Integrated air-space-ground management system based on multi-source data fusion

Abstract

This invention discloses an integrated air-space-ground management system based on multi-source data fusion, belonging to the field of smart agriculture information technology. It includes a bee colony data acquisition module, a time alignment module, a pollination comparison and analysis module, a flowering period verification module, and a pollination scheduling module. The bee colony data acquisition module uses fixed infrared bee counters at the beehive entrance and exit to continuously collect information on bee entry and exit times, ambient temperature, and geographical location according to a unified time marker, and generates a record of bee entry and exit time rhythms based on the collection results. This invention establishes a time correspondence between pollination behavior and crop physiological responses by integrating bee colony behavior, satellite vegetation index, and UAV imagery data, enabling real-time assessment and anomaly identification of pollination efficiency. Furthermore, it optimizes beehive placement, bee colony replenishment, and agricultural rhythm by combining flowering period imagery information, thereby improving pollination coverage and crop yield stability.

Description

Technical Field

[0001] This invention relates to the field of smart agriculture information technology, specifically to an integrated air-space-ground management system based on multi-source data fusion. Background Technology

[0002] A big data-driven integrated air-space-ground management system based on multi-source data fusion refers to a comprehensive management system that unifies and collaboratively analyzes vegetation index data (such as NDVI and EVI) acquired by satellite remote sensing, canopy structure and flowering period image data acquired by UAVs, and bee entry / exit frequency and environmental parameter data collected by infrared sensors deployed at beehive entrances and exits. This system, combined with a big data processing and computing framework, performs high-dimensional correlation modeling and temporal trend mining on multi-source information, thereby constructing a comprehensive management system for the relationship between crop growth status and pollination behavior. Satellite data is used to macroscopically identify changes in crop greening and bolting stages, UAV data is used to mesoscale characterize canopy flowering morphology and flower distribution, and bee colony entry / exit data serves as a real-time ground-based signal representing pollination activity. Through the mutual verification of multi-source data and big data processing results over time, the system achieves accurate determination of crop flowering initiation, peak flowering duration, and effective pollination windows, thus supporting the scheduling of farmland pollination resources, optimization of bee colony deployment, and refined management of the reproductive growth stage.

[0003] The existing technology has the following shortcomings:

[0004] In existing technologies, multi-source pollination monitoring typically uses beehive entry and exit counts as the primary basis for pollination intensity. When bee colony activity remains high for an extended period, the system assumes the pollination process is proceeding normally. However, during actual dynamic growth, vegetation indices may not show synchronous physiological responses, and drones may detect a decline in crop seed setting rates in later identification stages. This inconsistency in data across different stages is easily overlooked. Because existing systems lack the ability to continuously correlate bee colony behavior data with crop physiological changes, they are prone to misjudging seemingly active pollination behavior as effective pollination, thus failing to identify pollination failures or low pollination efficiency in a timely manner. This leads to the risk of yield reduction being masked in the early stages, resulting in missed intervention opportunities.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide an integrated air-space-ground management system based on multi-source data fusion to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: an integrated air-space-ground management system based on multi-source data fusion, comprising a bee colony data acquisition module, a time alignment module, a pollination comparison and analysis module, a flowering period verification module, and a pollination scheduling module:

[0008] The bee colony data acquisition module uses an infrared bee counter fixed at the entrance and exit of the beehive to continuously collect information on the number of times bees enter and exit, ambient temperature, and geographical location according to a unified time marker, and generates a record of the bee entry and exit time rhythm based on the collection results.

[0009] The time alignment module filters the daily peak periods of bee activity according to the bee entry and exit time rhythm records, obtains the corresponding time satellite vegetation index data within the peak bee activity period, and collects field drone image data within the peak bee activity period. The same time reference table is constructed by combining the obtained satellite vegetation index data and drone image data.

[0010] The pollination comparison analysis module analyzes the rate of change of satellite vegetation index data based on the same time reference table, and combines the bee activity intensity in the bee entry and exit time rhythm record to generate a pollination response comparison record, and marks the time period in the pollination response comparison record where the bee activity intensity and the rate of change of vegetation index do not match.

[0011] The flowering period verification module collects drone flowering period image data again within the corresponding time period for the marked time period, and extracts information on flower opening status, flower falling status and plant appearance changes from the drone flowering period image data, and supplements the pollination response comparison record with the extracted flowering period change information.

[0012] The pollination scheduling module identifies suspected ineffective pollination areas by comparing pollination response records after supplementing information on changes in flowering period, and adjusts the hive movement order, hive replenishment time, and agricultural operation avoidance time based on these suspected ineffective pollination areas.

[0013] Preferably, the steps for generating records of bee entry and exit time rhythms are as follows:

[0014] Infrared transmitters and receivers are fixed on both sides of the beehive entrance and exit to form a light curtain covering the channel with infrared beams. Optical collimation is used to keep the transmitted light beams and receiver probes coaxially aligned. At the same time, a temperature and humidity sensor is installed above the beehive and a geolocation device is installed at the bottom of the beehive.

[0015] After the sensing device is fixed, the infrared transmitter continuously emits invisible light waves. When the bee enters or exits the channel and blocks the light beam, a pulse signal is generated. The receiver detects the signal change and collects the number of times the bee enters or exits, the ambient temperature, and the geographical location information according to a uniform time mark.

[0016] The continuously collected data is organized in chronological order and supplemented with temperature, humidity and location information to generate a time slice data structure that includes time, environmental and spatial attributes.

[0017] Based on time slice data, the number of entry and exit events is counted, and a record of the bee entry and exit time rhythm is generated, which includes the number of entry and exit events, ambient temperature, and location coordinates.

[0018] Preferably, the transmission power, detection spacing, and sampling frequency of the infrared bee counter are adjusted according to the ambient light conditions to ensure that the infrared light curtain maintains a stable response under different climatic conditions. The collected bee entry and exit signals are recorded synchronously with the ambient temperature and geographical location data. Through continuous sampling, time series data covering the day and night cycle are formed, providing a stable collection basis for the generation of bee entry and exit time rhythm records.

[0019] Preferably, the steps for constructing a reference table at the same time are as follows:

[0020] Based on the generated records of bee entry and exit time rhythms, the daily activity curves are analyzed to determine the time range in which the bee colony activity intensity is highest and remains stable, and a table of concentrated activity time periods including start time, end time, duration and geographical coordinates is generated.

[0021] According to the time stamp of the concentrated period of activity, extract the corresponding satellite vegetation index data, obtain the NDVI and EVI values ​​of the pollination area, and align the time stamp of the data with the concentrated period of activity.

[0022] Based on the peak activity periods, drones were organized to collect field images during the corresponding time periods, and the image timestamps were matched with the satellite data timestamps to form time-aligned results;

[0023] By unifying and organizing peak bee colony activity periods, satellite vegetation index data, and drone imagery data, a unified time reference table is constructed, with time as the main axis and including behavioral, environmental, spatial, and physiological information.

[0024] Preferably, when extracting satellite vegetation index data, a geographic buffer is set with the beehive geographic location information as the center to obtain vegetation index raster data covering the pollination area. When constructing the same time reference table, the bee colony activity intensity, NDVI, EVI, ambient temperature and UAV image information are arranged synchronously according to a unified time label to ensure the consistency of multi-source data in time and space.

[0025] Preferably, the steps for generating pollination response comparison records are as follows:

[0026] Based on the established reference table for the same time, the satellite vegetation index data are organized in chronological order to obtain a sequence of vegetation index change rates over a continuous time period.

[0027] After obtaining the vegetation index change rate sequence, the bee entry and exit time rhythm records for the same time period were extracted from the same time reference table, and the bee activity intensity data were read and arranged in a corresponding manner with the environmental temperature and humidity information.

[0028] After completing the correlation between bee colony activity intensity and vegetation index change rate, pollination response comparison records were generated in chronological order, including time markers, bee colony activity intensity, NDVI change rate, EVI change rate, and geographic coordinates.

[0029] After generating pollination response comparison records, time periods where the intensity of bee colony activity did not match the rate of change of vegetation index were marked, and the type, duration, temperature, humidity and location information were recorded.

[0030] Preferably, when marking the time period in which the intensity of bee colony activity does not match the rate of change of vegetation index, the time period in which the intensity of bee colony activity remains at a high level and the rate of change of vegetation index is lower than the average of the continuous time period is marked as an abnormal phase of strong activity but weak physiological response, and the time period in which the rate of change of vegetation index increases but the intensity of bee colony activity does not increase is marked as a phase of insufficient pollination participation.

[0031] Preferably, the steps for generating pollination response comparison records are as follows:

[0032] Based on the time periods already marked in the pollination response comparison record, the time window and flight area for the UAV to collect data again are determined, and the time is aligned with the time tags of the pollination response comparison record through a time synchronization device.

[0033] After the UAV completes image acquisition according to the predetermined flight path, the acquired visible light images and multispectral images are organized according to time and space, and the geographic coordinates of the images are matched with the beehive monitoring points and satellite pixel positions;

[0034] After completing the processing of drone images of flowering period, information on flower opening status, flower falling status and plant appearance changes is extracted from the images, and these statuses are organized into structured information with text descriptions and numerical indicators.

[0035] The extracted information on changes in flowering period was added to the corresponding data rows according to the time markers of the pollination response comparison records. Fields for flower opening degree, flower drop ratio, and plant appearance change were added to each time period.

[0036] Preferably, the process of collecting images of flowering period by drone is further defined as follows: the drone is equipped with a multispectral and high-resolution visible light imaging device, and takes low-altitude pictures along a preset route at a set altitude and speed. During the flight, the drone maintains stable flight through attitude control and automatic route execution equipment to ensure that the collected image data corresponds completely with the pollination response comparison record in time and space.

[0037] Preferably, the steps for identifying suspected ineffective pollination areas based on pollination response comparison records after supplementing information on changes in flowering period, and adjusting the hive movement order, hive replenishment time, and agricultural operation avoidance time according to suspected ineffective pollination areas are as follows:

[0038] Based on the pollination response comparison records after supplementing information on flowering period changes, a spatiotemporal dataset containing time, geographical location, bee activity intensity, vegetation index change rate, flower opening status, flower fall status, and plant appearance changes was extracted, and a spatial correspondence between bee colony pollination activity intensity and crop vegetation response was established.

[0039] After obtaining a candidate set of suspected invalid pollination areas, the time series data are compared and combined with the changing trends of flower opening and falling status to determine stable suspected invalid pollination areas and form a list of areas.

[0040] After identifying suspected ineffective pollination areas, a plan for hive movement sequence and hive replenishment time is formulated based on regional distribution characteristics and bee colony activity patterns, so that the bee colony coverage radius extends to the ineffective pollination area.

[0041] After completing the hive movement sequence and bee colony replenishment schedule, adjust the agricultural operation avoidance time based on the time and spatial distribution of suspected ineffective pollination areas, and add it to the pollination response comparison record.

[0042] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0043] This invention achieves multi-source fusion of bee colony behavior data, satellite vegetation index data, and UAV imagery data within a unified timeframe, establishing a direct temporal correspondence between biological behavioral signals during pollination and crop physiological states. This method continuously tracks the dynamic relationship between bee activity intensity and crop photosynthetic activity and canopy structure changes, enabling timely detection of anomalies such as active bee colonies but asynchronous plant physiological responses during pollination. This allows for early identification of decreased pollination efficiency or pollination failure, preventing delayed exposure of yield loss risks.

[0044] This invention supplements pollination response comparison records with flowering period image information, further analyzing the correlation between physiological characteristics such as flower opening, flower fall, and changes in plant appearance and pollination behavior, thus providing physiological verification for pollination status judgment. Based on these correlation results, the system identifies suspected ineffective pollination areas and adjusts the beehive placement order, bee colony replenishment timing, and agricultural operation rhythm accordingly. This achieves dynamic optimization and spatial scheduling of pollination resources, improves pollination coverage and the matching degree between pollination behavior and flowering window, and promotes the stability of crop reproductive growth process and yield formation efficiency. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0046] Figure 1 This is a schematic diagram of the modules of the integrated air-space-ground management system based on multi-source data fusion of the present invention. Detailed Implementation

[0047] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0048] This invention provides, for example Figure 1 The integrated air-space-ground management system based on multi-source data fusion shown includes a bee colony data acquisition module, a time alignment module, a pollination comparison and analysis module, a flowering period verification module, and a pollination scheduling module.

[0049] The bee colony data acquisition module uses an infrared bee counter fixed at the entrance and exit of the beehive to continuously collect information on the number of times bees enter and exit, ambient temperature, and geographical location according to a unified time marker, and generates a record of the bee entry and exit time rhythm based on the collection results.

[0050] To achieve continuous data collection and circadian rhythm recording of bee activity at hive entrances and exits, a sophisticated deployment and data acquisition process can be implemented. This process uses an infrared bee counter as the core sensing unit, combined with an environmental temperature sensor and a geolocation device to form a continuous monitoring link, generating spatially attributed records of bee entry and exit circadian rhythms under a unified time marker. The entire process can be carried out as follows:

[0051] Within the implementation area, select beehives for pollination monitoring. Based on the geometry of the beehive entrance and exit and the bee flight path, determine the installation positions of the infrared transmitter and receiver. The infrared transmitter and receiver are fixed on both sides of the beehive entrance and exit, ensuring the infrared beam forms a continuous light curtain covering the entire width of the passage, guaranteeing that each bee entering or exiting interrupts the light signal. After installation, the infrared beam is optically collimated to ensure the emitted light and the receiving probe are coaxially aligned, avoiding signal interference caused by ambient light or bee reflections. Miniature temperature and humidity sensors are fixed above the beehive entrance and exit to synchronously record the ambient temperature and relative humidity during entry and exit times, facilitating subsequent analysis of the climatic background of bee activity. Simultaneously, a geolocation device is installed on the bottom of the beehive shell or its support platform, using global positioning technology to obtain the latitude and longitude information of the beehive's location, forming a monitoring unit with spatial coordinates. After sensor deployment, all sensors and the power management unit are connected via shielded cables to ensure electrical stability and reliable signal transmission. By adjusting parameters such as infrared detection spacing, transmission power, and sampling frequency, the light curtain maintains a continuous and stable response under different ambient light conditions.

[0052] After fixing and initially calibrating the sensing devices, the working status of the infrared bee counter is continuously sampled to establish a unified time reference. The time reference is provided by a high-precision real-time clock module and synchronized with an external data aggregation device wirelessly. The infrared transmitter continuously emits invisible light waves. When bees enter or exit the hive passage, their bodies block the light beam, creating a pulse signal. The signal change detected by the receiver is recorded as a single event. The system samples the signal at a fixed time resolution and assigns a timestamp and direction identifier to each entry / exit event. Temperature and humidity sensors synchronously collect current ambient temperature and humidity values ​​under the same time reference, and the positioning device continuously outputs location data and records it on the same timeline, ensuring all data has consistent time and spatial identifiers. Through this continuous acquisition process, a raw dataset containing time series, direction attributes, climate background, and spatial coordinates is obtained, laying the data foundation for subsequent rhythm recording. To prevent changes in external lighting or obstructions from affecting the accuracy of infrared detection, protective covers are installed on the infrared transmitter and receiver, ensuring they only respond to bee passage and guaranteeing a one-to-one correspondence between the collected signals and bee activity.

[0053] During continuous data collection, all timestamped data is organized and marked in chronological order to generate a preliminary sequence of bee entry and exit events. The data management unit counts the number of events within each collection cycle based on a time reference, distinguishing entry and exit directions to establish a correspondence between the number of bees entering and exiting per unit time. Temperature and humidity data, along with geographic location information, are synchronously appended to the corresponding time records, forming a complete time-slice data structure. This data structure allows for tracking the correlation between bee entry / exit activities and environmental changes within any time interval. To improve the continuity of the time series, synchronization identifiers are inserted between collection intervals to confirm consistency in continuous collection. If a short-term data interruption occurs during collection, the system automatically replenishes the identifiers based on the time reference to maintain the integrity of the time series. Throughout the process, the sampling frequency of the infrared signal remains consistent with the sampling frequency of temperature and humidity to ensure strict alignment of the time axis. Finally, through this continuous time-series process, a basic dataset of bee activity covering the diurnal cycle is obtained. Each record unit contains the number of entry / exit events, temperature and humidity values, and location coordinates, achieving a unified correlation between time, environment, and space.

[0054] Based on the processed time-series data, a circadian rhythm record of bee entry and exit is generated. First, the daily collected continuous time data is recombined chronologically so that the number of entry and exit events in each time period reflects the changing trend of bee activity intensity over time. Within each recording period, the number of bee entry and exit events is accumulated to form an activity intensity index. Temperature and humidity information is used to describe the environmental conditions at that time, and geolocation information is used to mark the specific spatial location where the activity occurred. Through multi-dimensional data fusion under a unified time marker, the changes in the bee colony's entry and exit rhythm under different climatic conditions can be displayed on the time axis, thus forming a continuously comparable circadian rhythm curve. This curve reflects the typical activity patterns of the bee colony during sunrise, noon, and evening, and further reveals the temporal characteristics of bee pollination behavior under different climatic environments and geographical locations. To maintain the continuity of the circadian rhythm record, the system maintains the precise consistency of the original time marker during data generation, ensuring that each record can be traced back to the corresponding infrared detection event and environmental state. In this way, the entire implementation process not only achieved the coordinated acquisition of infrared counting and environmental monitoring, but also formed a complete record of the bee entry and exit time rhythm with complete attributes in the time and space dimensions.

[0055] The time alignment module filters the daily peak periods of bee activity according to the bee entry and exit time rhythm records, obtains the corresponding time satellite vegetation index data within the peak bee activity period, and collects field drone image data within the peak bee activity period. The same time reference table is constructed by combining the obtained satellite vegetation index data and drone image data.

[0056] To achieve unified correlation of multi-source data across time and ensure that bee colony behavior data, satellite vegetation index data, and drone imagery data can be effectively correlated within the same time frame, a multi-stage acquisition and matching process based on bee entry and exit time rhythm records can be implemented. This process uses time rhythm data as the core guiding principle, achieving time-synchronized mapping from behavioral signals to physiological responses by selecting concentrated activity periods, extracting satellite vegetation indices, acquiring field drone imagery, and constructing a unified reference table. The entire process unfolds as follows:

[0057] Based on the generated bee entry and exit time rhythm records, the daily activity curves in the continuous records were analyzed to determine the concentrated periods of bee colony activity. These time rhythm records contain information such as the number of entry and exit events, directional indicators, ambient temperature, and geographical location, comprehensively reflecting the changes in the intensity of bee colony entry and exit at different time periods. By statistically comparing the event frequencies of each time interval in the records, the distribution pattern of bee entry and exit events in the diurnal cycle was determined, thereby identifying the time range with the highest and most stable activity intensity. To ensure the representativeness of the selected periods, the peak period of typical bee colony behavior under natural conditions was selected as a reference. Combined with the trend of ambient temperature changes and the duration of daylight, the concentrated periods of bee activity each day were finally defined. These concentrated periods not only reflect the highly active period of bee colony pollination behavior but also represent the time interval where crop pollen release and flower opening are most closely matched. Therefore, they play a crucial role as a time benchmark in subsequent multi-source data fusion. After screening, a table of concentrated activity periods containing start time, end time, duration, and geographical coordinates was generated, providing a precise time window for subsequent data extraction.

[0058] After identifying the peak activity periods of the day, satellite vegetation index data corresponding to these periods is extracted based on their time stamps. Satellite remote sensing platforms acquire multispectral images of the Earth's surface at fixed time intervals. After preprocessing, the image data includes reflectance information for multiple bands, including red, blue, and near-infrared, which can be used to calculate vegetation indices. Vegetation indices include the Normalized Difference Vegetation Index (NDVI) and the Enhanced Vegetation Index (EVI), both reflecting changes in crop chlorophyll content, canopy structure, and photosynthetic activity. Based on the time stamps of the peak bee activity periods, the closest satellite image data for that time period is retrieved from the cloud data center, and the NDVI and EVI values ​​for the corresponding geographic location pixels are extracted. During the extraction process, a geographic buffer zone is set around the beehive's geographic location information to acquire satellite vegetation index raster data covering the pollination area. To ensure temporal consistency, the extracted vegetation index values ​​are aligned with the peak activity periods according to their time stamps, and auxiliary information such as the specific time of data collection, observation angle, and atmospheric conditions are recorded, ensuring a correspondence between bee activity intensity and crop vegetation status on the same time scale. In this way, satellite data not only provides macroscopic information on crop physiological status, but also provides physiological reference for subsequent coupled analysis of flowering period images and pollination behavior.

[0059] After extracting satellite vegetation index data, drones were organized to collect field images during the corresponding time periods based on the same peak activity periods. The drones, equipped with high-resolution visible light and multispectral imaging equipment, conducted low-altitude photography of crop fields corresponding to the bee colony activity areas along preset flight paths. The shooting time was strictly controlled within the peak bee activity periods to ensure that the drone images matched the timing of bee pollination activities and satellite vegetation indices. During flight, the drones maintained stable flight through attitude sensing and positioning devices, and acquired continuous image data of the crop canopy using oversampling. The collected field images clearly showed detailed information such as flower distribution, canopy density, flowering status, and flower color changes, reflecting the morphological characteristics of the crop's reproductive growth stage at the microscopic level. After collection, the drone image data was stored according to timestamps and geographic coordinates, along with additional information such as shooting altitude, angle, and ambient light intensity. To ensure that the image data remained time-consistent with the aforementioned satellite vegetation index data, the drone image timestamps were matched with the satellite data timestamps, with the peak activity period as the main timeline, resulting in a time-aligned image and vegetation index. In this way, the microscopic images of peak bee colony activity can be synchronously correlated with the macroscopic vegetation index.

[0060] The obtained data on peak bee activity periods, satellite vegetation index data, and UAV field imagery were uniformly organized to construct a unified temporal reference table. This table uses time as the main axis and geographic location as the index unit, simultaneously recording bee activity intensity, NDVI and EVI values, ambient temperature, and canopy image descriptions at each time point. During construction, the number of consecutive events within peak bee activity periods was used as the behavioral intensity indicator, the vegetation index value of satellite pixels as the crop physiological indicator, and the flower quantity information extracted from UAV imagery as the flowering morphology indicator. These three data points were arranged synchronously over time, forming a data table structure with multi-dimensional attributes. Each record includes beehive location, time stamp, number of entry and exit events, NDVI, EVI, temperature and humidity, and UAV image file index, enabling cross-alignment of data from different sources at the same time scale. In this way, bee behavior data no longer exists independently but establishes a direct temporal mapping relationship with crop physiological state and flowering morphology. The constructed time reference table can serve as the basis for temporal fusion in subsequent analyses, comparing the synergy between bee colony activity and vegetation changes at different times, and providing input data within a unified time frame for subsequent pollination response comparisons. This reference table not only includes information from multiple dimensions such as behavior, environment, space, and physiology, but also possesses traceability and temporal continuity, enabling subsequent data processing at each stage to be correlated and expanded using a unified standard.

[0061] To ensure that all types of data have clear data sources, data structures, and processable attributes, the following descriptions are provided for the relevant data:

[0062] Satellite vegetation index data is a set of pixel-level data calculated based on satellite remote sensing images. Each pixel data is associated with a corresponding geographic location and contains a time identifier and a vegetation index value for the corresponding time. The vegetation index values ​​include normalized differential vegetation index values ​​and enhanced vegetation index values, which are used to characterize the chlorophyll content and canopy structure status of crops at the corresponding time nodes.

[0063] The bee colony behavior data is time-series data collected by infrared bee counters set at the entrance and exit of the beehive. The bee colony behavior data includes the number of bees entering and leaving the hive per unit time, and corresponds to the collection time and the geographical location of the beehive. It is used to characterize the entry and exit activities of bees in different time intervals.

[0064] Physiological information is a dataset extracted from field images collected by drones. Physiological information includes the number of flowers, the spatial distribution of flowers, and the morphology of the plant canopy. The number of flowers reflects the flowering status of the crop, the spatial distribution of flowers reflects the distribution of the number of flowers, and the morphology of the plant canopy reflects the growth status of the crop.

[0065] Behavioral information is time-series feature data formed based on bee colony behavior data. Behavioral information includes the change curve of bee entry and exit frequency over time and its corresponding time distribution, which is used to describe the changing pattern of bee activity in different time intervals.

[0066] All types of data are recorded according to a unified time stamp and stored using time stamps and geographic locations as indexes. This enables satellite vegetation index data, bee colony behavior data, and physiological information to establish corresponding relationships at the same time point and spatial location, thereby supporting collaborative analysis of multi-source data.

[0067] The pollination comparison analysis module analyzes the rate of change of satellite vegetation index data based on the same time reference table, and combines the bee activity intensity in the bee entry and exit time rhythm record to generate a pollination response comparison record, and marks the time period in the pollination response comparison record where the bee activity intensity and the rate of change of vegetation index do not match.

[0068] To achieve joint comparison and correlation of pollination responses based on multi-source data within the same time frame, multi-stage data processing and analysis can be conducted on the basis of the previously constructed time reference table, focusing on the correspondence between the temporal rate of change of vegetation indices and the intensity of bee activity. This process involves extracting the trend of vegetation index changes from continuous time series and synchronously comparing it with the activity intensity recorded in the bee entry and exit time rhythm. This generates a pollination response comparison record that reflects the correlation between pollination behavior and vegetation physiological responses. Furthermore, time periods where the two do not match are marked to identify potential pollination anomalies. The entire process can be carried out according to the following steps:

[0069] Based on the previously constructed time reference table, the satellite vegetation index data contained in the table were processed into a time series. The reference table already records information such as NDVI and EVI values, bee colony activity intensity, ambient temperature, and geographic coordinates for each moment. To form continuous time data that reflects the rate of crop physiological change, the vegetation index data of adjacent sampling periods were arranged sequentially according to time order, ensuring consistent time intervals and continuous coverage of concentrated activity periods. During the processing, the vegetation index value of each time period was differentially processed with the value of the previous time period, and the change per unit time was calculated using the collection time interval as a benchmark. The magnitude of the change reflects the trend of crop chlorophyll content and canopy structure changes in a short period of time. To avoid bias caused by different satellite observation angles, the solar altitude angle and ground reflection condition parameters in the image metadata were retained during processing for reference in subsequent comparisons. After time-series processing, a vegetation index change rate sequence was obtained within a continuous time period. This sequence can intuitively reflect the dynamic changes in crop photosynthetic activity and canopy structure during peak bee activity periods, providing a basis for subsequent pollination response comparisons.

[0070] After obtaining the temporal rate of change sequence of vegetation index, the corresponding bee entry and exit rhythm records for each time period were extracted from the same time reference table, and bee activity intensity data for the same time period were read from the records. This activity intensity data originated from the statistical analysis of entry and exit events collected by infrared sensors and was aligned with satellite observation time using a unified time stamp. The activity intensity was arranged in correspondence with time periods, ensuring a one-to-one correspondence between the frequency of bee colony entry and exit within different time windows and the rate of change of vegetation index. Simultaneously, the activity intensity data was extracted along with the ambient temperature and humidity information to describe the impact of the external environment on pollination behavior. In this way, each unified time node contains two core parameters: one is the rate of change of vegetation index at the crop level, and the other is the pollination activity intensity at the ground bee colony level. This correspondence ensures that the subsequently generated pollination response comparison records not only have temporal continuity but also achieve a physical time mapping from ground biological behavior to plant physiological responses. To ensure data continuity, the complete connection of the time index was maintained during the extraction process, ensuring that the activity intensity and vegetation change rate between adjacent time periods maintained the same time step, guaranteeing that they could be compared using the same time unit.

[0071] After compiling the correlation between bee colony activity intensity and vegetation index change rates, the time series of both were compared hourly to generate a pollination response comparison record. The core structure of the pollination response comparison record includes fields such as time markers, bee colony activity intensity, NDVI change rate, EVI change rate, ambient temperature, and geographic coordinates, used to describe the dynamic relationship between bee colony behavior and crop physiological responses. Using the time axis as the main line, bee colony activity intensity and vegetation index change rates at the same time point are recorded side-by-side, forming a continuous pollination response sequence through chronological order. This record not only reflects the synchronicity between changes in bee colony behavior and vegetation physiological responses but also reveals the activity level of the pollination process and the spatiotemporal response characteristics of plant energy absorption under specific environmental conditions. For example, when bee colony activity intensity increases and the NDVI or EVI change rate increases synchronously, it indicates a synergistic relationship between pollination behavior and crop photosynthetic activity; while when bee colony activity intensity remains at a high level but the vegetation index change rate does not increase accordingly, it indicates that the crop physiological response has not synchronized with pollination behavior, potentially indicating limited pollination efficiency or insufficient pollen activity. The pollination response comparison records generated in this way can reflect the dynamic matching degree between pollination activities and plant physiological states over time, forming a set of basic data for subsequent labeling and analysis.

[0072] After generating pollination response comparison records, mismatched time periods were marked and explained based on the comparison between bee colony activity intensity and the temporal rate of change of vegetation index. The marking process was conducted on a time-by-time basis, comparing the relative trends of bee colony activity intensity and the rates of change of NDVI and EVI in the records. When bee colony activity intensity remained at a high level and the rate of change of vegetation index was lower than the average of the continuous time period, this time period was marked as an abnormal phase of strong activity but weak physiological response. Conversely, when the rate of change of vegetation index increased but bee colony activity intensity did not increase, it was marked as a possible phase of insufficient pollination participation. Each marked time period was marked with an identification field indicating its type and duration, and the corresponding temperature, humidity, and location information were retained to analyze whether differences were caused by environmental conditions or spatial distribution. After marking, the pollination response comparison records not only included time periods of normal pollination but also clearly presented time zones where pollination behavior and crop physiological changes were asynchronous, making the data traceable and interpretable. This labeling method helps identify temporal discrepancies between bee colony behavior and vegetation status, providing accurate time references for subsequent flowering period verification and pollination scheduling. The labeled pollination response comparison records are stored in chronological order and can serve as input for multi-source data fusion analysis, further deriving the correlation between pollination status and crop growth stages.

[0073] To achieve a quantitative description of bee activity and crop flowering status, the calculation methods for each parameter are limited as follows:

[0074] First, the bee activity intensity is the sum of the number of times bees enter and leave the hive per unit time. Specifically, it is the cumulative value of the number of times bees enter and leave the hive recorded by an infrared bee counter within a preset time window. The preset time window is a fixed time interval, so that the data in different time periods have a consistent time scale, thereby forming continuous time series activity intensity data.

[0075] Secondly, the intensity of bee colony pollination activity is determined based on the activity intensity of bees, which is the frequency of bee activities participating in pollination per unit time. Its value is characterized by the change of the bee activity intensity over time, and is used to reflect the pollination activity level of the bee colony in the corresponding time period. In specific implementation, the intensity of bee colony pollination activity is directly represented by the number of times bees enter and exit per unit time.

[0076] Furthermore, the method for determining that the bee colony activity intensity remains at a high level is as follows: within multiple consecutive time windows, the bee activity intensity per unit time is higher than a preset activity intensity threshold; the preset activity intensity threshold is obtained by statistical analysis of historical data, or is set as a fixed reference value based on the bee colony size and crop type, thereby realizing the quantitative determination of the continuous active state.

[0077] Furthermore, the degree of flower opening is characterized by the ratio of the number of open flowers to the total number of flower buds per unit area. Specifically, in the flowering period images collected by drones, the number of open flowers to the total number of flower buds in the target area is counted and their ratio is calculated to obtain a quantitative result of the degree of flower opening; the change of the ratio over time is used to reflect the flowering period development process.

[0078] Meanwhile, the state of flower fall is characterized by the ratio of the number of fallen flowers per unit area to the initial number of flowers. Changes in plant appearance are described by changes in canopy height, leaf distribution morphology, and changes in image grayscale or spectral response, and are linked to time markers to form a continuous change record.

[0079] Through the above quantification methods, the intensity of bee activity, the intensity of bee colony pollination activity, and the degree of flower opening are all converted into measurable numerical parameters, and correlated with the rate of change of satellite vegetation index under a unified time marker, thereby realizing a quantitative correlation analysis between pollination behavior and crop physiological response.

[0080] The flowering period verification module collects drone flowering period image data again within the corresponding time period for the marked time period, and extracts information on flower opening status, flower falling status and plant appearance changes from the drone flowering period image data, and supplements the pollination response comparison record with the extracted flowering period change information.

[0081] To supplement the previously marked time periods with targeted data and further improve the pollination response comparison records, establishing a direct temporal correspondence between bee colony behavior signals and crop flowering status, this can be achieved through secondary acquisition of UAV flowering images and extraction of flowering change information. This process uses the marked time periods as the core basis, re-collecting and extracting features from the crop field under the same time conditions, thus establishing a traceable link between abnormal bee colony activity and changes in flowering status. The entire process is carried out according to the following steps:

[0082] Based on the marked time periods in the pollination response comparison records, the specific time window and flight area for the drone's next data collection were determined. The previous pollination response comparison records already included information such as bee activity intensity, vegetation index change rate, ambient temperature, and geographical location, clearly identifying time periods where bee activity intensity and vegetation index change rate did not match. Based on the start and end times of these marked time periods, a drone flight plan was developed to ensure that the shooting time strictly corresponded to the marked time intervals. The drone's flight area was centered on the bee colony monitoring point, and combined with the geographical coverage of satellite remote sensing, corresponding field flight boundaries were set to ensure that the shooting area covered the area affected by bee pollination activity. The drone was equipped with multispectral and high-resolution visible light imaging devices, as well as attitude control and automatic flight path execution equipment, to complete full-coverage field image acquisition at the set altitude and speed. After the flight mission was set, the time was synchronized with the time stamps in the pollination response comparison records using a time synchronization device to ensure that the drone's start time, flight process, and image acquisition were all consistent with the time nodes of the marked time periods, achieving complete temporal correspondence. In this way, the acquired image data can accurately reflect the true state of crop flowering during abnormal time periods.

[0083] After the drone completes image acquisition along its predetermined flight path, the acquired image data is processed temporally and spatially. The drone-acquired image data includes both visible light and multispectral images. Visible light images record the color, shape, and distribution of flowers in the crop canopy, while multispectral images record the reflectance characteristics of different wavelengths, reflecting the differences in the spectral response of petals, leaves, and fruit tissues. After acquisition, all image data are numbered according to the acquisition time sequence and matched one-to-one with the corresponding time markers in the pollination response comparison records. To ensure spatial positioning consistency, the geographic coordinates of the drone image data are spatially matched with the beehive monitoring points and satellite pixel locations, ensuring that each frame of image corresponds to a specific crop area. During the processing, the flight altitude, tilt angle, and illumination conditions are recorded simultaneously to accurately interpret image differences in subsequent analysis. The processed data files are indexed primarily by time tags, with each folder corresponding to all image materials within a marked time period, forming a data set that is aligned temporally and spatially. In this way, information on abnormal bee colony activity, vegetation index changes, and field flowering period images are uniformly correlated within each marked time period.

[0084] After processing the drone images of the flowering period, information on the opening and falling of crop flowers, as well as changes in plant appearance, was extracted from the images. The opening status of flowers can be determined by identifying the color, quantity, and distribution characteristics of flowers in the canopy. Changes in flower color reflect the flowering process, such as the color gradations from the initial opening stage to the peak bloom and then to the decline stage. The falling of flowers can be judged by the reduction in the number of flowers in the canopy, the traces of petal fall, and the proportion of remaining flower stalks. These changes represent the transformation of reproductive organ function after pollination. Changes in plant appearance are reflected in changes in canopy height, leaf morphology, and light reflection distribution, reflecting the structural adjustments of the crop during the reproductive growth stage. During the extraction process, images from different time periods in the same area were compared frame by frame according to the image time series, recording the trends of flower opening to closing and from new flowers to fallen flowers. These states were then organized into structured information using textual descriptions and numerical indicators. Each set of extraction results corresponds to the collection time and geographical location, forming a complete record of flowering period changes. This record not only reflects the flowering dynamics of crops during abnormal pollination periods, but also reveals the state transitions of plants during the photosynthetic and reproductive stages, providing a physiological basis for subsequent data supplementation.

[0085] The extracted flowering period change information was supplemented into the original pollination response comparison record to improve data quality and integrate information. During the supplementation, the time markers in the original pollination response comparison record were used as indexes, and flowering period change information for each marked time period was inserted into the corresponding data row. For each time period, three fields were added: flower opening degree, flower drop rate, and plant appearance change, to describe the crop's morphological state during that period. If multiple sets of UAV images existed within the same time period, the extracted results from these images were merged in chronological order, ensuring that the flowering period information in the record continuously reflects the crop's state change trend. After the supplementation, each data point in the pollination response comparison record simultaneously included bee colony activity intensity, vegetation index change rate, environmental parameters, and flowering period morphological information, achieving multi-dimensional information fusion from the behavioral to the physiological level. In this way, the pollination response comparison record was expanded from the original two-source comparison to a comprehensive record containing three types of data: bee colony behavior data, vegetation index data, and flowering period image data. The supplemented record maintains continuity on the timeline and retains the original marker fields, allowing subsequent pollination anomaly identification and management scheduling to be based on more comprehensive data. At the same time, this supplementary approach also makes the pollination response comparison record scalable, enabling the inclusion of more physiological or environmental monitoring data in subsequent implementations and achieving multi-source dynamic updates.

[0086] The pollination scheduling module identifies suspected ineffective pollination areas by comparing pollination response records after supplementing information on changes in flowering period, and adjusts the hive movement order, hive replenishment time, and agricultural operation avoidance time based on the suspected ineffective pollination areas.

[0087] To achieve spatial identification of pollination status and dynamic adjustment of production management based on multi-source data fusion, pollination response comparison records after supplementing flowering period change information in the previous stage can be used to identify and partition the spatial relationship between pollination behavior and crop physiological response. This identifies potentially ineffective pollination areas, and accordingly adjusts the beehive placement order, bee colony replenishment timing, and agricultural operation schedule. The entire process is based on the fusion of temporal, spatial, and physiological dimensions of multi-source information, achieving closed-loop management of the pollination process through identification, extraction, comparison, and adjustment steps.

[0088] Based on pollination response comparison records supplemented with flowering period change information, a spatiotemporal dataset was extracted, containing multi-dimensional information including time, geographical location, bee activity intensity, vegetation index change rate, flower opening status, flower fall status, and plant appearance changes. In this dataset, each record corresponds to a specific time node and spatial location, reflecting the combined performance of bee colony behavior and crop physiological response at that moment. To identify potential pollination anomaly areas, the recorded data was first organized in the geospatial dimension, establishing a spatial mapping range centered on the beehive location point, projecting the geographical coordinates of each recording unit onto the corresponding field grid location. This method establishes a spatial correspondence between bee colony pollination activity intensity and crop vegetation response. Further analysis of pollination response characteristics in different fields within the same time period, by comparing the differences in NDVI and EVI change rates among fields under similar bee colony activity intensities, preliminarily identified areas with active pollination behavior but weak crop physiological responses. These areas typically exhibit high bee colony activity intensity, gradual flowering period changes, and low vegetation index fluctuations, indicating that the pollination process failed to effectively promote physiological changes. These extracted areas were recorded as a candidate set of suspected invalid pollination areas, and their geographical coordinates, time range, and corresponding environmental information were preserved to provide a spatial basis for subsequent analysis.

[0089] After obtaining a candidate set of suspected ineffective pollination areas, the time-series data of these areas were further compared to confirm the persistence of their abnormal pollination status. This comparison was based on the time axis of the pollination response comparison records, longitudinally comparing the intensity of bee colony activity and the rate of change of vegetation indices at different time periods within the same area, while simultaneously analyzing the temporal evolution trends of flower opening and falling. When bee colony activity intensity remained consistently high while the rate of change of NDVI or EVI remained stable or decreased over a continuous time period, and flowering images showed delayed flower opening or a high proportion of fallen flowers, the area was determined to be a stable suspected ineffective pollination area. In this process, the environmental temperature variation also needed to be considered. When the environmental temperature was within the suitable range for bee activity and bee colony activity was high, while the crop's vegetation index showed no significant fluctuations, it could be further confirmed that the abnormal pollination response stemmed from insufficient pollination efficiency rather than environmental influences. By comparing continuous data from different time periods, the difference between short-term fluctuations and long-term trends could be effectively identified, thereby eliminating the interference of incidental factors and ensuring that the finally identified suspected ineffective pollination areas were persistent and representative. After the comparison is completed, a list of suspected invalid pollination areas is generated, which includes the area code, geographical range, abnormal time period, and description of pollination response characteristics.

[0090] After identifying suspected ineffective pollination areas, a hive relocation sequence and hive replenishment schedule are determined based on regional distribution characteristics and bee colony activity patterns. Bee colony activity records contain the activity intensity distribution of different hives at their respective locations. By overlaying the suspected ineffective pollination areas with the bee colony activity space, it is possible to determine which hive service areas have insufficient pollination coverage. For areas with insufficient coverage, the placement of adjacent hives is prioritized to extend their coverage radius towards the ineffective pollination area, thereby increasing the frequency of bee colony visits to that area. The relocation sequence is arranged according to the bee colony activity intensity gradient, moving hives with relatively low activity intensity first, and then arranging new locations based on temperature and light conditions during the pollination period. For areas that consistently show ineffective pollination over multiple monitoring periods, the specific timing of bee colony replenishment is determined based on the flowering period duration and crop variety characteristics, ensuring that the new bee colony deployment matches the peak pollen release period, thereby increasing the probability of successful pollination. Bee colony replenishment can be carried out after the original bee colony activity period has ended to avoid behavioral interference between bee colonies. All movement and replenishment operations were carried out within the time coordinate system of the pollination response comparison record to ensure that the adjustment measures were consistent with the monitoring time data.

[0091] After finalizing the hive movement sequence and colony replenishment schedule, adjust the avoidance time for agricultural operations based on the temporal and spatial distribution of suspected ineffective pollination areas. Since there is a risk of time overlap between the pollination process and manual agricultural activities (such as spraying, fertilizing, and mechanical operations), a pollination protection time window needs to be set in subsequent operation plans after identifying areas with low pollination efficiency. Based on the time distribution in the pollination response comparison record, determine the peak pollination period for that area and stagger agricultural operations during these time periods to prevent mechanical disturbance, chemical odors, or temperature fluctuations from affecting bee activity. Simultaneously, maintain a stable field environment during the pollination protection time, avoiding external noise or vibration that could interfere with bee colony activity. For pollination operations in multiple areas simultaneously, phased avoidance can be implemented based on the pollination time differences between areas, thereby ensuring that the pollination efficiency of the bee colony is improved in each area. After adjustments are completed, all movement, replenishment, and avoidance time data will be added to the pollination response comparison record, forming a comprehensive record including pollination anomalies, adjustment measures, and time arrangements, providing a complete reference for subsequent cycle analysis and management.

[0092] This invention achieves multi-source fusion of bee colony behavior data, satellite vegetation index data, and UAV imagery data within a unified timeframe, establishing a direct temporal correspondence between biological behavioral signals during pollination and crop physiological states. This method continuously tracks the dynamic relationship between bee activity intensity and crop photosynthetic activity and canopy structure changes, enabling timely detection of anomalies such as active bee colonies but asynchronous plant physiological responses during pollination. This allows for early identification of decreased pollination efficiency or pollination failure, preventing delayed exposure of yield loss risks.

[0093] This invention supplements pollination response comparison records with flowering period image information, further analyzing the correlation between physiological characteristics such as flower opening, flower fall, and changes in plant appearance and pollination behavior, thus providing physiological verification for pollination status judgment. Based on these correlation results, the system identifies suspected ineffective pollination areas and adjusts the beehive placement order, bee colony replenishment timing, and agricultural operation rhythm accordingly. This achieves dynamic optimization and spatial scheduling of pollination resources, improves pollination coverage and the matching degree between pollination behavior and flowering window, and promotes the stability of crop reproductive growth process and yield formation efficiency.

[0094] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. An integrated air-space-ground management system based on multi-source data fusion, characterized in that, It includes a bee colony data acquisition module, a time alignment module, a pollination comparison and analysis module, a flowering period verification module, and a pollination scheduling module: The bee colony data acquisition module uses an infrared bee counter fixed at the entrance and exit of the beehive to continuously collect information on the number of times bees enter and exit, ambient temperature, and geographical location according to a unified time marker, and generates a record of the bee entry and exit time rhythm based on the collection results. The time alignment module filters the daily peak periods of bee activity according to the bee entry and exit time rhythm records, obtains the corresponding time satellite vegetation index data within the peak bee activity period, and collects field drone image data within the peak bee activity period. The same time reference table is constructed by combining the obtained satellite vegetation index data and drone image data. The pollination comparison analysis module analyzes the rate of change of satellite vegetation index data based on the same time reference table, and combines the bee activity intensity in the bee entry and exit time rhythm record to generate a pollination response comparison record, and marks the time period in the pollination response comparison record where the bee activity intensity and the rate of change of vegetation index do not match. The flowering period verification module collects drone flowering period image data again within the corresponding time period for the marked time period, and extracts information on flower opening status, flower falling status and plant appearance changes from the drone flowering period image data, and supplements the pollination response comparison record with the extracted flowering period change information. The pollination scheduling module identifies suspected ineffective pollination areas by comparing pollination response records after supplementing information on changes in flowering period, and adjusts the hive movement order, hive replenishment time, and agricultural operation avoidance time based on these suspected ineffective pollination areas.

2. The integrated air-space-ground management system based on multi-source data fusion according to claim 1, characterized in that, The steps for generating records of bee entry and exit time rhythms are as follows: Infrared transmitters and receivers are fixed on both sides of the beehive entrance and exit to form a light curtain covering the channel with infrared beams. Optical collimation is used to keep the transmitted light beams and receiver probes coaxially aligned. At the same time, a temperature and humidity sensor is installed above the beehive and a geolocation device is installed at the bottom of the beehive. After the sensing device is fixed, the infrared transmitter continuously emits invisible light waves. When the bee enters or exits the channel and blocks the light beam, a pulse signal is generated. The receiver detects the signal change and collects the number of times the bee enters or exits, the ambient temperature, and the geographical location information according to a uniform time mark. The continuously collected data is organized in chronological order and supplemented with temperature, humidity and location information to generate a time slice data structure that includes time, environmental and spatial attributes. Based on time slice data, the number of entry and exit events is counted, and a record of the bee entry and exit time rhythm is generated, which includes the number of entry and exit events, ambient temperature, and location coordinates.

3. The integrated air-space-ground management system based on multi-source data fusion according to claim 2, characterized in that, The infrared bee counter's transmission power, detection spacing, and sampling frequency are adjusted according to ambient lighting conditions to ensure that the infrared light curtain maintains a stable response under different climatic conditions. The collected bee entry and exit signals are recorded synchronously with ambient temperature and geographical location data, and time series data covering the day and night cycle are formed through continuous sampling.

4. The integrated air-space-ground management system based on multi-source data fusion according to claim 2, characterized in that, The steps for constructing a reference table for the same time are as follows: Based on the generated records of bee entry and exit time rhythms, the daily activity curves are analyzed to determine the time range in which the bee colony activity intensity is highest and remains stable, and a table of concentrated activity time periods including start time, end time, duration and geographical coordinates is generated. According to the time stamp of the concentrated period of activity, extract the corresponding satellite vegetation index data, obtain the NDVI and EVI values ​​of the pollination area, and align the time stamp of the data with the concentrated period of activity. Based on the peak activity periods, drones were organized to collect field images during the corresponding time periods, and the image timestamps were matched with the satellite data timestamps to form time-aligned results; By unifying and organizing peak bee colony activity periods, satellite vegetation index data, and drone imagery data, a unified time reference table is constructed, with time as the main axis and including behavioral, environmental, spatial, and physiological information.

5. The integrated air-space-ground management system based on multi-source data fusion according to claim 4, characterized in that, When extracting satellite vegetation index data, a geographic buffer is set with the beehive geolocation information as the center to obtain vegetation index raster data covering the pollination area. When constructing the same time reference table, the bee colony activity intensity, NDVI, EVI, ambient temperature and UAV image information are arranged synchronously according to a unified time label.

6. The integrated air-space-ground management system based on multi-source data fusion according to claim 4, characterized in that, The steps for generating pollination response comparison records are as follows: Based on the established reference table for the same time, the satellite vegetation index data are organized in chronological order to obtain a sequence of vegetation index change rates over a continuous time period. After obtaining the vegetation index change rate sequence, the bee entry and exit time rhythm records for the same time period were extracted from the same time reference table, and the bee activity intensity data were read and arranged in a corresponding manner with the environmental temperature and humidity information. After completing the correlation between bee colony activity intensity and vegetation index change rate, pollination response comparison records were generated in chronological order, including time markers, bee colony activity intensity, NDVI change rate, EVI change rate, and geographic coordinates. After generating pollination response comparison records, time periods where the intensity of bee colony activity did not match the rate of change of vegetation index were marked, and the type, duration, temperature, humidity and location information were recorded.

7. The integrated air-space-ground management system based on multi-source data fusion according to claim 6, characterized in that, When marking periods where the intensity of bee colony activity does not match the rate of change of vegetation index, periods in which the intensity of bee colony activity remains high and the rate of change of vegetation index is lower than the average of consecutive periods are marked as abnormal phases of strong activity but weak physiological response, and periods in which the rate of change of vegetation index increases but the intensity of bee colony activity does not increase are marked as phases of insufficient pollination participation.

8. The integrated air-space-ground management system based on multi-source data fusion according to claim 6, characterized in that, The steps for generating pollination response comparison records are as follows: Based on the time periods already marked in the pollination response comparison record, the time window and flight area for the UAV to collect data again are determined, and the time is aligned with the time tags of the pollination response comparison record through a time synchronization device. After the UAV completes image acquisition according to the predetermined flight path, the acquired visible light images and multispectral images are organized according to time and space, and the geographic coordinates of the images are matched with the beehive monitoring points and satellite pixel positions; After completing the processing of drone images of flowering period, information on flower opening status, flower falling status and plant appearance changes is extracted from the images, and these statuses are organized into structured information with text descriptions and numerical indicators. The extracted information on changes in flowering period was added to the corresponding data rows according to the time markers of the pollination response comparison records. Fields for flower opening degree, flower drop ratio, and plant appearance change were added to each time period.

9. The integrated air-space-ground management system based on multi-source data fusion according to claim 8, characterized in that, The process of collecting images of flowering season by drone is further defined as follows: the drone is equipped with a multispectral and high-resolution visible light imaging device, and takes low-altitude pictures along a preset route at a set altitude and speed. During the flight, the drone maintains stable flight through attitude control and automatic route execution equipment.

10. The integrated air-space-ground management system based on multi-source data fusion according to claim 8, characterized in that, Based on the pollination response comparison records after supplementing information on changes in flowering period, suspected areas of ineffective pollination were identified. The following steps were taken to adjust the hive movement order, hive replenishment time, and agricultural operation avoidance time according to these suspected ineffective pollination areas: Based on the pollination response comparison records after supplementing information on flowering period changes, a spatiotemporal dataset containing time, geographical location, bee activity intensity, vegetation index change rate, flower opening status, flower fall status, and plant appearance changes was extracted, and a spatial correspondence between bee colony pollination activity intensity and crop vegetation response was established. After obtaining a candidate set of suspected invalid pollination areas, the time series data are compared and combined with the changing trends of flower opening and falling status to determine stable suspected invalid pollination areas and form a list of areas. After identifying suspected ineffective pollination areas, a plan for hive movement sequence and hive replenishment time is formulated based on regional distribution characteristics and bee colony activity patterns, so that the bee colony coverage radius extends to the ineffective pollination area. After completing the hive movement sequence and bee colony replenishment schedule, adjust the agricultural operation avoidance time based on the time and spatial distribution of suspected ineffective pollination areas, and add it to the pollination response comparison record.

Images