An adaptive wireless communication-based precision agriculture monitoring system

Abstract

The application relates to the technical field of precision agriculture monitoring, and discloses a precision agriculture monitoring system based on adaptive wireless communication. The system comprises a multi-source environment acquisition module, a communication protocol adaptation module and a crop state evaluation module. The multi-source environment acquisition module obtains soil humidity data, illumination intensity data and meteorological parameter data through wireless sensor nodes arranged in farmland areas, and synchronizes the timestamps of the data to ensure the consistency of the data in the time dimension. The communication protocol adaptation module dynamically selects a wireless transmission frequency band according to the synchronous data volume, allocates communication time slots based on data priority, and generates adaptive transmission channel parameters to cope with the communication requirements under the conditions of data volume fluctuation and complex environment. The crop state evaluation module receives the transmission channel parameters, analyzes real-time environment data streams, and divides the current crop growth state into a growth standard state, a growth lag state or a growth fluctuation state. The system provides support for fine management of agricultural production.

Description

Technical Field

[0001] This invention relates to the field of precision agriculture monitoring technology, specifically to a precision agriculture monitoring system based on adaptive wireless communication. Background Technology

[0002] In the process of modernizing agricultural production, accurate monitoring of the farmland environment and crop growth status is a crucial prerequisite for achieving scientific planting and efficient management. Traditional agricultural monitoring methods largely rely on manual inspections, which not only consumes a large amount of manpower and resources but also results in data that is delayed and subjective, making it difficult to meet the needs of large-scale, precision agricultural production. With the development of sensor technology and wireless communication technology, various agricultural monitoring systems have been gradually applied in practice, but existing systems still have many limitations.

[0003] Regarding environmental data acquisition, most systems employ a single type of sensor or fail to effectively synchronize data from multiple sensors. Differences in the collection times of various environmental factors, such as soil moisture, light intensity, and meteorological parameters, lead to deviations in subsequent data fusion and analysis, affecting the accurate assessment of the overall condition of the farmland environment. While some systems deploy multiple sensors, the lack of a unified timestamp synchronization mechanism prevents data from different nodes from being aligned in time, reducing the data's application value.

[0004] Wireless communication is crucial for data transmission in agricultural monitoring systems, but existing systems mostly use fixed communication protocols. When the number of sensor nodes in farmland increases or the amount of environmental data suddenly surges, the fixed transmission frequency bands are prone to channel congestion, leading to data loss or delays. Furthermore, existing systems do not differentiate between data of different priorities; important real-time environmental data is transmitted synchronously alongside non-critical data, further exacerbating the irrational allocation of communication resources and affecting the real-time performance and reliability of monitoring data.

[0005] In crop status assessment, existing systems mostly rely on single environmental factors or simple data thresholds to determine growth status, lacking comprehensive analysis of multi-source synchronous environmental data. This assessment method fails to fully reflect the true state of crop growth and often cannot accurately identify different states such as growth reaching targets, lagging, or fluctuating. This makes it difficult for agricultural producers to formulate targeted management measures, affecting the accuracy of resource input and the improvement of agricultural production efficiency. Summary of the Invention

[0006] The purpose of this invention is to provide a precision agricultural monitoring system based on adaptive wireless communication to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides a precision agriculture monitoring system based on adaptive wireless communication. The system includes: a multi-source environmental acquisition module, configured to acquire soil moisture data, light intensity data, and meteorological parameter data through wireless sensor nodes deployed in farmland areas, and to timestamp and synchronize the soil moisture data, light intensity data, and meteorological parameter data; a communication protocol adaptation module, configured to dynamically select a wireless transmission frequency band based on the amount of synchronized data output by the multi-source environmental acquisition module, allocate communication time slots based on data priority, and generate adaptive transmission channel parameters; and a crop status assessment module, configured to receive the transmission channel parameters output by the communication protocol adaptation module, parse the real-time environmental data stream, and classify the current crop growth status into a growth target state, a growth lag state, or a growth fluctuation state based on the parsing results.

[0008] Preferably, when the communication protocol adaptation module dynamically selects the wireless transmission frequency band, it includes: extracting the data type distribution characteristics and transmission urgency characteristics of the synchronization data volume; calculating the signal attenuation coefficient of the available frequency band using a frequency band occupancy prediction algorithm; mapping the data type distribution characteristics to a low-interference frequency band when the signal attenuation coefficient is lower than a preset attenuation threshold; and activating a multi-band aggregation transmission mechanism when the signal attenuation coefficient is higher than the preset attenuation threshold.

[0009] Preferably, when classifying crop growth states, the crop status assessment module includes: establishing a standard crop growth curve library; extracting soil moisture change gradient, cumulative light duration, and temperature fluctuation range from the real-time environmental data stream; comparing the soil moisture change gradient with the humidity threshold range of the standard crop growth curve library; comparing the cumulative light duration with the light threshold range of the standard crop growth curve library; generating a growth compliance state when all comparison results are within the threshold range; and marking abnormal parameters and activating the growth fluctuation state determination when any comparison result deviates from the threshold range.

[0010] Preferably, the system further includes: an abnormal area location module, configured to, when a growth fluctuation state is detected, divide the farmland area into gridded monitoring units, extract the environmental data deviation of each gridded monitoring unit; merge gridded monitoring units with similar environmental data deviation using a neighboring unit clustering algorithm; and delete gridded monitoring units with environmental data deviation below a preset deviation threshold.

[0011] Preferably, when the abnormal area location module extracts the environmental data deviation, it includes: calculating the soil moisture variation coefficient and light intensity dispersion of each gridded monitoring unit; performing a difference operation between the soil moisture variation coefficient and the historical variation benchmark value to generate a soil deviation factor; performing a difference operation between the light intensity dispersion and the historical dispersion benchmark value to generate a light deviation factor; and superimposing the soil deviation factor and the light deviation factor to generate the environmental data deviation.

[0012] Preferably, the system further includes: an intervention dataset generation module, configured to extract the environmental data deviation of the remaining gridded monitoring units and associate it with the crop growth stage parameters of the corresponding grid; when the environmental data deviation is greater than the tolerance upper limit corresponding to the growth stage parameter, construct a first intervention dataset; when the environmental data deviation is less than the tolerance lower limit corresponding to the growth stage parameter, construct a second intervention dataset.

[0013] Preferably, the system further includes: an intervention strategy association module, configured to sort the environmental data deviations in the first intervention dataset and the second intervention dataset in ascending order of numerical values; verify the transmission path correlation of two adjacent environmental data deviations using a wireless feature matching algorithm; mark the transmission path correlation as a mergeable intervention unit when it exceeds a preset association threshold; count the number of all mergeable intervention units and generate an intervention association topology map.

[0014] Preferably, the system further includes: a comprehensive regulation amount calculation module, configured to extract the deviation of unrelated independent environmental data in the intervention association topology map; calculate the comprehensive regulation coefficient of farmland area based on the number of mergeable intervention units and the distribution density of independent environmental data deviation; and superimpose the comprehensive regulation coefficient with a preset crop sensitivity weight to generate the final irrigation regulation amount.

[0015] Preferably, the system further includes: a growth trend prediction module, configured to infer the environmental parameter correction requirements based on the final irrigation regulation amount; simulate the corrected crop growth trajectory using a time-varying parameter prediction model; and trigger dynamic calibration of prediction model parameters when the simulated growth trajectory deviates from the tolerance threshold of the standard crop growth curve library.

[0016] Preferably, the system further includes: an execution device control module configured to convert the final irrigation regulation amount into the opening duration and flow parameters of the irrigation valve; adjust the irrigation timing allocation strategy according to the dynamic calibration results of the prediction model parameters; and drive the irrigation execution mechanism of the corresponding gridded monitoring unit based on the wireless control instruction set.

[0017] Compared with the prior art, the beneficial effects of the present invention are: the precision agricultural monitoring system based on adaptive wireless communication effectively solves the problems existing in the data acquisition, transmission and status assessment stages of the existing agricultural monitoring system through the synergistic effect of the multi-source environmental acquisition module, the communication protocol adaptation module and the crop status assessment module.

[0018] The multi-source environmental acquisition module comprehensively acquires key environmental data such as soil moisture, light intensity, and meteorological parameters through wireless sensor nodes deployed in farmland areas, and timestamps and synchronizes this data. This design ensures that sensor data from different types and locations maintains consistency over time, eliminating data bias caused by differences in acquisition time and providing a unified and reliable data foundation for subsequent crop growth status analysis. Through timestamping synchronization, various environmental factor data can be correlated and analyzed under the same time reference, more accurately reflecting the dynamic relationship between environmental changes and crop growth.

[0019] The communication protocol adaptation module dynamically selects the wireless transmission frequency band based on the synchronous data volume output by the multi-source environmental acquisition module, avoiding channel congestion issues that easily occur with fixed frequency bands when data volume fluctuates. Simultaneously, communication time slots are allocated based on data priority, ensuring that important environmental data is transmitted first, thus optimizing the allocation of communication resources. The generation of adaptive transmission channel parameters enables the system to flexibly respond to changes in communication needs in complex farmland environments. Whether it's the increase in data volume due to an increase in the number of sensor nodes or changes in channel quality caused by external electromagnetic interference, the system can maintain communication stability by adjusting transmission parameters, ensuring that monitoring data is transmitted to subsequent processing stages in a timely and complete manner.

[0020] The crop status assessment module receives transmission channel parameters from the communication protocol adaptation module, analyzes the real-time environmental data stream, and classifies crop growth status into four categories based on the analysis results: growth reaching target, growth lagging, or growth fluctuation. This comprehensive assessment method based on multi-source synchronous environmental data overcomes the limitations of single-factor judgment and can more comprehensively capture subtle changes in the crop growth process. By accurately classifying growth status, agricultural producers can intuitively understand the current growth status of crops and adjust planting management strategies according to different statuses. For example, they can optimize irrigation and fertilization plans for growth lagging status and strengthen environmental control for growth fluctuation status, making agricultural production management more targeted and scientific. Attached Figure Description

[0021] Figure 1 This is a timing diagram of the precision agriculture monitoring system based on adaptive wireless communication described in this invention.

[0022] Figure 2 Flowchart for selecting wireless transmission frequency bands for communication protocol adaptation modules;

[0023] Figure 3 A flowchart illustrating the growth status classification for the crop status assessment module;

[0024] Figure 4 This is a flowchart illustrating the association of intervention units within the intervention strategy association module. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Please see Figure 1 This invention provides a precision agriculture monitoring system based on adaptive wireless communication, the system comprising:

[0027] The system achieves precise monitoring and control of the farmland environment through multi-source environmental data acquisition, dynamic communication protocol adaptation, and crop growth status assessment. Deployed in farmland areas, the system consists of a network of wireless sensor nodes that transmit data via an adaptive wireless communication protocol. The multi-source environmental acquisition module periodically collects soil moisture, light intensity, and meteorological parameters, ensuring data consistency through timestamp synchronization. The communication protocol adaptation module dynamically adjusts frequency bands and time slot allocations based on data volume and transmission requirements, generating optimal transmission channel parameters. The crop status assessment module analyzes the real-time environmental data stream, compares it with a standard growth curve library, and outputs crop growth status classification results.

[0028] Example 1: See Figure 2 The communication protocol adaptation module implements dynamic wireless transmission frequency band selection through multi-dimensional data analysis and real-time decision-making mechanisms. This module continuously receives synchronous data streams from multi-source environmental acquisition modules, and these data are time-aligned using nanosecond-level timestamps. During the initialization phase, the module parses the metadata characteristics of the data stream: the data type distribution characteristics show high sampling density for soil moisture data, with a sampling interval typically set at 5 minutes; light intensity data exhibits periodic fluctuations, with a fixed update frequency of one minute; meteorological parameter data presents sudden transmission needs, triggering immediate reporting when wind speed changes or rainfall occurs. The calculation of transmission urgency characteristics relies on a crop growth stage weight table pre-installed in the embedded processor. This weight table is dynamically updated based on agronomic expert knowledge. For example, during the crop flowering stage, the priority coefficient of soil moisture data is automatically increased to the highest level, while the priority of light data during the normal growth period remains at a medium level.

[0029] The frequency band occupancy prediction algorithm is based on a distributed spectrum sensing network. Each wireless sensor node is equipped with a dual-channel RF unit; the main channel is used for service data transmission, and the secondary channel performs spectrum scanning. The algorithm periodically collects historical channel state information for each frequency band, including received signal strength indication, signal-to-noise ratio, and bit error rate records. By analyzing the channel attenuation pattern over the past 30 minutes, the signal attenuation coefficient of the available frequency band is calculated. This coefficient comprehensively reflects the superposition effect of multipath fading and co-channel interference. The preset attenuation threshold is set in layers according to the terrain characteristics of farmland: the threshold is set to -85dBm in flat areas and adjusted to -80dBm in hilly areas to compensate for terrain loss.

[0030] When the signal attenuation coefficient falls below a preset threshold, the module activates an intelligent frequency band mapping mechanism. Soil moisture data, due to its sensitivity to transmission interruptions, is allocated to the 868MHz low-frequency band for transmission, which possesses strong diffraction capabilities and low penetration loss. Illumination intensity data is mapped to the 2.4GHz band, utilizing its high bandwidth to meet the large data volume transmission requirements of the image sensor. Meteorological parameters are prioritized based on urgency: routine temperature and humidity data uses the 433MHz band, while sudden heavy rain warnings occupy high-priority time slots in the 2.4GHz band. The frequency band mapping table is stored in the node's flash memory using a hash algorithm, and the mapping relationship is updated each time a switch occurs.

[0031] When the signal attenuation coefficient exceeds a threshold, the multi-band aggregation transmission mechanism is immediately activated. The data splitter divides the original data packets into fixed-size data blocks, each with an error correction code and sequence number marker. The main controller coordinates the parallel transmission of LoRa and Wi-Fi dual channels: critical control commands and basic soil data are transmitted via LoRa's long-range mode, while high-resolution illumination data and weather radar information are transmitted via the Wi-Fi channel. The receiving end deploys a data reassembly engine, which reorders the data stream based on the timestamp and sequence number of the data blocks using a sliding window buffering technique. For data blocks with a transmission delay exceeding 200 milliseconds, the reassembly engine initiates a redundancy data request mechanism, requesting retransmission of a specific sequence number from the sending node. A dual verification mechanism is implemented during transmission: cyclic redundancy check ensures data integrity, and parity check ensures the correct order of data blocks.

[0032] The channel handover decision cycle is set to 15 minutes, but event-triggered instant handover is supported. When a node detects three consecutive packet transmission failures or a sudden drop in signal quality index of 30%, the current transmission session is forcibly interrupted. The session recovery mechanism first attempts to reconnect within the same frequency band; if a stable connection is not established within 5 seconds, cross-band handover is initiated. All frequency band handover actions are recorded in a local circular buffer on the node for optimizing subsequent handover decision models. The generation of transmission channel parameters includes seven dimensions: center frequency value, bandwidth configuration, modulation scheme selection, transmit power level, forward error correction level, time slot allocation scheme, and maximum number of retransmissions. These parameters are encapsulated into binary control frames and distributed to each sensor node via the physical layer broadcast channel.

[0033] The node response mechanism employs a hierarchical confirmation strategy: Level 1 confirmation verifies basic parameters at the MAC layer, while Level 2 confirmation verifies parameter compatibility with sensors at the application layer. For unsupported parameter combinations, the node returns an error code and maintains the original parameter operation. The parameter update process adheres to atomic operation principles, ensuring full network parameter synchronization is completed within a single communication cycle. To address sudden interference from agricultural metal machinery, the module reserves an emergency communication frequency band, which is only activated when the regular frequency band is completely blocked, triggering a full network encryption mode switch upon activation. The entire dynamic frequency band selection process forms a closed-loop control system: after each transmission, actual throughput, packet loss rate, and latency data are collected to revise the attenuation coefficient prediction model for the next transmission.

[0034] Example 2: See Figure 3 The crop status assessment module performs growth status classification based on spatiotemporal correlation analysis of multidimensional environmental parameters. This module receives real-time environmental data streams from the communication protocol adaptation module. The data stream encapsulation format includes timestamps, sensor location codes, and three environmental parameter values. The standard crop growth curve library adopts a hierarchical storage structure: the top layer is categorized by crop type, the middle layer is divided into growth stages, and the bottom layer stores the environmental parameter threshold ranges for each stage. Taking winter wheat at the jointing stage as an example, the humidity threshold range is set to 20%-25%, and the light threshold range is 8000-12000 lux / hour / day. The data parsing engine first verifies the validity of the input data stream, removing outliers exceeding the physical range, and then initiates the parameter feature extraction process.

[0035] The soil moisture change gradient was calculated using a dual-window sliding algorithm: a small window with a width of 10 minutes calculated the rate of change in humidity per minute; a large window with a width of 60 minutes calculated the moving average of the rate of change. Accumulated sunshine duration was collected using raw data from a quantum sensor. After Kalman filtering to eliminate cloud cover interference, the integrator accumulated the effective sunshine duration in segments based on sunrise and sunset times. A time-weighted mechanism was introduced to calculate the temperature fluctuation range, with the weight of daytime high-temperature data set at 1.5 times that of nighttime data. Feature extraction results were stored in a circular buffer with a depth set to the data from the most recent 6 hours.

[0036] The growth status determination unit executes a three-level comparison strategy: The first level matches the soil moisture change gradient with the moisture threshold range in the curve library frame by frame; an alert is triggered when five consecutive frames exceed the upper limit. The second level verifies whether the cumulative light duration reaches 90% of the daily requirement; if not, a photosynthetic deficiency flag is displayed. The third level checks whether temperature fluctuations exceed the variety's tolerance range. The determination logic is implemented using a state machine: the initial state is the growth standard compliance state; when a single parameter is abnormal for more than 30 minutes, it switches to the growth fluctuation state; if two or more parameters are abnormal simultaneously, it directly enters the growth lag state. The state transition signal includes an abnormal parameter type code, and a 4-byte status message is generated by the state encoder.

[0037] The abnormal area location module initiates gridding processing upon responding to growth fluctuation signals. The farmland area is divided into 10-meter-sided square grids based on latitude and longitude coordinates, with each grid assigned a unique ID code. The grid index table is stored in a spatial database, recording the coordinates of the grid center point and a list of included sensor nodes. The environmental data deviation calculator generates a deviation vector for each grid, containing three dimensions: soil moisture deviation index, light intensity deviation index, and temperature deviation index. The calculation process employs a rolling time window mechanism, using the current moment as a baseline and taking a 3-hour data window backward, dynamically comparing it with historical data windows from the same period.

[0038] Soil moisture deviation index is obtained using the coefficient of variation method: first, the coefficient of variation of all humidity sensor data within the grid is calculated, and then compared with the moving average of the coefficients of variation for the same period over the past 7 days to obtain the relative deviation rate. Light intensity deviation index uses dispersion analysis, calculating the interquartile range of light data within the grid and comparing it with historical discrete baseline values ​​to generate the deviation. Temperature deviation index focuses on the fluctuation range, calculating the difference between the standard deviation of temperature data within the grid and the historical standard deviation. After normalization, the deviation indices of each dimension are weighted and synthesized into a comprehensive environmental data deviation, with the weighting coefficients dynamically adjusted according to crop water requirement patterns.

[0039] The nearest neighbor clustering algorithm constructs a spatial relationship graph using a grid as the basic unit. During algorithm initialization, each grid is treated as an independent cluster, and the similarity of environmental features between adjacent grids is calculated. A modified Euclidean distance formula is used to measure similarity, comprehensively comparing the soil moisture gradient morphology, cumulative light curve, and temperature fluctuation pattern of two grids. When the similarity between grids exceeds 85% and the deviation difference is less than 5%, a cluster merging operation is performed. During the merging process, the cluster center position is updated to the weighted centroid of the member grids, with the weights determined by the deviation of each grid. The merging process is iteratively executed until no adjacent clusters satisfying the conditions exist, ultimately forming several continuous anomalous regions. The region boundary optimization module performs contour smoothing on the merged anomalous regions, using the Douglas-Puk algorithm to simplify boundary polylines and reduce jagged edges.

[0040] The data filtering mechanism is activated after clustering: it reads the comprehensive deviation value of each grid cluster and compares it with a preset deviation threshold. The deviation threshold is set according to the key growth stages of the crop; for example, the threshold is set to 0.15 for the heading stage and 0.12 for the grain-filling stage. Grid clusters below the threshold are marked as low-risk units and removed from the list of abnormal areas. The remaining grid cluster information is stored in a spatial database, recording boundary coordinates, area data, average deviation, and the types of major anomalies. The database updates the grid status view every 5 minutes, providing spatial data support for subsequent decision-making. The entire processing flow adopts a pipeline architecture, with each stage transmitting processing results through a message queue to achieve dynamic and balanced distribution of computational load.

[0041] Example 3: The environmental data deviation calculation of the anomaly area location module adopts a multi-dimensional spatiotemporal analysis method. The basic data structure of the gridded monitoring unit includes unit number, geographic coordinate range, sensor node list, and historical data cache. The calculation process of soil moisture variation coefficient first performs outlier detection on the current readings of all moisture sensors within the unit, removes data points exceeding three times the standard deviation, and calculates the ratio of the standard deviation to the mean of the remaining data. The calculation of light intensity dispersion adopts an improved interquartile range algorithm. After sorting the light sensor data within the unit by size, the difference between the 75th percentile and the 25th percentile is taken as the dispersion measure. The historical variation baseline value is obtained by querying the unit's dedicated historical database, which stores the statistical characteristic values ​​of the past 30 days by hour. The calculation uses the exponentially weighted moving average of the data for the same period of the most recent 7 days.

[0042] The deviation factor is generated using a differential accumulation method. The soil deviation factor is calculated by adding a time decay coefficient after determining the absolute difference between the current coefficient of variation and the historical baseline. The time decay coefficient is dynamically adjusted based on data freshness, with the weight of data from the most recent hour set at 0.6, and the weight of data from the last 1-3 hours linearly decreasing to 0.3. The calculation of the light deviation factor incorporates a day / night segmentation mechanism. During the daytime period (2 hours after sunrise to 2 hours before sunset), the difference between the current dispersion and the historical baseline is directly calculated. During the nighttime period (1 hour after sunset to 1 hour before sunrise), the cumulative light deficiency is used as a substitute indicator. The formula for synthesizing environmental data deviation is: in: Indicates the deviation of environmental data. Soil deviation factor, This is the illumination deviation factor. This is the temperature deviation factor. , , These are the dynamic weight coefficients for each factor, obtained from a preset matrix based on crop type and growth stage. For leafy vegetables in their vegetative growth stage, the typical weights are set to... , , The reproductive growth period for fruit-bearing crops has been adjusted to... , , Temperature deviation factor The calculation uses the sliding window method to compare the current 3-hour temperature fluctuation range with the historical values ​​for the same period.

[0043] The workflow of the intervention dataset generation module begins with scanning the deviation matrix. The module periodically iterates through the latest deviation values ​​of all grid cells, while simultaneously retrieving the corresponding growth stage code from the crop growth stage parameter library. The growth stage parameter library employs a hierarchical storage structure: the top layer contains the field number, the middle layer contains the crop variety, and the bottom layer contains the growth stage records. Each growth stage record contains six fields: stage code, start date, estimated end date, upper limit of humidity tolerance, lower limit of light tolerance, and temperature sensitivity coefficient. Tolerance parameters are expressed as percentages; for example, the upper limit of humidity tolerance during the flowering period might be set to +15%, indicating that a fluctuation exceeding the standard threshold by 15% is allowed.

[0044] The dataset construction logic implements a tiered triggering mechanism. When a certain grid... The generation process of the first intervention dataset is initiated when the value exceeds the tolerance limit for its growth stage. The dataset record structure contains seven fields: event number, grid coordinates, deviation value, main anomaly parameters, occurrence timestamp, crop stage code, and urgency score. The urgency score is calculated using a three-layer neural network, with the input layer receiving... The values, duration of the anomaly, and crop sensitivity coefficients are used to generate a score of 1-5 levels in the output layer. The triggering condition for the second intervention dataset is... If the value is below the tolerance limit for more than two consecutive testing cycles, its data structure additionally includes a water shortage prediction field, which is calculated by back-calculation using a soil moisture transport model.

[0045] Data storage employs a hybrid architecture combining time-series and relational databases. Real-time generated intervention records are first written to the time-series database, stored in time-partitioned locations for quick retrieval of recent events. A data migration operation is performed daily at dawn, transferring the previous day's complete records to the relational database and generating statistical summary information. The database indexing strategy implements multi-level optimization: the primary index is built on grid numbers, while secondary indexes include time range indexes, crop type indexes, and deviation value indexes. The query interface supports composite condition searches; for example, it can filter all grid cells in a specific field with a light deviation exceeding 10% in the past 24 hours.

[0046] The data association engine maintains the mapping relationship between grid cells and physical devices in real time. When each intervention record is generated, it automatically associates the irrigation valve number, fertilization equipment address, and environmental control device ID within that grid. This association information is stored in a distributed key-value database, using a consistent hashing algorithm for rapid location. When a new intervention record is inserted, a status check process for the associated devices is triggered to verify whether the target device is online and operational. Offline devices automatically trigger alarm events, while online devices preload control parameter templates to prepare for subsequent execution modules.

[0047] The historical data analysis module periodically performs pattern mining tasks. Weekly cluster analysis is performed on accumulated intervention data to identify the spatial distribution patterns of high-frequency anomaly areas. The analysis results generate heatmaps overlaid on the farmland electronic map, using color intensity to represent the frequency of anomalies. Long-term trend analysis employs time series decomposition, separating the deviation series of each grid into seasonal, trend, and residual components to predict potential future anomaly patterns. Analysis reports are automatically generated and pushed to the agricultural management platform, supporting decision-makers in adjusting planting plans or improving soil conditions.

[0048] The deviation calibration mechanism plays a continuous role in system operation. Historical baseline values ​​are reassessed quarterly, removing outdated data due to climate change. The calibration process uses a sliding window method, gradually eliminating data older than 180 days while incorporating new data within the normal fluctuation range. Dynamic weighting coefficients. , , The system is optimized and adjusted every two weeks based on an analysis of the contribution of various environmental parameters to recent crop growth. The entire deviation calculation system forms an adaptive closed loop, which can dynamically evolve in response to changes in the farmland ecological environment.

[0049] The dynamic adjustment function of the grid cells addresses changes in farmland layout. When fields are replanned or the sensor network is expanded, system administrators can modify the grid division scheme through configuration tools. The adjustment process maintains data continuity, with historical data from the original grids spatially interpolated and allocated according to the new grid boundaries. Grid merging and splitting operations are recorded in the version control system, supporting the ability to trace back to the grid state at any point in time. This flexibility allows the system to adapt to the monitoring needs of farms of different sizes, effectively covering everything from small experimental fields to large commercial planting areas.

[0050] Example 4: See Figure 4 The operation of the intervention strategy association module is based on a dual association criterion of spatial proximity and transmission characteristics. This module receives structured data from the intervention dataset generation module. Taking winter wheat field number F3-12 as an example, its first intervention dataset contains 17 anomalous grid cells, and its second intervention dataset contains 5 grid cells. During the data preprocessing stage, the deviation values ​​are standardized to eliminate dimensional differences between different parameters. The standardized deviations are then sorted in ascending order and stored in a priority queue. Each element in the queue records the grid number, deviation type (high / low), standardized value, and timestamp information.

[0051] The core of the wireless feature matching algorithm is to compare the transmission path characteristics of adjacent grid cells. The system maintains a transmission feature registration table, recording the physical layer characteristics of the most recent 10 data transmissions for each grid cell. Taking grids G35 and G36 as examples, their transmission feature comparison includes four dimensions: average received signal strength, data packet arrival time interval distribution, bit error rate variation pattern, and channel switching frequency. Similarity calculation uses the cosine similarity method of the angle between feature vectors. When the overall similarity exceeds a preset association threshold, a merging suggestion is generated. The association threshold is dynamically adjusted according to the wireless environment of farmland, set to 0.75 in densely populated areas and 0.65 in sparse areas.

[0052] The process for determining mergeable intervention units involves multi-level verification. Primary verification checks the spatial adjacency between grids, allowing only edge-adjacent or corner-adjacent grids to enter the merging process. Secondary verification analyzes the similarity of environmental deviation patterns, requiring that the dominant anomaly parameter types of the grids to be merged be consistent. Tertiary verification confirms the coordinability of equipment control paths, checking whether the irrigation valves of the target grids belong to the same control subnet. Grid groups that pass verification are marked as mergeable intervention units and assigned a unique merging group number.

[0053] The intervention-related topology graph is constructed using a graph data structure. Vertices represent merged intervention units, and edges represent control dependencies between units. The adjacency matrix is ​​filled according to the following rule: if two units need to execute control commands sequentially to avoid pipeline pressure fluctuations, the corresponding matrix element is set to 1; otherwise, it is set to 0. For example, the adjacency matrix representation of the local intervention-related topology of field F3-12 is shown in Table 1.

[0054] Merging Units CU-01 CU-02 CU-03 CU-04 CU-01 0 1 0 0 CU-02 0 0 1 0 CU-03 0 0 0 1 CU-04 0 0 0 0

[0055] The matrix shows that the execution of CU-01 takes precedence over CU-02, CU-03 depends on the network status after CU-02 is completed, and CU-04 can be executed independently. The topology sorting algorithm translates this dependency relationship into a specific execution sequence.

[0056] The process of extracting the deviation of independent environmental data involves scanning all vertices of the intervention-related topology graph. Vertices not connected to other vertices are identified as independent deviation units, which are typically distributed at field edges or in sensor-sparse areas. A density clustering algorithm analyzes the distribution characteristics of independent units, converting the number of units per square kilometer into density levels. The density levels are divided into five categories: extremely sparse (<3 units / km²), sparse (3-5), moderate (5-8), dense (8-12), and extremely dense (>12).

[0057] The calculation of the comprehensive regulation coefficient for farmland areas integrates multiple influencing factors. The base coefficient is derived from density level mapping values, with an assigned value of 0.2 for extremely sparse areas and 0.8 for extremely dense areas. The correction factor considers the duration of abnormalities, increasing the weight of abnormal units lasting more than 4 hours by 0.1. The meteorological compensation factor is adjusted based on the 24-hour weather forecast; when rainfall is forecast, the coefficient for units with high humidity deviations is reduced by 0.15. The final comprehensive regulation coefficient is obtained through a weighted average, retaining two decimal places of precision.

[0058] The crop sensitivity weight matrix is ​​pre-configured according to variety and growth stage. Taking spring maize as an example, its sensitivity weight changes as follows during key growth stages: seedling stage 0.3, jointing stage 0.6, tasseling stage 0.9, and grain-filling stage 0.7. The weight matrix is ​​stored in an embedded database, and each field is bound to a corresponding crop variety growth model. When the system detects changes in crop phenological stages, it automatically switches the weight configuration version.

[0059] The final irrigation control quantity is generated using a step-by-step calculation strategy. The first step multiplies the comprehensive control coefficient by the sensitivity weight to obtain the basic control intensity. The second step overlays the equipment calibration factor, taking into account the flow characteristic curves of the irrigation valves. The third step introduces a manual correction term, allowing agronomists to fine-tune the quantity via mobile terminals. The calculation results are converted into a standard irrigation instruction format, containing four fields: target grid group, water volume (cubic meters), duration (minutes), and execution priority (levels 1-5).

[0060] The instruction distribution system employs a reliable transmission protocol to ensure control reliability. Each irrigation instruction generates a unique transaction ID and is published to the corresponding device topic via the MQTT protocol. Upon receiving the instruction, the executing device returns an acknowledgment frame containing the transaction ID. If no acknowledgment is received within 10 seconds, the instruction enters the retransmission queue, with a maximum of three attempts. Successfully executed instructions are recorded in the operation log, with log entries including execution time, actual water volume, device status code, and other audit information.

[0061] A dynamic adjustment mechanism continuously monitors environmental feedback during irrigation. Soil moisture sensors activate high-frequency monitoring mode upon command execution, reporting data every minute. A deviation recalculation module compares actual moisture changes with the expected curve; when the difference exceeds the tolerance limit, compensatory irrigation is triggered. Compensation commands are prioritized to the highest level and marked as emergency interventions. All compensation operations are recorded in a separate log partition for subsequent analysis of system response performance.

[0062] Statistical analysis of historical intervention records supports long-term optimization. Weekly intervention effectiveness reports are generated, calculating the response success rate for various anomaly patterns. Key indicators in the reports include average response delay, water volume regulation accuracy, and compensation trigger frequency. These indicators help identify system weaknesses; for example, areas requiring frequent compensation irrigation may indicate leaks in the underground pipe network. Maintenance personnel adjust sensor deployment density or repair irrigation equipment based on the report results, forming a closed-loop management system for continuous improvement.

[0063] An anomaly association knowledge base automatically accumulates typical cases. Each successfully handled intervention case has its feature vector extracted and stored in the knowledge base. This vector includes the anomaly pattern, environmental context, execution strategy, and effect evaluation. The similarity search function supports rapid matching of historical cases, providing a reference for handling newly emerging anomalies. The knowledge base uses an incremental update mechanism; new cases are converted to official cases after a three-month verification period, while invalid cases are archived in the historical version repository.

[0064] The multi-terminal collaboration interface supports mobile device access. Agronomists use tablets with a dedicated application installed to receive real-time intervention alerts and topology map visualizations. The application provides a gesture-based interface, allowing direct adjustment of merged unit ranges or modification of execution order on the electronic map. All modifications generate discrepancy records, which are synchronized to a cloud database for version control. Conflicts between mobile terminals and fixed devices are resolved through a distributed locking mechanism, ensuring the consistency of control commands.

[0065] The system health monitoring module ensures stable operation. Core components are equipped with heartbeat sensors that report operational status every minute. A communication link quality dashboard displays the signal-to-noise ratio and packet loss rate for each frequency band in real time. When the controller's CPU load consistently exceeds 80% or signs of memory leaks are detected, a resource reclamation procedure is automatically triggered. Critical data channels implement a dual-backup strategy; in the event of a primary channel failure, switching to the backup channel within 200 milliseconds ensures the continuity of irrigation control.

[0066] Example 5: The operation of the growth trend prediction module begins with the reverse analysis of the final irrigation regulation amount. This module receives standardized regulation instructions from the integrated regulation amount calculation module, which include the identifier code of the target grid area and the regulation amount value. The analysis engine first decomposes the regulation amount into environmental parameter correction requirements, establishing a mapping relationship between water input and soil moisture response. The soil moisture increment calculation uses a water transport model, considering soil type, root depth, and groundwater level factors. The light compensation requirement is converted into the supplemental lighting duration based on the shading adjustment coefficient in the regulation amount. The correction requirement parameters are encapsulated as a prediction model input vector, with the vector dimensions including the target humidity value, light compensation amount, and temperature adjustment suggestion value.

[0067] The time-varying parameter prediction model is constructed using a recurrent neural network architecture. The model's input layer receives the current environmental parameter vector, the hidden layer contains long short-term memory units to capture time-dependent features, and the output layer generates a prediction of the crop growth trajectory for the next 72 hours. Key indicators of the growth trajectory include daily plant height increase, leaf area index change rate, and biomass accumulation. Model parameters include photosynthetic efficiency coefficient, water use efficiency coefficient, and nutrient conversion rate; these parameters are initially derived from a crop physiology database. The simulation is executed in hourly time steps, updating the impact of environmental parameters on morphological indicators at each step, and the growth trajectory points are calculated cumulatively.

[0068] The model calibration mechanism is dynamically activated during trajectory prediction. A standard crop growth curve library provides tolerance threshold bands for each growth stage, with each threshold band set to a range of plus or minus two standard deviations. The monitor compares the relative positions of simulated trajectory points to the threshold bands in real time, triggering the calibration process when three consecutive trajectory points exceed the threshold band. The calibration decision unit analyzes the deviation direction: if the trajectory remains above the threshold band, the water use efficiency coefficient is determined to be too high; if it remains below the threshold band, the photosynthetic efficiency coefficient may be underestimated. Parameter adjustment uses the gradient descent method, calculating the optimal correction magnitude based on historical actual growth data. Typical calibration operations include lowering the water use efficiency coefficient by 5% or increasing the photosynthetic efficiency coefficient by 3%, and then rerunning the prediction process.

[0069] The instruction conversion process of the execution equipment control module implements multi-level mapping. Irrigation control quantities are first decomposed into valve action parameters, and the basic conversion uses linear interpolation: mapping the control quantity per square meter to the valve opening duration, with an accuracy controlled within 0.1 seconds. Flow parameters are obtained by querying a valve characteristic curve table, which stores the flow coefficients under different pressures. The time-series allocation strategy receives calibration feedback from the prediction module and dynamically adjusts the execution order of different grids. When model calibration indicates a risk of growth lag in a certain area, the irrigation priority of that area is automatically increased by two levels. The allocation algorithm uses a scheduling strategy with time windows to ensure that high-priority areas are irrigated within the target time period.

[0070] The wireless control command set is designed with consideration for the electromagnetic environment characteristics of farmland. The command frame structure includes a frame header, grid group code, action type code, parameter packet, and checksum. The frame header identifies the command version and encryption flag, and the grid group code corresponds to the merged unit number in the intervention-related topology diagram. The action type code defines six basic operations: start irrigation, stop irrigation, adjust flow rate, start supplemental lighting, stop supplemental lighting, and calibrate sensors. The parameter packet uses a compact binary format; for example, irrigation duration is represented by two bytes in 0.1 seconds, and flow rate is represented by one byte in liters per minute. The checksum is generated using a cyclic redundancy check algorithm to ensure transmission integrity.

[0071] The transmission protocol implements an adaptive optimization mechanism. The basic transmission interval is set to 300 milliseconds, dynamically adjusted according to channel quality: when the receiver returns three consecutive successful acknowledgments, the interval is shortened to 200 milliseconds; if data packet loss is detected, the interval is extended to 500 milliseconds. The encryption module employs a hybrid encryption scheme: the command frame header uses elliptic curve cryptography to protect control commands, while the parameter packet uses symmetric encryption to ensure real-time performance. The key update cycle is set to 24 hours, with new keys distributed daily at midnight through a secure channel.

[0072] The device driver layer implements hardware abstraction. The irrigation actuator controller is equipped with a unified driver interface, which defines eight standard control primitives. The driver adapter converts wireless commands into device-specific control codes, supporting protocol conversion for mainstream irrigation valve models. The execution status feedback mechanism requires the device to return operation results after completion, including actual execution time, flow monitoring values, and fault codes. Feedback data is bound and stored with the original commands, forming a closed-loop control record.

[0073] The predictive model update system regularly optimizes core parameters. Monthly, it compares accumulated actual growth data with predicted data to calculate the model's prediction bias rate. The bias analysis module identifies systematic prediction errors and automatically initiates model parameter retraining. The retraining dataset includes environmental monitoring records and measured crop growth data from the past three months, and the training process is executed offline on edge computing nodes. New model parameters are deployed through a secure channel, undergoing two weeks of parallel validation testing before deployment.

[0074] Meteorological data fusion mechanisms enhance forecast reliability. Daily access to meteorological bureau forecast data triggers a scenario mode switch in the forecasting model when a heavy rainfall or high-temperature warning is forecast. The scenario mode adjusts the sensitivity of water use efficiency parameters; for example, under a high-temperature warning mode, the evaporation loss coefficient is automatically increased by 20%. Special weather events trigger real-time corrections to the forecast trajectory, with the correction magnitude determined based on the impact analysis of similar historical weather events.

[0075] Mobile terminals provide access for manual intervention. A tablet application used by agronomists displays a comparison chart of the predicted growth trajectory and the standard curve, and provides an interface for fine-tuning parameters. Adjustments include manually correcting environmental parameters, modifying model weights, or directly setting irrigation amounts. All manual interventions are logged in detail, noting the operator's identity and the reason for the modification. The system automatically analyzes the frequency of manual interventions, marking areas with frequent adjustments as high-uncertainty units.

[0076] The equipment health monitoring subsystem ensures reliable operation. Irrigation valves are equipped with current sensors to monitor motor operating status. Abnormal current patterns trigger preventative maintenance alarms; for example, persistently high starting current may indicate mechanical jamming. Supplemental lighting units are equipped with light intensity detection probes; a calibration process is triggered when the actual light output differs from the commanded light requirement by more than 15%. The sensor network implements a self-diagnostic protocol, with nodes reporting hardware status codes every 6 hours.

[0077] The data archiving system builds a long-term knowledge base. Complete forecast records, execution instructions, and environmental feedback are packaged and stored daily. Archived data is organized by crop growth cycle and supports searching historical growth patterns by variety. The knowledge discovery engine analyzes data correlations across multiple quarters to identify the spatiotemporal characteristics of optimal irrigation patterns under different soil conditions. These characteristics are used to initialize the forecasting model parameters for new planting areas, accelerating the system's adaptation process.

[0078] The cross-system integration interface supports connection to farm management platforms. Forecast results are output in a standard agricultural data exchange format, including a three-day growth trend curve and control recommendations. The execution command system is synchronized with the farm resource planning software, and irrigation operations are automatically included in the production cost module. The interface uses asynchronous communication via message queues, and flow control is implemented to prevent system overload.

[0079] An emergency fault response mechanism handles abnormal operating conditions. If the predictive model fails to calibrate more than three times consecutively, it switches to a conservative control mode. In this mode, irrigation volume is based on historical averages, and supplemental lighting duration follows a fixed schedule. Simultaneously, an expert alarm is triggered, requesting manual review of the model parameter configuration. In the event of a communication interruption, a fallback mechanism utilizes local cache control, executing the device to repeat operations according to the last valid instruction, maintaining basic irrigation functionality.

[0080] The visual monitoring interface integrates information from the entire process. An electronic map displays real-time environmental data, predicted growth status, execution equipment locations, and historical anomaly records in layers. A timeline view compares the predicted trajectory with the actual growth curve, using color-coded markers to indicate calibration event points. The equipment status panel centrally displays valve opening, supplementary lighting intensity, and communication quality indicators, supporting rapid location of malfunctioning equipment. The interface data refresh rate dynamically adjusts with system load to ensure the real-time performance of core functions.

[0081] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0082] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An adaptive wireless communication based precision agriculture monitoring system, characterized in that, The application relates to a crop growth state evaluation system based on multi-source environmental data, which comprises the following modules. A multi-source environmental acquisition module is configured to acquire soil humidity data, illumination intensity data and meteorological parameter data through wireless sensor nodes arranged in farmland areas, and to synchronize the soil humidity data, illumination intensity data and meteorological parameter data in time; A communication protocol adaptation module is configured to dynamically select a wireless transmission frequency band according to the synchronized data amount output by the multi-source environmental acquisition module, to allocate a communication time slot based on a data priority, and to generate adaptive transmission channel parameters; A crop state evaluation module is configured to receive the transmission channel parameters output by the communication protocol adaptation module, to analyze real-time environmental data streams, and to divide the current crop growth state into a growth standard state, a growth lag state or a growth fluctuation state according to the analysis result; When the communication protocol adaptation module dynamically selects a wireless transmission frequency band, the following steps are included. Data type distribution characteristics and transmission urgency characteristics of the synchronized data amount are extracted; A frequency band occupation rate prediction algorithm is used to calculate a signal attenuation coefficient of the available frequency band; When the signal attenuation coefficient is lower than a preset attenuation threshold, the data type distribution characteristics are mapped to a low-interference frequency band; When the signal attenuation coefficient is higher than the preset attenuation threshold, a multi-frequency band aggregation transmission mechanism is activated; The activated multi-frequency band aggregation transmission mechanism includes that a main controller coordinates parallel transmission of LoRa and Wi-Fi double channels: key control instructions and soil basic data are sent through a long-distance mode of LoRa, and high-resolution illumination data and meteorological radar information are transmitted through a Wi-Fi channel.

2. The precision agriculture monitoring system of claim 1, wherein, When the crop state evaluation module divides the crop growth state, the following steps are included. A standard crop growth curve library is established, and soil humidity variation gradients, illumination cumulative time lengths and temperature fluctuation ranges in real-time environmental data streams are extracted; The soil humidity variation gradients are compared with humidity threshold interval of the standard crop growth curve library; The illumination cumulative time lengths are compared with illumination threshold interval of the standard crop growth curve library; When all comparison results are within the threshold interval, a growth standard state is generated; When any comparison result deviates from the threshold interval, an abnormal parameter is marked and a growth fluctuation state is activated.

3. The precision agriculture monitoring system of claim 2, wherein, Further comprising: An abnormal area positioning module is configured to divide the farmland area into grid monitoring units when the growth fluctuation state is identified, and to extract environmental data deviation degrees of each grid monitoring unit; A neighboring unit clustering algorithm is used to combine grid monitoring units with similar environmental data deviation degrees; Grid monitoring units with environmental data deviation degrees lower than a preset deviation threshold are deleted.

4. The precision agriculture monitoring system of claim 3, wherein, When the abnormal area positioning module extracts the environmental data deviation degrees, the following steps are included. Soil humidity variation coefficients and illumination intensity dispersions of each grid monitoring unit are calculated; Difference operation is performed on the soil humidity variation coefficients and historical variation reference values to generate soil deviation factors; Difference operation is performed on the illumination intensity dispersions and historical dispersion reference values to generate illumination deviation factors; The soil deviation factors and the illumination deviation factors are superimposed to generate environmental data deviation degrees.

5. The precision agriculture monitoring system of claim 4, wherein, Further comprising: An intervention-required data set generation module is configured to extract environmental data deviation degrees of the remaining grid monitoring units and to associate crop growth stage parameters of corresponding grids. When the environmental data deviation degree is greater than the tolerance upper limit corresponding to the growth stage parameter, a first intervention data set is constructed; When the environmental data deviation degree is less than the tolerance lower limit corresponding to the growth stage parameter, a second intervention data set is constructed.

6. The precision agriculture monitoring system of claim 5, wherein, Further comprising: An intervention strategy association module configured to arrange the environmental data deviation degrees in the first intervention data set and the second intervention data set in ascending order of numerical value; A wireless feature matching algorithm is used to verify the transmission path correlation of the adjacent two environmental data deviation degrees; When the transmission path correlation exceeds a preset association threshold, it is marked as a mergable intervention unit; The number of all mergable intervention units is counted and an intervention association topology graph is generated.

7. The precision agriculture monitoring system of claim 6, wherein, Further comprising: A comprehensive regulation amount calculation module configured to extract independent environmental data deviation degrees that are not associated in the intervention association topology graph; According to the number of mergable intervention units and the distribution density of independent environmental data deviation degrees, a farmland area comprehensive regulation coefficient is calculated; The final irrigation regulation amount is generated by superimposing the comprehensive regulation coefficient and a preset crop sensitivity weight.

8. The precision agriculture monitoring system of claim 7, wherein, Further comprising: A growth trend prediction module configured to back-calculate environmental parameter correction requirements based on the final irrigation regulation amount; A time-varying parameter prediction model is used to simulate the corrected crop growth trajectory; When the simulated growth trajectory deviates from the tolerance threshold of the standard crop growth curve library, the prediction model parameter dynamic calibration is triggered.

9. The precision agriculture monitoring system of claim 8, wherein, Further comprising: An execution device control module configured to convert the final irrigation regulation amount into the opening duration and flow parameter of the irrigation valve; According to the prediction model parameter dynamic calibration result, the irrigation timing distribution strategy is adjusted; Based on the wireless control instruction set, the irrigation execution mechanism of the corresponding grid monitoring unit is driven.

Images