A poultry breeding environment adaptive intelligent control method and system

Abstract

The present invention proposes an adaptive intelligent control method and system for poultry breeding environment, the method comprising: collecting biosignals of individual poultry; constructing a health status model based on health index; collecting the position coordinates and velocity vector of poultry, constructing a time series model based on behavior pattern to monitor the changes of group behavior pattern in real time; dynamically adjusting the environmental control strategy according to the comprehensive environmental control index, and performing real-time feedback monitoring after adjustment; constructing a composite growth index for identifying the growth stage according to the biosignal, designing a comprehensive optimization mechanism for multi-stage environmental control, integrating data from different growth stages; designing an adaptive multi-objective optimization model to dynamically adjust the control strategy according to real-time data, dynamically adjust the control parameters in combination with historical data, and designing a multi-level feedback loop and emergency response mechanism to ensure that the system responds in time in emergencies. The present invention significantly improves the overall benefits of poultry breeding and has broad application prospects.

Description

Technical Field

[0001] The present invention belongs to the technical field of adaptive intelligent control, and in particular relates to an adaptive intelligent control method and system for a poultry breeding environment. Background Art

[0002] With the growth of the global population and the increasing demand for poultry products, modern and large-scale poultry farming has gradually become the mainstream. However, in this intensive farming model, environmental control has become a key factor in determining the health and production efficiency of poultry. Traditional poultry farming environmental control systems usually rely on preset parameters such as temperature, humidity, ventilation, and light, and achieve relative stability of the environment through timing or manual adjustment. However, with the complexity of environmental conditions and the diversification of poultry needs, this static environmental control method has become increasingly insufficient and difficult to cope with the ever-changing environment and dynamic needs of poultry in the actual farming process.

[0003] The current breeding environment control system has the following main shortcomings: First, most traditional systems are adjusted based on data input from environmental sensors (such as thermometers and hygrometers), ignoring the feedback of the poultry's own physiological state and behavioral patterns to the environment. Therefore, environmental control often lacks refined management of individuals or groups, which can easily lead to a disconnect between environmental regulation and the actual needs of poultry. For example, when the ambient temperature rises suddenly, the system may increase the ventilation volume, but ignores the stress response that some poultry may experience in this situation, which in turn affects their health and production efficiency.

[0004] Secondly, the traditional system lacks the ability to perceive group behavior. As a group animal, poultry's behavior patterns are very sensitive to environmental requirements. In existing technologies, environmental control systems often cannot capture changes in group behavior in real time, resulting in the system being unable to respond in time when crowding, local overheating or group stress reactions occur, which may cause problems such as disease transmission and increased stress reactions, further affecting breeding benefits.

[0005] Furthermore, the existing technology lacks personalized management for poultry at different growth stages. Poultry at different growth stages have different requirements for environmental conditions (such as light, temperature, sound, etc.). Existing environmental control systems usually adopt a unified standard and cannot be adjusted differently according to the growth stage of poultry. This may not only cause problems such as stunted development and decreased egg production in poultry groups with inconsistent growth stages, but also increase energy consumption and reduce the economic benefits of breeding. Summary of the invention

[0006] The purpose of the present invention is to design an adaptive intelligent control method and system for poultry breeding environment. The present invention not only needs to have the ability to perceive the environment and poultry status in real time, but also should be able to dynamically adjust environmental parameters according to actual needs to ensure that while optimizing poultry health and production efficiency, energy consumption is reduced and the overall benefits of breeding are improved.

[0007] In order to achieve the above object, in a first aspect of the present invention, a method for adaptive intelligent control of a poultry breeding environment is provided, the method comprising:

[0008] S1. Install micro sensors on individual birds to collect biological signals of individual birds, pre-process the collected data, and then design a multi-dimensional dynamic boundary anomaly detection algorithm to correct the anomalies and obtain data. Design a dynamic feature weighted extraction method for data Perform weighted feature extraction; the features include heart rate mean Average body temperature and activity-weighted variance

[0009] S2. Use the extracted features to build a health status model based on the health index. By calculating the health index H i (t) Thus, the deviation degree of health status evaluation ΔH is obtained i (t) Determine the past status of the poultry, and then design a feedback weight adaptive adjustment algorithm to adaptively adjust the weights in the health status model. Then, based on the health index H i (t) and deviation ΔH i (t) Classify the health status of poultry and adopt corresponding environmental control strategies;

[0010] Among them, by calculating the health index H i (t) Thus, the deviation degree of health status evaluation ΔH is obtained i (t) Determine the past status of the birds:

[0011] If ΔH i (t) exceeds the preset dynamic threshold Θ H , it indicates that the birds were in an abnormal or stressful state in the past;

[0012] S3, collect the position coordinate data and velocity vector data of the birds, and map the position coordinate data and velocity vector data to the behavior feature vector B in the feature space j (t), using the adaptive clustering algorithm based on the behavior feature vector B j (t) Identify the behavior patterns of poultry groups, serialize the behavior patterns and build a time series model based on the behavior patterns to monitor the changes in the behavior patterns of the groups in real time;

[0013] S4. Design of multimodal data fusion and weighted decision model for health index H i (t), behavior feature vector B j (t) is fused with the environmental sensor data to obtain the multimodal feature vector F(t), and then the comprehensive environmental control index D(t) is generated through the adaptive weighting function. The environmental control strategy is dynamically adjusted according to the comprehensive environmental control index D(t), and real-time feedback monitoring is performed after the adjustment;

[0014] S5. Constructing a composite growth index G for identifying growth stages based on biological signals i (t), according to the composite growth index G i (t) Design dynamic environment demand vector E (g) (t), and then design a differentiated weight allocation algorithm in the growth stage to dynamically adjust the dynamic environment demand vector E (g) The priority weight of (t) is determined, and after each adjustment, it is verified through a specific growth stage feedback mechanism. Finally, a comprehensive optimization mechanism for multi-stage environmental control is designed to integrate data from different growth stages.

[0015] S6. Design an adaptive multi-objective optimization model to dynamically adjust the control strategy according to real-time data, and then design an individual difference adaptive control model to assign personalized control parameters Θ to each individual according to biological signals. i (t), dynamically adjust control parameters based on historical data, and design multi-level feedback loops and emergency response mechanisms to ensure timely system response in emergency situations.

[0016] In one embodiment, the multi-dimensional dynamic boundary anomaly detection algorithm specifically includes:

[0017] According to the local mean μ of each sensor signal i (t) and standard deviation σ i (t), set a multi-dimensional dynamic threshold θ i (t), the detection conditions are as follows:

[0018]

[0019] Among them, γ and δ represent adjustment parameters, ρ ij (t) represents the correlation coefficient between signals i and j, X j (t) represents the sensor signal, μ i (t) represents the local mean, σ i (t) represents the standard deviation, θ i (t) represents the multi-dimensional dynamic threshold;

[0020] If the condition is met |x i (t)-μi (t)|>θ i (t), the point is considered abnormal and the abnormal point is corrected. The corrected value It is expressed as:

[0021]

[0022] Among them, S t Represents a set of normal data points within the time window, with weight w k It is determined by its similarity with other signals, k represents the kth sensor;

[0023] In the dynamic feature weighted extraction method, the weight ω is designed i (t) is an adjustment factor based on the activity intensity of the birds during that time period and is expressed as:

[0024]

[0025] Among them, η represents the adjustment parameter, which is used to control the impact of heart rate fluctuation on activity weight.

[0026] In one embodiment, the health index H i The representation is as follows:

[0027]

[0028] Among them, w HR represents the weight of heart rate, w T represents the weight of body temperature, w A represents the weight of the activity variance, represents the mean heart rate, represents the mean body temperature, represents the activity-weighted variance.

[0029] In one embodiment, the feedback weight adaptive adjustment algorithm is based on the recent health state deviation ΔH i The weight is adjusted dynamically based on the size of (t). The specific adjustment formula is:

[0030]

[0031]

[0032] Among them, α represents the learning rate, w HE (t) represents the weight of the heart rate at the previous moment, w T (t) represents the weight of the body temperature at the previous moment, w A (t) represents the weight of the variance of activity at the previous moment, ΔH i (t) represents the recent health status deviation, w HR(t+1) represents the weight of the adjusted heart rate, w T (t+1) represents the weight of the adjusted body temperature, w A (t+1) represents the weight of the adjusted activity variance.

[0033] In one embodiment, the adaptive clustering algorithm based on the behavior feature vector is expressed as follows:

[0034] The set of behavioral feature vectors of individual birds is Where N is the number of individuals in the group, and the specific clustering process is achieved by minimizing the following clustering objective function:

[0035]

[0036] Among them, K represents the number of clusters, C k represents the behavior pattern of the kth class, μ k represents the center of the class, the objective function J represents the sum of the distances from all individual behavior feature vectors to the center of their respective classes, and B j Represents the behavioral feature vector at a certain moment;

[0037] The time series model is modeled by a Markov behavior switching model.

[0038] In one embodiment, the environmental control strategy is dynamically adjusted according to the calculated comprehensive environmental control index D(t), wherein the comprehensive environmental control index D(t) is expressed as follows:

[0039]

[0040] Among them, W represents the weight vector, corresponding to the weight of each feature, and F(t) represents the multimodal feature vector;

[0041] Define different threshold intervals according to comprehensive environmental control indicators and dynamically adjust the adjustment strategy of environmental parameters;

[0042] The adjustment strategy of the environmental parameters after adjustment is compared with the adjustment strategy before adjustment. If the difference exceeds the preset tolerance range, the weight vector W is automatically adjusted, which is expressed as follows:

[0043]

[0044] Among them, α represents the learning rate, which controls the amplitude of weight adjustment; W(t+1) represents the updated weight, W(t) represents the current weight, D(t+1) represents the updated comprehensive environmental control index, and D(t) represents the comprehensive environmental control index at the previous moment.

[0045] In one embodiment, the composite growth index G i(t), the calculation formula is as follows:

[0046]

[0047] Among them, α W Represents the weight coefficient of body weight, α S Represents the weight coefficient of body shape, α H Represents the weight coefficient of heart rate, W i (t) represents body weight data, S i (t) represents body size data, Indicates heart rate data.

[0048] In one embodiment, the growth phase differentiated weight allocation algorithm is as follows:

[0049]

[0050] in, represents the environmental parameter value at the g-th growth stage, represents the specific parameter weight at this growth stage, E c (t) represents the current environment state, E (g) (t) represents the dynamic environment demand vector;

[0051] A multi-objective optimization function based on regularization terms is introduced to deal with the conflict between dynamic environment requirements. The multi-objective optimization function is expressed as follows:

[0052]

[0053] in, represents the loss function of environmental regulation, represents the regularization term between different environmental parameters, and λ is the regularization coefficient, which is used to balance the relationship between environmental parameters;

[0054] The growth stage feedback mechanism determines whether the adjustment of the current environment meets the growth target by monitoring the biological signals after the current environment is adjusted, as shown below:

[0055]

[0056] Where γ represents the feedback adjustment coefficient, ΔG i (t+1) represents the change in the adjusted growth index, ΔH i (t+1) represents the change of health status;

[0057] The objective function of the comprehensive optimization mechanism of the multi-stage environmental control is expressed as follows:

[0058]

[0059] in, represents the optimized environmental parameters, and represent the actual and expected growth indicators, represents the growth standard deviation of this stage, represents the regularization term between parameters.

[0060] In one embodiment, the adaptive multi-objective optimization model is constructed as follows:

[0061] The control objectives of the system are integrated into an adaptive multi-objective optimization model, and the loss function L of each control objective is set. n (Θ,t), where Θ is the parameter vector of the control strategy, n represents different control objectives, and the overall objective function L of multi-objective optimization is total (Θ,t) is represented as follows:

[0062]

[0063] Among them, λ n (t) represents the dynamic weight of each target, γ·∥Θ-Θ0∥ 2 represents a regular term, Θ0 represents the initial control parameter;

[0064] According to the individual's biological signal, each individual is assigned a personalized control parameter Θ i (t), expressed as follows:

[0065]

[0066] Among them, α represents the learning rate, J i (t) represents the Jacobian matrix of individual i, which is used to describe the relationship between the individual physiological state and the control parameters. represents the gradient of the individual loss function, Θ i (t+1) represents the updated control parameters.

[0067] In a second aspect of the present invention, there is provided a poultry breeding environment adaptive intelligent control system, the system comprising:

[0068] The data acquisition module is used to install micro sensors on individual birds to collect biological signals of individual birds, pre-process the collected data, and then design a multi-dimensional dynamic boundary anomaly detection algorithm to correct the anomalies and obtain data. Design a dynamic feature weighted extraction method for data Perform weighted feature extraction; the features include heart rate mean Average body temperature and activity-weighted variance

[0069] The data analysis modeling module is used to construct a health status model based on the health index using the extracted features. i (t) Thus, the deviation degree of health status evaluation ΔH is obtained i (t) Determine the past status of the poultry, and then design a feedback weight adaptive adjustment algorithm to adaptively adjust the weights in the health status model. Then, based on the health index H i (t) and deviation ΔH i (t) Classify the health status of poultry and adopt corresponding environmental control strategies;

[0070] Among them, by calculating the health index H i (t) Thus, the deviation degree of health status evaluation ΔH is obtained i (t) Determine the past status of the birds:

[0071] If ΔH i (t) exceeds the preset dynamic threshold Θ H , it indicates that the bird was in an abnormal or stressful state in the past

[0072] The group behavior monitoring and behavior pattern modeling module is used to collect the position coordinate data and velocity vector data of the birds, and map the position coordinate data and velocity vector data to the behavior feature vector B in the feature space. j (t), using the adaptive clustering algorithm based on the behavior feature vector B j (t) Identify the behavior patterns of poultry groups, serialize the behavior patterns and build a time series model based on the behavior patterns to monitor the changes in the behavior patterns of the groups in real time;

[0073] Multimodal data fusion and decision-making module, used to design multimodal data fusion and weighted decision-making model for health index H i (t), behavior feature vector B j (t) is fused with the environmental sensor data to obtain the multimodal feature vector F(t), and then the comprehensive environmental control index D(t) is generated through the adaptive weighting function. The environmental control strategy is dynamically adjusted according to the comprehensive environmental control index D(t), and real-time feedback monitoring is performed after the adjustment;

[0074] Environmental differentiation regulation module, used to construct a composite growth index G for identifying growth stages based on biological signals i (t), according to the composite growth index G i (t) Design dynamic environment demand vector E (g) (t), and then design a differentiated weight allocation algorithm in the growth stage to dynamically adjust the dynamic environment demand vector E (g)The priority weight of (t) is determined, and after each adjustment, it is verified through a specific growth stage feedback mechanism. Finally, a comprehensive optimization mechanism for multi-stage environmental control is designed to integrate data from different growth stages.

[0075] The system optimization module is used to design an adaptive multi-objective optimization model to dynamically adjust the control strategy according to real-time data, and then design an individual difference adaptive control model to assign personalized control parameters Θ to each individual according to biological signals. i (t) Dynamically adjust control parameters based on historical data, and design multi-level feedback loops and emergency response mechanisms to ensure timely system response in emergencies

[0076] The beneficial technical effects of the present invention are at least as follows:

[0077] (1) The present invention installs biosignal sensors on individual birds or their activity areas to monitor key physiological indicators of birds in real time, such as heart rate, body temperature, and activity level. The system uses deep learning algorithms to analyze these biosignals, accurately determine the health status and comfort of the birds, and dynamically adjust environmental parameters. Compared with traditional methods, this innovation significantly improves the accuracy of environmental control, can better respond to the actual needs of individual birds, and avoids stress reactions and health problems caused by improper environmental control.

[0078] (2) The present invention introduces a group behavior monitoring and modeling module. Through computer vision technology, the system can capture and analyze the group behavior of poultry (such as movement, aggregation, dispersion, etc.) in real time, and identify group behavior patterns and abnormal situations. The system coordinates and regulates environmental parameters based on the analysis results, such as increasing ventilation or adjusting temperature in areas with dense poultry populations to prevent local overheating or disease transmission. This innovation significantly improves the system's ability to respond to group needs and ensures the uniformity and stability of the environment.

[0079] (3) The present invention has designed a growth stage identification and differentiated management module. The system identifies the growth stage of the poultry by monitoring its weight, body shape and other data, and adjusts the light intensity, light cycle and sound environment in a targeted manner. For example, in the young bird stage, the system provides softer light and a quiet environment; in the egg-laying period, the light cycle is optimized and the sound environment is adjusted to promote egg-laying efficiency. This innovation not only improves the growth quality of poultry, but also reduces breeding costs by optimizing energy consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0081] Figure 1 This is a flow chart of a poultry breeding environment adaptive intelligent control method of the present invention.

[0082] Figure 2 This is a framework diagram of an adaptive intelligent control system for poultry breeding environment according to the present invention. DETAILED DESCRIPTION

[0083] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0084] In one or more embodiments, Figure 1 As shown, a poultry breeding environment adaptive intelligent control method is disclosed, and the method includes the following steps S1-S6:

[0085] S1. Install micro sensors on individual birds to collect biological signals of individual birds, pre-process the collected data, and then design a multi-dimensional dynamic boundary anomaly detection algorithm to correct the anomalies and obtain data. Design a dynamic feature weighted extraction method for data Perform weighted feature extraction; the features include heart rate mean Average body temperature and activity-weighted variance

[0086] Specifically, in order to maximize the accuracy and coverage of signal acquisition, the present invention installs micro sensors at multiple key locations of individual birds (such as the chest, wing roots, and legs), and these sensors are used to collect biological signals of different dimensions. Assume that a heart rate sensor (HR) is installed on the chest, a temperature sensor (T) is installed on the wing root, and an activity sensor (A) is installed on the leg. In addition, in order to increase the reliability of acquisition, redundant sensors (dual sensor redundant design) are also used at the same location to prevent the failure or error of a single sensor. The collected signal data is recorded as:

[0087] HR i (t),T i (t),A i (t)(1)

[0088] Where i represents the sensor position and t represents the time.

[0089] In view of the data drift and signal delay that may be generated by different sensors, the present invention designs a signal phase alignment algorithm. The algorithm uses the cross-correlation function (CCF) to align multiple sensor signals in the time domain. The formula is:

[0090]

[0091] Among them, τ ij is the optimal time offset between sensors i and j. After adjustment, the signals are synchronized to the same time reference.

[0092] In order to ensure the smoothness and continuity of the collected data, especially to deal with the problem of high noise in poultry breeding environment, this paper proposes an adaptive weighted Gaussian filter based on the traditional Gaussian filter, which dynamically adjusts the weight to deal with the noise level in different time periods. i ′(t) is expressed as:

[0093]

[0094] Among them, the weight w j (t) is adaptively adjusted according to the local noise level and bird activity intensity at each moment, and is calculated using the following formula:

[0095]

[0096] Here, β is the adjustment parameter, A avg (t) is the average activity in the current time period, μ and σ are the mean and standard deviation of the Gaussian distribution. i (t)-A avg (t)| is to dynamically adjust the filter response according to the intensity of bird activities to prevent over-smoothing of transient noise caused by intense activities.

[0097] In the process of collecting bio-signals of birds, abnormal data may appear due to sensor movement or external interference. In order to accurately identify and correct these abnormal points, the present invention proposes a multi-dimensional dynamic boundary abnormality detection algorithm. The algorithm determines a dynamic boundary to detect abnormal values ​​by analyzing the correlation between bio-signals of multiple dimensions (heart rate, temperature, activity):

[0098] First, the local mean μ of each sensor signal is calculated i (t) and standard deviation σ i (t), and then set a multidimensional dynamic threshold θ i (t), the test conditions are:

[0099]

[0100] Among them, γ and δ are adjustment parameters, ρ ij (t) is the correlation coefficient between signals i and j. If the condition |x is satisfied i (t)-μ i (t)|>θ i (t), then the point is considered abnormal.

[0101] After the anomaly is detected, the "multi-dimensional local weighted regression correction algorithm" is applied to correct the anomaly point. The corrected value It is expressed as:

[0102]

[0103] Among them, S t is a set of normal data points in the time window, with weight w k It is determined by its similarity with other signals, which is similar to the idea of ​​local weighted regression, but is optimized with multidimensional data.

[0104] After data preprocessing, it is necessary to extract features from the data in order to provide effective input for subsequent steps. This paper proposes a dynamic feature weighted extraction method that not only extracts the traditional mean and variance, but also performs weighting according to the intensity and stability of the signal. The signal features in each segment L are extracted as follows:

[0105] Heart rate average:

[0106]

[0107] Average body temperature:

[0108]

[0109] Activity-weighted variance:

[0110]

[0111] Among them, the weight ω i (t) is an adjustment factor based on the activity intensity of the birds during that time period, and the formula is:

[0112]

[0113] Among them, η is a regulating parameter used to control the impact of heart rate fluctuation on activity weight. This processing method allows the physiological response of birds to be fully considered during feature extraction, making the extracted features more practical.

[0114] S2. Use the extracted features to build a health status model based on the health index. By calculating the health index H i(t) Thus, the deviation degree of health status evaluation ΔH is obtained i (t) Determine the past status of the poultry, and then design a feedback weight adaptive adjustment algorithm to adaptively adjust the weights in the health status model. Then, based on the health index H i (t) and deviation ΔH i (t) Classify the health status of poultry and adopt corresponding environmental control strategies.

[0115] Among them, by calculating the health index H i (t) Thus, the deviation degree of health status evaluation ΔH is obtained i (t) Determine the past status of the birds:

[0116] If ΔH i (t) exceeds the preset dynamic threshold Θ H , it indicates that the birds were in an abnormal or stressful state in the past.

[0117] Specifically, the health status model based on the health index uses the core biological signal features extracted in the first step (such as heart rate mean Average body temperature and activity-weighted variance ), and establish a poultry health status model. Considering the different effects of these features on the health status of poultry, a multi-feature fusion method is used to construct a health index H by weighted summation. i . Health Index H i The formula is:

[0118]

[0119] Among them, w HR 、w T and w A are the weights of the variance of heart rate, body temperature and activity respectively. These weights are determined based on the regression analysis of historical data to ensure the optimal prediction ability in a specific environment. Through this formula, the present invention integrates the biological signal characteristics of different dimensions into a unified health status indicator, providing a direct basis for subsequent environmental regulation.

[0120] Furthermore, in each time step (for example, every minute), the health index H in that time period is calculated: i (t). Then, by comparing with the historical average health state Compare and evaluate the degree of deviation from the current health status. Deviation ΔH i (t) is defined as:

[0121]

[0122] in, It is the average health index calculated based on historical data within a certain period of time (for example, the past hour). i (t) exceeds the preset dynamic threshold Θ H (set according to the specific breeding environment and historical data), it indicates that the poultry may be in an abnormal or stressful state. At this time, the system will activate the early warning mechanism and prepare for the next environmental regulation decision.

[0123] Furthermore, in order to cope with the changes in the physiological state of birds and the dynamic adjustment of the environment, the weights w in the health state model need to be adjusted. HR 、w T and w A The present invention proposes a feedback weight adaptive adjustment algorithm, which is based on the recent health state deviation ΔH i The weight is adjusted dynamically based on the size of (t). The specific adjustment formula is:

[0124]

[0125] Among them, α is the learning rate, which controls the amplitude of weight adjustment. Through this dynamic adjustment, the system can optimize the contribution of each feature to the health status in real time, making the health index more in line with the current physiological state of the poultry.

[0126] Finally, based on the calculated health index H i (t) and deviation ΔH i (t), the health status of the poultry is divided into several levels (such as normal, slightly abnormal, and seriously abnormal). According to the different levels, the system adopts corresponding environmental control strategies, such as adjusting the temperature, increasing ventilation, or changing the lighting conditions. In order to classify, the present invention defines multiple thresholds Θ1, Θ2, ..., and divides ΔH i (t) is mapped to the corresponding state level.

[0127]

[0128] Each status level corresponds to specific environmental control measures to ensure that poultry can obtain a suitable breeding environment in different health states.

[0129] S3, collect the position coordinate data and velocity vector data of the birds, and map the position coordinate data and velocity vector data to the behavior feature vector B in the feature space j (t), using the adaptive clustering algorithm based on the behavior feature vector B j (t) Identify the behavioral patterns of poultry groups, serialize the behavioral patterns, and construct a time series model based on the behavioral patterns to monitor the changes in the group's behavioral patterns in real time.

[0130] Furthermore, in order to accurately capture the dynamics of poultry group behavior, multiple cameras and motion sensors are deployed in the farm to collect real-time behavioral data such as the movement trajectory, aggregation pattern, and activity intensity of the poultry. After preprocessing, these data are mapped to a unified behavioral feature space. Assume that the collected behavioral data includes the position coordinates (x j ,y j ) and velocity vector 9v x,j ,v u,j ), where j represents different individuals.

[0131] In order to conduct further analysis, the present invention maps the position data and speed data to the behavior feature vector B in the feature space respectively. j (t), which is expressed as:

[0132]

[0133] The behavior feature vector B j (t) It will serve as the basic input for group behavior analysis and for subsequent behavior pattern recognition and modeling.

[0134] Furthermore, in order to identify the behavior patterns of poultry groups, an adaptive clustering algorithm based on behavior feature vectors is adopted. Assume that the set of behavior feature vectors of poultry individuals is Where N is the number of individuals in the group. The goal of the clustering algorithm is to divide these behavioral feature vectors into several categories, each of which corresponds to a typical behavior pattern, such as aggregation, dispersion, and fast movement.

[0135] The specific clustering process is achieved by minimizing the following clustering objective function:

[0136]

[0137] Where K is the number of clusters, C k represents the behavior pattern of the kth class, μ k is the center of the class. The objective function J represents the sum of the distances from all individual behavior feature vectors to the center of their respective classes. By optimizing this function, different behavior patterns can be effectively identified. This clustering process dynamically adjusts the value of K to adapt to changes in group behavior in different time periods.

[0138] Furthermore, in order to further capture the dynamic changes and time dependence of group behavior, the present invention serializes the identified behavior patterns and constructs a time series model of the behavior patterns. Assume that the group behavior pattern at time t is C k (t), the present invention can be expressed as a discrete time series {C k (t)}. This time series is modeled by the “Markov Behavior Switching Model”.

[0139] The state transition probability matrix P of the Markov behavior transition model is expressed as:

[0140] P = [p ij ] (19)

[0141] Among them, p ij =P(C i (t+1)|C j (t)) indicates that the behavior mode C j Switch to Behavior Mode C i Based on this probability matrix, the model can predict future group behavior patterns and identify possible abnormal behavior pattern changes.

[0142] Furthermore, based on the results of time series modeling, the system monitors the changes in group behavior patterns in real time. ij The system will automatically trigger the abnormal detection response mechanism when the value is significantly lower than the normal level. For example, if a sudden high-density gathering or irregular movement is detected, the system will record the abnormal event and trigger the environmental control mechanism to adjust the temperature, ventilation or lighting conditions.

[0143] In addition, the system can also preset anomaly detection threshold Θ based on historical behavior pattern data B , when the transition probability p ij When this threshold is exceeded, emergency measures are taken immediately.

[0144] S4. Design of multimodal data fusion and weighted decision model for health index H i (t), behavior feature vector B j The multimodal feature vector F(t) is obtained by fusing the environmental sensor data, and then the comprehensive environmental control index D(t) is generated through an adaptive weighting function. The environmental control strategy is dynamically adjusted according to the comprehensive environmental control index D(t), and real-time feedback monitoring is performed after the adjustment.

[0145] Specifically, based on the health status indicator H generated in the previous steps i (t), behavioral characteristics B j (t) and environmental sensor data (such as temperature T e (t), humidity H e (t), light intensity L e (t), etc.), the present invention needs to fuse these multi-dimensional data to generate the best control decision suitable for the current breeding environment. In order to achieve this, a multi-modal data fusion and weighted decision model is proposed.

[0146] Specifically, first, define a multimodal feature vector F(t) to integrate all relevant data sources:

[0147]

[0148] Then, these features are weighted and summed through an adaptive weighting function to generate a comprehensive environmental control index D(t). This index is used to guide the specific parameter settings of environmental control, and the calculation formula is:

[0149] D(t)=W ·F(t) (21)

[0150] Where W is a weight vector, whose elements correspond to the weights of each feature. The initial value of the weight can be determined by the linear regression method of historical data, and then adaptively adjusted according to feedback during operation.

[0151] Furthermore, according to the calculated comprehensive environmental control index D(t), the present invention defines different threshold intervals to determine the adjustment strategy of the environmental parameters. For example, the following thresholds θ1 and θ2 can be set to divide different control levels:

[0152]

[0153] According to different control levels, the system adjusts environmental parameters such as temperature, humidity, ventilation and light. The specific control method can be achieved through linear or nonlinear functions. For example, the temperature control function can be designed as:

[0154] T c (t+1)=T c (t)+β T ·(D(t)-θ1) (23)

[0155] Among them, T c (t) is the current ambient temperature, β T is the temperature adjustment coefficient. Similarly, control parameters such as humidity, ventilation and lighting can also be adjusted according to the comprehensive environmental control index D(t) and the corresponding adjustment coefficient.

[0156] Furthermore, in order to ensure the effectiveness of the environmental control strategy, the system will conduct real-time feedback monitoring after each environmental parameter adjustment. The feedback data includes the new health status indicator H i (t+1) and updated behavior characteristics B j (t+1). Based on these feedback data, the system recalculates the comprehensive environmental control index D(t+1) and compares it with D(t) at the previous moment. If the difference exceeds the preset tolerance range, the system will automatically adjust the weight vector W to optimize future decisions. The specific weight adjustment mechanism is as follows:

[0157]

[0158] Among them, α is the learning rate, which controls the amplitude of weight adjustment. This feedback loop can ensure that the system can adjust adaptively to cope with the dynamic changes of environment and bird status, and achieve long-term environmental control optimization.

[0159] Furthermore, in order to improve the robustness and adaptability of the system, the present invention designs a multi-level environmental control mechanism. When it is detected that the system fails to adjust for multiple consecutive times (i.e., the feedback data shows that D(t+1) does not change in the expected direction), the system will automatically enter an emergency state and enable a backup control strategy. This strategy is usually based on successful cases in historical data, and maintains system stability by directly calling past environmental parameter settings until the problem is resolved or external intervention occurs. The control function in the emergency state can be expressed as:

[0160] F backup (t) = F history (t best ) (25)

[0161] Among them, F history (t best ) is the feature vector corresponding to the best control effect in historical data. Through this mechanism, the system can maintain the stability of the environment and ensure the health and breeding efficiency of poultry when facing complex or unknown problems.

[0162] S5. Constructing a composite growth index G for identifying growth stages based on biological signals i (t), according to the composite growth index G i (t) Design dynamic environment demand vector E (g) (t), and then design a differentiated weight allocation algorithm in the growth stage to dynamically adjust the dynamic environment demand vector E (g) The priority weight of the growth factor (t) is determined, and after each adjustment, it is verified through a specific growth stage feedback mechanism. Finally, a comprehensive optimization mechanism for multi-stage environmental control is designed to integrate data from different growth stages.

[0163] Specifically, in order to achieve differentiated environmental control based on the growth stage of poultry, it is first necessary to accurately identify the current growth stage of the poultry. In step 1, the biological signal features (such as body weight W i (t), body size S i (t), heart rate The present invention combines these features to construct a composite growth index G for identifying the growth stage. i (t). Specifically, G i The calculation formula of (t) is:

[0164]

[0165] Among them, α w , α S and α H are the weight coefficients of body weight, body shape and heart rate, respectively, obtained through regression analysis of historical data or calibration of experimental data. i (t) After standardization, it can be used to divide the growth stages (such as chick stage, rearing stage, egg-laying stage, etc.), and specific environmental control strategies will be adopted in each stage.

[0166] In order to closely link the identified growth stages with environmental requirements, a dynamic environmental requirement vector E is defined. (g) (t), which includes the temperature requirement T g (t), humidity requirement H g (t), light requirement L g (t) and ventilation demand V g (t), these requirements are modeled based on the specific needs of the current phase. For example:

[0167]

[0168] Among them, f(G i (t),g) is a nonlinear mapping function that transforms composite growth indicators and growth stages into specific environmental requirements.

[0169] Furthermore, after identifying the growth stage, the system compares the dynamic environmental demand with the current environmental state (represented by the environmental control index D(t) in step 4) and determines the environmental parameters that need to be adjusted. In order to achieve precise control, a "growth stage differentiated weight allocation algorithm" is proposed. This algorithm dynamically adjusts the priority weight of each environmental parameter according to the needs of the current growth stage to ensure that environmental control can meet the needs of a specific growth stage.

[0170] The adjustment formula of environmental control parameters is:

[0171]

[0172] in, is the environmental parameter value at the g-th growth stage, is the specific parameter weight at this growth stage, E c (t) is the current environmental state. Through this formula, the system adjusts the current environmental state to the state that best meets the needs of the growth stage. This adjustment process takes into account the priority of each parameter, thereby achieving refined management of environmental regulation.

[0173] Furthermore, in order to deal with possible conflicts between dynamic environmental requirements (e.g., high temperature and high humidity may have negative effects), a multi-objective optimization function based on a regularization term is introduced:

[0174]

[0175] in, is the loss function of environmental regulation, is the regularization term between different environmental parameters, and λ is the regularization coefficient, which is used to balance the relationship between environmental parameters. By minimizing this loss function, the system can find an optimal environmental control strategy in the case of multi-parameter conflicts.

[0176] Furthermore, in order to ensure the effectiveness and adaptability of environmental regulation, the system will verify it through a specific growth stage feedback mechanism after each regulation. This mechanism monitors the biological signals after the current environmental adjustment (such as the new heart rate body temperature ) and growth index G i (t+1) to determine whether the current environment adjustment meets the growth goals.

[0177] The specific feedback adjustment formula is:

[0178]

[0179] Where γ is the feedback adjustment coefficient, ΔG i (t+1) represents the change in the adjusted growth index, ΔH i (t+1) indicates the change of health status. Through this feedback mechanism, the system can dynamically adjust the priority weights of environmental parameters to ensure that the growth indicators develop in the expected direction.

[0180] Finally, in order to ensure the consistency and optimization of environmental control throughout the growth cycle, a comprehensive optimization mechanism for multi-stage environmental control is designed. This mechanism integrates data from different growth stages and uses historical data to optimize the control strategy for the current growth stage. By comparing the growth curves and environmental control parameters of different stages, the system can identify the best control path and apply it in the current environment.

[0181] The objective function of comprehensive optimization is:

[0182]

[0183] in, is the optimized environmental parameter, and are the actual and expected growth indicators, is the standard deviation of growth in this stage, is the regularization term between parameters. By minimizing this objective function, the system can achieve comprehensive optimization of different growth stages and ensure the continuity and consistency of environmental control.

[0184] S6. Design an adaptive multi-objective optimization model to dynamically adjust the control strategy according to real-time data, and then design an individual difference adaptive control model to assign personalized control parameters Θ to each individual according to biological signals. i (t), dynamically adjust control parameters based on historical data, and design multi-level feedback loops and emergency response mechanisms to ensure timely system response in emergency situations.

[0185] Specifically, in the previous steps, the system gradually established a complex and flexible intelligent control system by real-time collection of multi-dimensional data, construction of health status models, group behavior analysis, environmental control, and regulation of growth stages. However, as poultry grow and the external environment continues to change, the system needs to further improve its adaptive capabilities and dynamically adjust control strategies to optimize long-term breeding results. In the sixth step, an adaptive multi-objective optimization model is proposed to achieve this goal.

[0186] First, the present invention integrates multiple control objectives of the system (such as temperature, humidity, light intensity, etc.) into an adaptive multi-objective optimization model. Set the loss function L of each control objective k (Θ,t), where Θ is the parameter vector of the control strategy, and k represents different control objectives, such as temperature control, humidity regulation, etc. The overall objective function L of multi-objective optimization total (Θ,t) can be expressed as the weighted sum of each sub-goal:

[0187]

[0188] Among them, λ k (t) is the dynamic weight of each target, γ·∥Θ-Θ0∥ 2 is a regular term used to limit the range of parameter changes to avoid overly aggressive system regulation, and Θ0 is the initial control parameter. This optimization model integrates the multidimensional information obtained in the previous steps (such as the health status indicator H i (t), behavioral characteristics B j (t) and environmental control index D(t)), unifying the control objectives of the system into an optimization framework and achieving balance and coordination among the objectives. k (t), the system can dynamically adapt to changes in the environment and poultry needs to achieve long-term stable and efficient control.

[0189] Furthermore, in the breeding process, different individuals have different physiological states and environmental requirements, and the system needs to have personalized regulation capabilities. The present invention designs an "individual difference adaptive control model" based on the individual characteristics extracted above (such as heart rate body temperature Activity ), assigning personalized control parameters Θ to each individual i (t).

[0190] Specifically, the present invention uses a nonlinear feedback mechanism to adjust the control parameters of individuals. Assume that the current control parameter of individual i is Θ i (t), the feedback control formula is:

[0191]

[0192] Among them, α is the learning rate, J i (t) is the Jacobian matrix of individual i, which is used to describe the relationship between the individual physiological state and the control parameters. is the gradient of the individual loss function. This nonlinear feedback mechanism allows the system to adjust its control parameters based on the real-time feedback of individuals, achieving refined management. Through this model, the system can identify and adapt to the differences between individuals, especially for individuals with certain health conditions or growth stages. This mechanism is closely integrated with the biosignal monitoring and health status modeling in the previous steps, further improving the flexibility and responsiveness of the system.

[0193] Furthermore, in the long-term breeding process, the system needs to continuously learn from historical data to improve future control strategies. To this end, a "historical data-driven long-term learning mechanism" is introduced, which optimizes the system's control strategy by analyzing and summarizing historical breeding data. The present invention constructs a historical data loss function L history (Θ), expressed as:

[0194]

[0195] in, represents the expected operation, T is the historical time period, R(Θ) is the regularization term, and δ is its weight. By minimizing this historical loss function, the system can optimize the parameter Θ to ensure that the optimal control strategy is maintained in the long run. This mechanism combines the multidimensional data and control experience accumulated in the previous steps, and improves the effectiveness and robustness of future control strategies by summarizing and optimizing historical data. This learning mechanism based on historical data can capture the long-term trends of the environment and physiological state, providing a solid foundation for the long-term and stable operation of the system.

[0196] Furthermore, in order to ensure that the system can respond in time in emergencies, a multi-level feedback loop and emergency response mechanism is designed. This mechanism is based on the real-time data generated in the previous steps (such as health status change ΔH i (t), environmental state E p It automatically enters a feedback loop after each environmental adjustment to evaluate the effect of the current adjustment and decide whether to trigger an emergency response.

[0197] If multiple control failures are detected or environmental conditions deteriorate significantly (such as a sharp increase in temperature), the system will enter an emergency state and activate the backup strategy. The decision model for emergency response is expressed as:

[0198] Θ emergency =Θ history +ξ·Δ E (35)

[0199] Among them, Θ emergency is the control parameter in the emergency state, Θ history is the historical optimal strategy, ξ is the adjustment coefficient, Δ E is the emergency adjustment of environmental parameters. This formula ensures that the system can respond quickly to emergencies and maintain the stability of the environment. The multi-level feedback loop and emergency response mechanism can not only respond quickly to environmental changes, but also organically combine with the previous individual difference regulation, adaptive learning and historical data driven mechanism to form a complete closed-loop control system. It plays a key protective role in the entire patent solution, ensuring that the system can maintain efficient and stable operation under various conditions.

[0200] In one or more embodiments, Figure 2 As shown, a poultry breeding environment adaptive intelligent control system is disclosed, the system comprising:

[0201] The data acquisition module 101 is used to install micro sensors on individual birds to collect biological signals of individual birds, pre-process the collected data, and then design a multi-dimensional dynamic boundary anomaly detection algorithm to correct the anomalies to obtain data. Design a dynamic feature weighted extraction method for data Perform weighted feature extraction; the features include heart rate mean Average body temperature and activity-weighted variance

[0202] The data analysis and modeling module 102 is used to construct a health status model based on the health index by using the extracted features. i (t) Thus, the deviation degree of health status evaluation ΔH is obtained i(t) Determine the past status of the poultry, and then design a feedback weight adaptive adjustment algorithm to adaptively adjust the weights in the health status model. Then, based on the health index H i (t) and deviation ΔH i (t) Classify the health status of poultry and adopt corresponding environmental control strategies;

[0203] Among them, by calculating the health index H i (t) Thus, the deviation degree of health status evaluation ΔH is obtained i (t) Determine the past status of the birds:

[0204] If ΔH i (t) exceeds the preset dynamic threshold Θ H , it indicates that the bird was in an abnormal or stressful state in the past

[0205] The group behavior monitoring and behavior pattern modeling module 103 is used to collect the position coordinate data and velocity vector data of the birds, and map the position coordinate data and velocity vector data to the behavior feature vector B in the feature space. j (t), using the adaptive clustering algorithm based on the behavior feature vector B j (t) Identify the behavior patterns of poultry groups, serialize the behavior patterns and build a time series model based on the behavior patterns to monitor the changes in the behavior patterns of the groups in real time;

[0206] The multimodal data fusion and decision module 104 is used to design a multimodal data fusion and weighted decision model for the health index H i (t), behavior feature vector B j (t) is fused with the environmental sensor data to obtain the multimodal feature vector F(t), and then the comprehensive environmental control index D(t) is generated through the adaptive weighting function. The environmental control strategy is dynamically adjusted according to the comprehensive environmental control index D(t), and real-time feedback monitoring is performed after the adjustment;

[0207] The environment differentiation control module 105 is used to construct a composite growth index G for identifying the growth stage according to the biological signal. i (t), according to the composite growth index G i (t) Design dynamic environment demand vector E (g) (t), and then design a differentiated weight allocation algorithm in the growth stage to dynamically adjust the dynamic environment demand vector E (g) The priority weight of (t) is determined, and after each adjustment, it is verified through a specific growth stage feedback mechanism. Finally, a comprehensive optimization mechanism for multi-stage environmental control is designed to integrate data from different growth stages.

[0208] The system optimization module 106 is used to design an adaptive multi-objective optimization model to dynamically adjust the control strategy according to real-time data, and then design an individual difference adaptive control model to assign personalized control parameters θ to each individual according to biological signals. i (t), dynamically adjust control parameters based on historical data, and design multi-level feedback loops and emergency response mechanisms to ensure timely system response in emergency situations.

[0209] It is worth noting that the specific working process of the adaptive intelligent control system for poultry breeding environment provided by the embodiment of the present invention is the same as the working process of the adaptive intelligent control method for poultry breeding environment described in the above embodiment, and will not be repeated here.

[0210] Compared with the prior art, the poultry breeding environment adaptive intelligent control system provided by the embodiment of the present invention installs micro sensors on individual poultry to collect biological signals of individual poultry, pre-processes the collected data, and then designs a multi-dimensional dynamic boundary anomaly detection algorithm to correct the abnormal points to obtain data. Design a dynamic feature weighted extraction method for data Perform weighted feature extraction; the features include heart rate mean Average body temperature and activity-weighted variance The extracted features are used to construct a health status model based on the health index. i (t) Thus, the deviation degree of health status evaluation ΔH is obtained i (t) Determine the past status of the poultry, and then design a feedback weight adaptive adjustment algorithm to adaptively adjust the weights in the health status model. Then, based on the health index H i (t) and deviation ΔH i (t) Classify the health status of poultry and adopt corresponding environmental control strategies; among them, by calculating the health index H i (t) Thus, the deviation degree of health status evaluation ΔH is obtained i (t) Determine the past state of the bird: If ΔH i (t) exceeds the preset dynamic threshold Θ H , it indicates that the poultry was in an abnormal or stressful state in the past; collect the position coordinate data and velocity vector data of the poultry, and map the position coordinate data and velocity vector data to the behavior feature vector B in the feature space j (t), using the adaptive clustering algorithm based on the behavior feature vector B j (t) Identify the behavior patterns of poultry groups, serialize the behavior patterns and build a time series model based on the behavior patterns to monitor the changes in group behavior patterns in real time; design a multimodal data fusion and weighted decision model to predict the health index H i(t), behavior feature vector B j (t) is fused with environmental sensor data to obtain a multimodal feature vector F(t), and then a comprehensive environmental control index D(t) is generated through an adaptive weighting function. The environmental control strategy is dynamically adjusted according to the comprehensive environmental control index D(t), and real-time feedback monitoring is performed after the adjustment; a composite growth index G for identifying the growth stage is constructed based on biological signals i (t), according to the composite growth index G i (t) Design dynamic environment demand vector E (g) (t), and then design a differentiated weight allocation algorithm in the growth stage to dynamically adjust the dynamic environment demand vector E (g) The priority weight of (t) is determined, and after each regulation, it is verified through a specific growth stage feedback mechanism. Finally, a comprehensive optimization mechanism for multi-stage environmental control is designed to integrate data from different growth stages. An adaptive multi-objective optimization model is designed to dynamically adjust the control strategy based on real-time data. Then, an individual difference adaptive control model is designed to assign personalized control parameters Θ to each individual based on biological signals. i (t), dynamically adjust control parameters based on historical data, and design multi-level feedback loops and emergency response mechanisms to ensure timely system response in emergency situations.

[0211] The embodiment of the present invention also provides a poultry breeding environment adaptive intelligent control device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the steps in the above-mentioned poultry breeding environment adaptive intelligent control method embodiment are implemented, for example Figure 1 or, the processor implements the functions of each module in the above-mentioned system embodiments when executing the computer program.

[0212] Exemplarily, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program in the poultry breeding environment adaptive intelligent control device.

[0213] The poultry breeding environment adaptive intelligent control device can be a computing device such as a desktop computer, a notebook, a PDA, and a cloud server. The poultry breeding environment adaptive intelligent control device can include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the poultry breeding environment adaptive intelligent control device can also include input and output devices, network access devices, buses, etc.

[0214] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASAC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the poultry breeding environment adaptive intelligent control device, and various interfaces and lines are used to connect various parts of the entire poultry breeding environment adaptive intelligent control device.

[0215] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the poultry farming environment adaptive intelligent control device by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function, etc.; the data storage area can store data created according to the operation of the air conditioning controller, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMedaaCard, SMC), a secure digital (SecureDagatal, SD) card, a flash card (FlashCard), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0216] Wherein, if the module integrated in the poultry breeding environment adaptive intelligent control device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0217] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0218] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A poultry breeding environment adaptive intelligent control method, characterized in that: The method comprises: S1. Install micro sensors on individual birds to collect biological signals of individual birds, pre-process the collected data, and then design a multi-dimensional dynamic boundary anomaly detection algorithm to correct the anomalies and obtain data. Design a dynamic feature weighted extraction method for data Perform weighted feature extraction; the features include heart rate mean Average body temperature and activity-weighted variance S2. Use the extracted features to build a health status model based on the health index, and calculate the health index H i (t) thus obtaining the recent health status deviation ΔH i (t) Determine the past status of the poultry, and then design a feedback weight adaptive adjustment algorithm to adaptively adjust the weights in the health status model. Then, based on the health index H i (t) and recent health status deviation ΔH i (t) Classify the health status of poultry and adopt corresponding environmental control strategies; Among them, by calculating the health index H i (t) thus obtaining the recent health status deviation ΔH i (t) Determine the past status of the birds: If ΔH i (t) exceeds the preset dynamic threshold Θ H , it indicates that the birds were in an abnormal or stressful state in the past; S3, collect the position coordinate data and velocity vector data of the birds, and map the position coordinate data and velocity vector data to the behavior feature vector B in the feature space j (t), using the adaptive clustering algorithm based on the behavior feature vector B j (t) Identify the behavior patterns of poultry groups, serialize the behavior patterns and build a time series model based on the behavior patterns to monitor the changes in the behavior patterns of the groups in real time; S4. Design of multimodal data fusion and weighted decision model for health index H i (t), behavior feature vector B j (t) is fused with the environmental sensor data to obtain the multimodal feature vector F(t), and then the comprehensive environmental control index D(t) is generated through an adaptive weighting function. The environmental control strategy is dynamically adjusted according to the comprehensive environmental control index D(t), and real-time feedback monitoring is performed after the adjustment; S5. Constructing a composite growth index G for identifying growth stages based on biological signals i (t), according to the composite growth index G i (t) Design dynamic environment demand vector E (g) (t), and then design a differentiated weight allocation algorithm in the growth stage to dynamically adjust the dynamic environment demand vector E (g) The priority weight of (t) is determined, and after each adjustment, it is verified through a specific growth stage feedback mechanism. Finally, a comprehensive optimization mechanism for multi-stage environmental control is designed to integrate data from different growth stages. S6. Design an adaptive multi-objective optimization model to dynamically adjust the control strategy according to real-time data, and then design an individual difference adaptive control model to assign personalized control parameters Θ to each individual according to biological signals. i (t), dynamically adjust control parameters based on historical data, and design multi-level feedback loops and emergency response mechanisms to ensure timely system response in emergency situations.

2. The method for adaptive intelligent control of poultry breeding environment according to claim 1, characterized in that: The multi-dimensional dynamic boundary anomaly detection algorithm specifically includes: According to the local mean μ of each sensor signal i (t) and standard deviation σ i (t), set a multi-dimensional dynamic threshold θ i (t), the detection conditions are as follows: Among them, γ and δ represent adjustment parameters, ρ ij (t) represents the correlation coefficient between signals i and j, x j (t) represents the sensor signal, μ i (t) represents the local mean, σ i (t) represents the standard deviation, θ i (t) represents the multi-dimensional dynamic threshold; If the condition is met |x i (t)-μ i (t)|>θ i (t), the point is considered abnormal and the abnormal point is corrected. The corrected value It is expressed as: Among them, S t Represents a set of normal data points within the time window, with weight w k It is determined by its similarity with other signals, k represents the kth sensor; In the dynamic feature weighted extraction method, the weight ω is designed i (t) is an adjustment factor based on the activity intensity of the birds in the current time period and is expressed as: Among them, η represents the adjustment parameter, which is used to control the impact of heart rate fluctuation on activity weight.

3. The method for adaptive intelligent control of poultry breeding environment according to claim 1, characterized in that: The health index H i The representation is as follows: Among them, w HR represents the weight of heart rate, w T represents the weight of body temperature, w A represents the weight of the activity variance, represents the mean heart rate, represents the mean body temperature, represents the activity-weighted variance.

4. The method for adaptive intelligent control of poultry breeding environment according to claim 3, characterized in that: The feedback weight adaptive adjustment algorithm is based on the recent health state deviation ΔH i The weight is adjusted dynamically based on the size of (t). The specific adjustment formula is: Among them, α represents the learning rate, w HR (t) represents the weight of the heart rate at the previous moment, w T (t) represents the weight of the body temperature at the previous moment, w A (t) represents the weight of the variance of activity at the previous moment, ΔH i (t) represents the recent health status deviation, w HR (t+1) represents the weight of the adjusted heart rate, w T (t+1) represents the weight of the adjusted body temperature, w A (t+1) represents the weight of the adjusted activity variance.

5. The method for adaptive intelligent control of poultry breeding environment according to claim 1, characterized in that: The adaptive clustering algorithm based on behavioral feature vector is expressed as follows: The set of behavioral feature vectors of individual birds is Where N is the number of individuals in the group, and the specific clustering process is achieved by minimizing the following clustering objective function: Among them, K represents the number of clusters, C k represents the behavior pattern of the kth class, μ k represents the center of the class, the objective function J represents the sum of the distances from all individual behavior feature vectors to the center of their respective classes, and B j Represents the behavioral feature vector at a certain moment; The time series model is modeled by a Markov behavior switching model.

6. The method for adaptive intelligent control of poultry breeding environment according to claim 1, characterized in that: Dynamic adjustment of the environmental control strategy according to the calculated comprehensive environmental control index D(t), where the comprehensive environmental control index D(t) is expressed as follows: Among them, W represents the weight vector, corresponding to the weight of each feature, and F(t) represents the multimodal feature vector; Define different threshold intervals according to comprehensive environmental control indicators and dynamically adjust the adjustment strategy of environmental parameters; The adjustment strategy of the environmental parameters after adjustment is compared with the adjustment strategy before adjustment. If the difference exceeds the preset tolerance range, the weight vector W is automatically adjusted, which is expressed as follows: Among them, α represents the learning rate, which controls the amplitude of weight adjustment; W(t+1) represents the updated weight, W(t) represents the current weight, D(t+1) represents the updated comprehensive environmental control index, and D(t) represents the comprehensive environmental control index at the previous moment.

7. The method for adaptive intelligent control of poultry breeding environment according to claim 1, characterized in that: The composite growth index G i (t), the calculation formula is as follows: Among them, α W Represents the weight coefficient of body weight, α S Represents the weight coefficient of body shape, α H Represents the weight coefficient of heart rate, W i (t) represents body weight data, S i (t) represents body size data, Indicates heart rate data.

8. The method for adaptive intelligent control of poultry breeding environment according to claim 7, characterized in that: The growth phase differentiation weight allocation algorithm is as follows: in, represents the environmental parameter value at the g-th growth stage, represents the specific parameter weight at this growth stage, E c (t) represents the current environment state, E (g) (t) represents the dynamic environment demand vector; A multi-objective optimization function based on regularization terms is introduced to deal with the conflict between dynamic environment requirements. The multi-objective optimization function is expressed as follows: in, represents the loss function of environmental regulation, represents the regularization term between different environmental parameters, and λ is the regularization coefficient, which is used to balance the relationship between environmental parameters; The growth stage feedback mechanism determines whether the adjustment of the current environment meets the growth target by monitoring the biological signals after the current environment is adjusted, as shown below: Where γ represents the feedback adjustment coefficient, ΔG i (t+1) represents the change in the adjusted growth index, ΔH i (t+1) represents the change of health status; The objective function of the comprehensive optimization mechanism of the multi-stage environmental control is expressed as follows: in, represents the optimized environmental parameters, and represent the actual and expected growth indicators, represents the growth standard deviation of this stage, represents the regularization term between parameters.

9. The method for adaptive intelligent control of poultry breeding environment according to claim 1, characterized in that: The adaptive multi-objective optimization model is constructed as follows: The control objectives of the system are integrated into an adaptive multi-objective optimization model, and the loss function L of each control objective is set n (Θ,t), where Θ is the parameter vector of the control strategy, n represents different control objectives, and the overall objective function L of multi-objective optimization is total (Θ,t) is represented as follows: Among them, λ n (t) represents the dynamic weight of each target, γ·∥Θ-Θ0∥ 2 represents a regular term, Θ0 represents the initial control parameter; According to the individual's biological signal, each individual is assigned a personalized control parameter Θ i (t), expressed as follows: Among them, α represents the learning rate, J i (t) represents the Jacobian matrix of individual i, which is used to describe the relationship between the individual physiological state and the control parameters. represents the gradient of the individual loss function, Θ i (t+1) represents the updated control parameters.

10. An adaptive intelligent control system for poultry breeding environment, characterized in that: The system comprises: The data acquisition module is used to install micro sensors on individual birds to collect biological signals of individual birds, pre-process the collected data, and then design a multi-dimensional dynamic boundary anomaly detection algorithm to correct the anomalies and obtain data. Design a dynamic feature weighted extraction method for data Perform weighted feature extraction; the features include heart rate mean Average body temperature and activity-weighted variance The data analysis modeling module is used to construct a health status model based on the health index by using the extracted features. i (t) thus obtaining the recent health status deviation ΔH i (t) Determine the past status of the poultry, and then design a feedback weight adaptive adjustment algorithm to adaptively adjust the weights in the health status model. Then, based on the health index H i (t) and recent health status deviation ΔH i (t) Classify the health status of poultry and adopt corresponding environmental control strategies; Among them, by calculating the health index H i (t) thus obtaining the recent health status deviation ΔH i (t) Determine the past status of the birds: If ΔH i (t) exceeds the preset dynamic threshold Θ H , it indicates that the poultry was in an abnormal or stressful state in the past. The group behavior monitoring and behavior pattern modeling module is used to collect the position coordinate data and velocity vector data of the poultry, and map the position coordinate data and velocity vector data to the behavior feature vector B in the feature space. j (t), using the adaptive clustering algorithm based on the behavior feature vector B j (t) Identify the behavior patterns of poultry groups, serialize the behavior patterns and build a time series model based on the behavior patterns to monitor the changes in the behavior patterns of the groups in real time; Multimodal data fusion and decision-making module, used to design multimodal data fusion and weighted decision-making model for health index H i (t), behavior feature vector B j (t) is fused with the environmental sensor data to obtain the multimodal feature vector F(t), and then the comprehensive environmental control index D(t) is generated through an adaptive weighting function. The environmental control strategy is dynamically adjusted according to the comprehensive environmental control index D(t), and real-time feedback monitoring is performed after the adjustment; Environmental differentiation regulation module, used to construct a composite growth index G for identifying growth stages based on biological signals i (t), according to the composite growth index G i (t) Design dynamic environment demand vector E (g) (t), and then design a differentiated weight allocation algorithm in the growth stage to dynamically adjust the dynamic environment demand vector E (g) The priority weight of (t) is determined, and after each adjustment, it is verified through a specific growth stage feedback mechanism. Finally, a comprehensive optimization mechanism for multi-stage environmental control is designed to integrate data from different growth stages. The system optimization module is used to design an adaptive multi-objective optimization model to dynamically adjust the control strategy according to real-time data, and then design an individual difference adaptive control model to assign personalized control parameters Θ to each individual according to biological signals. i (t), dynamically adjust control parameters based on historical data, and design multi-level feedback loops and emergency response mechanisms to ensure timely system response in emergency situations.

Images