Plant growth cycle monitoring method and system based on edge calculation

Abstract

The invention provides a plant growth cycle monitoring method and system based on edge calculation. The method belongs to the technical field of intelligent agriculture and edge intelligence crossing. The method comprises the following steps: carrying out full-period environment parameter recording area division on a controllable environment planting area to generate environment recording partition data; according to the environment record partition data, deploying a full-period environment record device on the edge side, constructing an edge environment record network, and generating full-period environment original record data; by deploying a full-period environment recording network on the edge side, field complex and multi-dimensional environment time sequence characteristics such as spring temperature rise slope and autumn temperature drop amplitude are accurately captured, perennial plants can follow natural growth rhythms in a controllable environment, the problem of periodic signal loss caused by constant artificial environment parameters or simple circulation is avoided, and the yield of the perennial plants is improved. The plant growth quality is effectively improved, the quality of economic organs (roots and fruits) is guaranteed, and a solid foundation is laid for planting high-value economic crops.

Description

TECHNICAL FIELD

[0001] The application provides a plant growth cycle monitoring method and system based on edge computing, and belongs to the technical field of intelligent agriculture and edge intelligence. BACKGROUND

[0002] In the field of controlled environment planting, such as plant factories and greenhouses, traditional growth acceleration methods for perennial plants have many limitations. The existing technology mainly relies on high light intensity, long light, growth hormones and other means, and the core logic is that the photosynthesis rate determines the growth speed. However, for perennial plants such as ginseng, wolfberry and dendrobium, the quality of economic organs (roots, fruits) depends not only on biomass accumulation, but also on seasonal environmental cycle signals such as diurnal temperature difference, photoperiod change, humidity fluctuation, etc.

[0003] On the one hand, artificial environments mostly use constant or simple cyclic parameters, which cannot reproduce the complex multi-dimensional environmental timing characteristics in the wild throughout the year, resulting in the lack of cycle signals. On the other hand, only the photosynthesis rate is improved, ignoring the plant's internal perception of time, resulting in overgrowth, low effective ingredient content, and disturbed dormancy mechanism, etc. In addition, the existing data processing relies on the cloud, which has high delay, large bandwidth occupation, privacy risk and other problems, and is difficult to support real-time closed-loop control. Moreover, the manufacturers' publicity is mostly a general concept or a simple hardware combination, no one deeply couples the one-year wild environment full-cycle minute-level record with the acceleration playback simulation according to the plant physiological endurance, and realizes the related functions on the edge side, so it is urgent to solve these problems by innovative methods. SUMMARY

[0004] The application provides a plant growth cycle monitoring method and system based on edge computing, which solves the problems mentioned in the background art: The application provides a plant growth cycle monitoring method and system based on edge computing, which solves the problems mentioned in the background art: S1, dividing the full-cycle environmental parameter recording area of the controlled environment planting area, generating environmental recording partition data; deploying full-cycle environmental recording equipment on the edge side according to the environmental recording partition data, constructing an edge environmental recording network, and generating full-cycle environmental original recording data; S2, setting the environmental acceleration playback parameters, accelerating the playback simulation of the environment on the edge side, and collecting multi-source environmental data at the same time to obtain simulated environmental signal data and corresponding timestamp data; removing invalid environmental factors from the simulated environmental signal data to generate valid environmental signal data; S3. Based on the effective environmental signal data, analyze the plant physiological response to obtain plant biomass accumulation data and plant effective component content change data respectively; conduct a growth quality correlation assessment on the plant biomass accumulation data and plant effective component content change data, and determine the growth stage to obtain plant growth stage data. S4. Perform 3D laser scanning on the controllable environment planting area to generate 3D point cloud data of the planting area; analyze the working conditions of the 3D point cloud data of the planting area through plant growth stage data to generate plant growth working condition data; predict the risk of abnormal growth based on the plant growth working condition data to obtain growth abnormality risk prediction data. S5. Based on the growth abnormality risk prediction data, perform weighted growth stability index processing to generate a plant growth stability index; based on the plant growth stability index, perform risk warning processing to generate plant growth cycle warning data.

[0005] The plant growth cycle monitoring system based on edge computing proposed in this invention includes: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors are made to implement the method described in any one of the above.

[0006] The beneficial effects of this invention are as follows: By deploying a full-cycle environmental recording network at the edge, it accurately captures the complex and multidimensional temporal characteristics of the environment in the wild, such as the spring warming slope and the autumn cooling amplitude. This allows perennial plants to follow their natural growth rhythms even in a controlled environment, avoiding the problem of missing periodic signals caused by constant or simple cycles of artificial environmental parameters. This effectively improves plant growth quality, ensures the quality of economic organs (roots and fruits), and lays a solid foundation for the cultivation of high-value economic crops. Furthermore, by simulating accelerated environmental cycle playback at the edge based on plant physiological tolerance and eliminating ineffective environmental factors in real time, it achieves a scientific acceleration of the plant growth cycle while avoiding problems such as excessive vegetative growth, low content of effective components, and disordered dormancy mechanisms caused by only increasing the photosynthetic rate and neglecting the plant's internal perception of time. This improves the efficiency and effectiveness of controlled-environment cultivation. This system benefits the market's demand for high-quality perennial plants. By deeply coupling functions such as environmental recording, accelerated playback, elimination of invalid factors, control planting, and quality assessment, it forms an integrated system at the edge, enabling real-time monitoring and precise control of the plant growth cycle. Growers can adjust their planting strategies in a timely manner based on real-time data, reducing the risk of abnormal growth caused by information lag or improper handling, lowering planting costs, and enhancing the flexibility and adaptability of controlled environment planting. By combining 3D laser scanning with growth stage data for operational activity analysis and predicting the risk of abnormal growth, it can quickly and accurately identify potential problems in the plant growth process, classify them according to risk level, and provide targeted treatment suggestions, reducing the chance of missed or delayed treatment of abnormal growth events, enhancing the response efficiency to plant growth anomalies, and ensuring healthy plant growth. Attached Figure Description

[0007] Figure 1 This is a diagram illustrating the steps of the method described in this invention; Figure 2 This is a diagram of the edge environment recording network system architecture described in this invention; Figure 3 This is a diagram of the environmental accelerated playback simulation system described in this invention. Detailed Implementation

[0008] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0009] One embodiment of the present invention, such as Figure 1 As shown, a plant growth cycle monitoring method based on edge computing is described, the method comprising: S1. Divide the controllable environment planting area into full-cycle environmental parameter recording areas and generate environmental recording partition data; based on the environmental recording partition data, deploy full-cycle environmental recording devices on the edge side to construct an edge environmental recording network. This network can record complex multi-dimensional environmental time-series characteristics in the field throughout the year at the minute level, including the spring warming slope, autumn cooling amplitude, and rainy season humidity pulse, and generate full-cycle environmental raw record data. S2. Based on the full-cycle original environmental record data recorded by the edge environment recording network, environmental acceleration playback parameters are set, and the environmental cycle is accelerated and simulated on the edge side according to the plant's physiological tolerance. At the same time, multi-source environmental data is collected to obtain simulated environmental signal data and corresponding timestamp data. Invalid environmental factors are removed from the simulated environmental signal data to eliminate those factors that have no positive impact on plant growth or cause negative interference, and valid environmental signal data is generated. S3. Based on the effective environmental signal data, analyze the plant physiological response to obtain plant biomass accumulation data and plant effective component content change data respectively; conduct a growth quality correlation assessment on the plant biomass accumulation data and plant effective component content change data, and determine the growth stage to obtain plant growth stage data. S4. Perform 3D laser scanning processing on the controllable environment planting area using timestamp data to generate 3D point cloud data of the planting area; perform working condition activity analysis on the 3D point cloud data of the planting area using plant growth stage data to generate plant growth working condition data; perform growth abnormality risk prediction on the plant growth working condition data to obtain growth abnormality risk prediction data. S5. Based on the growth abnormality risk prediction data, a weighted growth stability index is processed to generate a plant growth stability index; based on the plant growth stability index, risk warning processing is performed to generate plant growth cycle warning data. At the same time, combined with the growth quality correlation assessment results, the plant growth cycle is scientifically adjusted and optimized to achieve scientific acceleration and quality assurance of the perennial plant growth cycle.

[0010] The working principle and effects of the above technical solution are as follows: By recording multi-dimensional temporal characteristics of the wild environment at the edge level (minute-level) and accelerating the playback simulation according to the plant's physiological tolerance, the decline in the quality of economic organs caused by traditional hormone / light saturation methods is reduced, the contradiction between environmental regulation and plant needs is avoided, and the scientificity and controllability of accelerating the growth cycle of perennial plants are improved; by systematically eliminating ineffective environmental factors and correlating changes in biomass and effective components, the ineffective consumption of energy, water, fertilizer, and other resources is reduced, avoiding the drawback of pursuing only growth rate while ignoring quality, and enhancing the resource utilization efficiency of controlled environment planting. Stability of growth quality; By generating point cloud data of the planting area through 3D laser scanning and combining it with the growth stage for working condition analysis and risk prediction, the omission or delay in handling of growth anomalies (such as temperature and humidity imbalance, stagnation of component accumulation) is reduced, avoiding planting losses caused by the expansion of anomalies and improving the efficiency of growth risk prediction and response; By setting risk weights, calculating the growth stability index and dynamically adjusting parameters through the analytic hierarchy process, the subjectivity and error of manual control are reduced, avoiding the problem of uncontrolled growth cycle caused by blind adjustment, enhancing the adaptability of the technology to different perennial plants, and making the whole cycle planting management more closed-loop and reliable.

[0011] One embodiment of the present invention, such as Figure 2 As shown, S1 includes: S11. Conduct a multi-dimensional comprehensive analysis of the wild native environment characteristics, topographical differences, plant distribution density and growth habits of the controllable environment planting area, and monitor key areas and accuracy requirements for environmental parameters; based on the analysis results, complete the division of the environmental parameter recording area for the whole cycle according to the principle of environmental heterogeneity, and generate environmental recording partition data, which includes the core planting area, transitional adaptation area, control reference area and edge buffer zone. S12. Based on the monitoring parameter types (light, temperature and humidity, air pressure, etc.) and the characteristics of the zoned environment, select suitable multi-sensor equipment (spectral sensors, soil temperature and humidity sensors, air pressure and wind speed monitors, etc.); according to the coverage and deployment requirements of the environmental recording zone data, deploy the selected full-cycle environmental recording equipment in a distributed manner on the edge side to build a decentralized edge environmental recording network. S13. Perform time synchronization and parameter calibration on all devices in the edge environment recording network to ensure the consistency and accuracy of data acquisition; start the network to collect multi-dimensional environmental time-series characteristics in the field at the minute level throughout the year. The multi-dimensional environmental time-series characteristics cover light duration / spectral distribution / luminous flux, atmospheric and soil temperature and humidity, air pressure, wind speed, rainfall and seasonal-specific parameters such as spring warming slope, autumn cooling amplitude, and rainy season humidity pulse, to generate full-cycle raw environmental record data.

[0012] The working principle and effects of the above technical solution are as follows: By analyzing the wild native environment and plant characteristics from multiple dimensions and dividing the monitoring area according to heterogeneity, monitoring blind spots caused by unreasonable area division are reduced, and the problem of the collection focus being out of sync with plant needs is avoided, thus improving the targeting and coverage uniformity of environmental records; By combining the type of monitoring parameters and the characteristics of the zoning to select and adapt sensors, and by constructing a decentralized network through distributed deployment, the impact of single-point equipment failure on overall monitoring is reduced, and data collection deviations caused by improper equipment adaptation are avoided, thus enhancing the stability and scenario adaptability of the environmental recording network; By performing time synchronization and parameter calibration on the equipment and collecting multi-dimensional time-series characteristics throughout the year at the minute level, consistency errors in data collection and omissions of seasonally specific parameters are reduced, thus avoiding the hidden danger of incomplete raw data affecting subsequent environmental simulations, and improving the accuracy and completeness of raw environmental record data throughout the entire cycle.

[0013] In one embodiment of the present invention, S13 includes: Develop a time synchronization and parameter calibration scheme for edge environment recording network devices, defining synchronization accuracy standards (e.g., millisecond-level synchronization), calibration indicators (sensor error thresholds, data transmission delay range), and operation procedures; initiate the device time synchronization mechanism according to the calibration scheme, perform time calibration on all sensor devices through the edge-side clock synchronization protocol, and generate device time synchronization confirmation results; Based on the synchronous confirmation results, the monitoring parameters of each sensor device are calibrated one by one, the device measurement deviation is corrected by comparing with the standard parameter values, and a sensor parameter calibration report is generated. The consistency of the data in the calibration report is verified. If there are any devices that do not meet the standards, the calibration operation is re-executed until all devices meet the data acquisition accuracy requirements and a network device calibration certificate is generated. The edge environment recording network is started based on the calibration certificate, and the data acquisition frequency is set at the minute level to continuously collect multi-dimensional environmental time-series characteristics in the field for one year (including basic parameters such as light-related parameters, temperature and humidity, and air pressure, as well as seasonal specific parameters). The collected time-series feature data is integrated and standardized in real time to generate full-cycle environmental raw record data with a well-structured structure and complete parameters.

[0014] The working principle and effects of the above technical solution are as follows: By establishing a clear time synchronization and parameter calibration scheme, and specifying millisecond-level synchronization accuracy and error thresholds, the arbitrariness of calibration operations is reduced, calibration deviations caused by the lack of standards are avoided, and the standardization and operability of equipment calibration are improved; Time synchronization and parameter calibration are performed on each sensor individually, and deviations are corrected by comparing with standard values, reducing data acquisition consistency errors between different devices, avoiding the cumulative effect of measurement deviations on subsequent analysis, and improving the accuracy of environmental data acquisition; Verification of compliance after calibration before restarting the network reduces the situation of unqualified equipment operating with defects, avoids the hidden danger of raw data distortion, and enhances the reliability of environmental recording network operation; Minute-level acquisition of multi-dimensional features throughout the year and standardized format processing reduce data omissions and format chaos, avoid the impact of irregular raw data on subsequent environmental simulation, and improve... One embodiment of the present invention, such as Figure 3 As shown, S2 includes: S21. Perform statistical analysis on the raw environmental records throughout the entire cycle, and extract the physiological tolerance thresholds (limit values ​​of temperature, humidity, light, etc.) of the target plants (such as melons and fruits, Chinese herbal medicines); set the environmental acceleration playback rate gradient parameters (such as time compression ratios of 1:2, 1:3, and 1:4) based on the physiological tolerance thresholds, and generate an environmental acceleration playback parameter configuration table. S22. Configure the simulation environment control equipment on the edge side according to the configuration table, complete the hardware adaptation and parameter initialization of the accelerated playback environment; execute the environmental cycle accelerated playback simulation, synchronously start the multi-source data acquisition module, acquire the simulation environment signal data and corresponding timestamp data in real time, and establish the environmental parameter and time mapping dataset. S23. Screen ineffective environmental factors (duration of no photosynthesis, extreme temperature and humidity, insufficient light period, etc.) based on plant growth mechanism, and formulate factor elimination rules; S24. Process the simulated environmental signal data according to the elimination rules, remove factors with no positive impact or negative interference, and generate valid environmental signal data.

[0015] The working principle and effects of the above technical solution are as follows: By statistically analyzing the original environmental data, extracting the physiological tolerance threshold of plants, and setting multiple acceleration gradients, plant damage caused by blindly setting acceleration parameters is reduced, ineffective acceleration beyond physiological tolerance is avoided, and the scientificity and adaptability of environmental acceleration playback are improved; hardware adaptation and parameter initialization are completed according to the configuration table, and simulated signals and timestamp data are collected synchronously, reducing simulation deviations caused by equipment and parameter mismatch, avoiding the problem of environmental data being disconnected from the time dimension, and enhancing the realism and data correlation of the simulated environment; based on the plant growth mechanism, ineffective factors are screened and elimination rules are formulated, reducing the interference of meaningless environmental periods on growth, avoiding the waste of energy and resources, and improving the accuracy of environmental regulation; negative and ineffective factors are eliminated according to the rules to generate effective signal data, reducing the interference of redundant data on subsequent analysis, avoiding ineffective regulation from slowing down the growth cycle, and enhancing the practicality of environmental signal data and growth promotion efficiency.

[0016] In one embodiment of the present invention, step S21 includes: The raw environmental records throughout the entire lifecycle are classified and organized according to parameter type (light, temperature and humidity, air pressure and wind speed, etc.), and abnormal entries with missing data rates exceeding the threshold are removed to generate a normalized environmental record classification dataset. Based on the normalized environmental record classification dataset, statistical analysis algorithms (such as extreme value analysis and distribution fitting) are used to calculate the annual fluctuation range, peak interval and duration of each environmental parameter to obtain the statistical analysis results of environmental parameters. Based on literature and preliminary experimental data on the growth and physiological characteristics of target plants (fruits and herbs), physiological tolerance thresholds are extracted from the statistical analysis results. These physiological tolerance thresholds include temperature tolerance limits, humidity adaptation ranges, and light thresholds, generating a list of physiological tolerance thresholds for target plants. Using physiological tolerance thresholds as constraints, multiple environmental acceleration playback rate gradients were designed, and the time compression ratios (1:2, 1:3, 1:4) and corresponding environmental parameter synchronous adjustment rules under different gradients were determined to form a rate gradient parameter scheme. The rate gradient parameter scheme is associated and matched with the adaptation requirements of various environmental parameters, and the equipment matching information is supplemented. The matching information includes operating parameters and data acquisition frequency, and the data format and field definition are standardized. All the above information is integrated to generate a complete and logically clear environmental acceleration playback parameter configuration table.

[0017] The working principle and effects of the above technical solution are as follows: By classifying and organizing the raw data according to parameter type and removing abnormal entries with high missing rates, the problems of data disorder and interference from abnormal entries in the analysis are reduced, avoiding distortion of subsequent statistical results and improving the regularity and reliability of environmental record data; by using statistical algorithms to analyze the annual fluctuation of parameters and combining plant physiological literature and pre-experiments to extract tolerance thresholds, the subjectivity and blindness of threshold setting are reduced, avoiding parameter definitions that are divorced from the actual needs of plants, and improving the accuracy and adaptability of physiological tolerance thresholds; by designing multiple acceleration gradients with thresholds as constraints and clarifying the time compression ratio and parameter adjustment rules, the problem of insufficient adaptability of a single acceleration rate is reduced, the risk of exceeding physiological tolerance is avoided, and the flexibility and safety of environmental acceleration playback are enhanced; by associating parameter adaptation requirements, supplementing equipment matching information and standardizing the format, the situation of incomplete configuration table information and logical confusion is reduced, avoiding the connection obstacles between subsequent hardware adaptation and simulation operation, and improving the practicality and executability of parameter configuration tables.

[0018] In one embodiment of the present invention, S3 includes: S31. Deploy plant physiological monitoring nodes according to plant distribution density in a controlled environment planting area. The plant physiological monitoring nodes include biomass monitoring units such as plant height, leaf area, and root growth, as well as effective component detection units. Based on the growth stage corresponding to the effective environmental signal data, start the physiological monitoring nodes to collect plant biomass index data and form a time-series biomass accumulation dataset. S32. Regularly collect the effective component content (such as active ingredients of traditional Chinese medicine, sugar content of fruits, etc.) of plant economic organs through professional testing equipment to generate a dataset of effective component content changes; construct a growth quality correlation assessment model, couple the biomass accumulation dataset with the effective component content change dataset for analysis, and quantify the correlation coefficient between the two. S33. Generate a growth quality correlation assessment report based on the coupling analysis results to determine the synergistic growth pattern of biomass and effective components; S34. Combining the assessment report with the physiological laws of plant growth, set the threshold for judging the growth stage, and complete the division of the growth stage through time series data trend analysis. The growth stage includes the budding period, the growth period, the vigorous growth period and the dormancy period, and obtain plant growth stage data.

[0019] The working principle and effects of the above technical solution are as follows: By deploying monitoring nodes covering biomass and effective components according to plant distribution density, and combining this with precise collection of time-series data at different growth stages, the problem of uneven monitoring coverage and single indicators is reduced, key growth information is avoided, and the comprehensiveness and timeliness of plant physiological data collection are improved. By collecting effective component data using specialized equipment and constructing models to couple and analyze the correlation between biomass and components, the one-sidedness of judging growth quality solely based on biomass is reduced, the drawback of neglecting the quality of economic organs is avoided, and the scientific rigor and objectivity of growth quality assessment are enhanced. Based on the coupling results, a correlation assessment report is generated, clarifying the synergistic growth pattern, reducing the ambiguity in understanding the internal logic of plant growth, avoiding the blindness of subsequent regulation due to a lack of data support, and improving the pertinence of growth stage division and regulation decisions. By combining the assessment report with physiological patterns to set thresholds and precisely divide growth stages, the problems of stage discrimination errors and boundary ambiguity are reduced, and inappropriate regulation caused by misjudgment of growth status is avoided, enhancing the reliability and application value of plant growth stage data.

[0020] In one embodiment of the present invention, S34 includes: Extract core information on synergistic growth patterns from growth quality correlation assessment reports (such as the peak synchronization interval of biomass and effective components, and the matching relationship of growth rates), and combine it with literature on plant growth physiology to compile a dataset of core criteria for dividing growth stages; Based on the core dataset, key discriminant dimensions for growth stage division are determined, including biomass growth rate, effective component accumulation rate, and physiological indicator mutation characteristics, generating a list of stage discriminant dimensions. For each discriminant dimension, differentiated discriminant threshold ranges are set based on the differences in growth characteristics of different plants (fruits and vegetables, Chinese herbal medicines) (such as biomass growth rate threshold during germination and effective component accumulation rate threshold during dormancy), generating a growth stage discriminant threshold table. We retrieved the time-series biomass accumulation dataset and the effective component content change dataset, and used trend extraction algorithms (such as sliding window fitting and slope analysis) to mine the time-series change characteristics of the data to generate a time-series trend feature dataset. We then matched the time-series trend feature dataset with the growth stage discrimination threshold table dimension by dimension, and used logical judgment rules (such as growth rate meeting the standard and duration meeting the requirements) to complete the preliminary division of the germination stage, growth stage, vigorous growth stage, and dormancy stage, generating the preliminary division results of the growth stages. The rationality of the stage transitions in the preliminary division results is verified, and corresponding problems are corrected, including cross-stage threshold conflicts and mismatches in temporal trends, to generate accurate and standardized plant growth stage data.

[0021] The working principle and effects of the above technical solution are as follows: Extracting core patterns from the evaluation report and combining them with physiological literature to form a basis for division reduces the problems of fragmented division criteria and lack of core support, avoids blind definitions divorced from the actual logic of plant growth, and improves the scientific rigor and relevance of growth stage division; clarifying multi-dimensional discrimination indicators and setting differentiated thresholds for different plants avoids the one-sidedness of single-dimensional evaluation and enhances the adaptability to different plants such as melons, fruits, and medicinal herbs; using trend extraction algorithms to mine time-series data features reduces errors and omissions in manual data analysis, avoids the neglect of key growth trends, and improves the accuracy and comprehensiveness of time-series feature capture; matching data and thresholds dimension by dimension and initially dividing according to logical rules reduces the subjectivity and arbitrariness of stage division, avoids the problem of blurred boundaries, and enhances the rationality of the initial division results; verifying the rationality of stage transitions and correcting conflicts and mismatches reduces loopholes such as cross-stage contradictions and time-series disconnections, avoids distortion of division results affecting subsequent applications, and improves the accuracy and standardization of plant growth stage data.

[0022] In one embodiment of the present invention, step S4 includes: S41. Extract the time information corresponding to key plant growth nodes (such as budding, flowering, and fruiting) from the timestamp data to determine the triggering time for the three-dimensional laser scan; start the three-dimensional laser scan equipment to perform a comprehensive scan of the controllable environment planting area and collect spatial information, including plant canopy structure, plant spacing, and planting area topography. S42. Preprocess the scanned data (denoise removal, stitching) to generate high-precision 3D point cloud data of the planting area; perform spatiotemporal registration between the plant growth stage data and the 3D point cloud data to establish the correspondence between growth stages and spatial morphology. S43. Based on the registration results, analyze the plant growth conditions (such as nutrient absorption efficiency, environmental adaptability, and growth rate) at different growth stages, and generate plant growth condition data. S44. Construct a growth abnormality risk prediction model, input risk characteristics, including growth condition data to identify temperature and humidity mismatch, stagnation of effective component accumulation, and abnormal growth rate, and obtain growth abnormality risk prediction data (including risk level, probability of occurrence and scope of impact).

[0023] The working principle and effects of the above technical solution are as follows: Extracting key growth nodes triggers 3D laser scanning and collects spatial information from all directions, reducing the omission of key morphological information due to improper scanning timing, avoiding the problem of one-sided spatial data collection, and improving the accuracy and timeliness of spatial information acquisition in the planting area; denoising and stitching the scanned data to generate a high-precision point cloud, and spatiotemporally registering the growth stage with the spatial morphology, reduces errors caused by data noise and coordinate deviation, avoids the disconnect between growth status and spatial characteristics, and enhances the accuracy and correlation of the two; analyzing the growth conditions at different stages based on the registration results reduces the vague understanding of plant nutrient absorption, environmental adaptation, and other states, avoids the one-sidedness of judging growth conditions solely by appearance, and improves the comprehensiveness and depth of growth condition assessment; constructing a model to input multi-dimensional risk characteristics to predict anomalies, clarifying the risk level and impact range, reducing the situation of missed or delayed handling of abnormal risks, avoiding planting losses caused by risk expansion, and enhancing the scientific nature and response efficiency of growth risk prediction.

[0024] In one embodiment of the present invention, S42 includes: S421. The raw data acquired by 3D laser scanning is classified and organized according to scanning angle and equipment number. Invalid entries with failed acquisition or incomplete data are removed to generate a normalized scanning dataset. Based on the normalized scanning dataset, an adaptive Gaussian filtering algorithm is used to filter out noise points (such as environmental interference points and equipment error points) in the data to generate denoised scanning data. S422. Using a multi-view point cloud stitching algorithm, coordinate alignment and fusion are performed on the denoised scan data from different positions and angles to fill data gaps and generate preliminary stitched scan data. The overlap and deviation values ​​of the preliminary stitched data are detected by the point cloud accuracy verification model, and stitching errors exceeding the threshold are corrected to generate high-precision three-dimensional point cloud data of the planting area. S423. Extract the spatial coordinate information of key time nodes (budding stage, peak growth period, etc.) from the plant growth stage data and 3D point cloud data to construct a spatiotemporal correlation basic dataset; S424. An improved ICP spatiotemporal registration algorithm is adopted to accurately match time nodes with spatial morphological data of corresponding stages, verify the consistency of matching logic, and establish a one-to-one correspondence between growth stages and spatial morphology data.

[0025] The working principle and effects of the above technical solution are as follows: Data is categorized and organized according to scanning angle and equipment number, and invalid entries are removed, reducing the problems of data disorder and invalid entries slowing down processing efficiency, avoiding interference from incomplete data in subsequent analysis, and improving the regularity and basic quality of scanned data; an adaptive Gaussian filtering algorithm is used to filter noise points, reducing the impact of environmental interference and equipment errors, avoiding noise masking real spatial features, and enhancing the purity and usability of scanned data; multi-view stitching fills data gaps, and accuracy verification corrects over-threshold deviations, reducing stitching loopholes and coordinate errors in point cloud data, avoiding low-precision point clouds affecting spatial morphology restoration, and improving the accuracy and completeness of 3D point cloud data in the planting area; key time nodes and spatial coordinates are extracted to construct a spatiotemporal correlated dataset, reducing the problem of time dimension and spatial data disconnection, avoiding obstacles to subsequent registration due to lack of basic correlation, and enhancing the spatiotemporal correlation of data; an improved ICP algorithm is used to accurately match time nodes and spatial morphology, verifying logical consistency, reducing registration deviations and stage-morphology correspondence confusion, avoiding matching logic conflicts, and improving the reliability and accuracy of the correspondence between growth stages and spatial morphology.

[0026] In one embodiment of the present invention, S422 includes: Extract spatial coordinate information and device attitude parameters from each viewpoint in the denoised scan data, establish a unified three-dimensional coordinate system benchmark, and generate coordinate benchmark calibration parameters. Based on the coordinate reference calibration parameters, a multi-view point cloud stitching algorithm is used to perform preliminary coordinate alignment on denoised data at different positions and angles, process conflict points in overlapping data areas, and generate preliminary aligned point cloud data. Spatial interpolation algorithm is used to fill in the gaps in the preliminary aligned point cloud data, filling in the information loss caused by scanning blind spots, and generating stitched data after gap filling. Start the point cloud accuracy verification model, calculate the accuracy index of the stitched data, including the region overlap and coordinate deviation, and generate a point cloud stitching accuracy test report. For the splicing error areas in the test report that exceed the preset threshold, a combination of local re-splicing and coordinate fine-tuning is used to correct the error, generating error-corrected splicing data. The error-corrected splicing data is then subjected to a second accuracy verification. After confirming that all indicators meet the high-precision requirements, the final high-precision 3D point cloud data of the planting area is generated.

[0027] The working principle and effects of the above technical solution are as follows: Establishing a unified three-dimensional coordinate system benchmark and generating calibration parameters reduces the problem of chaotic data coordinates from different perspectives, avoids systematic errors caused by inconsistent benchmarks during stitching, and improves the basic consistency of point cloud stitching; Preliminary data alignment based on calibration parameters and handling conflict points in overlapping areas reduces the connection contradictions between scanned data from different locations, avoids interference from conflict points in the stitching effect, and enhances the regularity of the initially aligned point cloud data; Spatial interpolation algorithms are used to fill in the gaps in the scanning blind areas, reducing incomplete spatial morphology restoration caused by missing data, avoiding the impact of missing blind area information on subsequent analysis, and improving the stitching data... The integrity of the data is ensured; the accuracy verification model is activated to calculate the overlap and deviation values ​​and generate a test report, reducing blind judgment on the stitching accuracy, avoiding the hidden danger of undiscovered errors, and enhancing the control over the stitching quality; local re-stitching and coordinate fine-tuning are used to correct errors in areas exceeding the threshold, reducing concentrated stitching deviations and avoiding the problem that a single correction method cannot eradicate errors, thus improving the accuracy compliance rate of the stitched data; after secondary verification confirms that all indicators meet the standards, the final point cloud data is generated, reducing the accuracy failure caused by residual errors, preventing low-precision data from flowing into subsequent stages, and enhancing the final reliability and application value of the 3D point cloud data of the planting area.

[0028] In one embodiment of the present invention, step S5 includes: S51. Based on the degree of influence of each risk factor on plant growth in the growth abnormality risk prediction data, the analytic hierarchy process (AHP) is used to set the weights of the risk factors; combined with the weights and growth condition data, the plant growth stability index is calculated to quantify the stability of the plant growth status. S53. Referencing the growth quality standards of the target plant, set a growth stability index warning threshold (high risk, medium risk, low risk); when the growth stability index is lower than the warning threshold, automatically trigger the risk warning mechanism and generate plant growth cycle warning data, which includes risk type, scope of impact, and response suggestions. S53. Combine the growth quality correlation assessment report and early warning data to dynamically adjust the environmental accelerated playback parameters and key growth factors (such as light duration and soil moisture); continuously monitor the adjusted plant growth status and iteratively optimize the growth cycle parameters.

[0029] The working principle and effects of the above technical solution are as follows: Using the analytic hierarchy process (AHP) to set risk factor weights and calculating the growth stability index based on growth condition data reduces the subjectivity of weight allocation and the one-sidedness of single-indicator evaluation, avoids vague understanding of growth stability, and improves the scientific rigor and accuracy of quantitative assessment of plant growth status. Setting multi-level early warning thresholds based on target plant quality standards, automatically triggering early warnings, and clearly defining risk types and response suggestions reduces missed or delayed handling of abnormal growth risks, avoids planting losses caused by risk expansion, and enhances the pertinence and timeliness of risk warnings. Dynamically adjusting environmental parameters based on growth quality assessment reports and early warning data, continuously monitoring and iteratively optimizing, reduces uncontrolled growth cycles or quality decline caused by blind regulation, avoids insufficient parameter adaptability, improves the controllability of accelerated growth cycles in perennial plants, ensures the quality stability of economic organs, and creates a closed-loop, efficient management model for planting regulation.

[0030] In one embodiment of the present invention, S51 includes: All risk factors (such as temperature and humidity mismatch and stagnation of effective component accumulation) are extracted from the growth abnormality risk prediction data and classified and sorted according to the impact dimensions (environmental adaptation, growth status, and component accumulation) to generate a risk factor classification list. Based on the risk factor classification list, a hierarchical structure model is constructed using the analytic hierarchy process (AHP) to clarify the hierarchical relationship between the target layer (growth stability assessment), the criterion layer (impact dimensions), and the scheme layer (specific risk factors), thus generating a hierarchical structure framework. Agricultural planting experts and data analysts were invited to compare the relative importance of each risk factor pairwise. A judgment matrix was constructed according to the scaling rules (1-9 scale) to generate a risk factor judgment matrix set. The judgment matrix set was tested for consistency, and the consistency index CI and random consistency ratio CR were calculated. If CR < 0.1, the judgment matrix was considered valid. Otherwise, the judgment matrix was revised until it met the standard, and a qualified consistency test result was generated. Based on a qualified judgment matrix, the weight values ​​of each risk factor are calculated using the eigenvalue decomposition method. After normalization, a risk factor weight allocation table is generated. Key evaluation indicators (such as nutrient absorption efficiency and growth rate compliance rate) are extracted from the growth condition data and matched with the risk factor weight allocation table. A weighted summation algorithm is used to calculate the plant growth stability index, quantifying the stability of plant growth status and generating plant growth stability index data.

[0031] The working principle and effects of the above technical solution are as follows: Classifying and organizing risk factors according to their impact dimensions reduces the problems of disorganized and overlapping factors, avoids the lack of a clear framework in subsequent analysis, and improves the systematicness and relevance of risk factor analysis; constructing a hierarchical structure model clarifies the relationship between the target layer, criterion layer, and solution layer, reducing evaluation confusion caused by ambiguity in hierarchical logic, avoiding disconnections between various stages, and enhancing the systematicness and logic of growth stability assessment; inviting experts and analysts to construct a judgment matrix through pairwise comparisons according to scaling rules reduces the subjective arbitrariness of weight setting, avoids the one-sidedness of a single perspective, and makes weight allocation more appropriate. It aligns with actual planting scenarios; it performs consistency checks on the judgment matrix and corrects non-compliant items, reducing weight bias caused by matrix logical conflicts, avoiding the impact of erroneous matrices on evaluation results, and improving the credibility and scientific rigor of weight allocation; it reduces errors and irrationality in weight calculation through eigenvalue decomposition and normalization, avoiding evaluation distortion caused by weight imbalance, and enhancing the accuracy of the weight allocation table; it correlates key growth condition indicators with weights and uses weighted summation to quantify the stability index, reducing vague qualitative judgments of growth status, avoiding the problem of inaccurate stability assessment, and improving the quantitative degree and practical value of plant growth status evaluation.

[0032] One embodiment of the present invention provides a plant growth cycle monitoring system based on edge computing, comprising: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors are made to implement the method described in any one of the above.

[0033] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for monitoring plant growth cycles based on edge computing, characterized in that, The method includes: S1. Divide the controllable environment planting area into full-cycle environmental parameter recording areas and generate environmental recording partition data; deploy full-cycle environmental recording devices on the edge side according to the environmental recording partition data, construct an edge environmental recording network, and generate full-cycle environmental raw recording data; S2. Set the environmental acceleration playback parameters, perform accelerated playback simulation of the environmental cycle at the edge, and simultaneously collect multi-source environmental data to obtain simulated environmental signal data and corresponding timestamp data; perform invalid environmental factor removal processing on the simulated environmental signal data to generate valid environmental signal data. S3. Based on the effective environmental signal data, analyze the plant physiological response to obtain plant biomass accumulation data and plant effective component content change data respectively; conduct a growth quality correlation assessment on the plant biomass accumulation data and plant effective component content change data, and determine the growth stage to obtain plant growth stage data. S4. Perform 3D laser scanning on the controllable environment planting area to generate 3D point cloud data of the planting area; analyze the working conditions of the 3D point cloud data of the planting area through plant growth stage data to generate plant growth working condition data; predict the risk of abnormal growth based on the plant growth working condition data to obtain growth abnormality risk prediction data. S5. Based on the growth abnormality risk prediction data, perform weighted growth stability index processing to generate a plant growth stability index; based on the plant growth stability index, perform risk warning processing to generate plant growth cycle warning data.

2. The plant growth cycle monitoring method based on edge computing according to claim 1, characterized in that, S1 includes: S11. Conduct a multi-dimensional comprehensive analysis of the wild native environment characteristics, topographic differences, plant distribution density and growth habits of the controlled environment planting area. Based on the analysis results, complete the division of the environmental parameter recording area for the whole cycle according to the principle of environmental heterogeneity, and generate environmental record partition data. S12. Based on the monitoring parameter types and the characteristics of the zoned environment, select suitable multi-sensor devices; according to the coverage and deployment requirements of the environmental recording zone data, deploy the selected full-cycle environmental recording devices in a distributed manner on the edge side to build a decentralized edge environmental recording network. S13. Perform time synchronization and parameter calibration on all devices in the edge environment recording network, start the network to collect multi-dimensional environmental time-series characteristics in the field at the minute level throughout the year, and generate full-cycle environmental raw record data.

3. The plant growth cycle monitoring method based on edge computing according to claim 1, characterized in that, The S2 includes: S21. Perform statistical analysis on the raw environmental records throughout the entire cycle to extract the physiological tolerance threshold of the target plant; set the environmental acceleration playback rate gradient parameter based on the physiological tolerance threshold and generate an environmental acceleration playback parameter configuration table. S22. Configure the simulation environment control equipment on the edge side according to the configuration table, complete the hardware adaptation and parameter initialization of the accelerated playback environment; execute the environmental cycle accelerated playback simulation, synchronously start the multi-source data acquisition module, acquire the simulation environment signal data and corresponding timestamp data in real time, and establish the environmental parameter and time mapping dataset. S23. Screen ineffective environmental factors based on plant growth mechanisms and formulate factor elimination rules; S24. Process the simulated environmental signal data according to the elimination rules, remove factors with no positive impact or negative interference, and generate valid environmental signal data.

4. The plant growth cycle monitoring method based on edge computing according to claim 3, characterized in that, S21 includes: The raw environmental records throughout the entire lifecycle are classified and organized according to parameter type. Abnormal entries with missing data rates exceeding the threshold are removed to generate a normalized environmental record classification dataset. Based on the normalized environmental record classification dataset, statistical analysis algorithms are used to calculate the annual fluctuation range, peak interval, and duration of each environmental parameter to obtain the statistical analysis results of the environmental parameters. Based on literature and preliminary experimental data on the growth and physiological characteristics of the target plants, physiological tolerance thresholds are extracted from the statistical analysis results to generate a list of physiological tolerance thresholds for the target plants. Using physiological tolerance thresholds as constraints, multiple environmental acceleration playback rate gradients are designed, and the time compression ratios and corresponding environmental parameter synchronous adjustment rules under different gradients are determined to form a rate gradient parameter scheme. The rate gradient parameter scheme is associated and matched with the adaptation requirements of various environmental parameters, supplementing equipment matching information and standardizing data format and field definition; all the above information is integrated to generate a complete and logically clear environmental acceleration playback parameter configuration table.

5. The plant growth cycle monitoring method based on edge computing according to claim 1, characterized in that, The S3 includes: S31. Deploy plant physiological monitoring nodes according to plant distribution density within the controllable environment planting area. Based on the growth stage corresponding to the effective environmental signal data, start the physiological monitoring nodes to collect plant biomass index data and form a time-series biomass accumulation dataset. S32. Regularly collect the effective component content of plant economic organs using detection equipment to generate a dataset of effective component content changes; construct a growth quality correlation assessment model, couple the biomass accumulation dataset with the effective component content change dataset, and quantify the correlation coefficient between the two. S33. Generate a growth quality correlation assessment report based on the coupling analysis results to determine the synergistic growth pattern of biomass and effective components; S34. Combining the assessment report with the physiological laws of plant growth, set the threshold for judging growth stages, and complete the division of growth stages through time series data trend analysis to obtain plant growth stage data.

6. The plant growth cycle monitoring method based on edge computing according to claim 1, characterized in that, The S4 includes: S41. Extract the time information corresponding to key plant growth nodes from the timestamp data to determine the triggering time for the 3D laser scan; start the 3D laser scanning equipment to perform a comprehensive scan of the controllable environment planting area and collect spatial information. S42. Preprocess the scanned data to generate high-precision 3D point cloud data of the planting area; perform spatiotemporal registration between the plant growth stage data and the 3D point cloud data to establish the correspondence between growth stages and spatial morphology. S43. Based on the registration results, analyze the plant growth conditions at different growth stages and generate plant growth condition data. S44. Construct a growth abnormality risk prediction model, input risk characteristics, and obtain growth abnormality risk prediction data.

7. The plant growth cycle monitoring method based on edge computing according to claim 6, characterized in that, S42 includes: S421. Classify and organize the raw data acquired by 3D laser scanning, remove invalid entries, and generate a normalized scanning dataset; based on the normalized scanning dataset, use an adaptive Gaussian filtering algorithm to filter out noise points in the data and generate denoised scanning data. S422. Using a multi-view point cloud stitching algorithm, coordinate alignment and fusion are performed on the denoised scan data from different positions and angles to fill data gaps and generate preliminary stitched scan data. The overlap and deviation values ​​of the preliminary stitched data are detected by the point cloud accuracy verification model, and stitching errors exceeding the threshold are corrected to generate high-precision three-dimensional point cloud data of the planting area. S423. Extract the spatial coordinate information of key time nodes and 3D point cloud data from plant growth stage data to construct a spatiotemporal correlation basic dataset. S424. An improved ICP spatiotemporal registration algorithm is adopted to accurately match time nodes with spatial morphological data of corresponding stages, verify the consistency of matching logic, and establish a one-to-one correspondence between growth stages and spatial morphology data.

8. The plant growth cycle monitoring method based on edge computing according to claim 7, characterized in that, S422 includes: Extract spatial coordinate information and device attitude parameters from each viewpoint in the denoised scan data, establish a unified three-dimensional coordinate system benchmark, and generate coordinate benchmark calibration parameters. Based on the coordinate reference calibration parameters, a multi-view point cloud stitching algorithm is used to perform preliminary coordinate alignment on denoised data at different positions and angles, process conflict points in overlapping data areas, and generate preliminary aligned point cloud data. Spatial interpolation algorithm is used to fill in the gaps in the preliminary aligned point cloud data, filling in the information loss caused by scanning blind spots, and generating stitched data after gap filling. Start the point cloud accuracy verification model, calculate the accuracy index of the stitched data, and generate a point cloud stitching accuracy test report; For the splicing error areas in the test report that exceed the preset threshold, a combination of local re-splicing and coordinate fine-tuning is used to correct the error, generating error-corrected splicing data. The error-corrected splicing data is then subjected to a second accuracy verification. After confirming that all indicators meet the high-precision requirements, the final high-precision 3D point cloud data of the planting area is generated.

9. The plant growth cycle monitoring method based on edge computing according to claim 1, characterized in that, The S5 includes: S51. Based on the degree of influence of each risk factor on plant growth in the growth abnormality risk prediction data, the analytic hierarchy process (AHP) is used to set the weights of the risk factors; combined with the weights and growth condition data, the plant growth stability index is calculated to quantify the stability of the plant growth status. S53. Referencing the growth quality standards of the target plant, set a warning threshold for the growth stability index; when the growth stability index is lower than the warning threshold, automatically trigger the risk warning mechanism and generate plant growth cycle warning data. S53. Combine the growth quality correlation assessment report and early warning data to dynamically adjust the environmental accelerated playback parameters and key growth factors; continuously monitor the adjusted plant growth status and iteratively optimize the growth cycle parameters.

10. A plant growth cycle monitoring system based on edge computing, including: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of claims 1 to 9.

Images