Big data cloud-side collaboration-based laying hen breeding accurate feeding adjustment system

Abstract

The invention relates to the technical field of laying hen breeding, and discloses a laying hen breeding accurate feeding adjusting system based on big data cloud-side collaboration. The system comprises an environmental data acquisition module, a cloud edge collaborative analysis module, a multi-dimensional difference evaluation module, an abnormal area positioning module, a feeding strategy adjustment module and an encryption communication link module. The environment data acquisition module acquires chicken flock multi-source data streams through various sensors, and standard chicken flock data is generated after processing; a cloud-edge collaborative analysis module constructs a dynamic feed supply model according to the dynamic feed supply model and outputs theoretical feed demand; the multi-dimensional difference evaluation module performs multi-dimensional comparison on the theoretical value and the actual consumption value to generate a feed difference coefficient matrix; the abnormal area positioning module generates an abnormal feeding probability distribution diagram in combination with farm topological information; the feeding strategy adjusting module generates a self-adaptive feeding control signal based on the graph; the encryption communication link module transmits a control signal to an actuator through a quantum encryption channel. The system realizes accurate adjustment and safety control of feeding for laying hen breeding.

Description

Technical Field

[0001] This invention relates to the field of egg-laying hen farming technology, specifically to a precision feed supply adjustment system for egg-laying hen farming based on big data cloud-edge collaboration. Background Technology

[0002] In the large-scale egg-laying hen farming industry, the precision of feed supply is one of the key factors determining the success of the operation. It not only affects the economic benefits of farming but is also closely linked to the overall health of the flock, egg production quality, and sustainable development. In traditional egg-laying hen farming, feed supply adjustment has long relied on the personal experience of the farmers. Farmers need to adjust the quantity and frequency of feed by subjective judgment, such as observing the flock's activity, droppings, feather luster, and regularly checking the amount of feed remaining in the troughs. This manual method has many insurmountable drawbacks. On the one hand, the frequency and detail of manual observation are limited, making it difficult to capture subtle changes in the flock in real time. When the flock's feed demand fluctuates due to sudden environmental changes or health risks, adjustments are often not made in time, which may lead to stunted growth in some chickens due to insufficient feed or digestive system diseases due to excessive feed. On the other hand, large-scale farms are usually divided into multiple breeding areas, and environmental factors such as temperature, humidity, light intensity, and ventilation conditions vary in different areas. At the same time, the age, breed, and health status of the flocks are also different. These factors together lead to significant differences in feed demand in different areas, and the traditional uniform feeding model cannot take into account these individual needs, often resulting in feed accumulation and waste in some areas and feed shortages in others. With the application and promotion of information technology in agriculture, the egg-laying hen farming industry is gradually transforming towards intelligentization. Some farms have begun to introduce Internet of Things (IoT) technology, deploying temperature sensors, humidity sensors, light sensors, feed weight sensors, and flock activity monitoring equipment to collect environmental parameters and flock-related data. However, current data processing methods still have significant limitations. Most farms use local computing terminals for data processing, which, limited by the terminal's computing power and storage capacity, can only perform simple statistical analysis and cannot delve into the underlying correlations, such as the dynamic relationship between changes in environmental temperature and humidity and flock feed consumption, or the potential connection between abnormal flock health indicators and changes in feed demand. This results in a large amount of valuable data being idle and difficult to transform into effective feed supply adjustment criteria. Meanwhile, some farms attempting to connect to cloud platforms face challenges due to a lack of robust cloud-edge collaboration mechanisms. Data transmission between the cloud and edge devices suffers from latency and data format incompatibility, preventing timely feedback of big data analysis results from the cloud to the edge feed control equipment, thus affecting the timeliness and accuracy of feed supply adjustments. In formulating feeding strategies, existing technologies mostly rely on single-dimensional data for judgment. For example, they adjust the next feed amount solely based on the remaining feed or modify feeding parameters based on changes in ambient temperature. This ignores the complex interactions between multiple factors such as the environment, the flock, and the feed, resulting in one-sided feeding strategies that deviate significantly from the actual needs of the flock. Simultaneously, the security of breeding data and feeding control signals during transmission has not received sufficient attention. Traditional encryption methods are easily cracked, and data and signals may be illegally intercepted, tampered with, or interfered with. This can not only lead to malfunctions in the feeding system and economic losses but also potentially leak core breeding data, posing potential risks to the enterprise. These problems hinder the improvement of feed utilization efficiency in layer hen farming, restrict the growth quality and egg production performance of the flock, and severely impede the transformation of the layer hen farming industry towards a more efficient, precise, and sustainable modern model. Summary of the Invention

[0003] The purpose of this invention is to provide a precision feed supply regulation system for egg-laying hen farming based on big data cloud-edge collaboration, so as to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, this invention provides a precision feed supply regulation system for layer hen farming based on big data cloud-edge collaboration, the system comprising: The environmental data acquisition module is equipped with chicken health sensors, feed consumption sensors, and environmental monitoring sensors. It performs initial processing on the multi-source data stream of the chicken flock acquired by the environmental data acquisition module to generate standard chicken flock data. The cloud-edge collaborative analysis module constructs a dynamic feeding model based on the standard chicken flock data and outputs the theoretical feed demand. The multidimensional difference assessment module compares and analyzes the theoretical feed requirement with the actual feed consumption in multiple dimensions to generate a feed difference coefficient matrix. The abnormal area location module generates an abnormal feed supply probability distribution map based on the feed difference coefficient matrix and farm topology information. The feeding strategy adjustment module generates an adaptive feeding control signal based on the abnormal feeding probability distribution map; The encrypted communication link module uses a quantum encryption channel to transmit the adaptive feeding control signal to the feed dispensing actuator.

[0005] Preferably, the environmental data acquisition module includes: The flock health sensor collects flock weight data, feeding frequency data, and egg production rate data. Temperature, humidity, and light intensity data are collected using the environmental monitoring sensors. Noise filtering is performed on the collected multi-source data streams of chickens, and abnormal data points are removed according to the preset data quality threshold. The filtered multi-source chicken data stream is time-series aligned to output the standard chicken data.

[0006] Preferably, the cloud-edge collaborative analysis module includes: The dynamic feeding model is trained based on historical flock operation data. The dynamic feeding model includes a time-series prediction unit and a feature compensation unit. The standard chicken flock data is acquired in real time via edge computing nodes; The standard chicken flock data is processed in batches using an incremental data batch processing algorithm to extract flock status features; The flock status characteristics are input into the dynamic feeding model, and the theoretical feed requirement is output.

[0007] Preferably, the multidimensional difference assessment module includes: The theoretical feed requirement and the actual feed consumption are calculated by performing a time-domain cumulative deviation calculation to generate a time-domain deviation vector. Frequency domain energy shift detection is performed on the theoretical feed demand and the actual feed consumption to generate a frequency domain shift vector; The sequence similarity between the theoretical feed demand and the actual feed consumption is evaluated based on the sequence structure matching algorithm, and a sequence similarity vector is generated. The time-domain deviation vector, frequency-domain offset vector, and sequence similarity vector are fused using tensors to generate the feed difference coefficient matrix.

[0008] Preferably, the abnormal area location module includes: A node-based topology network was constructed based on the layout of the farm, and the location information of the chicken houses and the impedance parameters of the feed delivery path were marked. Map the feed difference coefficient matrix to the corresponding nodes of the nodeized topology network; A graph neural network was used to deduce the propagation path of abnormal material supply, and the abnormal attenuation factor was calculated based on the node impedance parameters. Generate a probability distribution map of the abnormal feed supply covering the entire farm, and identify high-probability abnormal areas.

[0009] Preferably, the feeding strategy adjustment module includes: Configure the material supply verification parameters according to the abnormal material supply probability distribution map; A high-frequency monitoring mode is activated for the high-probability anomaly areas; A swarm intelligence optimizer is used to adjust the material supply control rules; The adaptive feeding control signal is generated based on the adjusted feeding control rules.

[0010] Preferably, the feeding strategy adjustment module further includes: The feed rate adjustment value is calculated using a temperature drop search algorithm; The feed rate adjustment value is converted into the adaptive feed control signal; Based on preset feed constraints, a gradient projection adjuster is used to optimize the feed distribution.

[0011] Preferably, the gradient projection adjuster includes: Define a set of constraints for material supply; Calculate the subgradient of the objective function based on the current material supply plan; Update the material supply plan through projection operations to meet all constraints; The optimized feeding scheme is output to the adaptive feeding control signal.

[0012] Preferably, the system further includes: The feedback detection module monitors the adjusted feed consumption data through a recursive state estimator. The monitoring data is fed back to the cloud-edge collaborative analysis module and the material supply strategy adjustment module; The material supply strategy is updated based on feedback data.

[0013] Preferably, the feedback detection module includes: Collect flock response data after feed distribution; The chicken flock response data is processed by a recursive state estimator to generate state estimation results. The state estimation results are input into the multidimensional difference assessment module for difference analysis iteration.

[0014] Compared with the prior art, the beneficial effects of the present invention are: By configuring multiple sensors in the environmental data acquisition module, the multi-source data streams of the chicken flock are initially processed and standard chicken flock data is generated. This enables comprehensive perception and standardized processing of the chicken flock's growth environment, health status, and feed consumption, transforming the originally scattered and messy data into effective information with unified standards, and providing reliable basic data for subsequent analysis. The cloud-edge collaborative analysis module builds a dynamic feeding model based on standard chicken flock data and outputs the theoretical feed demand. It combines the real-time nature of edge computing with the big data processing capabilities of cloud computing, enabling the feeding model to be dynamically adjusted according to real-time data and historical big data of the chicken flock. The output theoretical feed demand is more in line with the actual needs of the chicken flock, breaking the limitations of traditional local computing or single cloud analysis in terms of timeliness and data processing depth. The multidimensional difference assessment module compares and analyzes the theoretical feed requirement with the actual feed consumption in multiple dimensions and generates a feed difference coefficient matrix. It takes into account the influence of various factors such as environment and flock health on feed consumption. Through multidimensional comparative analysis, it clearly presents the difference between theoretical and actual consumption and the potential correlation that causes the difference, avoiding the one-sided conclusions that may be caused by single-dimensional analysis. The abnormal area location module generates an abnormal feed supply probability distribution map based on the feed difference coefficient matrix and farm topology information. This can accurately pinpoint areas in the farm where there are abnormal feed supply issues, allowing staff to intuitively understand the distribution range and probability of abnormal situations. This facilitates targeted measures and reduces the waste of manpower and time caused by blind investigations. The feeding strategy adjustment module generates an adaptive feeding control signal based on the abnormal feeding probability distribution map, enabling the feeding adjustment to dynamically adapt to the specific situation of the abnormal area. This achieves the personalization and precision of the feeding strategy and avoids the problem of excessive or insufficient feeding in some areas that may occur under the unified feeding mode. The encrypted communication link module uses a quantum encryption channel to transmit adaptive feeding control signals to the feed dispensing actuator. Utilizing the high security of quantum encryption technology, it ensures that the control signals will not be interfered with, stolen, or tampered with during transmission, guaranteeing the secure and reliable transmission of feeding system control commands and maintaining the stable operation of the entire feeding regulation system. Attached Figure Description

[0015] Figure 1 This is a timing diagram of the precision feed supply regulation system for egg-laying hen farming based on big data cloud-edge collaboration as described in this invention; Figure 2 A flowchart for data processing in the environmental data acquisition module; Figure 3 A flowchart for modeling and prediction in the cloud-edge collaborative analysis module; Figure 4 A flowchart for abnormal location of the abnormal area location module; Figure 5 A flowchart for optimizing the gradient projection adjuster. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figure 1This invention provides a precision feed adjustment system for egg-laying hen farming based on big data cloud-edge collaboration. The system includes: an environmental data acquisition module, a cloud-edge collaborative analysis module, a multi-dimensional difference assessment module, an abnormal area location module, a feed strategy adjustment module, and an encrypted communication link module.

[0018] The environmental data acquisition module, equipped with chicken health sensors, feed consumption sensors, and environmental monitoring sensors, performs initial processing on the acquired multi-source data streams from the chicken flock to generate standard flock data. The cloud-edge collaborative analysis module constructs a dynamic feeding model based on the standard flock data, outputting the theoretical feed requirement. The multi-dimensional difference assessment module performs multi-dimensional comparative analysis between the theoretical feed requirement and actual feed consumption, generating a feed difference coefficient matrix. The abnormal area location module generates an abnormal feeding probability distribution map based on the feed difference coefficient matrix and farm topology information. The feeding strategy adjustment module generates adaptive feeding control signals based on the abnormal feeding probability distribution map. The encrypted communication link module uses a quantum-encrypted channel to transmit the adaptive feeding control signals to the feed dispensing actuator. This system, through modular design, achieves collaborative operation of data acquisition, analysis, assessment, location, adjustment, and communication, ensuring the accuracy and security of the feeding process.

[0019] Example 1: See Figure 2 The environmental data acquisition module is implemented through a variety of sensor devices deployed within the egg-laying hen farm. Chicken health sensors are installed in a distributed manner, with monitoring points placed around perches, feeding areas, and laying boxes in the henhouse. Weight monitoring uses an embedded weighing platform; when a chicken stands on the platform, a pressure sensor collects the weight signal, and the data is transmitted wirelessly to the acquisition terminal via ZigBee. Feeding frequency monitoring uses an infrared beam array to form a monitoring area at the feed trough entrance, recording the number of times and the duration of the chicken's head passing through the beam. Each feeding event is timestamped and associated with an individual RFID tag. Egg production rate data acquisition is achieved through a photoelectric counter at the end of the egg conveyor belt. Each egg triggers a counting pulse when it passes through the detection area, while a weight sensor distinguishes abnormal eggs.

[0020] An environmental monitoring sensor network covers all functional areas of the chicken house. Temperature and humidity sensors are wall-mounted, 1.5 meters above the ground, with one monitoring point per 100 square meters. Sensor probes are protected by external covers to prevent dust contamination. Light intensity sensors are installed on the roof trusses of the chicken house, evenly distributed and avoiding direct lighting areas, collecting combined intensity values ​​of natural and artificial light. All sensor data is collected once per minute and initially aggregated through an industrial-grade IoT gateway.

[0021] The raw data stream first passes through a hardware filter to eliminate power frequency interference, and then a software filtering algorithm runs on the gateway processor. For weight data, a sliding window midpoint filter is used, with a window width set to 10 sampling points to eliminate instantaneous fluctuations caused by chicken movement. Feeding frequency data uses threshold filtering to remove abnormal records lasting less than 1 second or more than 300 seconds. Environmental data is smoothed using Kalman filtering to reduce the impact of random interference. The acceptable range for weight data is set at 1.5 to 4.5 kg; data points outside this range are marked as invalid. The reasonable threshold for feeding frequency is dynamically adjusted based on historical data, and outliers are excluded using the 3σ principle. Temperature and humidity data are validated based on a local climate model, with an acceptable temperature range of -5℃ to 45℃ and a humidity range of 30% to 85%. After all invalid data points are removed, the system automatically triggers a data re-collection mechanism.

[0022] The time series alignment process employs an NTP-based time synchronization scheme, with each sensor node undergoing time calibration with the central server every 24 hours, and the maximum permissible deviation controlled within 100 milliseconds. The data alignment algorithm uses the feeding event as the baseline time axis, resampling data streams from different sensors onto a unified time grid. The aligned data is organized at the minute granularity, forming a standard flock data packet containing timestamps, sensor types, data values, and quality indicators.

[0023] Raw sampling data is stored in a circular buffer on the edge computing node, retaining the full data for the most recent 72 hours. Preprocessed data is uploaded to the cloud platform's object storage system and stored in date-partitioned locations. Data indexing is based on a combination key of chicken coop number, date, and time period, supporting multi-dimensional fast queries. CRC checksums and retransmission mechanisms are used during data transmission to ensure data integrity. Daily scheduled sensor health checks are performed, including circuit self-tests, signal strength detection, and data consistency verification. Faulty sensors automatically trigger alarms and switch to backup nodes; system maintenance logs record all device status changes. Sensor calibration is performed every 30 days using standard calibration equipment on-site, and calibration data is entered into the device management database.

[0024] It provides a RESTful API interface for cloud-edge collaborative analysis modules to call, supporting combined queries by time range, sensor type, and spatial location. It communicates with the sensor network via the Modbus protocol, supporting real-time parameter configuration and data acquisition control. All data exchanges are encapsulated in JSON format, including metadata description information.

[0025] The system deployment takes into account the actual environmental conditions of the farm. Sensors are selected to meet IP67 protection standards, adapting to high dust and high humidity environments. Wireless transmission employs multi-band adaptive technology to avoid interference from industrial equipment. The power supply system uses a solar-assisted power solution to ensure continuous operation during cloudy or rainy weather. Equipment installation locations are avoided from areas where chickens directly contact the equipment, and anti-pecking designs are incorporated.

[0026] The data processing algorithms are optimized to adapt to the limitations of edge computing resources. The filtering algorithm uses fixed-point arithmetic to reduce computational overhead, and the lightweight LZ77 algorithm is used for data compression to reduce storage space. Memory management employs pooling technology to avoid frequent memory allocation, and real-time performance monitoring tracks processing latency and resource usage. When processing latency exceeds a threshold, the system automatically degrades to a simplified processing mode. Data quality control establishes a multi-level supervision mechanism: edge nodes perform real-time data verification, the cloud platform performs batch data quality analysis, and data quality reports are generated periodically. Anomaly root cause analysis tools help locate sensor malfunctions or environmental interference issues, and maintenance personnel perform equipment repairs or parameter adjustments based on system prompts. Historical data quality indicators are incorporated into the equipment performance evaluation system. Columnar storage is used to improve compression efficiency, a dedicated index structure supports time-range queries, and the data partitioning strategy is automatically managed based on a time sliding window. A data caching mechanism retains hot data in memory to reduce query response time. A data archiving strategy transfers data older than one year to a low-cost storage system.

[0027] Sensor nodes support hot-addition; new nodes automatically register with the system and begin data acquisition upon connection. Processing capacity supports horizontal scaling; as the scale of farming expands, the processing load can be shared by adding edge computing nodes. All configuration information is centrally managed, supporting batch deployment and unified configuration updates. Real-time display of sensor operating status and data quality indicators is provided, with abnormal states highlighted and handling suggestions offered. Historical data trend charts support multi-parameter overlay display, aiding in the analysis of the correlation between environmental parameters and flock behavior. Data export supports standard format output for easy use by third-party analysis tools. Data acquisition cycles avoid water mist disinfection periods to prevent sensor reading distortion. Equipment maintenance is synchronized with idle periods in the poultry house, minimizing impact on production activities. Wireless transmission power is optimized based on the poultry house layout, ensuring signal coverage while reducing energy consumption. All equipment casings are made of corrosion-resistant materials, suitable for use with cleaning agents in the farming environment.

[0028] Example 2: See Figure 3The implementation of the cloud-edge collaborative analysis module is based on a collaborative architecture of edge computing nodes deployed locally on the farm and a cloud server cluster. The edge computing nodes utilize industrial-grade embedded systems, equipped with multi-core processors and dedicated neural network accelerator cards, and are installed in the control cabinet of each chicken house. These nodes connect to the sensor network via gigabit Ethernet, receiving standard flock data transmitted in real time from the environmental data acquisition module. The cloud server cluster is deployed on a cloud computing platform, employing a distributed architecture, and includes model training nodes and inference service nodes.

[0029] The training process for the dynamic feeding model utilizes historical flock operation data, covering the entire production cycle of the past twelve months. The training dataset includes parameters sampled every minute, such as flock weight, feeding frequency, egg production rate, temperature and humidity, and light intensity, along with corresponding actual feed feeding records. Model training begins with feature engineering, extracting meaningful features from the raw data, including time-series features such as moving averages and trend slopes, and statistical features such as variance and peak value. During training, k-fold cross-validation is used to prevent overfitting, and model parameters are optimized using a backpropagation algorithm.

[0030] The time-series prediction unit employs a Long Short-Term Memory (LSTM) network structure, consisting of an input layer, three hidden layers, and an output layer. The input layer receives standardized time-series data, the hidden layers use LSTM units to capture long-term dependencies, and the output layer generates feed demand forecasts for the next 24 hours. Network training utilizes the Adam optimizer with an exponentially decaying learning rate. The feature compensation unit integrates a random forest algorithm to handle the non-linear relationship between environmental factors and feed demand. This unit receives real-time environmental data and outputs compensation coefficients to adjust the baseline prediction values.

[0031] Edge computing nodes acquire the latest standard flock data from the local sensor network every five minutes. Data acquisition uses a publish-subscribe model, with nodes registering as data consumers and pushing new data to the edge nodes immediately upon generation. The incremental data batch processing algorithm employs a sliding window mechanism, with a window size of 60 sampling points (corresponding to 1 hour of data) and a window sliding step of 5 minutes. Each data batch undergoes preprocessing, including missing value imputation and outlier handling, before being fed into the feature extraction pipeline. Flock status features are categorized into three main types: behavioral features, physiological features, and environmental features. Behavioral features extract the frequency, duration, and regularity of feeding activities; physiological features calculate weight change rate, egg production trend, and health indicators; environmental features comprehensively consider the influence of parameters such as temperature, humidity, and light. Feature extraction uses an autoencoder for dimensionality reduction, retaining the most important feature representations. The extracted feature vectors are then standardized before being fed into the model inference engine.

[0032] The time-series forecasting unit first processes the time-series data to generate a baseline feed demand forecast. The feature compensation unit simultaneously analyzes environmental parameters and calculates the degree of influence of environmental factors on feed demand. The outputs of the two units are combined through a weighted fusion module to produce the final theoretical feed demand. The theoretical value is expressed in kilograms at five-minute intervals, along with a confidence score.

[0033] The multidimensional variance assessment module is deployed on edge computing nodes, receiving theoretical feed requirements and actual values ​​from feed consumption sensors. Actual values ​​are measured by weight sensors and flow meters, with actual feed amounts recorded every five minutes. The time-domain cumulative deviation calculation uses a sliding window integration method with a 24-hour window size to calculate the cumulative difference between theoretical and actual values. This calculation produces a time-domain deviation vector containing deviation indices across multiple time scales, from minute-level to hour-level cumulative deviation.

[0034] Frequency-domain energy shift detection transforms the time-domain signal to the frequency domain using a Fast Fourier Transform (FFT). Theoretical demand and actual consumption data are first resampled to achieve the same time resolution. Then, the power spectral density of both is calculated, comparing the energy distribution differences across different frequency components. Frequency-domain analysis can reveal periodic supply-demand imbalance patterns that might be masked in time-domain analysis. The detection results generate a frequency-domain shift vector containing the amplitude and phase differences of the main frequency components.

[0035] The sequence structure matching employs a dynamic time warping algorithm, which can handle nonlinear deformations between time series. The algorithm first constructs a distance matrix between the theoretical and actual sequences, then searches for the optimal warping path. During the matching process, shape similarity and trend consistency of the sequences are considered to generate a similarity score. This score, after standardization, is added to the sequence similarity vector, which also includes local and global similarity indices. The tensor fusion process combines the time-domain bias vector, frequency-domain offset vector, and sequence similarity vector into a multi-dimensional array. The fusion algorithm uses an attention mechanism to assign adaptive weights to features of different dimensions. The weight coefficients are learned from historical data, reflecting the importance of each dimension's features to the final difference assessment. The fused tensor is normalized to generate a feed difference coefficient matrix. This matrix contains comprehensive difference information, reflecting the degree of inconsistency between theoretical demand and actual consumption from multiple perspectives.

[0036] Resource allocation for edge computing nodes is optimized to ensure real-time data processing and model inference. The model update mechanism supports online learning, automatically triggering model retraining when pattern changes are detected. Data stream processing employs a backpressure control mechanism to prevent processing delays caused by data congestion. All processing steps are logged and monitored for easy troubleshooting and performance optimization. The communication protocol uses a lightweight message queue to ensure efficient data transmission between edge nodes and the cloud. Security mechanisms include data encryption and authentication to prevent unauthorized access. The system supports smooth scaling; the number of edge nodes can be increased without affecting the overall architecture as the farming scale expands. Maintenance tools provide remote monitoring and fault diagnosis capabilities, reducing operational costs.

[0037] Example 3: See Figure 4 The implementation of the abnormal area location module is based on the construction of a digital twin model of the farm's physical layout. This module first divides the farm into several topological nodes, each corresponding to a physical area, typically a chicken house, and also including key control points along the feed delivery path. The node-based topological network is represented by a graph structure, where vertices represent spatial locations and edges represent connections. Each vertex attribute includes information such as chicken house number, spatial coordinates, and capacity specifications; each edge attribute includes impedance parameters such as pipe length, diameter, inclination angle, and number of bends. The calculation of impedance parameters incorporates comprehensive fluid dynamics characteristics, with the main resistance coefficient determined by pipe material, internal smoothness, and fluid properties. During network construction, survey data is used to establish a precise coordinate reference system to ensure consistency between the physical space and the digital model.

[0038] The mapping process of the feed variance coefficient matrix is ​​implemented through a spatial registration algorithm. Each topological node is associated with the corresponding feed variance coefficient based on the physical area it serves. The mapping algorithm uses the nearest neighbor matching principle to establish a correspondence between each variance value in the matrix and the spatially closest topological node. For regions spanning multiple nodes, a weighted allocation strategy is used, with the weights determined according to the proportion of the service area. After mapping, each topological node carries a variance coefficient value, which constitutes the node feature vector input to the graph neural network.

[0039] Graph neural networks employ a message-passing architecture to process topological networks. The network consists of an input layer, three graph convolutional layers, and an output layer. The graph convolutional layers update node representations using the following aggregation function:

[0040] in: Indicates the first Layer nodes eigenvectors, It is a node The set of neighboring nodes, It is a node and Edge weights between them It is a trainable weight matrix. It's an activation function. Edge weights. It is calculated from impedance parameters and reflects the physical characteristics of the connection between nodes.

[0041] The anomaly propagation simulation employs a random walk algorithm, which starts from nodes with high dissimilarity coefficients and propagates along topological edges. An anomaly decay factor is introduced during the propagation process. This factor is calculated based on the impedance parameter: ,in Represents a node arrive The path impedance value, This is the attenuation factor, obtained through training based on historical data. The propagation probability is inversely proportional to the attenuation factor; the higher the impedance, the lower the propagation probability.

[0042] The probability distribution map generation uses a kernel density estimation method, taking the node anomaly probability output by the graph neural network as input and employing a Gaussian kernel function for spatial smoothing. The bandwidth parameter of the kernel function is optimized based on the spatial scale of the farm to ensure the continuity and accuracy of the probability distribution. The generated probability distribution map is visualized as a heatmap, using color gradients to represent the anomaly probability, with darker areas indicating high-probability anomaly zones. The feeding strategy adjustment module formulates control strategies based on the probability distribution map, verifying parameter configurations including monitoring frequency, sampling accuracy, and response thresholds. For areas with a probability value exceeding 0.7, a high-frequency monitoring mode is activated, reducing the data acquisition interval from 5 minutes to 30 seconds. The monitoring range is extended to related areas, including upstream feeding paths and adjacent chicken houses.

[0043] The swarm intelligence optimizer employs an improved particle swarm optimization algorithm. During algorithm initialization, each particle represents a possible feed adjustment scheme, with the scheme dimension matching the number of topology nodes. The fitness function comprehensively considers the variance coefficient, nutrient requirements, and operating costs.

[0044] Where: F(x) represents the value of the fitness function, This is the feed rate vector, where each component xᵢ represents the system's suggested adjusted feed rate for the i-th node (region) in the topology network. This is the theoretical demand value. It is the lower limit of nutritional requirements. It is the unit cost coefficient. It is a weight parameter. Inertial weights and social learning mechanisms are introduced during the particle update process to gradually converge to the optimal solution. n represents the total number of nodes (i.e., chicken houses or key control areas) in the farm topology network.

[0045] The adaptive feeding control signals are generated using a digital instruction format, which includes a target area identifier, feeding quantity, execution timestamp, and priority flag. The instruction encoding uses JSON format, containing header information and a data payload. The header information records the instruction sequence number and generation time, while the data payload contains the specific control parameters. Signals are digitally signed and authenticated before transmission to ensure the integrity and authenticity of the instructions. When a high-probability abnormal area is detected, the system automatically triggers an early warning mechanism, notifying management personnel to conduct on-site verification. A detailed diagnostic report is also generated, including the anomaly type, possible causes, and handling suggestions. All operation records are stored in an audit log, supporting post-event analysis and process optimization.

[0046] The maintenance mechanism includes regular calibration and model updates. The topology network is updated monthly based on actual layout changes, and impedance parameters are dynamically adjusted according to pipeline usage. The graph neural network is retrained quarterly, incorporating the latest production data. The system supports remote upgrades and configuration updates, ensuring long-term reliability and accuracy. The monitoring interface provides interactive anomaly analysis tools, allowing users to view probability distribution maps through a visual interface, drill down to detailed data at specific nodes, and simulate the effects of different control strategies. The system offers historical comparison functionality, supporting comparative analysis between the current state and historical periods to help identify periodic anomaly patterns. Security mechanisms include access control and operation auditing; users with different permissions have different operational scopes, and critical operations require dual authentication. All configuration modifications are logged, supporting change tracking and accountability. Data backup employs an incremental backup strategy to ensure system data integrity and recoverability.

[0047] Example 4: See Figure 5 The implementation of Example 4 involves the coordinated operation of the temperature decrease search algorithm and the gradient projection adjuster in the feeding strategy adjustment module. This module receives an abnormal feeding probability distribution map from the abnormal region location module, which identifies three high-probability abnormal regions (regions A, B, and C) with abnormal probabilities of 0.82, 0.76, and 0.68, respectively. During system initialization, the parameters of the temperature decrease search algorithm are set: initial temperature T0 is 1000, temperature decay coefficient α is 0.95, and the maximum number of iterations is 200. The algorithm uses the current feeding scheme as the initial solution, which includes the baseline feeding amount for each region.

[0048] The temperature-decreasing search algorithm begins its iterative process. In the first iteration, the algorithm randomly generates a new solution, increasing the feed supply to area A by 5%, decreasing it to area B by 3%, and keeping it unchanged in area C. The objective function value corresponding to this solution is calculated, which is an evaluation value that comprehensively considers the difference coefficient matrix, nutritional requirements, and operating costs. If the objective function value of the new solution is better than the current solution, it is accepted directly; if it is worse, it is accepted with a certain probability according to the Metropolis criterion, and this probability decreases as the temperature decreases. After multiple iterations, the algorithm gradually converges to a better solution and outputs the adjusted feed supply values ​​for each area.

[0049] The conversion process employs a 12-bit precision digital-to-analog converter to convert digital signals into 4-20mA analog signals, corresponding to the fully closed to fully open states of the feed actuator. Signal transmission utilizes an industrial fieldbus protocol, including checksums and timing control information to ensure accurate delivery of commands to the actuator. A gradient projection adjuster then optimizes the allocation of these adjustment values. This adjuster sets three types of constraints: first, upper and lower limits for feed quantity, requiring each region's feed quantity to fall between the minimum nutritional requirement and maximum capacity; second, a total supply constraint, ensuring the sum of feed quantities across all regions does not exceed the current available feed silo capacity; and third, a priority constraint, giving priority to regions with high anomaly probabilities. The adjuster first calculates the subgradient of the objective function of the current feed scheme, then maps the scheme to the feasible region through a projection operation. The projection operation uses an iterative algorithm, moving the current solution along the feasible direction in each iteration to gradually satisfy all constraints. When a constraint violation is detected, the adjuster adjusts the allocation of each region according to weighted coefficients, prioritizing the feed demand of high-anomaly regions. After multiple projection iterations, an optimized feed scheme satisfying all constraints is finally obtained. The system continuously monitors its performance during actual operation. See Table 1 for the material supply adjustment plan.

[0050] Table 1: Material Supply Adjustment Plan.

[0051] Area code abnormal probability Current feed rate (kg / h) Adjustment value (kg / h) Optimized feed rate (kg / h) Status indicator A-12 0.82 125.6 +8.4 134.0 urgent B-07 0.76 98.3 +5.2 103.5 urgent C-15 0.68 112.8 +3.1 115.9 warn D-09 0.45 87.2 -2.8 84.4 normal E-21 0.31 76.5 -1.5 75.0 normal The regulator maintains a real-time constraint database during operation, containing dynamic parameters for each region: the minimum feed rate is dynamically calculated based on the flock's age and egg production rate; the maximum feed rate considers the feeder's physical capacity and the flow rate in the delivery pipeline; and priority weights are updated in real-time according to the probability of anomalies. The database is synchronized with the latest data every 5 minutes to ensure the timeliness of the constraints. The adaptive feed control signal generation adopts a multi-level structure. The first layer contains basic control instructions, specifying the target region and feed rate; the second layer adds execution parameters, including feed rate, duration, and smooth transition curve; the third layer contains safety verification information, such as the maximum allowable adjustment range and emergency stop conditions. The signal encapsulation uses the ASN.1 encoding format to ensure the reliability of data transmission and the consistency of parsing.

[0052] The system implementation took into account the operational needs of actual farms. The parameters of the temperature decay search algorithm can be adjusted according to different seasons and flock stages: a faster temperature decay rate is used in summer, and the number of iterations is increased during peak egg production. The constraint set of the gradient projection adjuster can be modified online, and farm managers can adjust nutritional requirement parameters or equipment capacity limits through a human-machine interface.

[0053] The anomaly handling mechanism includes multiple protection levels. When the adjuster detects that all constraints cannot be met, it first tries to reduce the material supply to low-priority areas. If this is still not feasible, a renegotiation mechanism is triggered to request the upstream module to revise the anomaly probability threshold. The final solution is to activate a tiered response strategy to prioritize ensuring the minimum material supply requirements of high-anomaly areas.

[0054] The system deployment employs a distributed architecture. The temperature decrease search algorithm runs on edge computing nodes for rapid response; the gradient projection adjuster is deployed on a regional server to coordinate feed distribution across multiple livestock sheds; and the control signal generation module resides within the field controller, directly driving the actuators. This architecture ensures both computational efficiency and system reliability. The monitoring system records the entire adjustment process. The candidate solutions, objective function values, and acceptance decisions for each iteration are recorded in the runtime log. The iterative process of the projection operation and constraint satisfaction are also recorded in detail. This data is used for subsequent analysis of algorithm performance and optimization of parameter settings. A visual interface displays the feed adjustment process in real time, including information such as current temperature, iteration count, and constraint satisfaction status.

[0055] The maintenance mechanism includes regular calibration and algorithm updates. The parameters of the temperature drop search algorithm are re-optimized quarterly, and the effects of different parameter combinations are tested using historical data. The constraints of the gradient projection adjuster are reviewed and updated monthly to reflect equipment modifications or process changes. The system supports remote diagnostics and parameter adjustment, reducing the need for on-site maintenance.

[0056] Example 5: The implementation of the feedback detection module is based on continuous monitoring of the system response after feed supply adjustments. This module collects flock response data after feed distribution through a sensor network deployed throughout the chicken house. This data includes multi-dimensional indicators such as changes in feeding behavior, weight gain trends, egg production performance, and group activity patterns. The sensor network consists of high-precision weighing devices, infrared activity monitors, sound collectors, and video analysis equipment, covering the entire breeding area in a distributed architecture. The data acquisition frequency is dynamically adjusted according to the magnitude of the feed supply adjustment. When a significant change in feed supply is detected, the system automatically increases the data sampling rate to once per minute to capture more detailed response characteristics.

[0057] The recursive state estimator uses a Kalman filter framework to process flock response data. This estimator maintains a state-space model containing latent variables such as flock health status, nutrient intake level, and environmental fitness. Each estimation round consists of two phases: prediction and update. The prediction phase extrapolates state change trends based on historical data, while the update phase integrates the latest observation data to correct the estimation results. The state estimator can handle noise and missing values ​​in sensor data, obtaining the most probable state sequence through probabilistic inference. The estimation results are output in the form of confidence intervals, reflecting the reliability of the estimation results.

[0058] The state estimation results are transmitted in real time to the cloud-edge collaborative analysis module and the feeding strategy adjustment module. The transmission process uses a lightweight messaging protocol to ensure the timeliness and integrity of data delivery. After receiving the state estimation results, the cloud-edge collaborative analysis module combines them with historical operational data for parameter updates to the dynamic feeding model. Model updates employ an online learning mechanism, adjusting neural network weights through incremental training to enable the model to adapt to changes in the flock's state. The feeding strategy adjustment module then uses the state estimation results to reassess the effectiveness of the current control strategy, triggering a strategy adjustment process when necessary.

[0059] Feedback data undergoes iterative analysis in the multidimensional difference assessment module. This process compares the state estimation results with the expected flock response and calculates a new difference coefficient matrix. The iterative analysis employs a sliding window mechanism, with the window size dynamically adjusted based on data characteristics, typically including data records from the most recent 24 hours. Special attention is paid to the time delay effect between feed adjustments and flock response during the analysis, establishing a delay compensation mechanism to accurately assess the effectiveness of control measures. The new difference coefficient matrix not only reflects the current supply-demand difference but also includes information on the effectiveness of control measures. The system establishes a closed-loop control architecture, enabling continuous optimization of the feed strategy based on real-time feedback. After each feed adjustment, the system initiates a complete monitoring-evaluation-adjustment cycle. This cycle typically lasts 6 to 12 hours, with the specific duration depending on the severity of the flock response and the sufficiency of data collection. During the cycle, the system records all relevant data, including changes in feed quantity, flock behavioral responses, and environmental parameter fluctuations, providing comprehensive data support for subsequent analysis.

[0060] Data management utilizes a time-series database to store all feedback data. The database design supports efficient time-range queries and streaming data processing, enabling rapid retrieval of flock response records within specific time periods. Data indexes are built based on timestamps and sensor types, supporting multi-dimensional data aggregation and analysis. All data records include quality identifiers indicating data reliability and completeness.

[0061] An anomaly detection mechanism continuously monitors the feedback data stream. When an abnormal response pattern is detected, the system automatically triggers an early warning process. Early warnings are categorized into multiple levels, from minor deviations to severe anomalies, each corresponding to a different handling strategy. For minor deviations, the system automatically fine-tunes parameters; for severe anomalies, manual intervention is required for in-depth analysis and handling. All early warning events are recorded in the event log, including the occurrence time, anomaly type, handling measures, and final result. System performance evaluation is based on statistical analysis of long-term operational data. The effectiveness of the control strategy is assessed by comparing flock status indicators before and after feed adjustment. Evaluation indicators include production performance parameters such as feed conversion ratio, egg production consistency, and flock uniformity. These evaluation results are used to optimize system parameters and improve the control algorithm, forming a virtuous cycle of continuous improvement.

[0062] The maintenance mechanism ensures the long-term stable operation of the system. Regular sensor calibration and data quality checks guarantee the accuracy of input data. The parameters of the recursive state estimator are retrained quarterly, incorporating the latest production data. The system provides comprehensive diagnostic tools to detect issues such as data flow anomalies, calculation biases, and communication failures. All maintenance operations are logged in the system log, supporting fault tracing and performance analysis.

[0063] The human-computer interface provides a visual display of feedback data. Farm managers can view flock response trend charts, status estimation results, and difference analysis reports. The interface supports drill-down queries, allowing users to access detailed records from aggregated data down to individual chicken houses. The system also provides automatic report generation, regularly generating operational summaries and performance evaluation reports. Security mechanisms protect the integrity and privacy of feedback data. Data transmission uses encryption protocols to prevent unauthorized access. The access control system restricts data viewing and operation permissions for different users. Data backup employs an incremental backup strategy to ensure rapid recovery in the event of system failure. All data operations are recorded in the audit log, supporting security incident investigations and accountability.

[0064] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0065] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A precision feed supply regulation system for layer hen farming based on big data cloud-edge collaboration, characterized in that, include: The environmental data acquisition module is equipped with chicken health sensors, feed consumption sensors, and environmental monitoring sensors. It performs initial processing on the multi-source data stream of the chicken flock acquired by the environmental data acquisition module to generate standard chicken flock data. The cloud-edge collaborative analysis module constructs a dynamic feeding model based on the standard chicken flock data and outputs the theoretical feed demand. The multidimensional difference assessment module compares and analyzes the theoretical feed requirement with the actual feed consumption in multiple dimensions to generate a feed difference coefficient matrix. The abnormal area location module generates an abnormal feed supply probability distribution map based on the feed difference coefficient matrix and farm topology information. The feeding strategy adjustment module generates an adaptive feeding control signal based on the abnormal feeding probability distribution map; The encrypted communication link module uses a quantum encryption channel to transmit the adaptive feeding control signal to the feed dispensing actuator.

2. The precision feed supply regulation system for egg-laying hen farming based on big data cloud-edge collaboration as described in claim 1, characterized in that, The environmental data acquisition module includes: The flock health sensor collects flock weight data, feeding frequency data, and egg production rate data. Temperature, humidity, and light intensity data are collected using the environmental monitoring sensors. Noise filtering is performed on the collected multi-source data streams of chickens, and abnormal data points are removed according to the preset data quality threshold. The filtered multi-source chicken data stream is time-series aligned to output the standard chicken data.

3. The precision feed supply regulation system for egg-laying hen farming based on big data cloud-edge collaboration as described in claim 2, characterized in that, The cloud-edge collaborative analysis module includes: The dynamic feeding model is trained based on historical flock operation data. The dynamic feeding model includes a time-series prediction unit and a feature compensation unit. The standard chicken flock data is acquired in real time via edge computing nodes; The standard chicken flock data is processed in batches using an incremental data batch processing algorithm to extract flock status features; The flock status characteristics are input into the dynamic feeding model, and the theoretical feed requirement is output.

4. The precision feed supply regulation system for egg-laying hen farming based on big data cloud-edge collaboration as described in claim 3, characterized in that, The multidimensional difference assessment module includes: The theoretical feed requirement and the actual feed consumption are calculated by performing a time-domain cumulative deviation calculation to generate a time-domain deviation vector. Frequency domain energy shift detection is performed on the theoretical feed demand and the actual feed consumption to generate a frequency domain shift vector; The sequence similarity between the theoretical feed demand and the actual feed consumption is evaluated based on the sequence structure matching algorithm, and a sequence similarity vector is generated. The time-domain deviation vector, frequency-domain offset vector, and sequence similarity vector are fused using tensors to generate the feed difference coefficient matrix.

5. The precision feed supply regulation system for egg-laying hen farming based on big data cloud-edge collaboration as described in claim 4, characterized in that, The abnormal region location module includes: A node-based topology network was constructed based on the layout of the farm, and the location information of the chicken houses and the impedance parameters of the feed delivery path were marked. Map the feed difference coefficient matrix to the corresponding nodes of the nodeized topology network; A graph neural network was used to deduce the propagation path of abnormal material supply, and the abnormal attenuation factor was calculated based on the node impedance parameters. Generate a probability distribution map of the abnormal feed supply covering the entire farm, and identify high-probability abnormal areas.

6. The precision feed supply regulation system for egg-laying hen farming based on big data cloud-edge collaboration as described in claim 5, characterized in that, The material supply strategy adjustment module includes: Configure the material supply verification parameters according to the abnormal material supply probability distribution map; A high-frequency monitoring mode is activated for the high-probability anomaly areas; A swarm intelligence optimizer is used to adjust the material supply control rules; The adaptive feeding control signal is generated based on the adjusted feeding control rules.

7. The precision feed supply regulation system for egg-laying hen farming based on big data cloud-edge collaboration as described in claim 6, characterized in that, The material supply strategy adjustment module also includes: The feed rate adjustment value is calculated using a temperature drop search algorithm; The feed rate adjustment value is converted into the adaptive feed control signal; Based on preset feed constraints, a gradient projection adjuster is used to optimize the feed distribution.

8. The precision feed supply regulation system for egg-laying hen farming based on big data cloud-edge collaboration as described in claim 7, characterized in that, The gradient projection adjuster includes: Define a set of constraints for material supply; Calculate the subgradient of the objective function based on the current material supply plan; Update the material supply plan through projection operations to meet all constraints; The optimized feeding scheme is output to the adaptive feeding control signal.

9. The precision feed supply regulation system for egg-laying hen farming based on big data cloud-edge collaboration as described in claim 1, characterized in that, Also includes: The feedback detection module monitors the adjusted feed consumption data through a recursive state estimator. The monitoring data is fed back to the cloud-edge collaborative analysis module and the material supply strategy adjustment module; The material supply strategy is updated based on feedback data.

10. The precision feed supply regulation system for egg-laying hen farming based on big data cloud-edge collaboration as described in claim 9, characterized in that, The feedback detection module includes: Collect flock response data after feed distribution; The chicken flock response data is processed by a recursive state estimator to generate state estimation results. The state estimation results are input into the multidimensional difference assessment module for difference analysis iteration.

Images