, ear temperature value, ear tag ID, detection time" data frame packaged by the MCU, transmitting it to the gateway, and then uploading it to the cloud through the gateway 4G module. III. Power supply module

[0037] The power supply module supplies power to the MCU, sensor, BLE module, and electromagnetic interference detection module. The sensor power supply uses a controllable switch that is only turned on during detection to reduce standby power consumption. IV. Electromagnetic interference detection module

[0038] The core element of the electromagnetic interference detection module is an EMI sensor, which is directly connected to the MCU through an I2C communication interface with a communication rate of 100 kHz. The EMI sensor collects the electromagnetic interference intensity in the pig house environment in real time and transmits the data to the MCU. V. MCU

[0039] The micro control unit includes a local storage module, a lightweight LSTM module, a probability determination module, a real-time completion module based on LSTM-attention mechanism, and a real-time clock module.

[0040] The MCU packages the "electromagnetic interference intensity value + estradiol concentration original value + ear temperature value + ear tag ID + detection time" into a JSON format data frame (data volume < 128 bytes), transmits it to the gateway through the BLE module, and uploads the data to the cloud after the gateway uploads the data to the cloud. It is used for "electromagnetic-temperature double factor cooperative correction" to eliminate the influence of environmental electromagnetic interference on estradiol concentration detection.

[0041] The ear tag BLE module is responsible for data transmission from the MCU to the gateway, and the gateway parses the data and generates an input sequence after receiving it, which is used for the local lightweight LSTM module to calculate the estrus probability.

[0042] The lightweight LSTM module reads the input sequence, extracts the short-term estradiol concentration change trend (such as the consecutive 2 concentration rise amplitude) through 1 layer hidden layer (32 neurons), and outputs the estrus probability (range 0-100%).

[0043] The probability determination module compares the estrus probability output by the lightweight LSTM module with the preset threshold (70%): if the probability ≥ 70%, it is marked as "suspected estrus data", and "ear tag ID + estradiol concentration value + detection time + estrus probability" is packaged and uploaded to the cloud through the 4G module; if the probability < 70%, it is marked as "non-suspected data" and only stored.

[0044] The real-time completion module based on LSTM-attention mechanism includes: Input layer: dimension 2x14, where "2" represents two types of input features (historical period data features, real-time activity data features), and "14" represents the input time step (take the detection data of the previous 7 days, every 2 hours, calculate the average value by day, form 7 time steps, each time step contains 2 feature values: daily average concentration, daily average activity), the input data needs to be standardized first (formula: ), to eliminate the dimension effect.

[0045] LSTM layer: Set 2 layers of hidden layers, 64 neurons in each layer, and use tanh as the activation function; set the forgetting gate threshold to 0.7 to retain key historical period information and forget ineffective fluctuation data; the output dimension of the first LSTM layer is 64x14, the second LSTM layer performs feature fusion on the output of the first layer, and the output dimension is 64x1, which extracts the long and short term dependence of the time series (such as the rising trend of concentration in the pre-estrus period, the correlation between activity and concentration).

[0046] Attention layer: uses additive attention (Additive Attention) mechanism to assign weights to the feature vectors output by the LSTM layer, and the core calculation formula is as follows: Feature similarity calculation: where, h j is the historical period data feature vector (dimension 64x1) output by the LSTM layer, u j is the real-time activity data feature vector (dimension 1x1), W 1 (64x32), and W 2 (1x32) are trainable weight matrices, v (32x1) is the attention vector. Weight normalization: , which constrains the historical period data weight to be fixed at 0.6 and the real-time activity data weight to be fixed at 0.4, ensuring that the weight distribution conforms to the physiological law (historical period has a greater impact on concentration trend).

[0047] Output layer: uses a fully connected layer with an input dimension of 64x1 (weighted features after attention layer) and an output dimension of 1, and uses sigmoid as the activation function. The output is the completed estradiol concentration value (unit: pM), and the actual concentration is restored through inverse normalization (formula: ).

[0048] When the MCU detects that the sampling data is missing for 2 consecutive times (interval of 2 hours, consistent with the preset detection period), the automatic trigger module is triggered. The historical estrus cycle data (daily average estradiol concentration) of the sow for the previous 7 days and the real-time activity data (average three-axis acceleration within 1 hour before the data is missing, converted to activity level 0-5) are retrieved from the ear tag local storage module. The input data is transmitted to the LSTM layer after standardization, and the trend characteristics are extracted; the attention layer is weighted and fused, and then input into the full connection layer, and the completed concentration value is output; finally, moving average smoothing (window size 3) is performed on the adjacent effective data (1 time before missing and 1 time after missing) to ensure that the completed data is consistent with the actual trend. The completed data must meet the condition that the deviation from the adjacent effective data is less than or equal to 5%; if the deviation exceeds the range, the model is called again and the historical data time step is adjusted (extended to 10 days), until the accuracy requirement is met. Six. Cloud data processing module

[0049] The BLE module receives ear tag data, which is checked for data integrity by the MCU, and then uploaded to the cloud server through the 4G module (using the MQTT protocol, QoS level 1, to ensure that data is not lost); in the cloud server, data processing is performed by the cloud data processing module. The data processing module includes a data preprocessing unit and a double-factor collaborative correction unit.

[0050] The data preprocessing unit performs outlier filtering: the collected estradiol concentration data is screened, and data that exceeds the detection range (<7.5 pM or >100 nM) or has abnormal fluctuations (single change amount >50%) is removed, marked as "invalid data" and recorded; smoothing: using the moving average method (window size 3 detection points) to smooth the effective data, eliminating random noise (such as signal fluctuations caused by insufficient sweat during detection); data association: associate the preprocessed estradiol concentration data with ear tag ID and sow basic information (breed, age, weight), and store it in the MySQL database.

[0051] Trend analysis model: model construction: based on historical estrus data (estradiol concentration change curve of sows with known estrus cycle), train the LSTM time series prediction model, input is the estradiol concentration sequence for 3 consecutive days (total 36 data points), output is the estrus probability (0-100%); parameter threshold: preset estrus determination threshold: when the model output estrus probability ≥80% and the estradiol concentration of 2 consecutive detections increases by ≥30% compared with the base value (average concentration for 7 days before estrus), trigger estrus warning; update mechanism: the system regularly (once a month) fine-tunes the model with new estrus data to optimize the prediction accuracy (aiming for accuracy ≥95%).

[0052] The "two-factor synergistic correction unit" completes the temperature-aptamer binding constant correction calculation, which is the core submodule of the cloud data processing module, and the specific implementation process is as follows: the cloud data processing module receives the data packet uploaded by the gateway through the MQTT protocol, and extracts the key parameters after analysis: ear tag ID, estradiol original concentration value (pM) converted by MCU, ear local temperature value (0-40℃, accuracy ±0.1℃) collected by ear tag temperature compensation resistance, and electromagnetic interference intensity value (dBμV / m). C raw The "two-factor synergistic correction unit" completes the temperature-aptamer binding constant correction calculation, which is the core submodule of the cloud data processing module, and the specific implementation process is as follows: the cloud data processing module receives the data packet uploaded by the gateway through the MQTT protocol, and extracts the key parameters after analysis: ear tag ID, estradiol original concentration value (pM) converted by MCU, ear local temperature value (0-40℃, accuracy ±0.1℃) collected by ear tag temperature compensation resistance, and electromagnetic interference intensity value (dBμV / m). T E

[0053] The "two-factor synergistic correction unit" completes the temperature-aptamer binding constant correction calculation, which is the core submodule of the cloud data processing module, and the specific implementation process is as follows: the cloud data processing module receives the data packet uploaded by the gateway through the MQTT protocol, and extracts the key parameters after analysis: ear tag ID, estradiol original concentration value (pM) converted by MCU, ear local temperature value (0-40℃, accuracy ±0.1℃) collected by ear tag temperature compensation resistance, and electromagnetic interference intensity value (dBμV / m). K 25 The "two-factor synergistic correction unit" completes the temperature-aptamer binding constant correction calculation, which is the core submodule of the cloud data processing module, and the specific implementation process is as follows: the cloud data processing module receives the data packet uploaded by the gateway through the MQTT protocol, and extracts the key parameters after analysis: ear tag ID, estradiol original concentration value (pM) converted by MCU, ear local temperature value (0-40℃, accuracy ±0.1℃) collected by ear tag temperature compensation resistance, and electromagnetic interference intensity value (dBμV / m). K d T K T K 25 T K T K 25

[0054] The "two-factor synergistic correction unit" completes the temperature-aptamer binding constant correction calculation, which is the core submodule of the cloud data processing module, and the specific implementation process is as follows: the cloud data processing module receives the data packet uploaded by the gateway through the MQTT protocol, and extracts the key parameters after analysis: ear tag ID, estradiol original concentration value (pM) converted by MCU, ear local temperature value (0-40℃, accuracy ±0.1℃) collected by ear tag temperature compensation resistance, and electromagnetic interference intensity value (dBμV / m). K T C corrected I I 0 S ​​​​​​​​​​​​​​​​​​​The sensor sensitivity (0.5nA / (pM), determined by calibration before the sensor leaves the factory; for example: when T=30℃, , if , then , the concentration calculation deviation caused by high temperature is corrected.

[0055] The corrected estradiol concentration value (E2) C corrected will be transmitted to the "data preprocessing unit" of the cloud data processing module, and the result after electromagnetic interference baseline correction (eliminate the influence of electromagnetic interference) is further fused, and finally the accurate concentration data that can be used for estrus determination is obtained, and stored in the MySQL database (retention time≥1 year).

[0056] The local computing module and the cloud data processing module form a "two-level determination" mechanism: the local module quickly screens suspected estrus data, reduces 90% of invalid data transmission, and reduces the network bandwidth demand of small-scale farms; the cloud module combines the historical estrus cycle data of the sow and the estrus characteristics of the group to perform secondary accurate analysis on the suspected data, and finally generates estrus warning and pushes it to the breeder APP, which not only ensures the determination accuracy (accuracy≥95%), but also adapts to the cost demand of small-scale breeding scenes.

[0057] The above examples are only for understanding the technical solutions of the present application and do not limit the protection scope of the present application.

Claims

1. A smart ear tag for monitoring estrus in gilts based on an estradiol sensor, characterized in that, The intelligent ear tag for monitoring estrus in gilts includes an estradiol detection element, a Bluetooth communication module, an electromagnetic interference detection module, a power supply module, a microcontroller unit, and a cloud data processing module. The estradiol detection element includes an electrochemical aptamer sensor and a three-electrode system. A temperature compensation resistor is disposed around the three-electrode system to detect the ear temperature of the sow. A polydimethylsiloxane breathable membrane is covered on the surface of the electrochemical aptamer sensor. The polydimethylsiloxane breathable membrane has a gradient pore structure, wherein the upper pore diameter is 5 μm and the lower pore diameter is 0.2 μm, and an estradiol guiding channel is constructed on the surface of the lower membrane. The power supply modules provide power to the microcontroller unit, the electrochemical aptamer sensor of the estradiol detection element, the Bluetooth communication module, and the electromagnetic interference detection module, respectively. The electromagnetic interference detection module includes an EMI sensor, which is connected to the microcontroller unit. The EMI sensor collects the electromagnetic interference intensity in the pig house environment in real time and transmits the data to the microcontroller unit. The microcontroller unit transmits the electromagnetic interference intensity value collected in real time by the EMI sensor of the electromagnetic interference detection module, the original value of estradiol concentration detected by the estradiol detection element, the ear temperature value, the ear tag ID, and the detection time to the gateway via the Bluetooth communication module, and then sends them to the cloud data processing module. The microcontroller unit includes a local storage module, a lightweight LSTM module, a probability determination module, a real-time completion module based on the LSTM-attention mechanism, and a real-time clock module. The ear tag ID is stored in the local storage module of the microcontroller unit. The detection time is generated by the real-time clock module built into the MCU. The lightweight LSTM module outputs the estrus probability by extracting short-term estradiol concentration change trends. The probability determination module compares the estrus probability output by the lightweight LSTM module with a preset threshold. The real-time completion module based on the LSTM-attention mechanism is used to complete the estradiol concentration value and output the actual estradiol concentration. The cloud-based data processing module includes a data preprocessing unit and a two-factor collaborative correction unit, wherein... The data preprocessing unit filters the collected estradiol concentration data, removes abnormal data, eliminates interference from valid data, and associates the preprocessed estradiol concentration data with ear tag ID and sow basic information. The dual-factor collaborative correction unit is used to correct the calculated temperature-aptamer binding constant and the calculated estradiol concentration. When the microcontroller unit experiences a continuous loss of sampled data, the dual-factor collaborative correction unit is triggered. The temperature-aptamer binding constant is calculated using the following formula: ， in, K 25 The standard binding constant between estradiol aptamer and estradiol at 25℃ is given by the formula. In the formula, T is the real-time local temperature of the sow's ear collected by the temperature compensation resistor of the estradiol detection element, which is directly used to correct the temperature-aptamer binding constant KT. like T <25℃, then K T < K 25 This indicates that low temperature reduces aptamer binding ability, and the concentration value needs to be corrected; if T >25℃, then K T > K 25 This indicates that high temperature enhances the aptamer binding ability, and the concentration value also needs to be corrected. Based on the modified binding constant K T Recalculate the accurate estradiol concentration value C corrected The calculation formula is as follows: , in, I The intensity of the original electrochemical signal uploaded by the microcontroller unit, in nA. I 0 The blank signal intensity, pre-calibrated in an estradiol-free environment, has a value of 10 nA. S For sensor sensitivity, Corrected estradiol concentration value C corrected The data preprocessing unit, which transmits the data to the cloud data processing module, is further integrated with the data to eliminate electromagnetic interference, ultimately yielding estradiol concentration data for estrus determination.

2. The intelligent ear tag for monitoring estrus in gilts based on an estradiol sensor according to claim 1, characterized in that, The thickness of the polydimethylsiloxane breathable membrane covering the surface of the electrochemical aptamer sensor is 50 μm.

3. The intelligent ear tag for monitoring estrus in gilts based on an estradiol sensor according to claim 1, characterized in that, Estradiol-directing channels were constructed on the surface of the lower membrane using the following method: The prepared bilayer gradient pore PDMS membrane was placed in an isopropanol solution, ultrasonically cleaned, and then dried. The pretreated PDMS film was etched using an inductively coupled plasma etching device. Ions were directed to bombard the estradiol-guided channels on the surface of the lower film, while a large number of hydroxyl groups were generated on the inner wall of the channels.

4. The intelligent ear tag for monitoring estrus in gilts based on an estradiol sensor according to claim 1, characterized in that, A gold nanoparticle modification layer with a diameter of 10 nm was constructed on the working electrode surface of the electrochemical aptamer sensor, and the gold nanoparticles were connected to the end of the electrochemical aptamer through thiol groups.

5. The method for monitoring the estrus status of gilts using the estradiol-based smart ear tag for estrus monitoring as described in claim 1, characterized in that, The method includes the following steps: Data collection steps: An estradiol detection element is worn on the ear of a gilt to detect estradiol in ear sweat. An EMI sensor is used to collect the electromagnetic interference intensity in the pig house environment in real time, obtain the ear tag ID and detection time, and transmit the electromagnetic interference intensity value collected in real time by the EMI sensor of the electromagnetic interference detection module, the original value of estradiol concentration detected by the estradiol detection element, the ear temperature value, the ear tag ID and detection time to the microcontroller unit. Data analysis and processing steps: The lightweight LSTM module of the microcontroller extracts short-term estradiol concentration trends and outputs the estrus probability. This probability is compared to a preset threshold. If the probability is ≥70%, it is marked as "suspected estrus data," and the electromagnetic interference intensity value, original estradiol concentration value, ear temperature value, estradiol detection element ID, and detection time are uploaded to the cloud. Simultaneously, "ear tag ID, estradiol concentration value, detection time, and estrus probability" are stored. If the probability is <70%, it is marked as "non-suspected data," and only the electromagnetic interference intensity value, original estradiol concentration value, ear temperature value, estradiol detection element ID, and detection time are stored. When the microcontroller detects missing continuous sampling data, it automatically triggers the real-time completion module of the LSTM-attention mechanism, and the completed data is consistent with the actual trend. Segmentation analysis steps: On the cloud server, the collected estradiol concentration data is screened, outliers are removed, and interference is eliminated from the valid data. The preprocessed estradiol concentration data is then associated with ear tag IDs and sow basic information. When continuous sampling data from the microcontroller unit is missing, the calculated temperature-aptamer binding constant and the calculated estradiol concentration are corrected. The temperature-aptamer binding constant is corrected using the following formula: ， in, K 25 The standard binding constant between estradiol aptamer and estradiol is given by KT, where T is the real-time local temperature of the sow's ear collected by the temperature compensation resistor of the estradiol detection element, and is directly used to correct the temperature-aptamer binding constant KT. like T <25℃, then K T < K 25 This indicates that low temperature reduces aptamer binding ability, and the concentration value needs to be corrected. T >25℃, then K T > K 25 This indicates that high temperature enhances the aptamer binding ability, and the concentration value also needs to be corrected. Based on the modified binding constant K T Recalculate the accurate estradiol concentration value C corrected The calculation formula is as follows: , in, I The intensity of the raw electrochemical signal uploaded by the MCU, in nA. I 0 The blank signal intensity, pre-calibrated in an estradiol-free environment, has a value of 10 nA. S For sensor sensitivity, Corrected estradiol concentration value C corrected The data is transmitted to the cloud for preprocessing and further integration to eliminate electromagnetic interference, ultimately yielding estradiol concentration data used for estrus determination.

Images