Boar mating control method and system based on deep learning network

Through the combination of deep learning network and NSGA-II algorithm, the problem of low breeding decision-making efficiency and accuracy of traditional breeding pigs is solved, and accurate prediction of the traits of offspring pigs is achieved and the combination selection of breeding combinations is improved, which is the benefits of pig farming industry and accuracy.

CN119005625AInactive Publication Date: 2024-11-22ZHENGDA KANGDI SHEKOU CO LTD +1
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411179479.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-27
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Traditional breeding decisions for breeders rely on the experience and intuition of breeders, making it difficult to accurately predict the growth performance, health status and economic benefits of offspring, especially when pursuing specific traits, efficiency and accuracy are limited.

Method used

The progeny trait index prediction model based on deep learning network and the NSGA-II multi-objective optimization algorithm are used to predict progeny traits by obtaining breeding demand data and breeding performance index data of boars and sows. , and through clustering processing and dynamic update data, selecting breeding combinations that best meet breeding needs.

Benefits of technology

It significantly improves the efficiency and economic benefits of the pig farming industry, accurately predicts the traits of offspring pigs, enhances the accuracy and practicality of predictions, reduces the amount of calculations, and improves the generalization ability of the model.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119005625A_ABST
    Figure CN119005625A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of artificial intelligence, in particular to a boar mating control method and system based on a deep learning network. The method comprises the following steps: acquiring breeding demand data, and acquiring breeding performance index data of breeding boars and breeding sows in different combinations; inputting the breeding performance index data into a trained offspring character index prediction model, and outputting offspring character data of a plurality of different growth stages, the offspring character index prediction model being obtained based on deep learning network training; and based on the breeding demand data, screening the multiple offspring character data, and determining a breeding pig combination corresponding to the screened data as a breeding paired breeding pig combination. By adopting the combination of the full-connection neural network and the GRU network, the model can learn a complex genetic mode and growth dynamics from the breeding performance index data of the breeding pigs, so that the characters of the offspring pigs are accurately predicted, and the breeding combination meeting the breeding requirements is screened out.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a breeding control method and system for breeding pigs based on a deep learning network. Background Art

[0002] In the traditional breeding industry, breeding decisions for sows often rely on the experience and intuition of breeders, as well as simple genetic principles. This process usually involves an assessment of the physical characteristics, health status, reproductive history, and genetic background of boars and sows. However, this traditional method has obvious limitations when faced with the complexity of genetic traits and the dynamic changes of environmental factors. It is difficult to accurately predict the growth performance, health status, and economic benefits of offspring, especially when pursuing specific traits (such as high growth rate, excellent meat quality, or efficient feed conversion rate), its efficiency and accuracy are severely limited. With the development of artificial intelligence technology and its application in the sow breeding industry, how to achieve sow breeding control based on deep learning networks has become a technical problem that needs to be solved urgently. Summary of the invention

[0003] In order to solve at least one of the technical problems raised above, the present invention provides a breeding control method for sows based on a deep learning network. By utilizing an offspring trait index prediction model based on deep learning and the NSGA-II multi-objective optimization algorithm, the present invention can accurately predict the traits of offspring pigs, and screen out the breeding combination of sows that best meets the breeding needs, thereby significantly improving the efficiency and economic benefits of the pig farming industry.

[0004] In one aspect, the present invention provides a method for controlling breeding of sows based on a deep learning network, the method comprising: Obtain breeding demand data and breeding performance index data of different combinations of boars and sows; Input the breeding performance index data into a trained offspring trait index prediction model, and output offspring trait data at multiple different growth stages, wherein the offspring trait index prediction model is obtained based on deep learning network training; Based on the breeding demand data, a plurality of offspring trait data are screened, and a breeding pig combination corresponding to the screened data is determined as a breeding pairing breeding pig combination.

[0005] Preferably, the offspring trait index prediction model includes: a fully connected neural network and a GRU network; The fully connected neural network receives the breeding performance index data, the number of neurons in the input layer of the fully connected neural network is equal to the number of features of the breeding performance index data, and the offspring trait data of the newborn offspring is output after hidden layer processing, and the number of neurons in the output layer of the fully connected neural network is equal to the number of features of the offspring trait data; The offspring trait data of the newly born offspring are input into the GRU network, and the offspring trait data at different growth stages are output.

[0006] Preferably, before obtaining the breeding performance index data of different combinations of boars and sows, the method further includes: Clustering the breeding performance index data of the sows to obtain multiple sow categories; Calculate the arithmetic mean of the breeding performance index data within the same sow category, and use the calculation result as the breeding performance index data corresponding to the sow category.

[0007] Preferably, the breeding performance index data of the sows are clustered to obtain a plurality of sow categories, including: Setting clustering parameters, including neighborhood radius and minimum number of points; Based on the clustering parameters, the breeding performance index data of the sows are clustered and analyzed according to the DBSCAN algorithm to obtain the sow category.

[0008] Preferably, the breeding performance index data include the age of the sows, the weight of the sows, the average weight of the offspring pigs at birth, the average weight of the offspring pigs during the growth stage, the average feed-to-meat ratio of the offspring pigs and the proportion of offspring pig types; the offspring trait data include the weight of the offspring pigs at birth, the weight of the offspring pigs during the growth stage, the feed-to-meat ratio of the offspring pigs and the offspring pig types; the offspring pig types include the sow types and the meat pig types.

[0009] Preferably, data from a recent preset number of matings are used to calculate the mean weight of the offspring pigs at birth, the mean weight of the offspring pigs at the growth stage and the mean feed-to-meat ratio of the offspring pigs in the breeding performance index data.

[0010] Preferably, the method further comprises training the offspring trait index prediction model, specifically comprising: Acquire breeding performance index data of multiple different combinations of pig breeds, and corresponding offspring trait data at different growth stages, to obtain training data; Using the breeding performance index data as input of the fully connected neural network and the offspring trait data at the early stage as labels, the fully connected neural network is trained; After the fully connected neural network training is completed, the GRU network is trained according to the output of the fully connected neural network and the offspring trait data.

[0011] In a second aspect, the present invention provides a sow breeding control system based on a deep learning network, the system comprising: The data acquisition module is used to obtain breeding demand data and breeding performance index data of different combinations of boars and sows; An offspring data prediction module is used to input the breeding performance index data into a trained offspring trait index prediction model, and output offspring trait data at multiple different growth stages, wherein the offspring trait index prediction model is obtained based on deep learning network training; The breeding pig combination determination module is used to screen the multiple offspring trait data based on the breeding demand data, and determine the breeding pig combination corresponding to the screened data as the breeding pairing breeding pig combination.

[0012] Preferably, the offspring trait index prediction model includes: a fully connected neural network and a GRU network; The fully connected neural network receives the breeding performance index data, the number of neurons in the input layer of the fully connected neural network is equal to the number of features of the breeding performance index data, and the offspring trait data of the newborn offspring is output after hidden layer processing, and the number of neurons in the output layer of the fully connected neural network is equal to the number of features of the offspring trait data; The offspring trait data of the newly born offspring are input into the GRU network, and the offspring trait data at different growth stages are output.

[0013] Preferably, the breeding performance index data include the age of the sows, the weight of the sows, the average weight of the offspring pigs at birth, the average weight of the offspring pigs during the growth stage, the average feed-to-meat ratio of the offspring pigs and the proportion of offspring pig types; the offspring trait data include the weight of the offspring pigs at birth, the weight of the offspring pigs during the growth stage, the feed-to-meat ratio of the offspring pigs and the offspring pig types; the offspring pig types include the sow types and the meat pig types.

[0014] The beneficial effects of the present invention are: (1) The present invention adopts a combination of a fully connected neural network and a GRU network. The model can learn complex genetic patterns and growth dynamics from the breeding performance index data of sows, thereby accurately predicting the traits of offspring pigs, including key indicators such as growth rate, feed conversion efficiency and type ratio; (2) The present invention clusters the breeding performance index data of sows, which not only reduces the amount of calculation but also improves the generalization ability of the model, making the prediction results more stable and reliable; (3) The present invention uses recent mating data to update the model, thereby ensuring that the prediction model can adapt to the natural changes in the breeding performance of sows, thereby enhancing the accuracy and practicality of the prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the background technology, the drawings required for use in the embodiments of the present invention or the background technology will be described below.

[0016] The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and, together with the specification, are used to illustrate the technical solutions disclosed in the present invention.

[0017] Figure 1 A schematic flow chart of a method for controlling breeding of sows based on a deep learning network provided in an embodiment of the present invention; Figure 2 A schematic diagram of the structure of a sow breeding control system based on a deep learning network provided in an embodiment of the present invention; Figure 3 A schematic diagram of the structure of an offspring trait index prediction model provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or devices.

[0020] The term "and / or" herein is only a description of the association relationship of the associated objects, indicating that there may be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" herein represents any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C can represent including any one or more elements selected from the set consisting of A, B, and C.

[0021] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0022] In addition, in order to better illustrate the present invention, numerous specific details are provided in the following specific embodiments. It should be understood by those skilled in the art that the present invention can be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present invention.

[0023] Existing breeding screening methods often rely on the breeder's experience and intuition, and it is difficult to accurately predict the growth performance, health status and economic benefits of offspring. Especially when pursuing specific traits, its efficiency and accuracy are severely limited.

[0024] Example 1, see Figure 1 The present invention provides a breeding control method for sows based on a deep learning network, the method comprising: S100, obtaining breeding demand data, and obtaining breeding performance index data of different combinations of boars and sows; Collect real-time data from the database or through IoT devices, including the age, weight, historical fertility records of the breeding pigs, the average initial weight of the offspring pigs, the average weight during the growth stage, the average feed conversion rate (feed-to-meat ratio), and the type ratio of the offspring pigs. These data may come from the farm's historical records or from real-time feedback from sensors and smart weighing systems. Breeding demand data include, but are not limited to, the type ratio of the target offspring pigs (breeding pigs and meat pigs), the expected growth rate, or feed efficiency indicators.

[0025] S200, inputting the breeding performance index data into a trained offspring trait index prediction model, and outputting offspring trait data at multiple different growth stages, wherein the offspring trait index prediction model is obtained based on deep learning network training; In this embodiment, the breeding performance index data is standardized to eliminate the impact of dimension, and then input into a pre-trained offspring trait index prediction model (based on a deep learning network). The offspring trait index prediction model is a composite structure, including a fully connected layer for processing static features and a GRU (gated recurrent unit) layer for capturing time series changes. The training process of the offspring trait index prediction model involves a large amount of breeding result data of known breeding pig combinations. These data are used to adjust the weights of the neural network to achieve accurate prediction of offspring traits.

[0026] S300, based on the breeding demand data, screening the plurality of offspring trait data, and determining the breeding pig combination corresponding to the screened data as the breeding pair breeding pig combination.

[0027] In order to select the best combination from the many possible breeding pig pairings, we can use multi-objective optimization algorithms, which can optimize multiple conflicting objectives at the same time, for example, maximizing the growth rate and feed conversion efficiency of offspring pigs while maintaining a specific ratio of breeding pig types.

[0028] The non-dominated sorting genetic algorithm (NSGA-II) is an evolutionary algorithm applied to multi-objective optimization problems. It evaluates the quality of solutions based on the concept of "non-domination", that is, whether a solution is not inferior to another solution in all objectives, but is better in at least one objective. NSGA-II updates the population through fast non-dominated sorting and crowding distance calculation, and can effectively find the Pareto optimal frontier.

[0029] The NSGA-II algorithm is used to solve the breeding pairing problem. First, the objective function is defined as the maximization of the average growth rate and feed conversion efficiency of offspring pigs, as well as the minimum deviation of the offspring pig sow ratio. A population containing N random sow pairs is initialized, and then the following steps are performed: Apply fast non-dominated sorting to stratify the population; Calculate the crowding distance of each individual as a measure of the density of solutions around the individual; Generate offspring populations using binary tournament selection, uniform crossover, and polynomial mutation for genetic operations; Merge the parent generation and the offspring generation, perform non-dominated sorting and crowding distance calculation again, and select the first N individuals as the next generation population; Repeat the above steps until the predetermined number of iterations is reached.

[0030] In the last generation of the population, the solutions on the Pareto optimal frontier represent the best breeding pig pairing plan to meet breeding needs.

[0031] Through the offspring trait index prediction model based on deep learning network pre-training, the traits of offspring pigs can be predicted more accurately, and the prediction results of different breeding combinations and breeding demand data can be matched based on the NSGA-II algorithm to screen out the optimal breeding combination that meets the breeding demand data conditions. Furthermore, the automated decision-making process greatly reduces the need for human intervention, improves breeding efficiency, and also reduces labor costs and error rates.

[0032] Preferably, the offspring trait index prediction model includes: a fully connected neural network and a GRU network; The fully connected neural network receives the breeding performance index data, the number of neurons in the input layer of the fully connected neural network is equal to the number of features of the breeding performance index data, and the offspring trait data of the newborn offspring is output after hidden layer processing, and the number of neurons in the output layer of the fully connected neural network is equal to the number of features of the offspring trait data; The offspring trait data of the newly born offspring are input into the GRU network, and the offspring trait data at different growth stages are output.

[0033] In this embodiment, the offspring trait index prediction model is composed of a fully connected neural network and a GRU network, wherein the fully connected neural network includes an input layer, a hidden layer, and an output layer. The number of neurons in the input layer of the fully connected neural network is equal to the number of features of the breeding performance index data, including but not limited to the genetic information of the parents (such as genotype), breeding history, growth rate, feed conversion rate, disease resistance, etc. Assuming that the breeding performance index data has N features, the input layer contains N neurons. The hidden layer of the fully connected neural network adopts a multi-layer perceptron structure, and each layer uses ReLU (Rectified Linear Unit) as an activation function to promote the nonlinear expression ability of the model. The number of hidden layers and the number of neurons in each layer are hyperparameters, which are optimized by grid search or random search. The number of neurons in the output layer of the fully connected neural network is equal to the number of features of the offspring trait data, such as growth rate, body size, meat quality, etc. The output layer uses a linear activation function to directly output continuous numerical results. The GRU network receives the trait data of the newborn offspring as input, and memorizes and updates the state through a gating mechanism (update gate and reset gate) to capture the trait characteristics that change over time. The dimension of the hidden layer of the GRU network can be adjusted according to the complexity of the training data and the model performance requirements. The output layer of the GRU network outputs the offspring trait data of the corresponding growth stage for each time step.

[0034] In this embodiment, the fully connected neural network can learn the complex relationship between breeding performance indicators and offspring traits, while the GRU network can effectively capture the dynamic characteristics of the evolution of traits over time, thereby significantly improving the prediction accuracy.

[0035] Preferably, before obtaining the breeding performance index data of different combinations of boars and sows, the method further includes: Clustering the breeding performance index data of the sows to obtain multiple sow categories; Calculate the arithmetic mean of the breeding performance index data within the same sow category, and use the calculation result as the breeding performance index data corresponding to the sow category.

[0036] Preferably, the breeding performance index data of the sows are clustered to obtain a plurality of sow categories, including: Setting clustering parameters, including neighborhood radius and minimum number of points; Based on the clustering parameters, the breeding performance index data of the sows are clustered and analyzed according to the DBSCAN algorithm to obtain the sow category.

[0037] In a large-scale pig farm, the number of sows may be very large, and the breeding performance indicator data of each sow is also quite rich, including but not limited to age, weight, birth records, genetic information, etc. Directly processing such a large data set not only requires huge computational workload, but also may lead to inefficient model training and even overfitting due to the complexity of the data.

[0038] The present invention proposes a method of first clustering and classifying sows and then calculating the average breeding performance index data in each classification to simplify the subsequent breeding performance prediction process. The specific process is as follows: Clean the breeding performance index data, handle missing values ​​and outliers, standardize or normalize continuous variables, and use the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm, which is a density-based clustering method that can identify clusters of any shape while excluding noise points. The DBSCAN algorithm requires setting two key parameters: neighborhood radius (Eps) and minimum number of points (MinPts). Select appropriate parameter values ​​through experiments or domain knowledge to achieve the ideal clustering effect. Run the DBSCAN algorithm to cluster the breeding performance index data of sows to obtain multiple sow categories; For each sow category, calculate the arithmetic mean of its internal breeding performance index data, including but not limited to sow age, sow weight, mean weight of offspring pigs at birth, mean weight of offspring pigs at the growth stage, mean feed-to-meat ratio of offspring pigs and proportion of offspring pig types, etc. The calculated mean value is used as the representative breeding performance index data of sows in this category for subsequent model training and prediction.

[0039] Through clustering processing, a large amount of sow data is summarized into a few categories, and the category average indicator is used instead of individual data, which significantly reduces the amount of data processed, thereby reducing the computational cost and time consumption of model training. Furthermore, the classification of sows helps farm managers to manage and allocate resources more effectively. For example, differentiated feeding strategies and health management measures can be adopted for different categories of sows.

[0040] Preferably, the breeding performance index data include the age of the sows, the weight of the sows, the average weight of the offspring pigs at birth, the average weight of the offspring pigs during the growth stage, the average feed-to-meat ratio of the offspring pigs and the proportion of offspring pig types; the offspring trait data include the weight of the offspring pigs at birth, the weight of the offspring pigs during the growth stage, the feed-to-meat ratio of the offspring pigs and the offspring pig types; the offspring pig types include the sow types and the meat pig types.

[0041] Preferably, data from a recent preset number of matings are used to calculate the mean weight of the offspring pigs at birth, the mean weight of the offspring pigs at the growth stage and the mean feed-to-meat ratio of the offspring pigs in the breeding performance index data.

[0042] In the pig farming industry, the reproductive performance of breeding pigs is affected by many factors, including but not limited to age, health status, feeding environment and genetic characteristics. As they age, the reproductive capacity of breeding pigs and the health status of their offspring may change, which directly affects the growth performance and economic benefits of offspring pigs. Therefore, using outdated breeding data for model training may cause the prediction results to deviate from the actual situation, affecting the accuracy and effectiveness of breeding decisions.

[0043] The present invention proposes a method for dynamically updating a training data set to reflect the latest changes in breeding performance of breeding pigs. The specific implementation steps are as follows: Regularly collect breeding performance index data of sows and their offspring, including but not limited to the mean birth weight, mean growth weight, mean feed-to-meat ratio, etc. of offspring. In order to ensure the accuracy and completeness of the data, necessary data cleaning and preprocessing are performed, such as outlier detection and missing value filling. Set a "recent" time window, such as the most recent mating cycles (which can be the most recent year, the most recent quarters, or the most recent mating events), and select breeding data during this period as input for model training to ensure the timeliness and relevance of the data. For each sow in the selected time window, calculate the mean birth weight, mean growth weight, and mean feed-to-meat ratio of its offspring, and use these mean data to replace the old breeding performance indicators to reflect the latest breeding performance of the sows.

[0044] Use updated breeding performance indicator data to retrain deep learning models, especially fully connected neural networks and GRU networks, regularly evaluate the predictive performance of the models, check the generalization ability of the models on new data through cross-validation and other techniques, and perform fine-tuning when necessary.

[0045] In this embodiment, by regularly updating the training data set, the model can capture the latest changes in the breeding performance of breeding pigs, thereby improving the accuracy of predicting the traits of offspring pigs.

[0046] Preferably, the method further comprises training the offspring trait index prediction model, specifically comprising: Acquire breeding performance index data of multiple different combinations of pig breeds, and corresponding offspring trait data at different growth stages, to obtain training data; Using the breeding performance index data as input of the fully connected neural network and the offspring trait data at the early stage as labels, the fully connected neural network is trained; After the fully connected neural network training is completed, the GRU network is trained according to the output of the fully connected neural network and the offspring trait data.

[0047] Collect breeding performance indicator data for various pig combinations, such as age, weight, health status, genetic information, etc. At the same time, collect trait data of corresponding offspring pigs at different growth stages, such as birth weight, growth rate, feed conversion rate, etc.

[0048] The number of neurons in the input layer of the fully connected neural network architecture is equal to the number of features of the breeding performance index data, and ReLU is used as the activation function. The data is cleaned, missing values ​​and outliers are processed, and necessary feature engineering is performed, such as feature encoding, normalization or standardization. The number of neurons in the output layer is equal to the number of features of the offspring trait data, which is used to predict the offspring traits in the early stage.

[0049] Use the collected breeding performance index data as input and the offspring trait data in the early stage as labels for model training. Use the backpropagation algorithm and gradient descent optimizer (such as Adam, SGD, etc.) to update the network weights to minimize the loss function (such as mean square error). Divide the data set into training set and validation set, monitor the loss and accuracy during training, and prevent overfitting.

[0050] On the basis of the training of the fully connected neural network, a GRU network is designed to capture the changing patterns of offspring traits over time. The input layer receives the output of the fully connected network, that is, the predicted values ​​of the traits of the newly born offspring, and the output layer predicts the offspring trait data at different growth stages. Using the output of the fully connected network and the offspring trait data, the GRU network is trained, its weights and gating parameters are optimized, the training process of the GRU network is monitored, and the hyperparameters are adjusted to optimize the model performance.

[0051] Example 2, see Figure 2 The present invention provides a breeding control system for sows based on a deep learning network, the system comprising: The data acquisition module 100 is used to acquire breeding demand data and breeding performance index data of different combinations of boars and sows; The offspring data prediction module 200 is used to input the breeding performance index data into a trained offspring trait index prediction model, and output offspring trait data at multiple different growth stages, wherein the offspring trait index prediction model is obtained based on deep learning network training; The breeding pig combination determination module 300 is used to screen the multiple offspring trait data based on the breeding demand data, and determine the breeding pig combination corresponding to the screened data as the breeding pairing breeding pig combination.

[0052] Preferably, see Figure 3 , the offspring trait index prediction model includes: a fully connected neural network and a GRU network; The fully connected neural network receives the breeding performance index data, the number of neurons in the input layer of the fully connected neural network is equal to the number of features of the breeding performance index data, and the offspring trait data of the newborn offspring is output after hidden layer processing, and the number of neurons in the output layer of the fully connected neural network is equal to the number of features of the offspring trait data; The offspring trait data of the newly born offspring are input into the GRU network, and the offspring trait data at different growth stages are output.

[0053] Preferably, the breeding performance index data include the age of the sows, the weight of the sows, the average weight of the offspring pigs at birth, the average weight of the offspring pigs during the growth stage, the average feed-to-meat ratio of the offspring pigs and the proportion of offspring pig types; the offspring trait data include the weight of the offspring pigs at birth, the weight of the offspring pigs during the growth stage, the feed-to-meat ratio of the offspring pigs and the offspring pig types; the offspring pig types include the sow types and the meat pig types.

[0054] The above is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features applied herein.

Claims

1. A method for controlling breeding of sows based on deep learning network, characterized in that: The method comprises: Obtain breeding demand data and breeding performance index data of different combinations of boars and sows; Input the breeding performance index data into a trained offspring trait index prediction model, and output offspring trait data at multiple different growth stages, wherein the offspring trait index prediction model is obtained based on deep learning network training; Based on the breeding demand data, a plurality of offspring trait data are screened, and a breeding pig combination corresponding to the screened data is determined as a breeding pairing breeding pig combination.

2. A method for controlling breeding of sows based on deep learning network according to claim 1, characterized in that: The offspring trait index prediction model includes: a fully connected neural network and a GRU network; The fully connected neural network receives the breeding performance index data, the number of neurons in the input layer of the fully connected neural network is equal to the number of features of the breeding performance index data, and the offspring trait data of the newborn offspring is output after hidden layer processing, and the number of neurons in the output layer of the fully connected neural network is equal to the number of features of the offspring trait data; The offspring trait data of the newly born offspring are input into the GRU network, and the offspring trait data at different growth stages are output.

3. The method for controlling breeding of sows based on deep learning network according to claim 1, characterized in that: Before obtaining the breeding performance index data of different combinations of boars and sows, the method further includes: Clustering the breeding performance index data of the sows to obtain multiple sow categories; Calculate the arithmetic mean of the breeding performance index data within the same sow category, and use the calculation result as the breeding performance index data corresponding to the sow category.

4. A method for controlling breeding of sows based on deep learning network according to claim 3, characterized in that: The breeding performance index data of the sows are clustered to obtain multiple sow categories, including: Setting clustering parameters, including neighborhood radius and minimum number of points; Based on the clustering parameters, the breeding performance index data of the sows are clustered and analyzed according to the DBSCAN algorithm to obtain the sow category.

5. The method for controlling breeding of sows based on deep learning network according to claim 2, characterized in that: The breeding performance index data include the age of breeding pigs, the weight of breeding pigs, the average weight of offspring pigs at birth, the average weight of offspring pigs at the growth stage, the average feed-to-meat ratio of offspring pigs and the proportion of offspring pig types; the offspring trait data include the weight of offspring pigs at birth, the weight of offspring pigs at the growth stage, the feed-to-meat ratio of offspring pigs and the offspring pig types; the offspring pig types include breeding pig types and meat pig types.

6. A method for controlling breeding of sows based on deep learning network according to claim 5, characterized in that: The data within the recent preset mating times are used to calculate the average weight of the offspring pigs at birth, the average weight of the offspring pigs at the growth stage and the average feed-to-meat ratio of the offspring pigs in the breeding performance index data.

7. The method for controlling breeding of sows based on deep learning network according to claim 5, characterized in that: The method further includes training the offspring trait index prediction model, specifically including: Acquire breeding performance index data of multiple different combinations of pig breeds, and corresponding offspring trait data at different growth stages, to obtain training data; Using the breeding performance index data as input of the fully connected neural network and the offspring trait data at the early stage as labels, the fully connected neural network is trained; After the fully connected neural network training is completed, the GRU network is trained according to the output of the fully connected neural network and the offspring trait data.

8. A sow breeding control system based on a deep learning network, using a sow breeding control method based on a deep learning network as described in any one of claims 1 to 7, characterized in that: The system comprises: The data acquisition module is used to obtain breeding demand data and breeding performance index data of different combinations of boars and sows; An offspring data prediction module is used to input the breeding performance index data into a trained offspring trait index prediction model, and output offspring trait data at multiple different growth stages, wherein the offspring trait index prediction model is obtained based on deep learning network training; The breeding pig combination determination module is used to screen the multiple offspring trait data based on the breeding demand data, and determine the breeding pig combination corresponding to the screened data as the breeding pairing breeding pig combination.

9. A sow breeding control system based on deep learning network according to claim 8, characterized in that: The offspring trait index prediction model includes: a fully connected neural network and a GRU network; The fully connected neural network receives the breeding performance index data, the number of neurons in the input layer of the fully connected neural network is equal to the number of features of the breeding performance index data, and the offspring trait data of the newborn offspring is output after hidden layer processing, and the number of neurons in the output layer of the fully connected neural network is equal to the number of features of the offspring trait data; The offspring trait data of the newly born offspring are input into the GRU network, and the offspring trait data at different growth stages are output.

10. A sow breeding control system based on deep learning network according to claim 9, characterized in that: The breeding performance index data include the age of breeding pigs, the weight of breeding pigs, the average weight of offspring pigs at birth, the average weight of offspring pigs at the growth stage, the average feed-to-meat ratio of offspring pigs and the proportion of offspring pig types; the offspring trait data include the weight of offspring pigs at birth, the weight of offspring pigs at the growth stage, the feed-to-meat ratio of offspring pigs and the offspring pig types; the offspring pig types include breeding pig types and meat pig types.

Citation Information

Patent Citations

  • Multi-objective optimization method for crop breeding

    CN111797499A

  • Beef cattle breeding optimization method and device based on machine learning

    CN114418182A

  • Breeding method and system for live pig hybrid commercial line

    CN118331965A

  • KR20230055449A