A Precision Feeding and Feed Conversion Ratio Intelligent Evaluation System and Method for Small Groups of Beef Cattle

Abstract

This invention discloses a precision feeding and intelligent feed conversion ratio (FCR) evaluation system and method for small-group beef cattle, belonging to the field of smart livestock farming technology. The system includes a water and weighing module, a TMR (Total Meat Ratio) fixed-point feeding module, a video feed residue monitoring module, a FCR calculation module, and a decision support module. This invention significantly improves the accuracy of FCR evaluation by refining it from a broad group level to a small group or even individual level. It replaces manual trough cleaning and weighing with video feed residue monitoring, achieving real-time automated collection of feed residue data, greatly improving monitoring frequency and data reliability. Furthermore, it uses water data to cross-validate feed intake, effectively correcting calculation errors and solving the data silo problem. In addition, by establishing a multi-dimensional FCR database, it can support four key decisions: raw material selection, formula optimization, breed selection, and culling of low-yielding cattle, helping farmers improve production efficiency from multiple levels.

Description

Technical Field

[0001] This invention relates to the field of intelligent livestock farming technology, and in particular to a precise feeding system and method for small groups of beef cattle and an intelligent evaluation system for feed conversion ratio. Background Technology

[0002] Beef cattle farming is a crucial component of animal husbandry, and its production efficiency directly impacts meat supply and economic benefits. In beef cattle farming, the feed conversion ratio (FCR) is a core indicator for measuring feed conversion efficiency. Accurate FCR evaluation is of significant guiding importance for breed selection, formula optimization, raw material screening, and culling of low-producing cattle. Currently, large-scale beef cattle farms commonly employ Total Mixed Ration (TMR) feeding technology, achieving a degree of precision feeding through feed weighing and control. Simultaneously, some farms are beginning to explore the introduction of video-based behavioral recognition technology or RFID-based individual identification technology to observe cattle feeding and drinking behaviors. However, these technologies largely operate independently and have not yet formed a complete technical system for accurately evaluating the FCR.

[0003] The existing beef cattle farming technology has the following problems: First, the feed conversion ratio is measured in a crude way. Traditional methods rely on the average weight gain of the population and the total feed consumption for estimation, which cannot be accurate to the individual or small group level, resulting in a lack of accurate data support for breeding and selection.

[0004] Second, the monitoring of leftover feed is lagging behind. The manual cleaning and weighing operation is infrequent, labor-intensive, and prone to human error, making it impossible to keep track of the dynamic changes in feed intake in real time.

[0005] Third, there is a serious problem of data silos. Water intake data, feeding data, and leftover data are scattered across different systems or manual records, lacking the ability to conduct joint analysis and making it difficult to improve the accuracy of feed intake calculation through cross-validation of multi-source data.

[0006] Fourth, there is insufficient basis for decision-making. Key production decisions such as raw material selection, formula optimization, breed selection, and culling of low-yielding cattle lack precise feed conversion ratio evaluation indicators as a basis.

[0007] Existing TMR (Total Feed Mixing) systems primarily focus on feed formulation accuracy and automated feeding, but lack the capability for precise closed-loop evaluation of feed conversion ratio (FCR) for small groups of beef cattle. Video recognition-based feeding monitoring technologies are mostly used for behavior or identity recognition and have not yet formed a complete integrated system with precise FCR calculation. Therefore, this invention proposes a precise feeding system and method for small groups of beef cattle with intelligent FCR evaluation to address the problems existing in the prior art. Summary of the Invention

[0008] To address the aforementioned problems, the present invention aims to propose a precise feeding system and method for small groups of beef cattle and an intelligent evaluation system for feed conversion ratio, thereby enabling precise decision support for raw material selection, formula optimization, breed selection, and culling of low-producing cattle during the beef cattle feeding process.

[0009] To achieve the objectives of this invention, the invention is implemented through the following technical solution: a precise feeding and feed conversion ratio (FCR) intelligent evaluation system for small groups of beef cattle, comprising a water and weighing module, a TMR (Total Meat Ratio) fixed-point feeding module, a video feed residue monitoring module, a FCR calculation module, and a decision support module, wherein the small group is a target group composed of a preset number of beef cattle, characterized in that: The drinking water weighing module is used to record the individual water consumption and corresponding timestamp of each beef cattle in the target group in real time. By obtaining the water consumption data of each beef cattle, a basis is provided for subsequent cross-validation of feed intake. The TMR (Total Mixed Ration) feeding module is used to precisely feed the target population into the feeding troughs according to a preset formula, and to record the amount of feed, formula information and feeding time. TMR is a total mixed ration. By recording the type, weight and ratio of feed, it ensures the accuracy of the source data for feed intake calculation. The video feed monitoring module is used to periodically collect images of the feeding trough, identify the feed area through deep learning semantic segmentation algorithm and calculate the feed area percentage, convert the feed area percentage into feed weight, and realize non-contact feed monitoring based on AI visual recognition of the remaining feed area (%) and the feed trough geometric parameters. The feed conversion ratio (FCR) calculation module receives individual water intake recorded by the water weighing module, feed intake and formulation information recorded by the TMR (Total Meat Ratio) feeding module, and residual feed weight calculated by the video residual feed monitoring module. It also acquires periodically measured weight gain data of the target population. Based on the feed intake and residual feed weight, it calculates the actual feed intake, and based on the actual feed intake and weight gain data, it calculates the FCR of the target population. Furthermore, it uses individual water intake to cross-validate and correct the actual feed intake. This module, as the core calculation unit, outputs the actual feed intake (kg) and FCR, and uses water intake (correlation coefficient R with feed intake) as the output. 2 >0.85) to verify the accuracy of feed intake; The decision support module is used to establish a feed conversion ratio database for the target group, perform multi-dimensional comparative analysis based on the feed conversion ratio data of the target group, and output suggestions for raw material selection, formula optimization, breed selection, and culling of low-yielding cattle, forming a closed loop of "data-evaluation-decision" to guide the optimization of breeding and production of small groups of beef cattle.

[0010] A further improvement is that the target group has a preset number of 10 to 20 head of cattle, that is, 10 to 20 cattle per pen, which facilitates module deployment and group management.

[0011] A further improvement is that the video waste material monitoring module includes an image acquisition unit, a semantic segmentation unit, and a waste material weight conversion unit, wherein: The image acquisition unit is used to acquire images above the feeding trough at regular intervals, using a fixed camera to take pictures at a preset frequency (such as every 2 hours); The semantic segmentation unit is used to perform pixel-level classification of the images acquired by the image acquisition unit using deep learning semantic segmentation algorithms, identify the residual material area and calculate the residual material area ratio. It uses algorithms such as DeepLabV3+ or U-Net to output the residual material area ratio (%). The residual material weight conversion unit is used to calculate the residual material weight based on the total volume of the feeding trough, the feed bulk density, the compaction coefficient, and the residual material area ratio. The calculation formula is: residual material weight = residual material area ratio × total volume of the trough × feed bulk density × compaction coefficient. The percentage area is converted into physical weight (kg), and the compaction coefficient is used to correct the feed bulk density.

[0012] A further improvement is that the feed conversion ratio (FCR) calculation module includes a feed intake calculation unit, a water consumption verification unit, a weight gain acquisition unit, and a FCR calculation unit, wherein: The feed intake calculation unit is used to calculate the actual feed intake of the target group. The calculation formula is: actual feed intake = TMR feed amount - residual feed weight. By simple difference, the actual daily feed consumption (kg) of the small group is obtained. The water intake verification unit is used to establish a correlation verification model between water intake and dry matter intake. When the deviation between an individual's water intake and actual intake exceeds a preset threshold, a correction signal is generated and the calculation parameters of the actual intake are adjusted. The preset threshold is, for example, 15%. If it exceeds the threshold, an abnormal alarm is triggered or the intake is recalculated. The weight gain acquisition unit is used to acquire the group weight data of the target group according to a preset weighing cycle and calculate the total weight gain of the group. The preset cycle is, for example, every 30 days. The group weight is acquired through the channel weighing system. The feed conversion ratio (FCR) calculation unit is used to calculate the FCR of the target population. The calculation formula is: FCR = Total actual feed intake within the preset period / Total weight gain of the population within the preset period. The lower the value, the higher the feed conversion efficiency.

[0013] A further improvement is that the decision support module includes a database unit, a comparative analysis unit, and a suggestion output unit, wherein: The database unit is used to store the feed conversion ratio data of the target group. The feed conversion ratio data is indexed according to raw material identifier, formula identifier, variety identifier, and individual identifier, and supports multi-dimensional data mining and comparative analysis. The comparative analysis unit is used to compare the feed conversion ratio (FCR) data corresponding to different raw material labels, different formula labels, different variety labels, and different individual labels within the same preset growth stage, thereby achieving four-level screening of raw materials, formulas, varieties, and individuals. The suggested output unit is used to output a suggestion to eliminate raw materials when the feed conversion ratio of raw materials is higher than a preset raw material elimination threshold, to output a suggestion to adopt the formula when the feed conversion ratio of the formula is lower than a preset formula preference threshold, to output a suggestion to prioritize the propagation of the variety when the feed conversion ratio of the variety is lower than a preset variety preference threshold, and to output a suggestion to eliminate the individual when the feed conversion ratio of an individual is higher than a preset low-yield elimination threshold. The preset low-yield elimination threshold is, for example, 120% of the average feed conversion ratio of the population.

[0014] A further improvement is that the drinking water weighing module includes a weighing sensor unit, an individual identification unit, and a drinking water volume recording unit, wherein: The weighing sensor unit is used to collect the weight change value of the drinking trough in real time. It uses a high-precision sensor to record the weight loss of each drinking session. Individual identification unit is used to identify individual cattle drinking water using RFID or visual recognition technology. It uses ear tag RFID or facial recognition to associate water consumption with specific cattle. The water consumption recording unit is used to associate and store the weight change values ​​collected by the weighing sensor unit with the corresponding individual identification and timestamp to generate individual water consumption data, so as to form a water consumption curve for each cow.

[0015] A further improvement is that the TMR fixed-point feeding module includes a weighing and feeding unit and a feeding information recording unit, wherein: The weighing and feeding unit is used to quantitatively feed the target group into the feeding trough according to the preset static error range and the preset dynamic error range through the weighing sensor configured on the TMR feed truck. The static error is ≤0.15‰ and the dynamic error is ≤±3‰, ensuring feeding accuracy. The feeding information recording unit is used to record the amount of feed, type and weight of roughage, proportion of concentrate, content of additives, and feeding time for each feeding. The recorded information is also linked to the current target population identifier to fully trace the formula composition of each feeding in each column.

[0016] A method for precise feeding and intelligent evaluation of feed conversion ratio in small groups of beef cattle includes the following steps: S1. Individual water intake collection The individual water consumption data and timestamp of each beef cattle in the target group were collected by the water weighing module to obtain the water consumption of each beef cattle (L / time), which served as the benchmark for feed intake verification. S2, Precise Feeding and Recording The TMR fixed-point feeding module accurately feeds the target group's feeding troughs according to the preset formula, and records the feeding amount, formula information and feeding time. By recording the total amount and composition of each feeding, it ensures that the amount of feed consumed by beef cattle each time has an initial value. S3. Leftover feed identification and feed intake calculation The video feed monitoring module collects images of the feeding trough at regular intervals, and uses a deep learning semantic segmentation algorithm to identify the proportion of feed area. The proportion of feed area is converted into feed weight, and the actual feed intake is calculated based on the amount of feed and the weight of feed. S4, Group Weight Gain Measurement The weight of the target group is measured according to the preset weighing cycle, and the total weight gain of the group within the preset cycle is calculated. For example, the preset cycle is 30 days. The weight gain of the group = the weight at the end of the period - the weight at the beginning of the period. S5. Calculation and Verification of Feed Conversion Ratio The feed conversion ratio (FCR) of the target population is calculated using the actual feed intake and total weight gain of the population through the feed conversion ratio calculation module. The actual feed intake is cross-validated using individual water intake data. FCR = total feed intake / total weight gain. If the deviation between water intake and feed intake exceeds 15%, a correction is triggered. S6, Decision Output The decision support module stores the feed conversion ratio data of the target group into the database, performs four decision analyses: raw material selection, formula optimization, breed selection, and culling of low-yielding cattle, and outputs corresponding suggestions. The final output is: culling inferior raw materials, recommending efficient formulas, selecting superior breeds, and accurately culling low-yielding cattle.

[0017] The beneficial effects of this invention are as follows: By refining the feed conversion ratio (FCR) evaluation from a broad group level to a small group or even individual level, the accuracy of the evaluation is significantly improved. Video monitoring of leftover feed replaces manual trough cleaning and weighing, enabling real-time automated collection of leftover feed data, greatly increasing monitoring frequency and data reliability. Furthermore, the system integrates three modules: drinking water weighing, precise feeding, and video leftover feed monitoring, forming a complete closed loop of "precise feeding - real-time monitoring - FCR evaluation - optimization decision-making." Cross-validation of feed intake using drinking water data effectively corrects calculation errors and solves the problem of data silos. In addition, by establishing a multi-dimensional FCR database, it can support four key decisions: raw material selection, formula optimization, breed selection, and culling of low-producing cattle, helping farmers improve production efficiency from multiple levels. Simultaneously, by accurately identifying and culling low-producing cattle, it can effectively improve the overall feed conversion efficiency of the herd and reduce feed waste during the breeding process. This system is flexible in deployment, suitable for both large-scale farms and small experimental stations, and has self-calibration capabilities for the leftover feed weight model, maintaining high estimation accuracy during long-term operation. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the framework structure of the intelligent evaluation system for precise feeding and feed conversion ratio of small groups of beef cattle according to the present invention; Figure 2 This is a flowchart illustrating the method for precise feeding of small groups of beef cattle and intelligent evaluation of feed conversion ratio according to the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] It should be noted that the technical means not described in detail in the following embodiments are all conventional means in the art, are not the key points of the invention, and will not be elaborated upon.

[0021] Example 1 according to Figure 1 As shown, this embodiment provides a precision feeding and feed conversion ratio (FCR) intelligent evaluation system for small groups of beef cattle, using a 10,000-head beef cattle farm as the application scenario. The cattle herd is divided into multiple small groups, each consisting of 10 to 20 cattle (20 cattle per pen in this embodiment). The system includes a water and weighing module, a TMR (Total Meat Retention) fixed-point feeding module, a video feed residue monitoring module, a FCR calculation module, and a decision support module. All modules are connected to a cloud server via industrial Ethernet or a 4G wireless network. The server deploys a MySQL database and a deep learning inference engine. Specifically: The drinking water weighing module is installed in each small group of cattle sheds and includes the following units: Weighing sensor unit: A cantilever beam weighing sensor with a range of 200kg and an accuracy of ±0.02kg is installed at the four corners of the bottom of the water trough. The sensor collects weight change values ​​in real time at a sampling rate of 100Hz. A drinking event is determined when the weight drops by more than 0.5kg and lasts for more than 3 seconds.

[0022] Individual identification unit: Each beef cattle wears a low-frequency RFID ear tag (134.2kHz). An RFID reader antenna is installed above the water trough, with a reading distance of 0.5m. When a cattle approaches the water trough, the reader reads the ear tag ID to identify the individual.

[0023] Water consumption recording unit: This unit associates the weight difference before and after drinking collected by the weighing sensor with the corresponding individual identification and timestamp to generate an individual water consumption record in the format: {cattle ID, drinking start time, drinking end time, water consumption (kg)}. The data is uploaded to the cloud server in real time.

[0024] The TMR point feeding module includes the following units: Weighing and Feeding Unit: The TMR feed truck is equipped with three independent weighing sensors (each with a static error ≤0.15‰ and a dynamic error ≤±3‰). A touchscreen terminal is located in the driver's cab. After the operator selects the target pen number and formula number, the system automatically controls the auger speed and discharge gate opening to achieve quantitative feeding. During the feeding process, the sensors monitor the weight changes in the hopper in real time and automatically stop when the preset feeding amount is reached.

[0025] Feeding Information Recording Unit: After each feeding, the system automatically records: feed amount (kg), type and weight of roughage (e.g., 120kg corn silage, 30kg sheep hay), concentrate ratio (e.g., 60% corn flour, 20% soybean meal, 15% wheat bran, 5% premix), additive content (e.g., 2kg sodium bicarbonate), and feeding time (accurate to the second). All information is uploaded to the cloud after being linked to the current small group identifier (pen number).

[0026] The video waste material monitoring module includes the following units: Image acquisition unit: A 2-megapixel network camera (1920×1080 resolution, 4mm focal length) is installed 1.5m directly above each feeding trough, covering the entire length of the feeding trough (3m). The camera automatically takes one image every 2 hours, for a total of 12 times per day. The images are compressed in JPEG format and uploaded to the edge computing server.

[0027] Semantic segmentation unit: The edge server loads a pre-trained DeepLabV3+ semantic segmentation model (with a ResNet-50 backbone). The model takes a 512×512 pixel image as input and outputs the class probability for each pixel, where all pixels in the "leftovers" category constitute the leftovers region. The percentage of leftovers region pixels out of the total number of pixels in the feeding trough region is calculated to obtain the leftovers area percentage (%).

[0028] Leftover material weight conversion unit: The feeding trough is a U-shaped concrete structure with a total volume of 0.8m³. 3 The feed bulk density was determined to be 0.65 t / m³ in the laboratory. 3 (i.e., 650kg / m) 3 The compaction coefficient is calibrated to 0.9 based on feed moisture and particle size. The formula for calculating the weight of residual material is: Residual material weight (kg) = Residual material area percentage (%) × 0.8m 3 ×650kg / m 3 ×0.9.

[0029] If the area of ​​leftover material is calculated to be 25% in a certain shooting, then the weight of leftover material = 0.25 × 0.8 × 650 × 0.9 = 117 kg.

[0030] The feed conversion ratio calculation module is deployed on a cloud server and includes the following units: Feed intake calculation unit: At 0:00 every day, the system summarizes the total amount of TMR feed input for the day and the average weight of residual feed at each time point of the video residual feed monitoring module, and calculates the actual feed intake for the day according to the following formula: Actual feed intake (kg) = TMR feed input (kg) - Average weight of remaining feed (kg) For example, if the total amount of feed fed to a certain pen on a given day is 400 kg, and the average weight of the remaining feed at different times is 80 kg, then the actual amount of feed consumed is 320 kg.

[0031] Water intake validation unit: Establish a linear regression model between water intake and dry matter intake. Historical data shows that for every 1 kg of water consumed by beef cattle, the corresponding dry matter intake is approximately 0.85 kg (R0). 2 =0.87). The system calculates the ratio of an individual's actual feed intake (estimated by allocating feed intake to water intake according to the group's proportion) to its actual water intake daily. When this ratio deviates from the group average by more than ±15% for an individual, a correction signal is generated, prompting a review of the cattle's health status or a check of the RFID reading accuracy.

[0032] Weight gain acquisition unit: Every 30 days (preset weighing cycle), a channel-type electronic floor scale (2000kg capacity, ±1kg accuracy) is used to weigh the small group. The total weight at the end of the period and the total weight at the beginning of the period are recorded, and the total weight gain of the group (kg) is calculated.

[0033] Feed conversion ratio (FCR) calculation unit: Calculates the feed conversion ratio (FCR) of the small population within a preset period using the following formula: Feed conversion ratio = Total actual feed intake within the preset period (kg) / Total weight gain of the population within the preset period (kg) For example: If the total actual feed intake of a certain pen in 30 days is 9600 kg and the total weight gain of the group is 1500 kg, then FCR = 9600 / 1500 = 6.4.

[0034] The decision support module includes the following units: Database Unit: A MySQL database is used to store feed conversion ratio (FCR) data. The data table contains the following fields: pen number, feed identifier (supplier batch number), formula identifier (formula number), breed identifier (Simmental / Angus, etc.), individual identifier (cattle RFID), start date, end date, total feed intake, total weight gain, and FCR.

[0035] Comparative analysis unit: within the same preset growth stage (e.g., 300-350kg stage): Analysis of variance (ANOVA) was performed on the feed conversion ratio data of different raw materials to rank and screen the optimal raw materials.

[0036] Paired t-tests were performed on the feed conversion ratio data of different formulations to determine the quality of the formulations.

[0037] Multiple comparisons were performed on the feed conversion ratio (FCR) data of different varieties to select the variety with the lowest FCR.

[0038] Within the same small group, individuals with different FCR (feed intake estimated by allocating feed intake based on water intake ratio) were sorted to identify individuals with high FCR.

[0039] Suggested output unit: Automatically generates suggestions based on preset thresholds. When the feed conversion ratio of a certain raw material is higher than the preset raw material elimination threshold (e.g., 7.0), the suggestion to "eliminate the raw material" is output.

[0040] When the feed conversion ratio of a certain formula is lower than the preset formula preference threshold (e.g., 6.0), the "use this formula" suggestion is output.

[0041] When the feed conversion ratio of a certain variety is lower than the preset variety preference threshold (e.g., 6.5), the suggestion to "prioritize the propagation of this variety" is output.

[0042] When an individual's feed conversion ratio exceeds the preset low-yield culling threshold (120% of the population average FCR), the suggestion to "cull the individual" is output.

[0043] Example 2 Regarding the intelligent evaluation system for precision feeding and feed conversion ratio in small groups of beef cattle provided in Example 1, see [link to example]. Figure 2 This embodiment provides a method for precise feeding and intelligent evaluation of feed conversion ratio in small groups of beef cattle. Taking a pen (20 Simmental cattle, weighing 350-400kg) in a 10,000-head beef cattle farm as an example, the specific implementation is as follows: S1. Individual water intake collection The water consumption data and timestamps for each beef cattle in the target group are collected using a water weighing module.

[0044] Specific operation: The system runs continuously during the 30-day evaluation period. Each time a cow drinks water, an RFID reader reads its ear tag ID, and a weighing sensor records the weight difference before and after drinking. The system records each drinking event; for example, one cow started drinking at 08:12:30 on March 1, 2025, and finished at 08:15:20, drinking 3.2 kg of water; another cow started drinking at 08:20:10 on March 1, 2025, and finished at 08:22:45, drinking 2.8 kg of water. All data is stored in a cloud database in real time, generating a daily cumulative water consumption record for each cow.

[0045] S2, Precise Feeding and Recording The TMR (Tracking Management) fixed-point feeding module accurately feeds the target group's feeding troughs according to the preset formula, and records the feeding amount, formula information, and feeding time.

[0046] Specific operations: Feeding is conducted twice daily, at 8:00 AM and 4:00 PM. The operator selects the current pen and corresponding feed formula on the TMR feed truck terminal. The system controls the feeding and records: Feed amount = 420 kg (210 kg in the morning + 210 kg in the afternoon), roughage: 260 kg corn silage, 40 kg hay; concentrate: 72 kg corn flour, 24 kg soybean meal, 18 kg wheat bran, 6 kg premix; additive: 0.1 kg monensin. Feeding time is recorded accurate to the second. After data upload, it is linked to the pen's identifier.

[0047] S3. Leftover feed identification and feed intake calculation The video feed monitoring module periodically collects images of the feeding trough, uses a deep learning semantic segmentation algorithm to identify the proportion of feed area, converts the proportion of feed area into feed weight, and calculates the actual feed intake based on the amount of feed fed and the weight of feed.

[0048] Specific operation: The camera takes pictures every 2 hours, for a total of 12 images per day. The edge server runs a DeepLabV3+ model to identify the percentage of residual material area at each time point and convert it into residual material weight. For example: at 8:00 AM before feeding, the residual material area percentage is 5%, and the residual material weight is 23.4 kg; at 10:00 AM, the residual material area percentage is 65%, and the residual material weight is 304.2 kg; at 12:00 PM, the residual material area percentage is 40%, and the residual material weight is 187.2 kg; at 2:00 PM, the residual material area percentage is 25%, and the residual material weight is 117.0 kg; at 4:00 PM before feeding, the residual material area percentage is 10%, and the residual material weight is 46.8 kg; at 6:00 PM, the residual material area percentage is 70%, and the residual material weight is 327.6 kg; at 8:00 PM, the residual material area percentage is 45%, and the residual material weight is 210.6 kg; and so on for subsequent time points. The average residual material weight for the day is the sum of the residual material weights at each time point divided by 12, which is approximately 165 kg. If the total feed intake for the day is 420 kg, then the actual feed intake for the day is 420 - 165 = 255 kg. The total actual feed intake over 30 consecutive days is 7650 kg.

[0049] S4, Group Weight Gain Measurement The weight of the target group is measured according to the preset weighing cycle, and the total weight gain of the group within the preset cycle is calculated.

[0050] Specific procedures: On the first day of the evaluation period (March 1, 2025), the total weight of the 20 cattle in the pen was weighed using a walkway scale and found to be 7200 kg. On the 30th day (March 30, 2025), the total weight was weighed again and found to be 8550 kg. The total weight gain of the herd = 8550 - 7200 = 1350 kg.

[0051] S5. Calculation and Verification of Feed Conversion Ratio The feed conversion ratio (FCR) of the target population is calculated using the actual feed intake and total population weight gain through the feed conversion ratio calculation module, and the actual feed intake is cross-validated using individual water intake data.

[0052] Specific operations: Feed conversion ratio calculation: FCR = Total actual feed intake (7650kg) / Total weight gain of the group (1350kg) = 5.67.

[0053] Cross-validation: The system calculates the average daily water intake of each cow during the period. For example, one cow drinks an average of 22.5L of water per day, while another cow drinks an average of 18.0L per day. The feed intake of the entire group is allocated according to the water intake ratio to estimate the individual feed intake. If an individual's estimated feed intake deviates from the group average by more than 15% (a preset threshold), the system outputs a validation anomaly alarm. This period did not exceed the limit, so the data is reliable.

[0054] S6, Decision Output The decision support module stores the feed conversion ratio data of the target group into the database, performs four decision analyses: raw material selection, formula optimization, breed selection, and culling of low-yielding cattle, and outputs corresponding suggestions.

[0055] Specific operations: Raw material screening: The system compares the feed conversion ratio (FCR) data of corn silage from different sources. For example, if the FCR of the current batch of corn silage is 5.67 and the FCR of another batch of corn silage is 6.23, the system will output a suggestion to "eliminate the batch with the higher FCR".

[0056] Formula optimization: The system compares the feed conversion ratio (FCR) data of different formulas. For example, if the FCR of a pen using the current formula is 5.67, and the FCR of a pen using another formula is 6.05, the system will output a suggestion to "use formulas with a lower FCR across the board".

[0057] Variety selection: Compare the feed conversion ratio (FCR) data of different varieties. For example, the average FCR of the Simmental variety is 5.70, while that of the Angus variety is 6.50. The system outputs the suggestion of "prioritizing the propagation of Simmental".

[0058] Low-producing cattle culling: Based on water consumption and body condition scores, the system identified a cow in this pen whose average daily water consumption was only 12 kg (group average 18 kg). Its estimated feed ration (FCR) was 7.2, which is 120% higher than the group average FCR (5.67) (i.e., 6.80). The system output a "cull this individual" recommendation. The farm culled the cow ahead of schedule to avoid further feed waste.

[0059] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A precise feeding and feed conversion ratio intelligent evaluation system for small groups of beef cattle, characterized in that, include: The drinking water weighing module is used to record the individual water consumption of each beef cattle in the target group and the corresponding timestamp in real time; The TMR fixed-point feeding module is used to accurately feed the target group's feeding troughs according to a preset formula, and record the feeding amount, formula information and feeding time. The video feed monitoring module is used to periodically collect images of the feeding trough, identify the feed waste area through a deep learning semantic segmentation algorithm, calculate the feed waste area ratio, and convert the feed waste area ratio into feed waste weight. The feed conversion ratio (FCR) calculation module receives individual water intake recorded by the water weighing module, feed intake and formula information recorded by the TMR fixed-point feeding module, and residual feed weight calculated by the video residual feed monitoring module. It also obtains weight gain data of the target group measured periodically, calculates the actual feed intake based on the feed intake and residual feed weight, calculates the FCR of the target group based on the actual feed intake and weight gain data, and performs cross-validation correction on the actual feed intake using individual water intake. The decision support module is used to establish a feed conversion ratio database for the target population, perform multi-dimensional comparative analysis based on the feed conversion ratio data of the target population, and output suggestions for raw material selection, formula optimization, breed selection, and culling of low-yielding cattle.

2. The intelligent evaluation system for precise feeding and feed conversion ratio of small groups of beef cattle according to claim 1, characterized in that: The target group is expected to have 10 to 20 head of beef cattle.

3. The intelligent evaluation system for precise feeding and feed conversion ratio of small groups of beef cattle according to claim 1, characterized in that: The video waste monitoring module includes: The image acquisition unit is used to periodically acquire images of the area above the feeding trough. The semantic segmentation unit is used to perform pixel-level classification of the images acquired by the image acquisition unit using a deep learning semantic segmentation algorithm, identify the waste material area and calculate the proportion of waste material area. The residual material weight conversion unit is used to calculate the residual material weight based on the total volume of the feeding trough, the feed bulk density, the compaction coefficient, and the residual material area ratio. The calculation formula is: Residual material weight = Residual material area ratio × Total volume of the trough × Feed bulk density × Compaction coefficient.

4. The intelligent evaluation system for precise feeding and feed conversion ratio of small groups of beef cattle according to claim 1, characterized in that: The feed conversion ratio calculation module includes: The feed intake calculation unit is used to calculate the actual feed intake of the target population. The calculation formula is: Actual feed intake = TMR feed input - Residual feed weight; The water intake verification unit is used to establish a correlation verification model between water intake and dry matter intake. When the deviation between an individual's water intake and actual intake exceeds a preset threshold, a correction signal is generated and the calculation parameters of the actual intake are adjusted. The weight gain acquisition unit is used to acquire the group weight data of the target group according to a preset weighing cycle and calculate the total weight gain of the group. The feed conversion ratio calculation unit is used to calculate the feed conversion ratio of the target population. The calculation formula is: Feed conversion ratio = Total actual feed intake within the preset period / Total weight gain of the population within the preset period.

5. The intelligent evaluation system for precise feeding and feed conversion ratio of small groups of beef cattle according to claim 1, characterized in that: The decision support module includes: A database unit is used to store feed conversion ratio data for the target population, and the feed conversion ratio data is indexed according to raw material identifier, formula identifier, variety identifier, and individual identifier; The comparative analysis unit is used to compare the feed conversion ratio (FCR) data corresponding to different raw material labels, different formula labels, different variety labels, and different individual labels within the same preset growth stage. The suggested output unit is used to output a suggestion to eliminate raw materials when the feed conversion ratio of raw materials is higher than the preset raw material elimination threshold, to output a suggestion to adopt the formula when the feed conversion ratio of the formula is lower than the preset formula preference threshold, to output a suggestion to prioritize the propagation of the variety when the feed conversion ratio of the variety is lower than the preset variety preference threshold, and to output a suggestion to eliminate the individual when the feed conversion ratio of an individual is higher than the preset low-yield elimination threshold.

6. The intelligent evaluation system for precise feeding and feed conversion ratio of small groups of beef cattle according to claim 1, characterized in that: The drinking water weighing module includes: The weighing sensor unit is used to collect the weight change value of the drinking trough in real time; Individual identification unit, used to identify individual cattle drinking water using RFID or visual identification technology; The drinking water recording unit is used to associate and store the weight change values ​​collected by the weighing sensor unit with the corresponding individual identification and timestamp to generate individual drinking water data.

7. The intelligent evaluation system for precise feeding and feed conversion ratio of small groups of beef cattle according to claim 1, characterized in that: The TMR fixed-point feeding module includes: The weighing and feeding unit is used to quantitatively feed the target group into the feeding trough according to the preset static error range and the preset dynamic error range by using the weighing sensor configured on the TMR feed truck. The feeding information recording unit is used to record the amount of feed, type and weight of roughage, proportion of concentrate, content of additives, and feeding time for each feeding, and to bind the recorded information to the current target group identifier.

8. A method for precise feeding and intelligent evaluation of feed conversion ratio in small groups of beef cattle according to any one of claims 1-7, characterized in that, Includes the following steps: S1. Collect individual water consumption data and timestamps for each beef cattle in the target group through the water weighing module; S2. The TMR fixed-point feeding module accurately feeds the target group's feeding troughs according to the preset formula, and records the feeding amount, formula information and feeding time. S3. The video feed monitoring module periodically collects images of the feeding trough, uses a deep learning semantic segmentation algorithm to identify the proportion of feed area, converts the proportion of feed area into feed weight, and calculates the actual feed intake based on the amount of feed and the weight of feed. S4. Measure the weight of the target group according to the preset weighing cycle and calculate the total weight gain of the group within the preset cycle; S5. The feed conversion ratio (FCR) of the target population is calculated using the actual feed intake and total weight gain of the population through the feed conversion ratio calculation module, and the actual feed intake is cross-validated using individual water consumption data. S6. The feed conversion ratio data of the target group is stored in the database through the decision support module. Four decision analyses are performed: raw material selection, formula optimization, breed selection, and culling of low-yielding cattle. Corresponding suggestions are then output.

Images