Intelligent and accurate live pig feeding system and method based on internet-of-things management and control and mutual feeding of ingestion

Abstract

The invention relates to the technical field of intelligent management, in particular to an intelligent precise pig feeding system and method based on internet-of-things management and control and mutual feeding of ingestion. The multi-modal data acquisition unit is used for acquiring environmental parameters, individual characteristics, behavior videos and food consumption; the ingestion mutual feedback analysis unit is used for processing and analyzing ingestion behavior videos in real time, acquiring real-time data of the live pigs in combination with data fed back by the weighing trough, and establishing an ingestion efficiency evaluation model of the individual live pigs by utilizing the real-time data of the live pigs; and the feeding parameter generation unit is used for acquiring the optimal feeding period and feeding amount according to the ingestion efficiency index of each live pig. According to the system and method, the actual feeding state of each pig can be evaluated more accurately, the accuracy of feeding behavior analysis is improved, misjudgment caused by neglecting individual differences and environmental factors is avoided, the problem of high-frequency noise interference caused by head posture changes can be effectively solved, and the accuracy of feeding behavior analysis is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent management, and more specifically, to a pig intelligent precise feeding system and method based on Internet of Things control and feeding feedback. Background Art

[0002] In modern animal husbandry, with the continuous expansion of the breeding scale and the progress of technology, the demand for precise monitoring and management of the growth performance and health status of pigs is increasing day by day. To meet this demand, intelligent pig breeding technology has emerged, which uses the Internet of Things (IoT), big data analysis, artificial intelligence (AI), and advanced sensor technology to improve breeding efficiency and animal welfare. Through the application of these technologies, real-time monitoring of the pigsty environment can be achieved, including automatic adjustment of key parameters such as temperature, humidity, and ammonia concentration, ensuring that pigs are in the most suitable living environment. At the same time, the intelligent system can also track the growth data of each pig, such as weight changes, feed consumption, and activity patterns, providing scientific decision-making support for farmers. However, despite the significant progress of intelligent pig breeding technology, some existing solutions still have deficiencies. Existing technologies often use a unified standard to evaluate the feeding status of all pigs, ignoring that pigs of different weights and body sizes have different neck movement ranges and head posture changes, and thus have different feeding efficiency indexes, resulting in feed waste and uneven growth. Therefore, a pig intelligent precise feeding system and method based on Internet of Things control and feeding feedback are provided. Summary of the Invention

[0003] The purpose of the present invention is to provide a pig intelligent precise feeding system and method based on Internet of Things control and feeding feedback to solve the problem proposed in the above background art that a unified standard is used to evaluate the feeding status of all pigs, ignoring that pigs of different weights and body sizes have different neck movement ranges and head posture changes, and thus have different feeding efficiency indexes, resulting in feed waste and uneven growth.

[0004] To achieve the above purpose, on the one hand, the present invention aims to provide a pig intelligent precise feeding system based on Internet of Things control and feeding feedback, including: A multimodal data acquisition unit, which is used to monitor the environmental parameters of the pigsty through environmental sensors, track the individual characteristics of individual pigs using RFID ear tags and body temperature sensors, capture the feeding behavior videos of pigs with a camera, collect the feed intake of pigs using a weighing feed trough, and align the environmental parameters, individual characteristics, behavior videos, and feed intake in time; Feeding interaction analysis unit, which is used to process and analyze the feeding behavior video in real time, obtain the real-time data of live pigs by combining the data fed back by the weighing feeder, establish a feeding efficiency evaluation model for each live pig by using the real-time data of live pigs, and the feeding efficiency evaluation model outputs the feeding efficiency index of each live pig; during the process of obtaining the real-time data of live pigs, the influence of live pig weight and feeder height is considered and the pitch angle dynamic compensation is introduced for optimization; Feeding parameter generation unit, which is used to obtain the optimal feeding cycle and feeding amount according to the feeding efficiency index of each live pig.

[0005] As a further improvement of this technical solution, the real-time data of the live pigs includes head posture, chewing frequency, and feed intake. The head posture and chewing frequency are obtained by using the optical flow method for the feeding behavior video, and the feed intake is obtained by the weighing feeder.

[0006] As a further improvement of this technical solution, the specific steps for obtaining the head posture of the live pig are as follows; S21. Locate the key points of the live pig's head in the video. The key points include the tip of the nose, left ear, right ear, left side of the mandible, and right side of the mandible; S22. Denote the midpoint of the line connecting the two ears as the origin, and the line connecting the tip of the nose to the origin as the axis, and the line connecting the left ear to the right ear is denoted as axis; the direction perpendicular to the plane is denoted as axis; S23. Obtain the pitch angle of each live pig at each moment: ; In the formula, is the pitch angle of the live pig at time; is the origin coordinate of the live pig at time; is the tip-of-nose coordinate of the live pig at time;

[0007] S24. Determine whether the live pig is eating effectively: set the range of the effective eating pitch angle. When is within this pitch angle range, it is marked as effective eating, otherwise it is marked as ineffective eating.

[0008] As a further improvement of this technical solution, during the process of obtaining the pitch angle of the live pig in S23, the neck movement ranges of live pigs with different body sizes are different, and as the feed in the feeding trough changes, the reference height when the live pig lowers its head to eat will change. Therefore, considering the influence of the live pig's weight and the height of the feeding trough, specifically: ; In the formula, is the optimized pitch angle of the live pig at time; is the weight compensation coefficient of the live pig at time; is the feeding trough height compensation angle of the live pig at time; ; In the formula, is the height of the feed pile of the live pig at time; is the distance from the tip of the nose of the live pig to the feeding trough at time; Furthermore, it is judged whether the live pig is effectively eating: According to the weight of the live pig, different ranges of effective eating pitch angles are set. When the weight of the live pig is within the pitch angle range corresponding to this weight and continuously frames are all within this range, it is marked as effectively eating, otherwise it is marked as not effectively eating.

[0009] As a further improvement of this technical solution, the specific steps for obtaining the chewing frequency of the live pig are as follows; S25. According to the left mandible and right mandible located in S21, output the coordinates and , where is the left mandible coordinate; is the right mandible coordinate; and the displacement vectors of the left and right mandibles are tracked by the coefficient optical flow method, specifically: ; In the formula, is the gradient of the image in the direction; is the gradient of the image in the direction; is the time gradient; S26. Obtain the displacement vectors of the left and right endpoints of the mandible based on the result of S25, and then calculate the opening and closing amplitude based on the displacement vectors of the left and right endpoints of the mandible: ; In the formula, is the opening and closing amplitude of the mandible of the live pig at time ; is the displacement vector of the left endpoint of the mandible; is the displacement vector of the right endpoint of the mandible; S27. Perform band-pass filtering on the mandibular opening and closing amplitude sequence to remove high-frequency noise and low-frequency components: ; In the formula, is the mandibular opening and closing amplitude sequence of the live pig after filtering at time ; is the function for performing the band-pass filtering operation; is the lowest cut-off frequency; is the highest cut-off frequency; S28. Extract the main frequency from the filtered mandibular opening and closing amplitude sequence to obtain the chewing frequency of the live pig: ; In the formula, is the chewing frequency of the live pig at time ; is the signal length; is the frequency index; ; is all frequency values.

[0010] As a further improvement of this technical solution, during the process of obtaining the chewing frequency, when the pitch angle changes rapidly, high-frequency noise is generated, which in turn interferes with the calculation of the chewing frequency. Therefore, dynamic compensation of the pitch angle is introduced into the original mandibular opening and closing amplitude , and the specific steps are as follows: Introduce the optimized pitch angle of the live pig at time into the original mandibular opening and closing amplitude to obtain the corrected mandibular opening and closing amplitude: ; In the formula, is the corrected mandibular opening and closing amplitude; Then, adjust the cut-off frequency of the band-pass filtering according to the optimized pitch angle of the live pig at time: ; In the formula, is the cut-off frequency of the band-pass filter; is the maximum frequency adjustment amount; Based on the adjusted cut-off frequency, perform band-pass filtering on the mandibular opening and closing amplitude sequence : ; In the formula, is the mandibular opening and closing amplitude sequence of the live pig after filtering optimization at time ; Extract the main frequency from the filtered mandibular opening and closing amplitude sequence to obtain the finally optimized chewing frequency: ; In the formula, is the chewing frequency of the optimized live pig at time ;

[0011] As a further improvement of this technical solution, in the foraging interaction analysis unit, an individual live pig foraging efficiency evaluation model is established using the real-time data of the live pig, and the foraging efficiency evaluation model outputs the foraging efficiency index of each live pig. The specific steps are as follows: Calculate the pitch angle effectiveness score, chewing frequency score, and feed intake score, and use the pitch angle effectiveness score, chewing frequency score, and feed intake score as the feature vectors input to the foraging efficiency evaluation model: ; In the formula, is the head pitch angle score of the live pig ; is the optimal pitch angle; is the effective range of the pitch angle; ; In the formula, is the chewing frequency score of the live pig ; is the lowest effective chewing frequency; is the optimal chewing frequency; is the highest effective chewing frequency; ; In the formula, is the feed intake score per unit time of the live pig ; is the feed intake per unit time of the live pig ; is the lowest effective feed intake; is the optimal feed intake; is the highest effective feed intake; ; wherein, is the input feature vector; Use ReLU as the activation function to process the input feature vector at the current moment, obtaining a real-time feature vector to capture the key features of the current state of the live pig: ; wherein, is the real-time feature vector; is the activation function; is the third weight matrix; is the second weight matrix; is the first weight matrix; is the first bias vector; is the second bias vector; is the third bias vector; Adopt LSTM to process the data of the past time points, and extract the historical feature vector reflecting the feeding behavior pattern of the live pig: ; wherein, is the historical feature vector; is the LSTM network; Concatenate the real-time feature vector and the historical feature vector, and then construct a feeding efficiency evaluation model through the linear layer of the Sigmoid activation function, thereby generating a preliminary feeding efficiency index: ; wherein, is the feeding efficiency index of the live pig ; is the Sigmoid function; is the fourth weight matrix; is the fourth bias vector.

[0012] As a further improvement of this technical solution, a temperature compensation coefficient is introduced into the feeding efficiency index for adjustment to generate the final feeding efficiency index: ; wherein, is the final feeding efficiency index of the live pig ; is the temperature compensation coefficient of the live pig at time ; among them, , is the ambient temperature where the live pig is located, is the optimal temperature.

[0013] As a further improvement of this technical solution, the feeding parameter generation unit obtains the optimal feeding cycle and feeding amount specifically as follows: Optimal feeding cycle: ; In the formula, is the feeding cycle of live pigs ; is the reference feeding cycle; is the adjustment coefficient for controlling the change range of the feeding cycle; is the target feeding efficiency index; Optimal feeding amount: ; In the formula, is the feeding amount of live pigs ; is the reference feeding amount; is the adjustment coefficient for controlling the change range of the feeding amount.

[0014] On the other hand, the present invention provides a method for intelligent precise feeding of live pigs based on Internet of Things control and feeding feedback, which is used for the above-mentioned intelligent precise feeding system of live pigs based on Internet of Things control and feeding feedback, and includes the following steps: S1. Monitor the environmental parameters of the pigsty through environmental sensors, track the individual characteristics of individual live pigs using RFID ear tags and temperature sensors, capture the video of the feeding behavior of live pigs with a camera, and collect the feed intake of live pigs using a weighing trough; S2. Process and analyze the video of the feeding behavior in real time, obtain the real-time data of live pigs by combining the data fed back by the weighing trough, establish an evaluation model of the feeding efficiency of each live pig using the real-time data of live pigs, and the evaluation model of the feeding efficiency outputs the feeding efficiency index of each live pig; S3. Obtain the optimal feeding cycle and feeding amount according to the feeding efficiency index of each live pig.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. In the intelligent precise feeding system and method of live pigs based on Internet of Things control and feeding feedback, considering the differences in the neck movement ranges of live pigs with different weights and the influence of the change in the feed height in the trough on the reference height when lowering the head to eat, the effective recognition of the head posture is ensured. The system can more accurately evaluate the actual feeding status of each pig, not only improving the accuracy of the feeding behavior analysis, avoiding misjudgment caused by ignoring individual differences and environmental factors, but also further optimizing the feeding strategy, improving the breeding efficiency and animal welfare.

[0016] 2. In the intelligent and precise pig feeding system and method based on IoT control and feeding feedback, in order to accurately obtain the chewing frequency, a pitch angle dynamic compensation mechanism is introduced to correct the original mandibular opening amplitude, which can effectively cope with the high-frequency noise interference problem caused by head posture changes, improve the accuracy of eating behavior analysis, and ensure that even in a complex feeding environment, the true chewing activities can be reliably identified. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a block diagram of the overall process of the present invention; Figure 2 It is a flowchart of the method of the whole present invention; The meanings of the various reference numerals in the figure are as follows: 1. Multimodal data acquisition unit; 2. Feeding feedback analysis unit; 3. Feeding parameter generation unit. SPECIFIC EMBODIMENTS

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0019] Embodiment 1: Please refer to Figure 1 As shown, an intelligent and precise pig feeding system based on IoT control and feeding feedback is provided, including a multimodal data acquisition unit 1, a feeding feedback analysis unit 2, and a feeding parameter generation unit 3; Among them, the multimodal data acquisition unit 1 is used to monitor the environmental parameters of the pigsty through environmental sensors, track the individual characteristics of individual pigs using RFID ear tags and body temperature sensors, capture the eating behavior videos of pigs with a camera, collect the feed intake of pigs using a weighing feeder, and align the environmental parameters, individual characteristics, behavior videos, and feed intake in time; Specifically: The environmental sensors include a digital temperature and humidity sensor (such as SHT35), an electrochemical ammonia sensor (such as MQ-137), and a light intensity sensor. 3-5 nodes are arranged at the top of each pigsty to form a grid monitoring network. The sampling frequency is 1 time per minute, and the accuracy is ±0.5°C (temperature), ±2%RH (humidity), ±5ppm (ammonia); The RFID ear tag is a ultra-high frequency RFID ear tag (ISO18000-6C protocol), and the body temperature sensor is a patch type body temperature sensor (accuracy ±0.1°C). The ear tag and the body temperature sensor are integrated through a flexible circuit and wirelessly transmitted to the base station via LoRa; The camera is specifically a wide-angle infrared camera (resolution 1920×1080@30fps), with a built-in waterproof and anti-fog optical cover. It uses YOLOv5s to detect the pig's head in real time, and the ROI area is cropped and then input into the analysis model; The structure of the weighing type feeding trough is a cantilever beam type weighing sensor (range 0-50kg, accuracy ±2g), with a stainless steel moisture-proof structure, aligned with the camera timestamp, and records the weight change every 0.5 seconds. A denoising module is set in the weighing type feeding trough, and the denoising module eliminates the noise data caused by the pig's trampling vibration based on the Kalman filter algorithm.

[0020] The feeding interaction analysis unit 2 is used to process and analyze the feeding behavior video in real time, obtain the real-time data of the pigs by combining the data fed back by the weighing type feeding trough, establish an eating efficiency evaluation model for each pig using the real-time data of the pigs, and the eating efficiency evaluation model outputs the eating efficiency index of each pig; During the process of obtaining the real-time data of the pigs, the influence of the pig's weight and the height of the feeding trough is considered and the pitch angle dynamic compensation is introduced for optimization; The real-time data of the pigs includes the head posture, chewing frequency, and food intake. The head posture and chewing frequency are obtained by using the optical flow method for the feeding behavior video, and the food intake is obtained by the weighing type feeding trough; The food intake is obtained by the weighing type feeding trough. Specifically, the instantaneous food intake of the pig at time is obtained by subtracting the weight of the feeding trough after the pig eats from the weight of the feeding trough before the pig eats, and then the total instantaneous food intakes in the statistical time period are accumulated to obtain the total food intake in a certain time period. The specific steps for obtaining the pig's head posture are as follows;

[0021] Specifically, the steps for obtaining the pig's head posture are as follows; S21. Locate the key points of the pig's head in the video. The key points include the tip of the nose, the left ear, the right ear, the left side of the mandible, and the right side of the mandible; S22. Denote the midpoint of the line connecting the two ears as the origin, and the line connecting the tip of the nose to the origin as the axis, and the line connecting the left ear to the right ear is denoted as axis; The direction perpendicular to the plane is denoted as axis; S23. Obtain each pig Pitch angle at a moment: ; In the formula, is the live pig at the pitch angle at time; is the origin coordinate of the live pig at time; is the nose tip coordinate of the live pig at time; is the left ear coordinate of the live pig at time; is the right ear coordinate of the live pig at time; The pitch angle represents the up and down tilt angle of the live pig's head, reflecting its feeding posture. When the head is down, it means the live pig is approaching the feed trough and may be feeding; when the head is raised, it means the live pig may be observing or resting.

[0022] S24. Determine whether the live pig is effectively feeding: Set the range of the effective feeding pitch angle. When is within this pitch angle range, it is marked as effective feeding; otherwise, it is marked as ineffective feeding. Ineffective feeding includes observing, turning the head, or resting, etc. Exclude the situation where the live pig is near the feed trough but not feeding, such as sniffing, playing, or social behavior. Only record data when the live pig is actually feeding to improve the accuracy of feed intake calculation and avoid mis-triggering the feeding system due to the non-feeding state of the live pig.

[0023] In the process of obtaining the pitch angle of the live pig in S23, the neck movement ranges of live pigs of different body sizes are different, and with the change of the feed in the feed trough, the reference height when the live pig lowers its head to feed will change. Specifically, pigs with larger body weights usually have shorter necks, smaller head movement ranges, and smaller pitch angle change amplitudes; pigs with smaller body weights have more flexible necks, larger head movement ranges, and larger pitch angle change amplitudes. And if the change of the feed height is not considered, the feeding state of the live pig may be misjudged. Therefore, consider the influence of the live pig's weight and the feed trough height, specifically: ; In the formula, is the optimized pitch angle of the live pig at time; is the weight compensation coefficient of the live pig at time. Pigs with larger body weights have smaller head movement ranges, and by reducing the pitch angle change amplitude; pigs with smaller body weights have larger head movement ranges, and by Increase the range of pitch angle change; For live pigs At The height compensation angle of the feeding trough at time; ; In the formula, For live pigs At The height of the feed pile at time (calculated from the change in the weight of the feeding trough). When the feed in the feeding trough decreases, the feed height decreases; when the feed in the feeding trough increases, the feed height increases; For live pigs At The distance from the tip of the nose to the feeding trough at time (calibrated by the camera); when the height of the feeding trough changes (such as the accumulation of feed), the pitch angle reference value is automatically adjusted; Furthermore, it is judged whether the live pig is effectively eating: According to the weight of the live pig, different ranges of effective eating pitch angles are set. When the live pig The weight is within the pitch angle range corresponding to this weight, and continuously Frames are all within this range, then it is marked as effective eating, otherwise it is marked as ineffective eating; to avoid misjudgment due to weight differences: pigs with larger weights have smaller angle changes, so a looser range is set. Pigs with smaller weights have larger angle changes, so a stricter range is set. To avoid misjudgment caused by short-term head swings: continuous Frame judgment can filter out noise and prevent misjudgment as eating.

[0024] Considering the differences in the neck movement ranges of live pigs with different weights and the influence of the change in the feed height in the feeding trough on the reference height when lowering the head to eat, the system can more accurately evaluate the actual eating state of each pig. Specifically, the system dynamically adjusts the calculation of the pitch angle by integrating the live pig weight information and real-time monitoring of the feeding trough height, ensuring effective recognition of the head posture. This personalized monitoring method not only improves the accuracy of eating behavior analysis, avoids misjudgment caused by ignoring individual differences and environmental factors, but also further optimizes the feeding strategy, improving the breeding efficiency and animal welfare.

[0025] The specific steps for obtaining the chewing frequency of live pigs are as follows; S25. According to the left mandible and right mandible located in S21, output the coordinates And , where Is the left mandible coordinate; Is the right mandible coordinate; and the displacement vectors of the left and right mandibles are tracked using the coefficient optical flow method , specifically: ; wherein, is the gradient of the image in the direction; is the gradient of the image in the direction; is the time gradient; S26. Obtain the displacement vectors of the left and right endpoints of the mandible according to the result of S25, and then calculate the opening and closing amplitude based on the displacement vectors of the left and right endpoints of the mandible: ; wherein, is the opening and closing amplitude of the live pig's at time ; is the displacement vector of the left endpoint of the mandible; is the displacement vector of the right endpoint of the mandible; if the mandible opens, the distance between the left and right endpoints increases; if it closes, the distance between the left and right endpoints decreases.

[0026] S27. Perform band-pass filtering on the mandible opening and closing amplitude sequence to remove high-frequency noise and low-frequency components: ; wherein, is the mandible opening and closing amplitude sequence of the live pig after filtering at time , retaining the frequency band of 0.5 - 3 Hz, because the chewing frequency of live pigs is usually within the range of 0.5 - 3 Hz, and signals outside this range may be noise or other irrelevant actions; is the function for performing the band-pass filtering operation; is the lowest cut-off frequency; is the highest cut-off frequency; its parameters are the signal, the lowest cut-off frequency (0.5 Hz), and the highest cut-off frequency (3 Hz) respectively; S28. Extract the main frequency from the filtered mandible opening and closing amplitude sequence to obtain the chewing frequency of the live pig: ; wherein, is the chewing frequency of the live pig at time ; is the signal length; is the frequency index; ; select the frequency component with the largest amplitude in the Fourier transform result as the main chewing frequency of the live pig; is all frequency values, representing that specific frequency value with the largest amplitude among all frequencies; During the acquisition of chewing frequency, when the pitch angle changes rapidly, high-frequency noise is generated, which in turn interferes with the calculation of chewing frequency. Specifically, when the head is raised (such as eating high feed), the projection of mandibular movement is reduced, resulting in an underestimate of the chewing amplitude; when the head is horizontal, the mandibular opening and closing amplitude data is true and reliable; when the head is fully raised (such as licking, observing), the mandibular opening and closing almost loses its practical meaning and needs to be eliminated. The pitch angle dynamic compensation is introduced in the following steps: Introducing the optimized pig into the original jaw opening and closing range exist The pitch angle at time is corrected to obtain the jaw opening and closing range: ; In the formula, is the corrected mandibular opening and closing range; When the pig's head is tilted upward When , the projection of the actual mandibular movement amplitude in the vertical direction decreases and is attenuated by the cosine function; when (horizontal attitude), the compensation coefficient is 1, no attenuation; when (fully tilt your head back), the compensation coefficient is 0, completely eliminating invalid data.

[0027] According to the optimized pig exist The pitch angle of time adjusts the cutoff frequency of the bandpass filter to suppress the high-frequency noise introduced by abnormal head posture: ; In the formula, is the cutoff frequency of the bandpass filter; is the maximum frequency adjustment; when When the head posture changes significantly, the high cutoff frequency is increased to suppress the high-frequency signal caused by abnormal posture; otherwise, it is kept at 3Hz without excessive adjustment to ensure normal feeding behavior analysis; this can effectively suppress the high-frequency interference caused by changes in head posture and make the filtered mandibular movement signal more stable; Based on the adjusted cutoff frequency, the jaw opening and closing amplitude sequence is To perform bandpass filtering: ; In the formula, Pigs after filtering optimization In time The sequence of jaw opening and closing amplitudes at From the filtered jaw opening and closing amplitude sequence Extract the main frequency: ; Wherein, is the optimized chewing frequency of live pigs at time to obtain the finally optimized chewing frequency; This compensation mechanism can effectively address the problem of high-frequency noise interference caused by changes in head posture. Especially when the pig's head is raised or fully tilted back, due to the reduction or loss of practical significance of the mandibular movement projection, it may lead to an underestimation of the chewing amplitude or data distortion. By real-time monitoring and adjusting the compensation factor calculated based on the pitch angle, the system can provide a more accurate estimate of the chewing frequency when the head is in different positions, such as during actions like eating high-place feed, normal horizontal eating, and raising the head to observe.

[0028] In the feeding interaction analysis unit 2, an individual live pig feeding efficiency evaluation model is established using the real-time data and weight change curve of the live pig. The feeding efficiency evaluation model outputs the feeding efficiency index of each live pig. The specific steps are as follows: Calculate the pitch angle effectiveness score, chewing frequency score, and feed intake score, and use the pitch angle effectiveness score, chewing frequency score, and feed intake score as the feature vectors input to the feeding efficiency evaluation model: ; Wherein, is the head pitch angle score of the live pig ; is the optimal pitch angle; is the effective range of the pitch angle; this index is used to measure whether the head posture of the live pig is appropriate during feeding. By comparing the actually observed pitch angle with the ideal angle and considering an allowable angle range, a score between 0 and 1 can be obtained, indicating the effectiveness of the current feeding posture of the live pig, which helps to identify those live pigs that may not be able to eat normally due to physical discomfort or environmental problems.

[0029] ; Wherein, is the chewing frequency score of the live pig ; is the lowest effective chewing frequency; is the optimal chewing frequency; is the highest effective chewing frequency; the actual chewing frequency of the live pig is compared with its optimal value, and the score is determined by setting the lowest effective frequency and the highest effective frequency. Such a scoring mechanism can help monitor the health status of the live pig, because abnormal chewing frequency may be an early sign of disease or other health problems; ; Wherein, is the feed intake score per unit time of the live pig ; For live pigs Feed intake per unit time; Is the minimum effective feed intake; Is the optimal feed intake; Is the maximum effective feed intake; This score reflects the appetite of live pigs within a specific time period. By comparing the actual feeding speed with the ideal minimum, maximum, and optimal feeding rates, a score reflecting the health of the live pig's feeding behavior can be obtained; ; Wherein, Is the input feature vector; Use ReLU as the activation function to process the input feature vector at the current moment to obtain a real-time feature vector, capturing the key features of the current state of the live pig: ; Wherein, Is the real-time feature vector; Is the activation function; Is the third weight matrix; Is the second weight matrix; Is the first weight matrix; Is the first bias vector; Is the second bias vector; Is the third bias vector; Among them, ; , , ; , , ; Adopt LSTM to process the data of the past time points, and extract the historical feature vector reflecting the feeding behavior pattern of live pigs: ; Wherein, Is the historical feature vector; Is the LSTM network; Concatenate the real-time feature vector and the historical feature vector, and then construct a feeding efficiency evaluation model through the linear layer of the Sigmoid activation function, so as to generate a preliminary feeding efficiency index: ; Wherein, Is the feeding efficiency index of the live pig , ; Is the Sigmoid function, mapping the output to range; Is the fourth weight matrix; is the fourth offset vector; A three-layer fully connected network (64→32→16 nodes, ReLU activation) processes real-time features ; An LSTM network (16 units) processes the time series features of live pigs (such as the sequence in the past 1 hour) . After concatenating the dual-channel outputs, the feeding efficiency index of live pigs is output through the Sigmoid function ; All in all, the model construction part uses a three-layer fully connected neural network (FCN) with a ReLU activation function to process features obtained from three different dimensions: pitch angle effectiveness score, chewing frequency score, and feed intake score. In addition, an LSTM network is used to capture the dynamic changes in the time series, that is, the feeding pattern in the past period of time. After combining these two sets of feature vectors, they are mapped to the interval [0,1] through the Sigmoid function to form the feeding efficiency index; A temperature compensation coefficient is introduced into the feeding efficiency index for adjustment to generate the final feeding efficiency index: ; In the formula, is the final feeding efficiency index of live pigs ; is the temperature compensation coefficient of live pigs at time ; Among them, , is the ambient temperature where the live pig is located, is the optimal temperature; The feeding efficiency evaluation model can output the feeding efficiency index (0-1, 1 represents the best state) by inputting the head pitch angle (judging whether the feeding is effective), chewing frequency (evaluating the feeding speed), and feed intake per unit time (g / s) of live pigs; when it is detected that the feeding efficiency of a certain live pig is <0.5 for three consecutive times, an alarm is triggered and the camera is linked to capture a video clip for manual review of the health status; The feeding parameter generation unit 3 is used to obtain the optimal feeding cycle and feeding amount according to the feeding efficiency index of each live pig; Optimal feeding cycle: ; In the formula, is the feeding cycle of live pigs ; is the reference feeding cycle, set to four hours; is the adjustment coefficient that controls the change range of the feeding cycle; is the target feeding efficiency index; When happens, it indicates that the current feeding status of this pig is good and it may not need to be fed frequently. Therefore, the feeding cycle can be appropriately extended to avoid overfeeding. When happens, it indicates that there may be health problems with this pig or it needs more nutrition. At this time, the feeding cycle should be shortened to ensure that the live pig ingests enough feed.

[0030] Optimal feeding amount: ; In the formula, is the feeding amount of the live pig ; is the reference feeding amount; is the adjustment coefficient for controlling the change range of the feeding amount; Similarly, if the feeding efficiency index of a certain pig is higher than the target value, its feeding amount can be appropriately reduced because this indicates that the pig has reached a good growth state and does not need additional feed supply. On the contrary, if the feeding efficiency index is lower than the target value, the feeding amount needs to be increased to meet the higher nutritional requirements of the live pig and help it return to the ideal state; Through the above two formulas, the intelligent feeding system can achieve personalized management of each pig, not only improving the feed conversion rate, reducing the breeding cost, but also effectively enhancing the growth rate and health level of the live pig. Embodiment 2

[0031] Please refer to Figure 2 as shown. This embodiment provides a method for intelligent and precise feeding of live pigs based on Internet of Things control and feeding feedback, which is used for the above-mentioned intelligent and precise feeding system of live pigs based on Internet of Things control and feeding feedback, and includes the following steps: S1. Monitor the environmental parameters of the pigsty through environmental sensors, track the individual characteristics of individual live pigs using RFID ear tags and body temperature sensors, capture the video of the feeding behavior of the live pigs with a camera, and collect the feed intake of the live pigs using a weighing feed trough; S2. Process and analyze the feeding behavior video in real time, obtain the real-time data of the live pigs by combining the data fed back by the weighing feed trough, establish an evaluation model for the feeding efficiency of each live pig using the real-time data of the live pigs, and the evaluation model for the feeding efficiency outputs the feeding efficiency index of each live pig; S3. Obtain the optimal feeding cycle and feeding amount according to the feeding efficiency index of each live pig.

[0032] The basic principles, main features and advantages of the present invention have been shown and described above. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and the descriptions in the specification are only preferred examples of the present invention, which are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed.

Claims

1. A smart and precise pig feeding system based on IoT control and feeding-feedback, characterized in that, Including: A multi-modal data acquisition unit (1), which is used to monitor the environmental parameters of the pigsty through environmental sensors, track the individual characteristics of individual live pigs using RFID ear tags and body temperature sensors, capture the feeding behavior videos of live pigs with a camera, collect the feed intake of live pigs using a weighing feeder, and align the environmental parameters, individual characteristics, behavior videos, and feed intake in time; A feeding interaction analysis unit (2), which is used to process and analyze the feeding behavior videos in real time, obtain the real-time data of live pigs by combining the data fed back by the weighing feeder, establish a feeding efficiency evaluation model for each live pig using the real-time data of live pigs, and the feeding efficiency evaluation model outputs the feeding efficiency index of each live pig; during the process of obtaining the real-time data of live pigs, the influence of the live pig weight and the feeder height is considered and the pitch angle dynamic compensation is introduced for optimization; A feeding parameter generation unit (3), which is used to obtain the optimal feeding cycle and feeding amount according to the feeding efficiency index of each live pig.

2. The intelligent and precise pig feeding system based on Internet of Things control and feeding reciprocity according to claim 1, wherein: The real-time data of the live pigs includes the head posture, chewing frequency, and feed intake. The head posture and chewing frequency are obtained by using the optical flow method for the feeding behavior videos, and the feed intake is obtained by the weighing feeder.

3. The intelligent and precise pig feeding system based on IoT control and feeding and foraging interaction according to claim 2, characterized in that, The specific steps for obtaining the head posture of the live pigs are as follows; S21. Locate the key points of the live pig's head in the video. The key points include the nose tip, left ear, right ear, left side of the mandible, and right side of the mandible; S22. Denote the midpoint of the line connecting the two ears as the origin, and the line connecting the tip of the nose to the origin as axis, and denote the line connecting the left ear to the right ear as axis; The direction perpendicular to the plane is denoted as axis; S23. Obtain the pitch angle of each live pig at each moment: ； In the formula, is the live pig at the pitch angle at time; is the live pig at the origin coordinates at time; is the live pig at the nose tip coordinates at time; Determine whether the live pig is in effective feeding: Set the range of the effective feeding pitch angle. When within this range of the pitch angle, it is marked as effective feeding; otherwise, it is marked as ineffective feeding.

4. The intelligent and precise pig feeding system based on Internet of Things control and feeding interaction according to claim 3, characterized in that: During the process of obtaining the pitch angle of the live pig in S23, the neck movement ranges of live pigs with different body sizes are different, and as the feed in the feeder changes, the reference height when the live pig lowers its head to eat will change. Therefore, the influence of the live pig weight and the feeder height is considered, specifically: ； In the formula, is the pitch angle of the optimized live pig at time; is the weight compensation coefficient of the live pig at time; is the feeding trough height compensation angle of the live pig at time; ； In the formula, is the feed stacking height of live pigs at time; is the distance from the tip of the nose of live pigs at time to the feeding trough; Set different ranges of effective feeding pitch angles according to the weight of live pigs. When the live pig is within the pitch angle range corresponding to this weight and is continuously frames are all within this range, it is marked as effective feeding; otherwise, it is marked as ineffective feeding.

5. The intelligent and precise pig feeding system based on IoT control and feeding and foraging interaction according to claim 4, wherein The specific steps for obtaining the chewing frequency of the live pigs are as follows; S25. Output coordinates based on the left mandible and right mandible located in S21 and , where are the coordinates of the left mandible; are the coordinates of the right mandible; and use the coefficient optical flow method to track the displacement vectors of the left and right mandibles , specifically: ； In the formula, is the gradient of the image in the direction; is the gradient of the image in the direction; is the temporal gradient; S26. Obtain the displacement vector of the left and right endpoints of the mandible according to the result of S25, and then calculate the opening and closing amplitude based on the displacement vector of the left and right endpoints of the mandible: ； In the formula, is the opening and closing amplitude of the lower jaw of live pigs at time ; is the displacement vector of the left end point of the lower jaw; is the displacement vector of the right end point of the lower jaw. S27. Perform band-pass filtering on the mandibular opening and closing amplitude sequence to remove high-frequency noise and low-frequency components: ； In the formula, is the filtered sequence of the opening and closing amplitude of the lower jaw of the live pig at time ; is the function for performing the band-pass filtering operation; is the lowest cut-off frequency; is the highest cut-off frequency; S28. Extract the main frequency from the filtered mandibular opening and closing amplitude sequence to obtain the chewing frequency of the live pig: ； In the formula, is the chewing frequency of live pigs at time ; is the signal length; is the frequency index; ; is all frequency values.

6. The intelligent and precise pig feeding system based on IoT control and feeding interaction according to claim 5, characterized in that: During the process of obtaining the chewing frequency, when the pitch angle changes rapidly, high-frequency noise is generated, which in turn interferes with the calculation of the chewing frequency. Therefore, dynamic compensation for the pitch angle is introduced in the original mandibular opening and closing amplitude The specific steps are as follows: Introduce the optimized live pigs into the original mandibular opening and closing amplitude At The mandibular opening and closing amplitude with the pitch angle corrected at a certain time: ； In the formula, is the corrected mandibular opening and closing amplitude; Then, according to the optimized live pigs At Adjust the cut-off frequency of the band-pass filter according to the pitch angle of time: ； In the formula, is the cut-off frequency of the band-pass filter; is the maximum frequency adjustment amount; Based on the adjusted cut-off frequency, for the mandibular opening and closing amplitude sequence Perform band-pass filtering: ； In the formula, is the sequence of the opening and closing amplitudes of the lower jaw of the pigs after filtering optimization at time ; Extract the main frequency from the filtered mandibular opening and closing amplitude sequence to obtain the finally optimized chewing frequency: ； In the formula, is the optimized chewing frequency of live pigs at time and 7. The intelligent and precise pig feeding system based on IoT control and feeding reciprocity according to claim 6, characterized in that: In the feeding interaction analysis unit (2), a feeding efficiency evaluation model for individual live pigs is established using the real-time data of live pigs, and the feeding efficiency evaluation model outputs the feeding efficiency index of each live pig. The specific steps are as follows: Calculate the pitch angle effectiveness score , chewing frequency score and feed intake score , and use the pitch angle effectiveness score, chewing frequency score, and feed intake score as the feature vectors input to the feed intake efficiency evaluation model ; Among them, ； Use ReLU as the activation function to process the input feature vector at the current moment to obtain the real-time feature vector and capture the key features of the current state of the live pig: ； Wherein, is the real-time feature vector; is the activation function; is the third weight matrix; is the second weight matrix; is the first weight matrix; is the first bias vector; is the second bias vector; is the third bias vector; Use LSTM to process the data of the past time points, and extract historical feature vectors that reflect the feeding behavior patterns of live pigs: ； In the formula, is the historical feature vector; is the LSTM network; Concatenate the real-time feature vector with the historical feature vector, and then construct a feeding efficiency evaluation model through the linear layer of the Sigmoid activation function to generate a preliminary feeding efficiency index: ； In the formula, is the feeding efficiency index of live pigs ; is the Sigmoid function; is the fourth weight matrix; is the fourth bias vector.

8. The intelligent and precise pig feeding system based on IoT control and feeding interaction according to claim 7, characterized in that: Introduce a temperature compensation coefficient into the feeding efficiency index for adjustment to generate the final feeding efficiency index: ； In the formula, is the live hog final feeding efficiency index; is the live hog at time temperature compensation coefficient; where , is the live hog ambient temperature at the location of, is the optimal temperature.

9. The intelligent and precise pig feeding system based on IoT control and feeding interaction according to claim 8, characterized in that, The feeding parameter generation unit (3) obtains the optimal feeding cycle and feeding amount specifically as follows: Optimal feeding cycle: ； In the formula, is the feeding cycle of live pigs ; is the reference feeding cycle; is the adjustment coefficient for controlling the change range of the feeding cycle; is the target feeding efficiency index; Optimal feeding amount: ； In the formula, is the feeding amount of live pigs ; is the reference feeding amount; is the adjustment coefficient for controlling the variation range of the feeding amount.

10. A method for intelligent and precise feeding of live pigs based on IoT control and feeding interaction feedback, which is used for the intelligent and precise feeding system of live pigs based on IoT control and feeding interaction feedback as described in any one of claims 1-9, characterized in that, Including the following steps: S1. Monitor the environmental parameters of the pigsty through environmental sensors, track the individual characteristics of individual live pigs using RFID ear tags and body temperature sensors, capture the feeding behavior videos of live pigs with a camera, and collect the feed intake of live pigs using a weighing feeder; S2. Analyze the video of the feeding behavior in real time, combine the data fed back by the weighing feeder to obtain the real-time data of the live pigs, and establish an evaluation model for the feeding efficiency of each live pig by using the real-time data of the live pigs. The evaluation model for the feeding efficiency outputs the feeding efficiency index of each live pig; S3. Obtain the optimal feeding cycle and feeding amount according to the feeding efficiency index of each live pig.

Images