Queen status monitoring method and apparatus

Abstract

This invention discloses a method and device for monitoring the status of queen bees. The method includes: acquiring real-time data using an RFID tag attached to the queen bee and sensors attached to the beehive; extracting individual queen bee movement characteristics, bee colony behavior characteristics, and environmental auxiliary characteristics; aligning and stitching these characteristics to generate a real-time multidimensional feature vector; inputting the queen bee spatial trajectory time-series data, bee colony inflow and outflow time-series data, beehive internal environmental information, and the real-time multidimensional feature vector into a pre-trained queen bee status assessment model, and outputting the queen bee status assessment result; and issuing a dynamic early warning based on the queen bee status assessment result, real-time data, and real-time multidimensional feature vector. This invention provides a continuous, objective, and multi-dimensional information-integrated queen bee status monitoring solution, improving accuracy and efficiency, avoiding interference with the bee colony, and achieving dynamic early warning of the queen bee status in multi-comb beehives.

Description

Technical Field

[0001] This invention relates to the field of intelligent beekeeping technology, and in particular to a method and device for monitoring the status of queen bees. Background Technology

[0002] In beekeeping, the queen bee, as the core role in the bee social structure, bears the dual mission of colony reproduction and stable order. Her age and egg-laying capacity directly determine the scale of colony development and production efficiency. The queen bee's age and egg-laying status are important factors influencing colony development, honey production, and the sustainable development of the colony.

[0003] Currently, beekeepers mainly rely on periodic manual inspections of the hive to monitor and assess the condition of queen bees, making judgments based on their appearance, behavior, and egg-laying patterns. This method has the following significant drawbacks: Firstly, it is inefficient and disruptive to the bee colony: frequent opening of the hive for inspection is time-consuming and labor-intensive, and it causes intrusion and serious disturbance to the bee colony (for example, disrupting the hive temperature and humidity, affecting the normal order of the bee colony), which may cause stress to the bee colony or even cause the queen bee to stop laying eggs or be surrounded by queens.

[0004] Secondly, manual assessment is highly subjective and lacks accuracy: the judgment of the queen bee's age and egg production depends heavily on the beekeeper's experience and lacks objective, quantitative data support, making it easy to make misjudgments. This is especially true for multi-comb hives, where it is difficult to quickly locate and continuously observe the queen bee, making manual location and observation of the queen bee extremely difficult.

[0005] Furthermore, manual assessment methods suffer from information lag and lack early warning capabilities: manual inspections are discrete and periodic, making continuous monitoring impossible. By the time queen bee aging, decreased egg production, or loss is detected, irreversible damage to the colony's development has often already occurred, missing the optimal opportunity for artificial queen rearing or replacement.

[0006] Although some technologies exist for monitoring queen bees using single-type data such as hive temperature and images, multi-source data fusion analysis is still lacking: the individual behavior trajectory of the queen bee is not deeply integrated and correlated with the overall activity of the bee colony, resulting in a single evaluation dimension and one-sided conclusions.

[0007] Overall, the shortcomings of existing technologies include discrete monitoring, strong subjectivity, insufficient accuracy, low efficiency, easy interference with bee colonies, information lag and inability to provide early warnings, single evaluation dimensions, and one-sided conclusions. Summary of the Invention

[0008] This invention provides a method for monitoring the status of queen bees, offering a continuous, objective, and multi-dimensional information-integrated queen bee status monitoring solution. This improves accuracy and efficiency, avoids disturbing the bee colony, and enables dynamic early warning of the queen bee status in multi-comb hives. The method includes: Real-time data is acquired by using RFID (Radio Frequency Identification) tags attached to the queen bee and sensors attached to the beehive; the real-time data includes time-series data of the queen bee's spatial trajectory, time-series data of the bee colony's inflow and outflow, and information on the internal environment of the beehive. Extract individual queen bee movement characteristics, collective bee colony behavior characteristics, and environmental auxiliary characteristics from queen bee spatial trajectory time series data, bee colony inflow and outflow time series data, and beehive internal environment information; Align and stitch together the individual movement characteristics of the queen bee, the collective behavior characteristics of the bee colony, and the environmental auxiliary characteristics to generate a real-time multidimensional feature vector; The queen bee spatial trajectory time-series data, bee colony inflow and outflow time-series data, hive internal environment information, and real-time multidimensional feature vectors are input into a pre-trained queen bee state assessment model, and the output is the queen bee state assessment result. The queen bee state assessment model is obtained by training a classification model using historical data labeled with queen bee state tags and historical multidimensional feature vectors. The queen bee state tags include one or any combination of juvenile, middle-aged, and senile stages. The queen bee state assessment result includes the probability that the queen bee belongs to each queen bee state tag. Dynamic early warnings are issued based on the queen bee status assessment results, real-time data, and real-time multidimensional feature vectors.

[0009] Another aspect of the present invention provides a queen bee status monitoring device, which provides a continuous, objective, and multi-dimensional information-integrated queen bee status monitoring solution, improves accuracy and efficiency, avoids disturbing the bee colony, and realizes dynamic early warning of the queen bee status in multi-comb hives. The device includes: The real-time data acquisition module is used to acquire real-time data using RFID tags attached to the queen bee and sensors attached to the beehive. The real-time data includes time-series data of the queen bee's spatial trajectory, time-series data of the bee colony's inflow and outflow, and information on the internal environment of the beehive. The feature extraction module is used to extract individual queen bee movement features, collective bee colony behavior features, and environmental auxiliary features from queen bee spatial trajectory time series data, bee colony inflow and outflow time series data, and beehive internal environment information. The real-time multidimensional feature vector generation module is used to align and stitch together the individual movement features of the queen bee, the collective behavior features of the bee colony, and environmental auxiliary features to generate real-time multidimensional feature vectors. The queen bee status assessment result output module is used to input the queen bee spatial trajectory time series data, bee colony inflow and outflow time series data, hive internal environment information, and real-time multidimensional feature vectors into a pre-trained queen bee status assessment model, and output the queen bee status assessment result. The queen bee status assessment model is obtained by training a classification model using historical data labeled with queen bee status tags and historical multidimensional feature vectors. The queen bee status tags include one or any combination of juvenile, middle-aged, and senile stages. The queen bee status assessment result includes the probability that the queen bee belongs to each queen bee status tag. The dynamic early warning module is used to issue dynamic early warnings based on the queen bee status assessment results, real-time data, and real-time multi-dimensional feature vectors.

[0010] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described queen bee status monitoring method.

[0011] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described queen bee status monitoring method.

[0012] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described queen bee status monitoring method.

[0013] Compared with existing technologies, this invention utilizes an RFID tag on the queen bee and sensors on the beehive to acquire real-time data. This real-time data includes queen bee spatial trajectory time-series data, bee colony inflow / outflow time-series data, and beehive internal environment information. From these data, individual queen bee movement features, bee colony collective behavior features, and environmental auxiliary features are extracted. These features are then aligned and stitched together to generate a real-time multidimensional feature vector. Finally, the real-time multidimensional feature vector is integrated with the queen bee spatial trajectory time-series data, bee colony inflow / outflow time-series data, beehive internal environment information, and real-time multidimensional feature vector. The feature vector is input to a pre-trained queen bee status assessment model and outputs the queen bee status assessment result. The queen bee status assessment model is obtained by training a classification model using historical data labeled with queen bee status tags and historical multidimensional feature vectors. The queen bee status tags include one or any combination of juvenile, middle-aged, and senile stages. The queen bee status assessment result includes the probability that the queen bee belongs to each queen bee status tag. Based on the queen bee status assessment result, real-time data, and real-time multidimensional feature vectors, dynamic early warnings are issued, realizing a continuous, objective, and multi-dimensional information-integrated queen bee status monitoring solution, improving accuracy and efficiency, avoiding interference with bee colonies, and realizing dynamic early warning of queen bee status in multi-comb hives. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a flowchart of the queen bee status monitoring method in an embodiment of the present invention; Figure 2 This is a flowchart illustrating a specific example of the queen bee status monitoring method in this invention. Figure 3 This is a schematic diagram of the queen bee status monitoring device in an embodiment of the present invention; Figure 4 This is a schematic diagram of a specific example of the queen bee status monitoring device in this invention. Figure 5 This is a schematic diagram of a computer device in an embodiment of the present invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0016] This invention belongs to the field of intelligent beekeeping technology, specifically involving the refined monitoring and management of beekeeping. More precisely, this invention provides a bee colony status monitoring system and method based on Internet of Things (IoT) sensing technology and data analysis. By integrating bee colony entry and exit behavior activity data with queen bee individual spatial movement trajectory data based on radio frequency identification (RFID) technology, an evaluation model is constructed to achieve automated assessment of the queen bee's age stage, egg-laying performance, and colony health status in multi-comb hives.

[0017] Existing technology has constructed a highly integrated "transparent laboratory beehive" for scientific research observation. The device uses acrylic (a transparent material) to form the outer wall of the beehive body, creating an experimental environment that facilitates direct visual observation. The beehive integrates a high-definition macro camera and a thermal imaging camera for visual observation and individual identification, and is designed for researchers to conduct real-time observation or recording and analysis.

[0018] While the aforementioned existing technologies achieve video presentation by integrating high-definition cameras, they are more akin to "observation instruments." Their core design focuses on providing comprehensive data recording for laboratory environments, which is severely out of step with mainstream standard wooden beehive structures. Consequently, they cannot be directly applied to actual large-scale beekeeping scenarios, resulting in low practicality. Furthermore, this solution only supports manual observation and interpretation, failing to achieve automated assessment and early warning of queen bee age, egg-laying performance, or health status. In addition, its positioning technology suffers from ambiguous accuracy and insufficient reliability in real, crowded, and opaque complex beehive environments. Frequent adjustments and maintenance of the device may still interfere with the normal activities of the bee colony, further limiting its promotion and application in commercial beekeeping.

[0019] Existing technology also provides a historical archive analysis method based on static image comparison. This method involves periodically taking images of beehives, identifying and extracting the morphological characteristics of the queen bee (such as body size, color, and other appearance indicators), and then comparing these characteristics with the "standard growth images" of queen bees of the same breed at different age groups in the database to infer the age group of the target queen bee and issue an alert when it is determined to be aging.

[0020] The aforementioned existing technologies mainly rely on periodically photographing beehives, extracting static morphological features of the queen bee (such as body size, color, and texture), and comparing them with standard images in a pre-built database to infer her age. The data dimensions are singular and all information is static, which cannot effectively capture the queen bee's real-time three-dimensional movement trajectory, activity rhythm, and dynamic correlation with the overall entry and exit behavior of the bee colony. This results in a one-sided assessment dimension, insufficient reliability of the conclusions, and susceptibility to interference from environmental factors such as shooting light, angle, and obstruction. This method is highly dependent on a large, accurate database of queen bee morphological images covering multiple species and regions. The database construction and maintenance are difficult and costly, and its universality is poor. The monitoring process is not continuous and real-time, and its functions are mainly limited to preliminary judgment of queen bee age and aging warning. It lacks in-depth assessment of egg-laying performance, immediate detection of queen bee loss, and intelligent management capabilities for multi-source data fusion, making it difficult to meet the actual needs of modern beekeeping for dynamic, comprehensive, and timely queen bee status monitoring.

[0021] In general, existing methods suffer from limitations such as one-sided evaluation dimensions, outdated information, and strong interference, failing to achieve non-intrusive, continuous, and multi-source fusion-based automated monitoring. This invention addresses these shortcomings by providing an innovative solution.

[0022] The present invention aims to solve the above problems and provide a non-invasive, automated, continuous, objective system and method for queen bee status monitoring and age assessment that integrates multi-dimensional information, so as to achieve accurate judgment of the key life stages of queen bees in multi-comb hives and early warning of the health status of the bee colony.

[0023] This invention provides a method for monitoring the status of queen bees. Figure 1 This is a flowchart of the queen bee status monitoring method in an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes: Step 101: Use the RFID tag set on the queen bee and the sensor set on the beehive to acquire real-time data; the real-time data includes the queen bee's spatial trajectory time series data, the bee colony's inflow and outflow time series data, and the beehive's internal environment information. Step 102: Extract the queen bee's individual movement characteristics, the bee colony's collective behavior characteristics, and environmental auxiliary characteristics from the queen bee's spatial trajectory time series data, the bee colony's inflow and outflow time series data, and the beehive's internal environmental information. Step 103: Align and stitch together the individual movement characteristics of the queen bee, the collective behavior characteristics of the bee colony, and the environmental auxiliary characteristics to generate a real-time multidimensional feature vector. Step 104: Input the queen bee spatial trajectory time series data, bee colony inflow and outflow time series data, hive internal environment information, and real-time multidimensional feature vectors into the pre-trained queen bee state assessment model, and output the queen bee state assessment results; the queen bee state assessment model is obtained by training a classification model using historical data and historical multidimensional feature vectors labeled with queen bee state tags; queen bee state tags include one or any combination of youth, middle age, and old age; the queen bee state assessment results include the probability that the queen bee belongs to each queen bee state tag; Step 105: Based on the queen bee status assessment results, real-time data, and real-time multidimensional feature vectors, issue a dynamic early warning.

[0024] Compared with existing technologies, the embodiments of the present invention provide a continuous, objective, and multi-dimensional information-integrated queen bee status monitoring solution through the above steps, improving accuracy and efficiency, avoiding interference with bee colonies, and enabling dynamic early warning of queen bee status in multi-comb hives.

[0025] Figure 2 The flowchart below shows a specific example of the queen bee status monitoring method in this invention. Figure 2 As shown, the method may specifically include: Step 1: Synchronous acquisition and preprocessing of multi-source data.

[0026] The system synchronously and continuously collects time-series data on the queen bee's spatial trajectory, hive entrance flow, and environmental data. The trajectory data undergoes noise reduction, interpolation, and smoothing; the hive entrance flow and temperature / humidity data are aggregated according to time series.

[0027] Step 2: Multi-dimensional feature engineering extraction.

[0028] Two main categories of features are extracted from the preprocessed data: (1) Individual movement characteristics of queen bees: Based on trajectory data, calculate the daily average movement speed, activity range (three-dimensional spatial coverage volume), stay time in the brooding area, regularity of movement trajectory (such as entropy value), activity rhythm (daytime vs. nighttime activity ratio), etc.

[0029] (2) Collective behavior characteristics of bee colonies: Based on the inflow and outflow data, the total number of bees entering and leaving the hive daily, the peak time for entering and leaving the hive, and the daily activity rate change rate are calculated. The internal environment data of the hive is used as an auxiliary feature.

[0030] Step 3: Integrate feature dataset construction with model training.

[0031] Individual movement characteristics and collective behavior characteristics within the same time period are aligned and fused to form a "feature-label" dataset. Beekeeping staff then label the dataset based on the queen bee's emergence records, indicating the queen's age stage (e.g., juvenile stage <1 year, peak laying stage 1-2 years, and senescence stage >2 years). After this dataset is appropriately partitioned, it is trained using a gradient boosting decision tree (XGBoost) model to obtain a queen bee state assessment and early warning model.

[0032] Step 4: Real-time assessment and early warning of the queen bee's condition.

[0033] The feature vectors collected and extracted in real time are input into a pre-trained evaluation model, which outputs the predicted probability of the queen bee's age and condition. At the same time, the system sets threshold rules (such as a sudden drop in activity over several consecutive days and a sharp reduction in the queen bee's activity range) to trigger an alert that "the queen bee may be lost or have abnormal health."

[0034] Step 5: Visualization and report generation.

[0035] The platform provides a visual dashboard that displays the real-time location heatmap of the queen bee, historical trajectory playback, colony activity curve, model evaluation results and early warning information, and can generate periodic analysis reports.

[0036] The core of this invention's algorithm lies in the deep fusion and analysis of multi-source heterogeneous sensor data through localization, feature extraction, and machine learning models, ultimately outputting a quantitative assessment of the queen bee's condition. The algorithm mainly consists of four key parts: raw sensor data monitoring, data synchronization and preprocessing, queen bee localization and trajectory reconstruction algorithm, and queen bee condition early warning model.

[0037] In one embodiment, acquiring real-time data using an RFID tag on the queen bee and sensors on the beehive may include: employing a signal strength algorithm or a time difference of arrival algorithm, using an RFID reader antenna deployed inside the beehive to continuously read the identification number of the UHF passive RFID tag on the queen bee, determining the queen bee's movement trajectory within the beehive in real time during the reading process, and generating queen bee spatial trajectory time-series data; using a capacitive sensing counter on the beehive to count the number of bees entering and leaving the beehive per unit time by measuring the capacitance and / or voltage changes caused by bees entering and leaving the beehive, and generating bee colony inflow and outflow time-series data; and using sensors on the beehive to acquire information about the internal environment of the beehive.

[0038] This invention also provides a queen bee status monitoring and age assessment system based on bee colony behavior and spatial trajectory. Its core lies in synchronously collecting bee colony collective behavior data and queen bee individual movement data through Internet of Things sensing technology, and constructing a data fusion analysis model to achieve indirect, non-destructive, and intelligent assessment of the queen bee's status.

[0039] The hardware architecture of this system mainly includes: Queen Bee Individual Identification and Tracking Unit: Live queen bees are tagged with ultra-lightweight UHF passive RFID tags, and multiple RFID reader antennas are deployed in standard multi-comb hives to form a grid-like or honeycomb-like positioning network, enabling continuous three-dimensional trajectory tracking of the queen bee in complex three-dimensional space.

[0040] Lightly attach or install ultra-high frequency (UHF) passive RFID tags on the queen bee's body (e.g., on her back). Deploy multiple RFID reader antennas in key locations inside the hive (especially between combs in the brood rearing area) to form a honeycomb or grid-like sensing network. Continuously read the identification numbers (IDs) of the UHF passive RFID tags and calculate the queen bee's movement trajectory in the three-dimensional hive space in real time using signal strength index (RSSI) or time difference of arrival (TDOA) algorithms.

[0041] Bee colony entry and exit behavior monitoring unit: A capacitance sensor counter is installed at the entrance of the beehive. By measuring the capacitance and voltage changes caused by bees entering and leaving the hive, the number of bees entering and leaving the hive per unit time can be accurately counted, and the activity level of the bee colony can be calculated.

[0042] Beehive Environmental Information Acquisition Unit: Temperature and humidity sensors and other devices are deployed inside the beehive to monitor the microenvironment information inside the beehive.

[0043] Data aggregation and processing unit: The data acquisition gateway with built-in microprocessor is responsible for receiving, temporarily storing, and preprocessing all the above-mentioned sensor data, and uploading the data to the cloud or edge server through the wireless communication module (4G / 5G).

[0044] Cloud server data analysis and evaluation platform: Receives and stores data, runs core algorithm models, performs feature extraction, model calculation, status evaluation, and result visualization.

[0045] The specific hardware deployment is as follows: Select a standard Langstroth 10-frame beehive. Seven days after the queen emerges from her cell, attach a miniature UHF RFID tag (weighing <5mg) to the dorsal plate of her thorax using special adhesive.

[0046] Six RFID reader antennas are embedded in a grid pattern inside the side walls and partitions of the beehive to ensure coverage of all combs, especially the central brooding area.

[0047] A pair of capacitive counting sensors are installed at the entrance of the hive to count the number of bees entering and leaving.

[0048] A temperature and humidity sensor is installed at the top center of the box.

[0049] All sensors are connected to a data acquisition gateway installed on the side of the enclosure.

[0050] Data collection and uploading: The RFID reader network scans once per minute, recording the antenna ID and signal strength of the detected queen bee tag, and calculates the queen bee's real-time coordinates (x, y, z) using a positioning engine.

[0051] The gate counter counts the number of entries and exits every second, and the gateway summarizes the data every minute.

[0052] The temperature and humidity sensor records data once per minute.

[0053] Every hour, the gateway uploads compressed data packets to the cloud platform via the NB-IoT (Narrow Band Internet of Things) network.

[0054] In one embodiment, a signal strength algorithm or a time difference of arrival algorithm is used to continuously read the identification number of an ultra-high frequency passive RFID tag attached to the queen bee using an RFID reader antenna deployed inside the beehive. During the reading process, the queen bee's movement trajectory within the beehive is determined in real time, generating queen bee spatial trajectory time-series data. This can include: using an RFID reader antenna deployed inside the beehive to test and determine the first signal strength received by the RFID reader antenna when reading the identification number of an ultra-high frequency passive RFID tag set at a preset distance; using multiple RFID reader antennas deployed inside the beehive to continuously read the identification number of an ultra-high frequency passive RFID tag attached to the queen bee and obtain the second signal strength received by each RFID reader antenna; determining the distance between the ultra-high frequency passive RFID tag and each RFID reader antenna based on the first signal strength, the second signal strength, the path loss exponent related to the environment, and zero-mean Gaussian noise under random shading effect; and determining the queen bee's spatial position based on the distance between the ultra-high frequency passive RFID tag and each RFID reader antenna, generating queen bee spatial trajectory time-series data.

[0055] In this embodiment, this step aims to convert discrete RFID signal data into a continuous three-dimensional motion trajectory of the queen bee.

[0056] Data input: Signal strength (RSSI) data collected at time point t from 6 RFID reader antennas deployed at different locations within the beehive.

[0057] Distance estimation: Signal attenuation during propagation follows a log-normal shaded model. i Signal strength received by the reader antenna RSSI i Distance from antenna to tagd i The relationship between them can be modeled as follows:

[0058] In one embodiment, determining the distance between the UHF passive RFID tag and each RFID reader antenna based on a first signal strength, a second signal strength, an environment-related path loss exponent, and zero-mean Gaussian noise under random shading effects may include: determining the distance between the UHF passive RFID tag and each RFID reader antenna according to the following formula: ; in, for The second signal strength received by the i-th RFID reader antenna; The distance between the ultra-high frequency passive RFID tag and the i-th RFID reader antenna; The first signal strength received by the RFID reader antenna when reading the identification number of an ultra-high frequency passive RFID tag set at a preset distance; n is the environment-related path loss index; The preset distance; It is zero-mean Gaussian noise under random shading effect.

[0059] This model allows for measurement... RSSI i Reverse the derivation of UHF passive RFID tags and the first i The distance between the antennas of the RFID reader d i .

[0060] Location determination: The system will receive the signal strength of the queen bee tag ( RSSI i The installation position of the largest reader antenna is directly used as the estimated position p(t) of the queen bee at that scanning moment.

[0061] In one embodiment, in step 102, individual queen bee movement characteristics, collective bee behavior characteristics, and environmental auxiliary characteristics are extracted from queen bee spatial trajectory time-series data, bee colony entry and exit flow time-series data, and hive internal environment information. The queen bee's individual three-dimensional spatial movement trajectory data and the bee colony's hive entrance entry and exit activity data are synchronously collected, time-aligned, and fused for analysis, serving as the core basis for assessing the queen bee's physiological state.

[0062] In this embodiment, extracting individual queen bee movement characteristics, collective bee colony behavior characteristics, and environmental auxiliary characteristics from queen bee spatial trajectory time-series data, bee colony inflow / outflow time-series data, and beehive internal environment information may include: performing noise reduction, interpolation, and smoothing processing on the queen bee spatial trajectory time-series data, bee colony inflow / outflow time-series data, and beehive internal environment information; aligning and aggregating the noise-reduced, interpolated, and smoothed queen bee spatial trajectory time-series data, bee colony inflow / outflow time-series data, and beehive internal environment information according to time; and then performing time-based alignment and aggregation on the aligned and aggregated queen bee spatial trajectory time-series data. Extract individual movement characteristics of the queen bee; these characteristics include one or any combination of the following: total daily movement distance, average daily movement speed, activity range, duration of stay in the brood rearing area of ​​the hive, variance of movement speed, entropy of movement trajectory, and activity rhythm. Based on the aligned and aggregated hive inflow and outflow time series data, extract collective hive behavior characteristics; these characteristics include one or any combination of the following: total number of bees entering and leaving the hive daily, peak time for bees entering and leaving the hive daily, daily activity change rate, total number of foraging bees daily, and average attendance rate during peak time for bees entering and leaving the hive daily.

[0063] In this embodiment, the positioning trajectory, hive entrance count, and environmental data are analyzed to extract key feature vectors that characterize the queen bee's state and the activity of the bee colony.

[0064] Data preprocessing: imputing missing values ​​in trajectory data; aggregating nest entrance count data according to a preset time window (e.g., 1 minute) to calculate inflow and outflow; taking statistical values ​​(e.g., daily average temperature and humidity) from environmental data within a period.

[0065] Feature extraction is performed, including: (1) Individual movement characteristics of queen bees: calculated based on smoothed trajectory data and daily cycle.

[0066] Activity level indicators: total daily distance traveled and average daily speed. For example, total distance traveled and average speed traveled within 24 hours.

[0067] Spatial distribution indicators: the percentage of time spent in the pre-designed brooding area and the three-dimensional activity range. For example, the percentage of time spent in the central brooding area of ​​the central 5 frames.

[0068] Behavioral rhythm indicators: the ratio of diurnal activity intensity and the regularity of movement trajectories (e.g., measured by calculating trajectory entropy). For example, the variance of movement speed and the number of activities at night (8 pm to 6 am).

[0069] (2) Collective behavior characteristics of bee colonies: calculated on a daily cycle based on time-series data of hive entrance flow.

[0070] Activity metrics: total number of bees at work each day, and average attendance rate during peak hours. For example, the average attendance rate between 9 and 11 a.m.

[0071] Trend indicator: The rate of change of daily activity relative to the previous day.

[0072] (3) Environmental auxiliary characteristics: temperature characteristics and humidity characteristics.

[0073] Feature Fusion: Construct a real-time multidimensional feature vector, including factors such as total daily movement distance, percentage of time spent in the brood rearing area, variance of movement speed, number of nighttime activities, total number of attending bees per day, morning attendance rate, and average daily temperature. Align and concatenate the individual queen bee movement characteristics, collective bee colony behavior characteristics, and environmental auxiliary characteristics (average daily temperature, etc.) for the same date to form a multidimensional feature vector, which serves as the input for subsequent models.

[0074] In one embodiment, before inputting the queen bee spatial trajectory time-series data, bee colony inflow / outflow time-series data, hive internal environment information, and the real-time multidimensional feature vector into a pre-trained queen bee state assessment model and outputting the queen bee state assessment result, the method may further include: collecting historical data from the complete life cycle of more than a preset number of bee colonies over multiple consecutive beekeeping seasons; the historical data includes queen bee spatial trajectory time-series data, bee colony inflow / outflow time-series data, and hive internal environment information; extracting and generating historical multidimensional feature vectors from the historical data; labeling the historical data and historical multidimensional feature vectors with queen bee state tags according to the queen bee's emergence date in each bee colony; the emergence date of queen bees labeled as juveniles is no more than a first preset time interval from the labeling time; the emergence date of queen bees labeled as middle-aged is greater than or equal to the first preset time interval and less than or equal to a second preset time interval from the labeling time; the emergence date of queen bees labeled as seniles is more than a second preset time interval from the labeling time; and using the Extreme Gradient Boosting (XGBoost) algorithm to train a classification model using the historical data and historical multidimensional feature vectors labeled with queen bee state tags.

[0075] This step utilizes a machine learning model to learn the complex mapping relationship between feature vectors and the physiological state of the queen bee.

[0076] Model Input and Output: Input Metric: Constructed real-time multidimensional feature vector.

[0077] Output metric: The probability distribution of the queen bee belonging to each predefined age state category (such as "young high-yielding period", "middle-aged peak-yielding period", "old and low-yielding period").

[0078] Model training: Supervised learning was performed using a dataset collected historically and labeled with status tags by experts based on the queen bee's emergence date.

[0079] The XGBoost (Extreme Gradient Boosting) algorithm is used for training. This model iteratively constructs a series of decision trees (weak learners) using an additive strategy to minimize an objective function that includes a loss function and a regularization term. Its general form can be expressed as:

[0080] Where n is the number of samples. For the model, the true value for the i-th sample, Let be the model's predicted value for the i-th sample. The loss function (such as cross-entropy) measures the difference between the predicted value and the true label, where K is the number of decision trees. Let be the k-th decision tree, and Ω be a regularization term that controls model complexity to prevent overfitting. After training, a classifier capable of inferring the queen bee state probability based on the input feature vector is obtained.

[0081] In this embodiment, complete lifecycle data of over 100 bee colonies were collected over three consecutive beekeeping seasons. Researchers labeled each stage of the dataset with queen bee status tags (e.g., high-producing youth, peak-producing middle age, and low-producing aging). The basic discrimination logic is as follows: The peak productive period for young bees (< 1 year): from queen emergence to the first winter. The core biological characteristics of the queen bee are a rapid increase and stabilization of egg production, active behavior, and robust physique. Identification is primarily based on the queen's emergence date recorded in beekeeping records. In training data, behavioral data from queens of this age group will be used to establish a baseline model for "young" bees.

[0082] Mid-life peak production period (1-2 years): After a complete overwintering period, the queen bee enters its most physiologically mature stage, with the highest and most stable egg-laying capacity. Judgment logic: Also based on the queen bee's emergence date records. The model learns and identifies typical and efficient egg-laying behavior patterns based on the dataset from this stage.

[0083] Low-production period due to aging (>2 years): The queen bee's physiological functions decline significantly. Biological and behavioral markers include: a significant decrease in egg production, irregular egg production, a shrinking egg-laying circle, a slower egg-laying speed, and sluggish movement. The identification logic is based on the date the queen bee emerges from her cell.

[0084] After properly dividing the labeled queen bee state dataset, a classification model is trained using the XGBoost algorithm. The input is the fused feature vector, and the output is the probability of each state label.

[0085] In one embodiment, issuing a dynamic early warning based on the queen bee status assessment results, real-time data, and real-time multidimensional feature vectors may include: triggering an early warning message for replacing the queen bee when the queen bee status assessment results show the highest probability of being in the aging stage for a consecutive preset number of days, and the probability of the aging stage exceeds a preset threshold; and triggering an early warning message for queen bee loss or health abnormality when real-time data and real-time multidimensional feature vectors show abnormalities in one or any combination of queen bee spatial trajectory, bee colony inflow / outflow, and beehive internal environment.

[0086] This step generates the final evaluation conclusion and early warning signal based on the model output and rule logic.

[0087] State decision: Input the feature vector extracted in real time into the deployed XGBoost model, and take the category with the highest output probability as the evaluation result of the queen bee's state for the day.

[0088] Warning Trigger: Set dynamic rules to issue warnings for abnormal situations: Aging warning rule: If the probability of "low productivity of aging" output by the model exceeds the set threshold (e.g., 80%) for a preset number of days (e.g., 7 days), a king replacement suggestion warning will be triggered.

[0089] Loss Anomaly Warning Rule: If the system fails to detect a valid queen bee trajectory for a preset duration (e.g., 12 hours) and the total daily attendance of the bee colony drops sharply to a certain percentage (e.g., 50%) of the historical normal level during the same period, an alarm will be triggered indicating that the queen bee may be lost or has health abnormalities.

[0090] This algorithm system realizes an automated closed loop from raw data to state knowledge and then to decision support, providing key technical support for the core invention objectives.

[0091] In this embodiment, after the model is deployed, the system automatically calculates the feature vectors of the previous 24 hours daily and inputs them into the model. If the model outputs a probability of the queen bee being in an aging and low-producing period exceeding 80% for a consecutive week, a notification to replace the queen bee is sent to the user's mobile phone. If the system detects that the queen bee's trajectory has not been updated for 12 consecutive hours, and the worker bees' entry and exit activities at the hive entrance are abnormal (e.g., the flow rate at the entrance and exit is too low), an alert is sent to the user that the queen bee may be lost.

[0092] This invention also provides a queen bee status monitoring device, as described in the following embodiments. Since the principle by which this device solves the problem is similar to that of the queen bee status monitoring method, the implementation of this device can refer to the implementation of the queen bee status monitoring method; repeated details will not be elaborated further.

[0093] Figure 3 This is a schematic diagram of the queen bee status monitoring device in an embodiment of the present invention, as shown below. Figure 3 As shown, the device includes: The real-time data acquisition module 301 is used to acquire real-time data using an RFID tag set on the queen bee and a sensor set on the beehive; the real-time data includes the queen bee's spatial trajectory time series data, the bee colony's inflow and outflow time series data, and the beehive's internal environmental information; The feature extraction module 302 is used to extract individual movement features of the queen bee, collective behavior features of the bee colony, and environmental auxiliary features from the queen bee spatial trajectory time series data, bee colony inflow and outflow time series data, and beehive internal environment information. The real-time multidimensional feature vector generation module 303 is used to align and stitch together the individual movement features of the queen bee, the collective behavior features of the bee colony, and the environmental auxiliary features to generate a real-time multidimensional feature vector. The queen bee status assessment result output module 304 is used to input the queen bee spatial trajectory time series data, the hive inflow and outflow time series data, the hive internal environment information, and real-time multidimensional feature vectors into a pre-trained queen bee status assessment model, and output the queen bee status assessment result. The queen bee status assessment model is obtained by training a classification model using historical data and historical multidimensional feature vectors labeled with queen bee status tags. The queen bee status tags include one or any combination of juvenile, middle-aged, and senile stages. The queen bee status assessment result includes the probability that the queen bee belongs to each queen bee status tag. The dynamic early warning module 305 is used to issue dynamic early warnings based on the queen bee status assessment results, real-time data, and real-time multi-dimensional feature vectors.

[0094] In one embodiment, the real-time data acquisition module 301 is specifically used for: Using signal strength algorithm or time difference of arrival algorithm, the RFID reader antenna deployed inside the beehive continuously reads the identification number of the UHF passive RFID tag set on the queen bee. During the reading process, the movement trajectory of the queen bee in the beehive is determined in real time, and the spatial trajectory time series data of the queen bee is generated. By using a capacitive sensor counter installed on the beehive, the number of bees entering and leaving the beehive per unit time is counted by measuring the capacitance and / or voltage changes caused by bees entering and leaving the beehive, generating bee colony inflow and outflow time-series data; by using sensors installed on the beehive, information about the internal environment of the beehive is obtained.

[0095] In one embodiment, the real-time data acquisition module 301 is specifically used for: The first signal strength received by the RFID reader antenna when reading the identification number of an UHF passive RFID tag set at a preset distance was determined by using an RFID reader antenna deployed inside the beehive. By using multiple RFID reader antennas deployed inside the beehive, the identification number of the ultra-high frequency passive RFID tag set on the queen bee is continuously read, and the second signal strength received by each RFID reader antenna is obtained. Based on the first signal strength, the second signal strength, the path loss exponent related to the environment, and the zero-mean Gaussian noise under random shadowing effect, the distance between the UHF passive RFID tag and each RFID reader antenna is determined. Based on the distance between the UHF passive RFID tag and each RFID reader antenna, the spatial location of the queen bee is determined, and the temporal data of the queen bee's spatial trajectory is generated.

[0096] In one embodiment, the real-time data acquisition module 301 is specifically used for: The distance between the UHF passive RFID tag and each RFID reader antenna is determined using the following formula: ; in, for The second signal strength received by the i-th RFID reader antenna; The distance between the ultra-high frequency passive RFID tag and the i-th RFID reader antenna; The first signal strength received by the RFID reader antenna when reading the identification number of an ultra-high frequency passive RFID tag set at a preset distance; n is the environment-related path loss index; The preset distance; It is zero-mean Gaussian noise under random shading effect.

[0097] In one embodiment, the feature extraction module 302 is specifically used for: Noise reduction, interpolation and smoothing were performed on the temporal data of queen bee spatial trajectory, temporal data of bee colony inflow and outflow, and information on the internal environment of beehive. The time-series data of queen bee spatial trajectory, bee colony inflow and outflow, and beehive internal environment information after noise reduction, interpolation and smoothing are aligned and aggregated according to time. Based on the aligned and aggregated temporal data of the queen bee's spatial trajectory, the individual movement characteristics of the queen bee are extracted. The individual movement characteristics of the queen bee include one or any combination of the following: total daily movement distance, average daily movement speed, activity range, duration of stay in the brood rearing area in the hive, variance of movement speed, entropy value of movement trajectory, and activity rhythm. Based on the aligned and aggregated bee colony inflow and outflow time series data, bee colony collective behavior characteristics are extracted. These characteristics include one or any combination of the following: total number of bees entering and leaving the colony daily, peak time for bees entering and leaving the colony daily, daily activity change rate, total number of bees at work daily, and average attendance rate during peak time for bees entering and leaving the colony daily.

[0098] Figure 4 This is a schematic diagram of a specific example of the queen bee status monitoring device in this invention, as shown below. Figure 4 As shown, in one embodiment, the queen bee status monitoring device may further include: a training module 401, used for: Collect historical data from multiple consecutive beekeeping seasons, covering the entire life cycle of bee colonies exceeding a preset number; historical data includes queen bee spatial trajectory time-series data, bee colony inflow and outflow time-series data, and beehive internal environment information. Extract and generate historical multidimensional feature vectors from historical data; Based on the queen bee's emergence date in each colony, historical data and historical multidimensional feature vectors are labeled with queen bee status tags; the emergence date of queen bees labeled as juvenile is no more than a first preset time interval from the labeling time; the emergence date of queen bees labeled as middle-aged is greater than or equal to the first preset time interval and less than or equal to the second preset time interval from the labeling time; the emergence date of queen bees labeled as senile is more than the second preset time interval from the labeling time. The XGBoost algorithm with extreme gradient boosting is used to train a classification model using historical data labeled with queen bee state tags and historical multidimensional feature vectors.

[0099] In one embodiment, the dynamic early warning module 305 is specifically used for: If the queen bee status assessment results show that the most probable state is aging for a consecutive preset number of days, and the probability of aging exceeds a preset threshold, an early warning message to replace the queen bee is triggered. When real-time data and real-time multidimensional feature vectors show anomalies in one or any combination of the queen bee's spatial trajectory, the flow of bees in and out of the hive, or the internal environment of the hive, an early warning message is triggered indicating that the queen bee is missing or has an abnormal health condition.

[0100] Figure 5 This is a schematic diagram of a computer device in an embodiment of the present invention. Based on the foregoing inventive concept, as follows... Figure 5 As shown, the present invention also proposes a computer device 500, including a memory 501, a processor 502, and a computer program 503 stored in the memory 501 and executable on the processor 502. When the processor 502 executes the computer program 503, it implements the aforementioned queen bee status monitoring method.

[0101] Based on the aforementioned inventive concept, the present invention proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the aforementioned queen bee status monitoring method.

[0102] Based on the aforementioned inventive concept, the present invention proposes a computer program product, which includes a computer program that, when executed by a processor, implements a method for monitoring the queen bee's status.

[0103] Compared with existing technologies, this invention utilizes an RFID tag on the queen bee and sensors on the beehive to acquire real-time data. This real-time data includes queen bee spatial trajectory time-series data, bee colony inflow / outflow time-series data, and beehive internal environment information. From these data, individual queen bee movement features, bee colony collective behavior features, and environmental auxiliary features are extracted. These features are then aligned and stitched together to generate a real-time multidimensional feature vector. Finally, the real-time multidimensional feature vector is integrated with the queen bee spatial trajectory time-series data, bee colony inflow / outflow time-series data, beehive internal environment information, and real-time multidimensional feature vector. The feature vector is input to a pre-trained queen bee status assessment model and outputs the queen bee status assessment result. The queen bee status assessment model is obtained by training a classification model using historical data labeled with queen bee status tags and historical multidimensional feature vectors. The queen bee status tags include one or any combination of juvenile, middle-aged, and senile stages. The queen bee status assessment result includes the probability that the queen bee belongs to each queen bee status tag. Based on the queen bee status assessment result, real-time data, and real-time multidimensional feature vectors, dynamic early warnings are issued, realizing a continuous, objective, and multi-dimensional information-integrated queen bee status monitoring solution, improving accuracy and efficiency, avoiding interference with bee colonies, and realizing dynamic early warning of queen bee status in multi-comb hives.

[0104] The key technical point of this invention lies in the non-invasive collection and fusion of individual queen bee spatial movement trajectory data and overall colony entry and exit behavior data to achieve automated assessment of the queen bee's age stage, egg-laying performance, and colony health status. The main protection points include: 1. The three-dimensional spatial movement trajectory data of the queen bee and the data of the hive entrance and exit activities of the bee colony are collected synchronously, time-aligned and fused for analysis, which serves as the core basis for evaluating the physiological state of the queen bee.

[0105] 2. Fit live queen bees with ultra-lightweight UHF passive RFID tags and deploy multiple RFID reader antennas in a standard multi-comb hive to form a grid-like or honeycomb-like positioning network, enabling continuous three-dimensional trajectory tracking of the queen bee in a complex three-dimensional space.

[0106] 3. Extract individual queen bee movement characteristics (including daily movement distance, average speed, percentage of time spent in the brood-rearing area, regularity of movement trajectory, and ratio of daytime and nighttime activity) and collective bee colony behavior characteristics (including total daily attendance, peak attendance rate, and activity change rate) from the fused data to construct a multidimensional feature vector.

[0107] 4. Label the age stages based on the queen bee's emergence date, and train a machine learning model (preferably XGBoost) to establish a mapping relationship between multidimensional feature vectors and the queen bee's different ages and egg-laying states (juvenile stage, peak egg-laying stage, and senile stage).

[0108] 5. Based on the state probability output by the model, combined with preset threshold rules (such as the aging probability exceeding the threshold for several consecutive days, or the long-term absence of the trajectory accompanied by a sharp drop in the activity of the bee colony), an automated early warning is achieved for queen bee aging, decreased egg-laying capacity, or possible loss.

[0109] 6. The overall system architecture consists of a non-invasive queen bee individual movement trajectory acquisition unit (RFID tag and antenna array), a bee colony collective entry and exit activity acquisition unit (hive entrance capacitive sensor counter), an environmental information auxiliary acquisition unit, and a data fusion analysis unit.

[0110] 7. The methodology includes multi-source data synchronous acquisition and preprocessing, feature extraction and vector construction, model training and real-time evaluation, early warning triggering, and visualization report generation.

[0111] 8. A technical solution for monitoring queen bee status by fusing the above-mentioned behavioral data without modifying the standard beehive structure, relying on transparent materials, or using direct image recognition.

[0112] The embodiments of the present invention can achieve the following technical effects: 1. By using the RFID sensing network and hive entrance sensor inside the beehive, the queen bee's trajectory and bee colony activity data can be collected in real time without frequent opening of the hive. This reduces interference with the normal activities of the bee colony and avoids problems such as temperature and humidity fluctuations, bee colony stress, and queen bee cessation of production caused by manual inspection, which is conducive to maintaining the stability and production performance of the bee colony.

[0113] 2. Based on multi-source sensor data, quantitative features (such as queen bee movement speed, activity range, and hive entrance and exit flow) are extracted and fused through machine learning models, overcoming the subjectivity and one-sidedness of traditional human experience judgment.

[0114] 3. Through continuous monitoring and real-time analysis, early warnings can be issued in the early stages of decreased queen egg-laying capacity, abnormal behavior, or loss, prompting beekeepers to take queen-rearing or queen-replacing measures in advance, avoiding the decline of bee colony development due to information lag, and ensuring the continuous and healthy development of bee colonies.

[0115] 4. By simultaneously collecting data on the individual behavior of the queen bee and the overall activity of the bee colony, the limitations of insufficient evaluation dimensions due to a single data type can be overcome. This allows for a more comprehensive and in-depth reflection of the intrinsic relationship between the queen bee's condition and the health of the bee colony, thereby improving the systematic nature and reliability of the evaluation.

[0116] 5. Automated data collection, cloud analysis, and visualization significantly reduce the physical labor and management time costs for beekeepers, enabling them to remotely and in real-time monitor the status of multiple bee colonies and achieve refined, data-driven beekeeping management.

[0117] 6. The algorithm adopts a modular design, and can be optimized and iterated according to different bee species, regions and breeding methods.

[0118] 7. Long-term, multi-dimensional bee colony behavior data can provide valuable data resources for scientific research on queen bee aging mechanisms and bee colony social behavior.

[0119] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0120] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0121] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0122] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0123] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for monitoring the status of queen bees, characterized in that, include: Real-time data is acquired using RFID tags attached to the queen bee and sensors attached to the beehive; the real-time data includes queen bee spatial trajectory time-series data, bee colony inflow and outflow time-series data, and beehive internal environment information; Extract individual queen bee movement characteristics, bee colony behavior characteristics, and environmental auxiliary characteristics from the queen bee spatial trajectory time series data, the bee colony inflow and outflow time series data, and the beehive internal environment information. The individual movement characteristics of the queen bee, the collective behavior characteristics of the bee colony, and the environmental auxiliary characteristics are aligned and stitched together to generate a real-time multidimensional feature vector. The queen bee spatial trajectory time-series data, the bee colony inflow and outflow time-series data, the beehive internal environment information, and the real-time multidimensional feature vector are input into a pre-trained queen bee state assessment model, and the queen bee state assessment result is output. The queen bee state assessment model is obtained by training a classification model using historical data labeled with queen bee state tags and historical multidimensional feature vectors. The queen bee state tags include one or any combination of juvenile, middle-aged, and senile stages. The queen bee state assessment result includes the probability that the queen bee belongs to each queen bee state tag. Based on the queen bee status assessment results, the real-time data, and the real-time multidimensional feature vector, a dynamic early warning is issued.

2. The method as described in claim 1, characterized in that, Real-time data is acquired using RFID tags attached to the queen bee and sensors attached to the beehive, including: Using signal strength algorithm or time difference of arrival algorithm, the RFID reader antenna deployed inside the beehive continuously reads the identification number of the UHF passive RFID tag set on the queen bee. During the reading process, the movement trajectory of the queen bee in the beehive is determined in real time, and the spatial trajectory time series data of the queen bee is generated. Using a capacitive sensor counter installed on the beehive, the number of bees entering and leaving the beehive per unit time is counted by the capacitance and / or voltage changes caused by bees entering and leaving the beehive, generating bee colony inflow and outflow time-series data. Sensors installed on the beehive are used to obtain information about the internal environment of the beehive.

3. The method as described in claim 2, characterized in that, Using either a signal strength algorithm or a time difference of arrival algorithm, and leveraging an RFID reader antenna deployed inside the beehive, the system continuously reads the identification number of the UHF passive RFID tag attached to the queen bee. During the reading process, the system determines the queen bee's movement trajectory within the beehive in real time, generating temporal data of the queen bee's spatial trajectory, including: The first signal strength received by the RFID reader antenna when reading the identification number of an UHF passive RFID tag set at a preset distance was determined by using an RFID reader antenna deployed inside the beehive. By using multiple RFID reader antennas deployed inside the beehive, the identification number of the ultra-high frequency passive RFID tag set on the queen bee is continuously read, and the second signal strength received by each RFID reader antenna is obtained. Based on the first signal strength, the second signal strength, the path loss index related to the environment, and the zero-mean Gaussian noise under random shadowing effect, the distance between the ultra-high frequency passive RFID tag and each RFID reader antenna is determined. Based on the distance between the ultra-high frequency passive RFID tag and each RFID reader antenna, the spatial location of the queen bee is determined, and the temporal data of the queen bee's spatial trajectory is generated.

4. The method as described in claim 3, characterized in that, Based on the first signal strength, the second signal strength, the environment-related path loss exponent, and zero-mean Gaussian noise under random shading effects, the distance between the UHF passive RFID tag and each RFID reader antenna is determined, including: The distance between the UHF passive RFID tag and each RFID reader antenna is determined using the following formula: ； in, for The second signal strength received by the i-th RFID reader antenna; The distance between the ultra-high frequency passive RFID tag and the i-th RFID reader antenna; The first signal strength received by the RFID reader antenna when reading the identification number of an ultra-high frequency passive RFID tag set at a preset distance; n is the environment-related path loss index; The preset distance; It is zero-mean Gaussian noise under random shading effect.

5. The method as described in claim 1, characterized in that, From the queen bee spatial trajectory time-series data, bee colony entry and exit flow time-series data, and hive internal environment information, extract queen bee individual movement characteristics, bee colony collective behavior characteristics, and environmental auxiliary characteristics, including: The temporal data of the queen bee's spatial trajectory, the temporal data of the bee colony's inflow and outflow, and the information of the beehive's internal environment are subjected to noise reduction, interpolation, and smoothing. The time-series data of the queen bee's spatial trajectory, the time-series data of the bee colony's inflow and outflow, and the information of the beehive's internal environment after noise reduction, interpolation, and smoothing are aligned and aggregated according to time. Based on the aligned and aggregated temporal data of the queen bee's spatial trajectory, the individual movement characteristics of the queen bee are extracted; the individual movement characteristics of the queen bee include one or any combination of the following: total daily movement distance, average daily movement speed, activity range, duration of stay in the brood rearing area in the hive, variance of movement speed, entropy value of movement trajectory, and activity rhythm. Based on the aligned and aggregated bee colony inflow and outflow time series data, bee colony collective behavior characteristics are extracted; the bee colony collective behavior characteristics include one or any combination of the daily total number of bees entering and leaving, the daily peak time for bees entering and leaving, the daily activity change rate, the daily total number of attending bees, and the average attendance rate during the daily peak time for bees entering and leaving.

6. The method as described in claim 1, characterized in that, Before inputting the queen bee spatial trajectory time-series data, the bee colony inflow and outflow time-series data, the beehive internal environment information, and the real-time multidimensional feature vector into the pre-trained queen bee state assessment model and outputting the queen bee state assessment result, the following steps are also included: Collect historical data from multiple consecutive beekeeping seasons, covering the entire life cycle of bee colonies exceeding a preset number; the historical data includes queen bee spatial trajectory time-series data, bee colony inflow and outflow time-series data, and beehive internal environment information; Historical multidimensional feature vectors are extracted and generated from the historical data; Based on the emergence date of the queen bee in each colony, the historical data and the historical multidimensional feature vector are labeled with queen bee status tags; the emergence date of the queen bee labeled as juvenile is no more than a first preset time interval from the labeling time; the emergence date of the queen bee labeled as middle-aged is greater than or equal to the first preset time interval and less than or equal to the second preset time interval from the labeling time; the emergence date of the queen bee labeled as senile is more than the second preset time interval from the labeling time. The XGBoost algorithm with extreme gradient boosting is used to train a classification model using historical data labeled with queen bee state tags and historical multidimensional feature vectors.

7. A queen bee status monitoring device, characterized in that, include: The real-time data acquisition module is used to acquire real-time data using RFID tags installed on the queen bee and sensors installed on the beehive; the real-time data includes queen bee spatial trajectory time-series data, bee colony inflow and outflow time-series data, and beehive internal environment information; The feature extraction module is used to extract individual movement features of the queen bee, collective behavior features of the bee colony, and environmental auxiliary features from the queen bee spatial trajectory time series data, the bee colony inflow and outflow time series data, and the beehive internal environment information. The real-time multidimensional feature vector generation module is used to align and stitch together the individual movement features of the queen bee, the collective behavior features of the bee colony, and the environmental auxiliary features to generate a real-time multidimensional feature vector. The queen bee status assessment result output module is used to input the queen bee spatial trajectory time-series data, the bee colony inflow and outflow time-series data, the beehive internal environment information, and the real-time multidimensional feature vector into a pre-trained queen bee status assessment model, and output the queen bee status assessment result. The queen bee status assessment model is obtained by training a classification model using historical data labeled with queen bee status tags and historical multidimensional feature vectors. The queen bee status tags include one or any combination of juvenile, middle-aged, and senile stages. The queen bee status assessment result includes the probability that the queen bee belongs to each queen bee status tag. The dynamic early warning module is used to issue dynamic early warnings based on the queen bee status assessment results, the real-time data, and the real-time multidimensional feature vector.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 6.

Images