Livestock and poultry house environment parameter self-optimization regulation and control method based on group intelligence

Abstract

The invention discloses a livestock and poultry house environment parameter self-optimization regulation and control method based on swarm intelligence, and the method comprises the steps: collecting multi-dimensional environment parameters, such as temperature and humidity, through arranging a sensor array in a livestock and poultry house, and constructing an environment parameter sensing matrix; and converting the matrix into hypergraph structure data, inputting the hypergraph structure data into an optimized hypergraph neural network, mining high-order correlation and dynamic change characteristics among parameters, and clustering environment states through a swarm intelligence optimization strategy. A multi-dimensional environmental data regulation and control model is constructed based on the information, a parameter regulation and control target interval is set in combination with livestock and poultry growth requirements, a hypergraph network search strategy is regulated and controlled after deviation is calculated, equipment is driven to adjust environmental parameters, and closed-loop feedback is formed. According to the method, the hypergraph neural network and the swarm intelligence technology are utilized, the complex relation of the environmental parameters is accurately analyzed, the growth requirements of livestock and poultry are dynamically met, efficient self-optimization regulation and control of the environmental parameters of the livestock and poultry house are achieved, and the breeding environment quality and the production benefits are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of livestock and poultry breeding environment management, and in particular to a livestock and poultry house environment parameter self-optimization regulation method based on swarm intelligence. BACKGROUND

[0002] Under the background of modern livestock and poultry breeding towards scale and intensification, accurate regulation of livestock and poultry house environment parameters is crucial to ensure the healthy growth of livestock and poultry and improve breeding efficiency. Temperature, humidity, ammonia concentration, carbon dioxide concentration and other environmental parameters are interrelated and dynamically changing. The traditional regulation method relying on manual experience has been difficult to meet the demand. With the development of sensor technology and information technology, digital and intelligent regulation means have been gradually applied in the field of livestock and poultry breeding. Through real-time monitoring and analysis of environmental parameters, automatic regulation has become a new direction of industry development.

[0003] However, the existing livestock and poultry house environment parameter regulation technology still has obvious defects. Firstly, in the aspect of environmental parameter correlation analysis, most technologies treat each environmental parameter separately, without fully considering the complex high-order correlation between temperature, humidity, gas concentration and other parameters. For example, humidity changes not only directly affect the comfort of livestock and poultry, but also interact with ammonia concentration to affect air quality. However, existing technologies cannot accurately capture this complex correlation, resulting in lack of overall and effectiveness of regulation strategy. Secondly, in terms of dynamic adaptability, existing regulation methods are difficult to adapt to the dynamic changes of livestock and poultry house environment and the differentiated needs of livestock and poultry growth. The requirements of livestock and poultry at different growth stages for environmental parameters are different, and environmental parameters fluctuate with factors such as season and weather. However, existing technologies mostly use fixed regulation mode, which cannot be adjusted in time according to actual situation, resulting in significant reduction of regulation effect and difficulty in maintaining the best environment state of livestock and poultry house. SUMMARY

[0004] In order to overcome the shortcomings and deficiencies of the existing technology, the present application provides a livestock and poultry house environment parameter self-optimization regulation method based on swarm intelligence.

[0005] The technical solution adopted by the present application is a livestock and poultry house environment parameter self-optimization regulation method based on swarm intelligence, comprising the following steps:

[0006] Step S1: By distributing sensor arrays in livestock and poultry house, real-time collection of multi-dimensional environmental parameters is carried out in time series as dimension, and the collected data is integrated in order to form an environmental parameter perception matrix with time sequence characteristics;

[0007] Step S2: Convert the environmental parameter perception matrix into hypergraph structure data and input it into the optimized hypergraph neural network to deeply mine and extract high-order correlation relationships between multi-dimensional environmental parameters and dynamic change characteristics in the time dimension, and generate a feature expression vector containing environmental parameter correlation characteristics and change trends;

[0008] Step S3: Based on the swarm intelligence optimization strategy, the clustering operation is performed on the feature expression vector according to its distribution characteristics and similarity in the feature space, and the environmental state of the livestock and poultry house is divided into multiple state clusters with similar environmental parameter change patterns and correlation characteristics;

[0009] Step S4: For each environmental state cluster obtained by division, a multi-dimensional environmental data regulation and control model is constructed, and by analyzing the correlation relationships and change rules between environmental parameters in each state cluster, an environmental parameter regulation and control hypergraph network representing the mutual influence mechanism of environmental parameters is established;

[0010] Step S5: In combination with the specific needs of the livestock growth stage and the standard specifications of the breeding process, a corresponding regulation and control target interval is set for each environmental parameter;

[0011] Step S6: Calculate the deviation vector between the current environmental parameter and the set regulation and control target interval, and use the deviation vector as a guide to execute a path search algorithm in the environmental parameter regulation and control hypergraph network to obtain a regulation and control strategy set suitable for the current environmental state;

[0012] Step S7: According to the obtained regulation and control strategy set, the corresponding regulation and control equipment in the livestock and poultry house is driven to adjust the environmental parameters, and the adjusted environmental parameters are fed back to the optimized hypergraph neural network to form a closed-loop environmental parameter self-optimization regulation and control process.

[0013] Further, in the step S2, the optimized hypergraph neural network adopts the following hyperedge weight update formula:

[0014]

[0015] wherein, represents the weight of the hyperedge e ij at time t, which is used to measure the strength of the correlation between environmental parameters connected by the hyperedge; σ is an activation function; e ij is the hyperedge connecting nodes v i and v j ; is the attention weight of node v k relative to hyperedge e ij , reflecting the contribution degree of the node to the calculation of the hyperedge weight; is the feature vector of node v k at the previous time t-1; is the hyperedge eij bias term; k e e ij denotes node v k belongs to hyperedge e ij the set of connected nodes.

[0016] Further, the step S4, the environmental parameter regulation hypergraph network constructed by the multi-dimensional environmental data regulation model, adopts the following formula to calculate the influence strength between nodes:

[0017]

[0018] wherein, I ij denotes the influence strength between the nodes i and j corresponding to the environmental parameters; e ij is the set of hyperedges connecting nodes i and j; w e is the weight of the hyperedge e; v k is the other node connected by the hyperedge e except nodes i and j; is the feature vector of node v k ; e i , e j are the sets of hyperedges connecting nodes i and j.

[0019] Further, the step S3, when the feature expression vector is clustered based on the swarm intelligence optimization strategy, adopts the density clustering algorithm with hypergraph structure constraint, and determines the clustering core node through the following formula:

[0020]

[0021] wherein, C v is the clustering core node corresponding to node v; p u is the local density of node u, representing the aggregation degree of the node in the feature space; N(v) is the neighborhood node set of node v; d vu is the feature distance between node v and node u; f u is the feature vector of node u.

[0022] Further, the step S6, when the path search is performed in the environmental parameter regulation hypergraph network, adopts the heuristic search algorithm based on the hypergraph topology, and the probability of selecting the hyperedge e ij is calculated through the following formula:

[0023]

[0024] wherein, denotes the probability of selecting the hyperedge e ij ; p is the pheromone concentration on the hyperedge e ij ; a is the pheromone importance factor; heij is the heuristic information related to the potential of environmental parameter regulation between the connection nodes of the hyperedge e ij ; β is the importance factor of the heuristic information; is the set of all reachable hyperedges from node i.

[0025] Further, the step S1, the collected multi-dimensional environmental parameters further include dust concentration and noise intensity parameters, by adding corresponding type sensors, after obtaining data, it is integrated into the environmental parameter perception matrix for unified processing.

[0026] Further, the step S5, when setting the environmental parameter regulation target interval, the livestock physiological index feedback parameter is introduced, and the target interval boundary is dynamically adjusted through the following formula:

[0027]

[0028] wherein, are the lower and upper limits of the target interval of the environmental parameter under the scene s considering the physiological index feedback, respectively; are the initially set lower and upper limits of the target interval; ΔT s，p is the target interval adjustment amount under the scene s based on the livestock physiological index feedback.

[0029] Further, the step S7, when feeding back the adjusted environmental parameters to the optimized hypergraph neural network, a weighted feedback mechanism based on the correlation degree of the hyperedge is adopted, and the parameter feedback weight of each parameter is determined through the following formula:

[0030]

[0031] wherein, ω i is the feedback weight of the environmental parameter i; ε ij is the set of hyperedges simultaneously connecting the nodes corresponding to the environmental parameter i and the environmental parameter j; w e is the weight of the hyperedge e; n is the total number of environmental parameters.

[0032] Further, the step S2, the optimized hypergraph neural network is also configured with a parameter association dynamic evaluation module, and the feature propagation weight of different environmental parameter nodes is dynamically adjusted through the following formula:

[0033]

[0034] wherein, γ i is the feature propagation weight of the node corresponding to the environmental parameter i; Δx it , Δx jt are the change amounts of the environmental parameter i and the environmental parameter j at time t relative to the last time, respectively; n is the total number of environmental parameters; T is the total length of the collected data.

[0035] Beneficial effects: The present application proposes a livestock and poultry house environmental parameter self-optimization regulation method based on swarm intelligence. In the aspect of environmental parameter correlation analysis, the method uses an optimized hypergraph neural network to convert multi-dimensional environmental parameters into hypergraph structure data, and through a hyperedge weight updating mechanism and a node feature propagation algorithm, deeply mines high-order correlation relationships among parameters such as temperature, humidity, and ammonia concentration. Compared with the traditional isolated processing mode, the method can more comprehensively and accurately grasp the interaction among environmental parameters. In terms of dynamic adaptability, the method dynamically adjusts the environmental parameter regulation target interval by introducing livestock and poultry physiological index feedback parameters, and clusters the environmental state based on a swarm intelligence optimization strategy, and combines the environmental parameter regulation hypergraph network constructed by the multi-dimensional environmental data regulation model, so that the regulation strategy can be dynamically adjusted according to the needs of different growth stages of livestock and poultry, seasonal changes, etc. The method constructs a perception matrix by real-time collection of multi-dimensional environmental parameters through a sensor array, and then performs feature extraction, cluster analysis, model construction, and strategy search, etc., to drive the regulation equipment and form a closed loop feedback. The method not only can adapt to environmental changes in real time, but also can accurately meet the differentiated needs of livestock and poultry growth, significantly improving the intelligent level and effectiveness of livestock and poultry house environmental regulation, and providing strong support for guaranteeing the healthy growth of livestock and poultry and improving breeding efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 The method flowchart of the present application;

[0037] Figure 2 The method implementation unit composition diagram of the present application. DETAILED DESCRIPTION

[0038] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict, and the present application will be further described in detail below in combination with the drawings and specific embodiments.

[0039] As shown in the drawings, Figure 1 A livestock and poultry house environmental parameter self-optimization regulation method based on swarm intelligence includes the following steps:

[0040] Step S1: A sensor array is distributed in a livestock and poultry house, and multi-dimensional environmental parameters such as temperature, humidity, ammonia concentration, carbon dioxide concentration, light intensity, and wind speed are collected in time series as a dimension, and the collected data is integrated in order to form an environmental parameter perception matrix with time sequence characteristics;

[0041] Specifically, this step is achieved by distributing sensor arrays in the livestock and poultry house, collecting real-time data of temperature, humidity, ammonia concentration, carbon dioxide concentration, light intensity, and wind speed, etc. in time series, and integrating the collected data to form an environmental parameter perception matrix with time sequence characteristics. This step uses a distributed sensor array to set sensor nodes at different heights and regions in the livestock and poultry house, forming a three-dimensional monitoring network in space to ensure comprehensive collection of environmental parameters. The sensor collection frequency is set according to the characteristics of environmental parameters. The collection frequency of slowly changing parameters such as temperature and humidity is 1 time per minute, and the collection frequency of sudden parameters such as ammonia concentration and wind speed is increased to 1 time per second to capture subtle changes in environmental parameters. The collected data is arranged in chronological order to form a multi-dimensional time series data matrix, with each row representing a time point and each column corresponding to an environmental parameter, thereby constructing a complete environmental parameter perception matrix to provide basic data support for subsequent analysis.

[0042] The implementation of this step strictly follows the principles of data collection integrity and accuracy. The selection and placement of sensors are strictly calculated and verified. The temperature sensor uses a high-precision digital sensor with a measurement range of -40℃ to 80℃ and an accuracy of ±0.1℃. The humidity sensor has a measurement range of 0% to 100% RH with an accuracy of ±3% RH. The ammonia concentration sensor uses an electrochemical sensor with a measurement range of 0 to 100 ppm and a resolution of 0.1 ppm. All sensors are calibrated and calibrated to ensure the reliability of the collected data. Data transmission uses a combination of wired and wireless transmission methods. Key area sensors use wired transmission to ensure data stability, and remote areas use low-power wireless transmission modules with a transmission frequency of 2.4 GHz and a transmission distance of not less than 100 meters to ensure real-time and accurate data transmission to the data processing center, providing a high-quality data foundation for subsequent environmental state analysis.

[0043] Step S2: Convert the environmental parameter perception matrix into hypergraph structure data and input it into the optimized hypergraph neural network. Use the specially designed hyperedge weight update mechanism and node feature propagation algorithm in the network to deeply mine and extract the high-order correlation between multi-dimensional environmental parameters and the dynamic change characteristics in the time dimension, generating a feature expression vector containing environmental parameter correlation characteristics and change trends.

[0044] Specifically, this step converts the environmental parameter perception matrix into hypergraph structure data and inputs it into the optimized hypergraph neural network. The network uses a specially designed hyperedge weight update mechanism and node feature propagation algorithm to deeply mine and extract the high-order correlation relationships between multi-dimensional environmental parameters and the dynamic change characteristics in the time dimension, generating a feature expression vector containing the correlation characteristics and change trends of environmental parameters. This step first converts the environmental parameter perception matrix into a hypergraph representation, with each environmental parameter as a node in the hypergraph and the correlation between parameters represented by hyperedges. Unlike traditional edges, hyperedges can connect multiple nodes and more accurately represent the complex correlations between multi-dimensional environmental parameters. The optimized hypergraph neural network uses a multi-layer architecture, with each layer containing a node feature extraction layer, a hyperedge weight update layer, and a feature propagation layer. The node feature extraction layer converts the features of each environmental parameter node, the hyperedge weight update layer dynamically adjusts the hyperedge weights based on the correlation strength between nodes, and the feature propagation layer transfers and integrates node features through hyperedges, thereby deeply mining the high-order correlation relationships between environmental parameters.

[0045] The implementation of this step is based on the special architecture design of the hypergraph neural network, and the network parameters are optimized through a large amount of historical data training. During training, the random gradient descent algorithm is used for parameter update, with a learning rate of 0.001, a batch size of 64, and 1000 training rounds. To prevent overfitting, L2 regularization is introduced with a regularization coefficient of 0.0001. The hyperedge weight update mechanism dynamically adjusts the weight based on the degree of collaborative change between nodes, with higher collaborative change resulting in larger hyperedge weights. The feature propagation algorithm uses an attention mechanism to assign different attention weights based on node importance, highlighting the impact of key environmental parameters. Through this step, the complex correlation relationships between environmental parameters are converted into low-dimensional feature expression vectors, with a vector dimension of 128, containing both static features and dynamic change trends of environmental parameters, providing effective feature representation for subsequent environmental state clustering.

[0046] Step S3: Based on the swarm intelligence optimization strategy, the feature expression vectors are clustered according to their distribution characteristics and similarity in the feature space, and the environmental state of the livestock and poultry house is finely divided into multiple state clusters with similar environmental parameter change patterns and correlation characteristics.

[0047] Specifically, this step is based on swarm intelligence optimization strategy, and according to the distribution characteristics and similarity of feature expression vectors in feature space, clustering operation is performed, and the environmental state of livestock and poultry house is finely divided into multiple state clusters with similar environmental parameter change mode and correlation characteristics. This step uses an improved particle swarm optimization algorithm to cluster the feature expression vectors. Each particle in the particle swarm represents a possible clustering center. The optimal clustering scheme is found through iterative optimization. In the algorithm, hypergraph structure constraints are introduced to guide the clustering process by using the correlation between nodes in the hypergraph. The environmental states in the same cluster not only have similar parameter correlation patterns, but also have similar distances in the feature space. The number of clusters is dynamically determined according to the actual changes of the livestock and poultry house environment. By evaluating the silhouette coefficients under different cluster numbers, the optimal cluster number is selected to ensure that the clustering results can reflect the differences in environmental states and not be too fragmented.

[0048] The implementation of this step combines the global search ability of swarm intelligence algorithm and the correlation constraint characteristics of hypergraph structure. The parameter settings of the particle swarm algorithm are as follows: the number of particles is 50, the maximum number of iterations is 200, the inertia weight is linearly decreased from 0.9 to 0.4, and the acceleration constants c1 and c2 are both 2.0. In each iteration process, each particle updates its position based on its own experience and group experience, while considering the hypergraph structure constraints, so that the particle moves to the area with similar correlation patterns. In the clustering process, the cosine similarity is used as the distance measure between feature vectors, which can better capture the similarity of environmental parameter change trends. After clustering, each state cluster has a unique environmental parameter change pattern and correlation characteristics, providing a basis for subsequent construction of targeted regulation models.

[0049] Step S4: For each environmental state cluster obtained by division, a multi-dimensional environmental data regulation model is constructed, and a hypergraph network of environmental parameter regulation is established to represent the mutual influence mechanism of environmental parameters by analyzing the complex correlation between environmental parameters in each state cluster;

[0050] Specifically, this step constructs a multi-dimensional environmental data regulation model for each environmental state cluster obtained by division, and establishes a hypergraph network of environmental parameter regulation to represent the mutual influence mechanism of environmental parameters by analyzing the complex correlation between environmental parameters in each state cluster. In each environmental state cluster, an environmental parameter regulation model is constructed based on the hypergraph structure. The nodes in the hypergraph represent environmental parameters, and the hyperedges represent the correlation between parameters. By analyzing a large amount of historical data, the influence strength and direction between parameters are determined, and each hyperedge is assigned a corresponding weight and direction. The model considers the nonlinear interaction between environmental parameters, and learns the complex mapping relationship between parameters through a multilayer perceptron. In the construction process, the cross-validation method is used to evaluate the performance of the model to ensure that the model can accurately reflect the actual correlation between environmental parameters.

[0051] The implementation of this step is based on a data-driven modeling method by in-depth analysis of environmental parameter data within each state cluster. First, the data is standardized to eliminate the influence of different parameter dimensions. Then, the random forest algorithm is used to evaluate the importance of each parameter and determine the key parameters and their relationships. The determination of hyper-edge weights combines statistical analysis and machine learning methods by calculating mutual information, Granger causality, and other indicators between parameters to quantify the strength of the relationship between parameters. The structure of the multilayer perceptron is designed with the number of input layer nodes equal to the number of environmental parameters, 3 hidden layers with 64, 32, and 16 nodes respectively, and the number of output layer nodes equal to the number of environmental parameters. The model training uses the Adam optimizer with mean squared error loss as the loss function, and the early stopping strategy is used during training to prevent overfitting. The environmental parameter regulation hypergraph network constructed through this step can accurately describe the interaction mechanism between environmental parameters and provide a theoretical basis for environmental parameter regulation.

[0052] Step S5: Set the corresponding regulation target interval for each environmental parameter based on the specific needs of the livestock and poultry growth stage and the standard specifications of the breeding process, to clearly define the ideal value range of the environmental parameters.

[0053] Specifically, this step considers the specific needs of the livestock and poultry growth stage and the standard specifications of the breeding process to set the corresponding regulation target interval for each environmental parameter, thereby clearly defining the ideal value range of the environmental parameters. This step determines the target value and allowable fluctuation range of each environmental parameter based on the livestock and poultry breed, growth stage, and breeding process requirements. There are significant differences in the requirements of livestock and poultry at different growth stages for environmental parameters, for example, the temperature requirement for young chicks is higher, generally 32-35°C, while adult chickens can adapt to a temperature range of 20-25°C. At the same time, considering the breeding process standards such as ventilation requirements, lighting system, etc., the environmental parameters are comprehensively constrained. The setting of the target interval not only considers the optimal value of a single parameter, but also considers the synergistic effect between parameters to ensure that the combination of environmental parameters can meet the best needs of livestock and poultry growth.

[0054] The implementation of this step is based on the knowledge of breeding experts and statistical analysis of historical data. First, collect the physiological demand data of livestock and poultry at different growth stages, and determine the benchmark target value of each environmental parameter according to the breeding process standard. Then analyze the relationship between environmental parameters and livestock growth performance in historical breeding data, determine the influence of parameter fluctuation on growth performance through regression analysis, and determine the reasonable fluctuation range. For example, the temperature fluctuation range is set to ±1℃ in the young chick period and ±2℃ in the adult period. For interrelated parameters, a multi-objective optimization method is used to determine the optimal combination to ensure that the livestock growth demand is met while reducing the control cost. The final control target interval has dynamic adaptability and can automatically adjust according to the change of livestock growth stage, providing clear target guidance for environmental parameter control.

[0055] Step S6: Calculate the deviation vector between the current environmental parameters and the set control target interval, and execute the path search algorithm in the environmental parameter control hypergraph network guided by the deviation vector to obtain a control strategy set suitable for the current environmental state;

[0056] Specifically, this step calculates the deviation vector between the current environmental parameters and the set control target interval, and executes the path search algorithm in the environmental parameter control hypergraph network guided by the deviation vector to obtain a control strategy set suitable for the current environmental state. This step first calculates the deviation between the current environmental parameter vector and the target interval center vector, and each dimension of the deviation vector corresponds to the deviation value of an environmental parameter. Then, guided by the deviation vector, the optimal control path is searched in the environmental parameter control hypergraph network. The path search algorithm considers the correlation between parameters and the control cost, and selects the optimal control strategy set by evaluating the control effect and cost of different paths. In the search process, heuristic information is used to guide the search direction, and paths that can quickly reduce the deviation and have low control cost are preferentially selected.

[0057] The implementation of this step is based on the hypergraph topology structure and heuristic search algorithm. The calculation of the deviation vector uses the Euclidean distance metric to ensure that the difference between the current environmental state and the target state can be accurately reflected. The path search algorithm uses an improved ant colony algorithm to find the optimal control path in the hypergraph network. The algorithm introduces hyperedge weights and heuristic information to guide the moving direction of ants, and the greater the hyperedge weight, the higher the probability of ants selecting the hyperedge. The heuristic information is pre-calculated according to the correlation between parameters and the control efficiency, guiding the ants to move in the direction that can effectively reduce the deviation. In the search process, the pheromone concentration is dynamically adjusted to enhance the attractiveness of excellent paths while avoiding local optimization. The control strategy set obtained through this step considers the correlation between environmental parameters and the control cost, and can provide effective decision support for actual control.

[0058] Step S7: According to the obtained set of control strategies, drive the corresponding control equipment in the livestock and poultry house to adjust the environmental parameters, and feed back the adjusted environmental parameters to the optimized hypergraph neural network, forming a closed-loop self-optimizing control process of environmental parameters.

[0059] Specifically, according to the obtained set of control strategies, the corresponding control equipment in the livestock and poultry house is driven to adjust the environmental parameters, and the adjusted environmental parameters are fed back to the optimized hypergraph neural network, forming a closed-loop self-optimizing control process of environmental parameters. This step converts the control strategy into specific device control instructions to drive the ventilation system, temperature control system, lighting system and other control equipment to adjust the environmental parameters. The control instructions contain information such as the on / off state and operating parameters of the equipment to ensure that the equipment can be accurately controlled according to the predetermined strategy. After adjustment, the adjusted environmental parameters are collected in real time by sensors and fed back to the optimized hypergraph neural network to update the network parameters, so that the network can adapt to the new environmental state, thereby forming a closed-loop control system.

[0060] The implementation of this step is based on a distributed control system architecture, which uses modular design to achieve collaborative control of different control equipment. The control instructions are transmitted to each device controller through industrial Ethernet with a transmission rate not less than 100 Mbps to ensure real-time performance of the instructions. The device controller uses a programmable logic controller (PLC) with high reliability and anti-interference capability to accurately execute the control instructions. In the control process, a PID control algorithm is used to accurately adjust the device operating parameters, and the control parameters are dynamically adjusted according to the real-time feedback of the environmental parameters to ensure control accuracy. The feedback mechanism uses a weighted feedback strategy to assign different feedback weights according to the importance of each environmental parameter, giving more attention to changes in key parameters. Through this closed-loop control process, the system can continuously learn and adapt to environmental changes, continuously optimize control strategies, and improve the intelligent level and stability of livestock and poultry house environmental control.

[0061] Preferably, in step S2, the optimized hypergraph neural network uses the following hyperedge weight update formula:

[0062]

[0063] wherein, represents the weight of the hyperedge e ij at time t, which is used to measure the strength of the correlation between the environmental parameters of the connected nodes; σ is an activation function; e ij is the hyperedge connecting node v i and v j ; is the attention weight of node v k relative to hyperedge e ij at time t, reflecting the contribution of the node to the calculation of the hyperedge weight. for node v k feature vector at last time t-1; for hyperedge e ij bias term; k e e ij denotes node v k connected node set of hyperedge e ij .

[0064] Specifically, the optimized hypergraph neural network can dynamically measure the correlation strength of the environmental parameters between the nodes connected by the hyperedge through a specially designed hyperedge weight updating mechanism. This mechanism outputs the hyperedge weight after activation function processing according to the feature vector of each node connected by the hyperedge at the last time, combined with the attention weight of the node relative to the hyperedge and the bias term of the hyperedge. In implementation, during the network training process, the hyperedge weight will be continuously adjusted with the changes of environmental parameters and the updates of node features, ensuring that the network can accurately capture the complex correlation relationship between temperature, humidity, ammonia concentration and other multi-dimensional environmental parameters, providing a foundation for subsequent deep mining of high-order correlation and dynamic change characteristics of environmental parameters.

[0065] Preferably, in the step S4, the environmental parameter regulation hypergraph network constructed by the multi-dimensional environmental data regulation model is calculated according to the following formula:

[0066]

[0067] wherein, I ij denotes the influence strength between the nodes i and j corresponding to the environmental parameters; e ij is the hyperedge set connecting nodes i and j; w e is the weight of hyperedge e; v k is the other node connected by the hyperedge e except nodes i and j; f vk is the feature vector of node v k ; e i , e j are the hyperedge sets connecting nodes i and j, respectively.

[0068] Specifically, the environmental parameter regulation hypergraph network constructed by the multi-dimensional environmental data regulation model determines the influence strength between nodes through a specific calculation method. This calculation comprehensively considers the hyperedge set connecting two nodes, the hyperedge weight, and the feature vector of other nodes connected by the hyperedge, and obtains the influence strength between nodes through standardization processing. In actual implementation, this calculation process is based on a large amount of historical environmental parameter data, which can quantify the degree of mutual influence between environmental parameters, thereby helping to construct a more practical environmental parameter regulation hypergraph network, clearly presenting the mutual influence mechanism between environmental parameters, and providing an important basis for formulating regulation strategies.

[0069] Preferably, in step S3, when clustering the feature expression vectors based on the swarm intelligence optimization strategy, a density clustering algorithm with hypergraph structure constraints is adopted to determine the cluster core nodes by the following formula:

[0070]

[0071] Among them, C v is the cluster core node corresponding to node v; ρ u is the local density of node u, which represents the degree of node aggregation in the feature space; N(v) is the set of neighboring nodes of node v; d vu is the characteristic distance between node v and node u; f u is the feature vector of node u.

[0072] Specifically, when clustering based on a swarm intelligence optimization strategy, a density clustering algorithm constrained by a hypergraph structure is used to determine the core nodes of the cluster. This algorithm calculates the core nodes based on the local density of other nodes in the node's neighborhood, the characteristic distance from neighboring nodes, and the node's characteristic vector. During implementation, the clustering process is constrained by the relationships between nodes in the hypergraph structure, ensuring that environmental states within the same cluster are not only close in characteristic space distance but also have similar environmental parameter correlation patterns, enabling a more refined and accurate classification of livestock and poultry housing environmental conditions.

[0073] Preferably, in step S6, when searching for paths in the environment parameter control hypergraph network, a heuristic search algorithm based on hypergraph topology is used to calculate and select the hyperedge e by the following formula: ij Probability of:

[0074]

[0075] in, Indicates the selection of hyperedge e ij probability; is the hyperedge e ij The pheromone concentration on the surface; α is the pheromone importance factor; η eij is the hyperedge e ij The heuristic information is related to the potential for regulating environmental parameters between hyperedge-connected nodes; β is the importance factor of the heuristic information; is the set of all reachable hyperedges starting from node i.

[0076] Specifically, in the process of searching for a path in the environment parameter regulation hypergraph network, the heuristic search algorithm based on hypergraph topology plans the path by calculating the probability of selecting a hyperedge. This probability calculation comprehensively considers the pheromone concentration on the hyperedge, the heuristic information related to the environmental parameter regulation potential between the nodes connected by the hyperedge, and the importance factors corresponding to the two. In implementation, the algorithm simulates the ant colony pathfinding process, dynamically guides the search direction based on the above factors, and preferentially selects the hyperedge path that can efficiently reduce the deviation of the current environmental parameter from the target zone and has a lower regulation cost, thereby quickly obtaining a regulation strategy set that adapts to the current environmental state.

[0077] Preferably, in the step S1, the collected multi-dimensional environmental parameters further include dust concentration and noise intensity parameters. By adding corresponding type sensors, the data is integrated into the environmental parameter perception matrix for unified processing after being obtained.

[0078] Specifically, the range of environmental parameter collection is further expanded by adding dust concentration and noise intensity detection sensors. These new sensors and the original temperature and humidity, gas concentration, etc. sensors together form a more complete sensor array. After collecting data, the dust concentration and noise intensity data are integrated into the environmental parameter perception matrix together with other environmental parameters. In implementation, through a unified data processing process and format, the multi-dimensional environmental parameters are integrated, so that the perception matrix contains more comprehensive environmental information, providing a richer data basis for subsequent environmental analysis and regulation.

[0079] Preferably, in the step S5, when setting the environmental parameter regulation target interval, the livestock physiological index feedback parameter is introduced, and the target interval boundaries are dynamically adjusted by the following formula:

[0080]

[0081] wherein, are the lower and upper limits of the target interval of the environmental parameter in scenario s after considering the physiological index feedback; are the initially set lower and upper limits of the target interval; ΔT s，p is the target interval adjustment amount in scenario s based on the livestock physiological index feedback.

[0082] Specifically, when setting the environmental parameter regulation target interval, the livestock physiological index feedback parameter is introduced to achieve dynamic adjustment. According to the initially set upper and lower limits of the target interval, combined with the adjustment amount in different scenarios based on the livestock physiological index feedback, the target interval boundaries are re-determined. In the implementation process, by monitoring the physiological indexes of livestock such as body temperature and respiratory rate in real time, the changes of these indexes are mapped as adjustment signals for the target interval of the environmental parameter, so that the regulation target interval can dynamically change according to the actual physiological needs of livestock, improving the accuracy and adaptability of environmental parameter regulation.

[0083] Preferably, in the step S7, the adjusted environmental parameters are fed back to the optimized hypergraph neural network using a weighted feedback mechanism based on the correlation degree of hyperedges, and the weight of each parameter feedback is determined by the following formula:

[0084]

[0085] wherein ω i is the feedback weight of the environmental parameter i; ε ij is the set of hyperedges simultaneously connecting the nodes corresponding to the environmental parameter i and the environmental parameter j; w e is the weight of the hyperedge e; and n is the total number of environmental parameters.

[0086] Specifically, in the step S7, the adjusted environmental parameters are fed back to the optimized hypergraph neural network using a weighted feedback mechanism based on the correlation degree of hyperedges to determine the weight of each parameter feedback. The weight is calculated based on the total weight of the hyperedges connecting the nodes corresponding to the environmental parameters, and the proportion of the total weight of the hyperedges connecting all parameter nodes. In implementation, different weights are given to the feedback data of different environmental parameters through this mechanism. For parameters with high correlation and greater influence in the interaction of environmental parameters, higher feedback weights are given to enable the network to pay more attention to the changes of key parameters and optimize the response and learning ability of the network to environmental changes.

[0087] Preferably, in the step S2, the optimized hypergraph neural network is further configured with a parameter correlation dynamic evaluation module to dynamically adjust the feature propagation weight of different environmental parameter nodes by the following formula:

[0088]

[0089] wherein γ i is the feature propagation weight of the node corresponding to the environmental parameter i; Δx it and Δx jt are the change amounts of the environmental parameter i and the environmental parameter j at time t relative to the previous time; n is the total number of environmental parameters; and T is the total duration of data collection.

[0090] Specifically, the parameter correlation dynamic evaluation module of the optimized hypergraph neural network dynamically adjusts the feature propagation weight of different environmental parameter nodes by analyzing the change amount difference of environmental parameters in time series. In the implementation process, the module continuously monitors the changes of each environmental parameter at different times, compares the change amount difference between parameters, and gives higher feature propagation weights to parameter nodes with high change coordination and close correlation with other parameters, to ensure that the network can highlight the role of key environmental parameters in the feature extraction and propagation process and more accurately mine high-order correlation features between environmental parameters.

[0091] For example, Figure 2As shown, a livestock and poultry house environment parameter self-optimization regulation method based on swarm intelligence, the method is realized by the following units, including:

[0092] Livestock and poultry house multi-dimensional environment parameter real-time perception and collection unit, for distributed deployment of sensors in livestock and poultry house, real-time collection of temperature, humidity and other multi-dimensional environment parameters, and integration to form an environment parameter perception matrix;

[0093] Hypergraph neural network feature deep extraction unit, connected with livestock and poultry house multi-dimensional environment parameter real-time perception and collection unit, receiving the environment parameter perception matrix, converting it into hypergraph structure data, and extracting the environment parameter correlation and change characteristics through the optimized hypergraph neural network;

[0094] Swarm intelligence driven feature clustering and division unit, connected with hypergraph neural network feature deep extraction unit, clustering the extracted feature expression vector based on swarm intelligence optimization strategy, and dividing livestock and poultry house environment state clusters;

[0095] Multi-dimensional environment data regulation hypergraph construction unit, connected with swarm intelligence driven feature clustering and division unit, constructing multi-dimensional environment data regulation model for each environment state cluster to form an environment parameter regulation hypergraph network;

[0096] Growth demand adaptation target interval setting unit, connected with multi-dimensional environment data regulation hypergraph construction unit, setting the environment parameter regulation target interval in combination with livestock and poultry growth demand;

[0097] Deviation-oriented regulation strategy search unit, connected with growth demand adaptation target interval setting unit and multi-dimensional environment data regulation hypergraph construction unit respectively, calculating the environment parameter deviation and searching for regulation strategies in the regulation hypergraph network;

[0098] Closed-loop feedback regulation device driving unit, connected with deviation-oriented regulation strategy search unit and hypergraph neural network feature deep extraction unit, driving the regulation device to adjust the environment parameters, and feeding back the adjustment results to the hypergraph neural network feature deep extraction unit.

[0099] A livestock and poultry house environment parameter self-optimization regulation method based on swarm intelligence, aiming at the problem that traditional technology is difficult to mine high-order correlation of environment parameters, the method converts multi-dimensional environment data such as temperature, humidity, ammonia concentration into hypergraph structure, and through the special designed hyperedge weight update mechanism and node feature propagation algorithm, accurately analyzes the complex interaction relationship between parameters. For example, when analyzing the influence of humidity on ammonia concentration, not only the direct correlation between the two is considered, but also the synergistic effect of other parameters such as light intensity and wind speed is integrated to construct a regulation hypergraph network that fully reflects the mutual influence mechanism of environment parameters, changing the limitation of isolated processing of parameters in the past, making the regulation strategy more systematic and scientific.

[0100] In terms of dynamic adaptability, the method introduces livestock physiological index feedback, combines livestock growth stage characteristics and breeding process standards, and dynamically adjusts the regulation target interval of environmental parameters. At the same time, based on the swarm intelligence optimization strategy, the environmental state is clustered to divide the state cluster with similar environmental change mode. When the environment changes, the system calculates the deviation of the current environmental parameters from the target interval, performs path search in the regulation hypergraph network, and quickly generates the regulation strategy adapted to the current environmental state. In addition, the adjusted environmental parameters are fed back to the hypergraph neural network in real time, forming a closed-loop regulation process to ensure that the system can continuously adapt to seasonal changes, weather changes and environmental needs of different stages of livestock growth.

[0101] Compared with traditional regulation technology, the method realizes precise, dynamic and collaborative regulation of livestock house environmental parameters through multi-dimensional data hypergraph modeling, dynamic target interval adjustment, swarm intelligence clustering analysis and closed-loop feedback mechanism, significantly improves the environmental regulation efficiency and breeding refinement level, and provides a more efficient and intelligent technical solution for modern livestock breeding.

[0102] In the description of the present application, it should be pointed out that, unless otherwise specified and limited, the terms "arrangement", "installation", "connection", "connection", "fixing" should be understood broadly, for example, it can be fixed connection, or detachable connection, or integral connection; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through intermediate medium, or the communication inside two elements. For ordinary skilled in the art, the specific meaning of the above terms in the present application can be understood through specific circumstances.

[0103] Although the embodiments of the present application have been shown and described, it can be understood by those skilled in the art that various equivalent changes, modifications, replacements and variations of the embodiments can be made without departing from the principles and spirits of the present application, and the scope of the present application is defined by the appended claims and their equivalent scope.

Claims

1. A method for self-optimization and control of livestock and poultry house environmental parameters based on swarm intelligence, characterized in that: The following steps are involved: Step S1: A sensor array is distributed in the livestock and poultry house to collect multi-dimensional environmental parameters in real time using time series as the dimension, and the collected data is sequentially integrated to construct an environmental parameter perception matrix with time series characteristics; Step S2: The environmental parameter perception matrix is ​​converted into hypergraph structure data and input into the optimized hypergraph neural network to deeply mine and extract the high-order correlation between multi-dimensional environmental parameters and the dynamic change characteristics in the time dimension, generating a feature expression vector containing the correlation characteristics and change trends of environmental parameters; Step S3: Based on the swarm intelligence optimization strategy, a clustering operation is performed on the feature expression vectors according to their distribution characteristics and similarity in the feature space, and the livestock and poultry house environmental state is divided into multiple state clusters with similar environmental parameter change patterns and associated characteristics; Step S4: Constructing a multidimensional environmental data control model for each environmental state cluster obtained by division, and establishing an environmental parameter control hypergraph network that characterizes the mutual influence mechanism of environmental parameters by analyzing the correlation and change rules between environmental parameters in each state cluster; Step S5: setting a corresponding control target range for each environmental parameter based on the specific needs of livestock and poultry growth stages and breeding process standards and specifications; Step S6: Calculate the deviation vector between the current environmental parameter and the set control target interval, and use the deviation vector as a guide to execute a path search algorithm in the environmental parameter control hypergraph network to obtain a set of control strategies suitable for the current environmental state; Step S7: Based on the acquired control strategy set, drive the corresponding control equipment in the livestock and poultry house to adjust the environmental parameters, and feed the adjusted environmental parameters back to the optimized hypergraph neural network.

2. The method for self-optimization and control of livestock and poultry house environmental parameters based on swarm intelligence according to claim 1 is characterized in that: In step S2, the optimized hypergraph neural network adopts the following hyperedge weight update formula: in, Indicates that at time t the edge e ij The weight is used to measure the strength of the association between the environmental parameters of the nodes connected by the hyperedge; σ is the activation function; e ij For connecting node v i With v j The superedge of is node v at time t k Relative to the hyperedge e ij The attention weight reflects the contribution of the node to the calculation of the hyperedge weight; For node v k The eigenvector at the previous moment t-1; is the hyperedge e ij Bias term; k∈e ij Represents node v k Belongs to the hyperedge e ij The collection of connected nodes.

3. The method for self-optimization and control of livestock and poultry house environmental parameters based on swarm intelligence according to claim 1 is characterized in that: In step S4, the environmental parameter control hypergraph network constructed by the multi-dimensional environmental data control model uses the following formula to calculate the influence strength between nodes: Among them, I ij Indicates the influence strength between the environmental parameters corresponding to nodes i and j; ε ij is the set of hyperedges connecting node i and node j at the same time; w e is the weight of hyperedge e; v k are the other nodes connected by hyperedge e except nodes i and j; For node v k The eigenvector of i , ε j are the sets of hyperedges connecting node i and node j respectively.

4. The method for self-optimization and control of livestock and poultry house environmental parameters based on swarm intelligence according to claim 1, characterized in that: In step S3, when clustering the feature expression vectors based on the swarm intelligence optimization strategy, a density clustering algorithm with hypergraph structure constraints is used to determine the cluster core nodes using the following formula: Among them, C v is the cluster core node corresponding to node v; ρ u is the local density of node u, which represents the degree of node aggregation in the feature space; N(v) is the set of neighboring nodes of node v; d vu is the characteristic distance between node v and node u; f u is the feature vector of node u.

5. The method for self-optimization and control of livestock and poultry house environmental parameters based on swarm intelligence according to claim 1 is characterized in that: In step S6, when searching for paths in the environment parameter control hypergraph network, a heuristic search algorithm based on the hypergraph topology is used to calculate and select the hyperedge e by the following formula: ij Probability of: in, Indicates the selection of hyperedge e ij probability; is the hyperedge e ij The pheromone concentration on the surface; α is the pheromone importance factor; η eij is the hyperedge e ij The heuristic information is related to the potential for regulating environmental parameters between hyperedge-connected nodes; β is the importance factor of the heuristic information; is the set of all reachable hyperedges starting from node i.

6. The method for self-optimization and control of livestock and poultry house environmental parameters based on swarm intelligence according to claim 1, characterized in that: In step S1, the multi-dimensional environmental parameters collected further include dust concentration and noise intensity parameters. By adding corresponding types of sensors, the data is acquired and integrated into the environmental parameter perception matrix for unified processing.

7. The method for self-optimization and control of livestock and poultry house environmental parameters based on swarm intelligence according to claim 1, characterized in that: In step S5, when setting the target range for environmental parameter regulation, the livestock and poultry physiological index feedback parameters are introduced to dynamically adjust the target range boundary using the following formula: in, are the lower and upper limits of the target range of environmental parameters in scenario s after considering physiological indicator feedback; are the lower and upper limits of the initially set target range; ΔT s，p is the target interval adjustment amount in scenario s based on the feedback of livestock and poultry physiological indicators.

8. The method for self-optimization and control of livestock and poultry house environmental parameters based on swarm intelligence according to claim 1, characterized in that: In step S7, when feeding back the adjusted environmental parameters to the optimized hypergraph neural network, a weighted feedback mechanism based on hyperedge relevance is adopted, and the feedback weight of each parameter is determined by the following formula: Among them, ω i is the feedback weight of environmental parameter i; ε ij is the set of hyperedges that connect the nodes corresponding to environment parameter i and environment parameter j at the same time; w e is the weight of the hyperedge e; n is the total number of environment parameters.

9. The method for self-optimization and control of livestock and poultry house environmental parameters based on swarm intelligence according to claim 1, characterized in that: In step S2, the optimized hypergraph neural network is further configured with a parameter association dynamic evaluation module to dynamically adjust the feature propagation weights of different environment parameter nodes using the following formula: Among them, γ i is the characteristic propagation weight of the node corresponding to the environmental parameter i; Δx it , Δx jt are the changes of environmental parameters i and j at time t relative to the previous time; n is the total number of environmental parameters; T is the total time of data collection.

Images