Pear tree ecological index assessment method based on comprehensive data analysis
By constructing a digital twin system of resource allocation game optimizer, the resource competition and collaboration within the pear tree ecosystem are quantified. Through multi-dimensional data collection and resource allocation optimization models, the resource allocation problem within the pear tree ecosystem in existing technologies is solved, and the precise management and dynamic regulation of the pear tree ecosystem are realized.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- INST OF HORTICULTURE JIANGXI ACAD OF AGRI SCI
- Filing Date
- 2026-01-15
- Publication Date
- 2026-05-12
AI Technical Summary
Existing technologies lack precise quantitative assessment of the dynamic processes and multi-objective synergy within the pear tree ecosystem, cannot provide dynamic regulation suggestions based on real-time data, and fail to deeply explore the game processes and regulation patterns within the ecosystem behind the data.
A digital twin system with a resource allocation game optimizer at its core is constructed. Through multi-dimensional data collection, resource allocation game optimization model and reinforcement learning algorithm, the internal resource competition and collaboration of each subsystem of the pear tree are quantified, the long-term balance of multiple objectives is optimized, and ecological management strategies are generated.
It has enabled precise management of the pear tree ecosystem, significantly improved diagnostic accuracy and tree recovery speed, reduced the use of chemical fertilizers and pesticides, increased the rate of high-quality fruit and the tree's overwintering survival rate, and optimized resource allocation vectors to maintain yield and ecological balance.
Smart Images
Figure CN122022286A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of smart agriculture technology, and in particular to a method for evaluating ecological indicators of pear trees based on comprehensive data analysis. Background Technology
[0002] As one of the major economic fruit trees, the ecological and sustainable cultivation of pear trees is key to ensuring fruit safety and improving the economic and ecological benefits of orchards. Traditional pear orchard management relies heavily on farmers' experience or single agronomic indicators (such as yield and leaf nutrition), lacking precise quantitative assessment of the dynamic processes within the pear tree ecosystem and the synergy of multiple objectives.
[0003] Currently, existing technologies for pear tree ecological assessment and management mainly focus on the relationship between external inputs (such as water and fertilizer) and final outputs (if actual yield). However, they lack effective monitoring and control methods for the dynamic allocation mechanism of photosynthetic products among different functional organs (if fruit, leaves, roots, and defense tissues) within the tree, thus failing to provide actionable dynamic control recommendations based on real-time data. Furthermore, although technologies such as the Internet of Things and sensors have been applied to orchard information collection, the massive amounts of data generated are mostly used for status monitoring and simple early warning, failing to deeply explore the internal game processes and regulatory patterns of the ecosystem reflected in the data. Existing models are mostly focused on growth prediction or disease early warning, with few studies simulating the dynamic game of various functional subsystems within the tree competing for limited resources from the perspective of the entire ecosystem, and generating optimized management strategies accordingly. Therefore, this paper proposes a pear tree ecological index assessment method based on comprehensive data analysis. Summary of the Invention
[0004] The main objective of this invention is to provide a method for evaluating pear tree ecological indicators based on comprehensive data analysis. By constructing a digital twin system with a resource allocation game optimizer as its core, the invention quantifies the internal resource competition and collaboration of each subsystem of the pear tree, optimizes the long-term balance of multiple objectives, and achieves data-driven precision management, which can effectively solve the problems in the background technology.
[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: The pear tree ecological index assessment method based on comprehensive data analysis includes the following steps: Establish a multi-dimensional data collection system for pear trees to acquire tree physiological data, environmental stress data, and resource input data in real time; The physiological system of pear trees is divided into a fruit growth subsystem. Root development subsystem Disease and pest defense subsystem Symbiotic Support Subsystem A resource allocation game optimization model is constructed with each subsystem as a game participant. Based on the collected multi-dimensional data, the current capital status of each game participant is dynamically calculated. and environmental stress index ; Based on the total resource budget of pear trees As constraints, a reinforcement learning algorithm is used to solve the resource allocation game optimization model to obtain the optimal resource allocation vector. ; By comparing the optimal resource allocation vector Compared with the measured assigned vector Calculate the ecological coordination index of pear trees ; Based on the aforementioned ecological coordination indicators Optimal resource allocation vector and the current capital status of each of the game participants. It generates and outputs corresponding ecological management strategies.
[0006] Furthermore, the construction process of the resource allocation game optimization model includes: Define the resource allocation ratio of each subsystem within each time period t, and the sum of the ratios is 1; Establish the payoff function of the i-th subsystem at stage t. ; The resource allocation game optimization model is constructed with the objective of maximizing the discounted present value of the total revenue of each subsystem within growth stage T. The optimization objective is defined as follows: ,in The time discount factor, Let be the time-varying weighting coefficient of the i-th subsystem.
[0007] Furthermore, the revenue function Depending on the amount of resources allocated, the resource allocation of other subsystems, and the level of environmental stress, it includes at least: Based on the logarithmic revenue term of resource allocation, its revenue decreases as resources increase; The magnitude of the competition loss term, which is related to competition with other functional subsystems, is determined by the competition intensity matrix. An environmental stress penalty is imposed, and the penalty is triggered when the capital of a functional subsystem falls below a maintenance threshold. The profit function Defined as: ;in, For stage-related priority coefficients, The resource utilization efficiency coefficient. This is the competition sensitivity coefficient. For elements of the competition intensity matrix, This represents the environmental stress penalty coefficient. To maintain resource thresholds for the functionality of each participant. To take 0 and Maximum value operation in the context of .
[0008] Furthermore, the elements of the competition intensity matrix The settings are dynamically configured based on the collaborative or competitive relationships between the subsystems, and are dynamically adjusted according to the following rules: ,in, The benchmark competition intensity is given by k, which is an adjustment coefficient. Capital status scaling factor; Furthermore, based on the synergistic relationship between the fruit growth subsystem and the symbiotic support subsystem, the following is set: ; Based on the strong competitive relationship between the root development subsystem and the disease and pest resistance subsystem, the following is set: .
[0009] Furthermore, the time-varying weighting coefficients The adjustments are made dynamically based on the degree of environmental stress, and the adjustment rules are as follows: ; in, Let i be the base weight of the i-th game participant. and For stress response parameters, For indicator functions, Let be the coercion response threshold for the i-th game participant.
[0010] Furthermore, the step of using a reinforcement learning algorithm to solve the resource allocation game optimization model includes: The upper-level central coordinator takes the capital status, environmental stress index, phenological period, and total resources of each subsystem as inputs, and its state space... Encoding phenological periods using resource allocation vectors For output; the reward function of the upper-level central coordinator. The definition is based on a comprehensive consideration of the current revenue, expected future revenue, and stability of each capital state of each functional subsystem. The lower-level subsystem strategy network takes the capital status, environmental stress, previous stage allocation and benefits of each subsystem as inputs, and outputs the desired resource ratio.
[0011] Furthermore, the reinforcement learning algorithm iteratively updates the parameters of the upper-level central coordinator and the lower-level subsystem policy network through a fixed-policy alternating optimization method.
[0012] Furthermore, the reward function of the upper-level central coordinator Defined as: ; in, Weights for expected future earnings. This is the penalty coefficient for capital volatility. It is a function of standard deviation. Let be the value function, representing the estimated cumulative expected return that the system can obtain from the current time step to a future period.
[0013] Furthermore, the pear tree ecological harmony index The calculation is performed using the following formula: ,in, The measured assigned vector represents the Euclidean norm. Obtained by carbon isotope tracing or by inferring the growth of each organ; Furthermore, when When this is determined to be an imbalance in ecological coordination, a corresponding ecological management strategy is generated and output.
[0014] Furthermore, the steps for generating the ecological management strategy include: Calculate the allocation bias of the i-th game participant. ; The pre-defined prescription rules are matched according to the allocation deviation; The prescription is adjusted and quantified in a personalized manner, taking into account tree age, variety characteristics, and agricultural feasibility. The final prescription is converted into augmented reality guidance instructions to guide field operations; Furthermore, the prescription rules include at least: Rule 1: When and At that time, a root-promoting management prescription is generated. The target capital state for the second game participant; Rule 2: When When the disease index in D(t) is greater than 0.6, a defense enhancement prescription is generated; Rule 3: When When the value is less than 0.1 for three consecutive stages, a symbiotic system repair prescription is generated.
[0015] A pear tree ecological indicator assessment system based on comprehensive data analysis, used to implement the aforementioned pear tree ecological indicator assessment method based on comprehensive data analysis, includes: A multi-source data acquisition module is used to acquire tree physiological, environmental stress, and resource input data in real time; The game modeling and solving module includes: This is a unit for constructing a resource allocation game optimization model with the objective of maximizing the discounted present value of the total revenue of each subsystem within growth stage T. This is used to solve a resource allocation game optimization model using reinforcement learning algorithms, in order to obtain the optimal resource allocation vector. Hierarchical reinforcement learning to solve units; The ecological index calculation module is used to compare the optimal resource allocation vector. Compared with the measured assigned vector Calculate the ecological coordination index of pear trees ; The prescription generation module is used to manage prescriptions based on ecological harmony indicators. Optimal resource allocation vector and the current capital status of each game participant Generate and output corresponding ecological management strategies; The visualization and interaction module is used to show users the assessment results of pear tree ecological indicators and provide augmented reality task guidance.
[0016] A computer-readable storage medium includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements the pear tree ecological index evaluation method based on resource allocation game theory and dynamic optimization.
[0017] The present invention has the following beneficial effects: Compared with existing technologies, this solution uses a game theory model to quantify in real time the distribution ratio of photosynthetic products among four key functional subsystems: fruit growth, root development, disease resistance, and symbiotic support, and outputs specific distribution ratio values. This breaks through the limitations of traditional agronomy's understanding of the "black box" of resource allocation within the tree.
[0018] Compared with existing technologies, this solution transforms the abstract tree imbalance into a quantitative score by calculating ecological coordination indicators. By using ecological coordination indicators to set the conditions for judging ecological imbalance, it can automatically diagnose the causes of imbalance, thereby achieving accurate identification and early warning of the root causes of ecosystem imbalance. Compared with traditional experience-based diagnosis, it can advance the imbalance warning by 1-2 phenological stages, significantly improving the accuracy of diagnosis.
[0019] Compared with existing technologies, this solution guides tree investment in long-term capital by solving the optimal resource allocation vector and generating an agronomic resource scheduling strategy, thereby enhancing the resilience of the pear tree system. Pear orchards managed using this method show significantly faster tree recovery when faced with the same degree of drought or disease stress, and the amount of emergency use of chemical fertilizers and pesticides is significantly reduced.
[0020] Compared with existing technologies, this solution, through the reward function of the central coordinator, enables the model to automatically find the optimal balance point of multiple objectives such as yield, quality, tree health, and resource efficiency, avoiding the sacrifice of other objectives for a single objective (such as maximum yield). While maintaining yield, it can improve the rate of high-quality fruit and the overwintering survival rate of trees, and significantly improve the overall ecological index of the orchard. Attached Figure Description
[0021] Figure 1 This is a flowchart illustrating the pear tree ecological index assessment method based on comprehensive data analysis according to the present invention. Figure 2 This is a schematic diagram of the pear tree ecological index assessment system based on comprehensive data analysis, as described in this invention. Detailed Implementation
[0022] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. Example 1
[0023] See Figure 1 The flowchart shown is a method for evaluating pear tree ecological indicators based on comprehensive data analysis according to the present invention, which includes the following steps: Step 1: Establish a multi-dimensional data collection system for pear trees to acquire tree physiological data, environmental stress data, and resource input data in real time.
[0024] In one possible implementation, the following specific steps can be taken: Step 1.1: Network Planning and Equipment Deployment Step 1.11: Grid-based monitoring unit division 1) Establishment of geographical baselines: The entire park was surveyed using RTK drones at the centimeter level, generating a digital elevation model (DEM) and orthophotos.
[0025] Based on topography, soil maps, and variety zoning, the pear orchard is divided into rectangular or hexagonal monitoring grids with an area of 3-5 mu.
[0026] Assign a unique code to each grid (e.g.) (and mark the GPS coordinates of the center point).
[0027] 2) Selection of representative sample tree: Within each grid, select 5-8 pear trees that are growing normally and can represent the average condition of that grid as fixed monitoring sample trees in a "Z" shape.
[0028] Each sample tree is associated with a unique electronic ID tag (RFID or QR code).
[0029] Step 1.12: Layered Deployment of Hardware Devices Standardize deployment within each monitoring grid according to the following checklist: a) Aerial monitoring nodes (1 set per grid, flying as needed) Multi-rotor drone (equipped with multispectral / hyperspectral camera and thermal infrared camera) Fixed flight paths and altitudes ensure consistent image resolution.
[0030] b) Canopy and environmental monitoring nodes (1 set per grid, fixed installation) Micro-weather station: Installed between rows, 2.5 meters high. Monitors: air temperature and humidity, light intensity, wind speed and direction, and rainfall.
[0031] Canopy multispectral camera: overhead view of the canopy, fixed angle. Monitoring: Normalized Difference Vegetation Index (NDVI), Photochemical Reflectance Index (PRI), and Water Stress Index (NDWI).
[0032] Insect acoustic sensor / intelligent insect monitoring lamp: monitors the acoustic spectrum of insect activity or automatically identifies and attracts images.
[0033] c) Tree trunk and fruit monitoring nodes (1 set per sample tree) Tree trunk flow meter: installed at the base of the tree trunk to monitor instantaneous and cumulative evapotranspiration water consumption.
[0034] Fruit enlargement sensor: Install one standard fruit in each of the four cardinal directions of the sample tree to monitor the daily changes in fruit diameter.
[0035] Tree tilt sensor: monitors changes in tree posture caused by load or wind.
[0036] d) Rhizosphere soil monitoring nodes (3-5 points per grid, buried in layers) Soil three-parameter sensor: buried at depths of 10cm, 30cm, and 50cm to monitor soil volumetric water content, temperature, and electrical conductivity (EC).
[0037] Soil solution sampler: Buried in the main root layer, it can be connected to an automatic sampler to periodically collect soil solution for ion analysis.
[0038] e) Resource Input Record Node Intelligent water and fertilizer integrated machine: A flow meter and EC / pH monitor are installed at the irrigation head to automatically record the amount of water, fertilizer and time of each irrigation.
[0039] Agricultural operation handheld terminal: Equipped to each operator, used to scan codes to record details of operations such as pruning, spraying, and fruit thinning (time, location, operation type, dosage).
[0040] Step 1.13: Construction of Communication and Power Supply Networks 1) Communication Network: Deploy LoRaWAN as the backbone network for long-distance, low-power sensor data transmission. Deploy gateways at the center of the campus or high points. Key data nodes (such as weather stations) can be supplemented with 4G / 5G backhaul.
[0041] 2) Power supply system: Fixed nodes will be powered primarily by solar panels and batteries. The power generation during the month with the lowest local sunshine hours will be calculated to ensure continuous equipment operation.
[0042] 3) Edge computing nodes: Deploy an edge computing gateway in each grid or every few grids to be responsible for the initial processing, caching and protocol conversion of local data.
[0043] Step 1.2: Standardized Data Acquisition Process (Automation and Periodic Execution) Step 1.21: Automating high-frequency data streams (real-time / daily) The following process is executed automatically by the device, and data is uploaded periodically via wireless network: 1) Environmental and tree physiological data (collected every 15-30 minutes) Trigger: System clock or threshold trigger (such as sudden temperature change).
[0044] Action: All sensors are awakened and data is read in sequence.
[0045] Contents: air / soil temperature and humidity, light intensity, stem flow, fruit diameter, soil EC.
[0046] Transmission: Data is packaged and sent to the edge gateway via LoRaWAN.
[0047] 2) Canopy image data (automatically collected 1 hour before and after noon each day, under clear and cloudless conditions) Triggered: Light intensity > And there was no rainfall.
[0048] Action: The canopy multispectral camera automatically takes pictures, and the drone automatically takes off and takes pictures along a preset route.
[0049] Processing: The edge gateway automatically calculates vegetation indices such as NDVI and PRI.
[0050] Transmission: Thumbnails and index data are uploaded, while the original image is temporarily stored locally.
[0051] 3) Insect infestation data (automatically collected daily at dusk) Trigger: Light intensity is below the threshold.
[0052] Action: The insect monitoring light is turned on, automatically taking pictures and using AI to identify and count them.
[0053] Transmission: Upload the identification results (type and quantity of pests).
[0054] Step 1.22: Manually assisted standardized sampling (periodic) The process is carried out by technicians according to a fixed plan, and recorded using a dedicated app. 1) Leaf nutrient diagnostic sampling (once every 14 days, 8-10 am) Location: Within each grid, select the nutrient branches in the middle of the outer canopy of the sample tree that are of moderate growth.
[0055] Method: Collect 8-10 leaves from each tree and mix them into one sample. Use the APP to scan the code and record the tree ID, grid number, and sampling time.
[0056] Processing: The sample is quickly sent to the laboratory for determination of nitrogen, phosphorus, potassium and trace elements content.
[0057] 2) Rhizosphere soil and microbial sampling (every 30 days, or during key phenological periods) Location: Near the drip line of the sample tree, use a soil drill to collect soil samples from the 0-20cm layer.
[0058] Method: Three samples were taken from each grid and mixed. One sample was used for routine physicochemical analysis (pH, organic matter, etc.), and the other was stored in a low-temperature incubator for microbial DNA sequencing.
[0059] Records: The app records photos, locations, and times of sampling points.
[0060] 3) Tree capital status verification sampling (key phenological periods: full bloom, fruit enlargement, and post-harvest) Defense substances: Collect specific tissues (such as bark and leaves) and determine the content of secondary metabolites such as phenols and flavonoids.
[0061] Symbiotic state: Fine root samples were collected and mycorrhizal infection rate was examined under a microscope.
[0062] Carbon allocation verification: passed 13 C-pulse labeling experiments (1-2 times per season) to track the fate of photosynthetic products.
[0063] Step 1.23: Recording agricultural operations and resource input data (event triggering) Before operation: Technicians select the job type (such as "spraying", "fertilizing", "pruning") on the handheld terminal APP, and the system automatically generates a work order with a QR code.
[0064] During operation: Upon reaching the work grid, scan the grid QR code or tree ID to confirm the location. While performing the operation, a Bluetooth-connected smart device (such as a flow meter) automatically records the usage, or the usage can be manually entered.
[0065] After the operation: The APP automatically uploads the record, including: operation type, execution time, specific location (grid / tree ID), input name and quantity, and operator.
[0066] Step 1.3: Data aggregation, quality control, and preprocessing (automatically executed at the platform level) Step 1.31: Data Access and Cleaning (Real-time) Access: The edge gateway converts the received data into JSON format and uploads it to the cloud platform or local server via the MQTT protocol.
[0067] Cleaning rules: Range test: Check whether the data is within a reasonable physical range (e.g., soil moisture: 0-100%, temperature: 20-60℃).
[0068] Mutation detection: Using sliding window statistics, data points with sudden and drastic changes are removed.
[0069] Missing items handling: Short-term missing items are handled by linear interpolation, while long-term missing items are marked and trigger device check alarms.
[0070] Step 1.32: Spatiotemporal Alignment and Fusion (Daily Scheduled Task) Time alignment: unify the timestamps of all data to Beijing time and resample to a standard time series (e.g., one data point per hour).
[0071] Spatial matching: Point sensor data (such as soil moisture) are used to generate a continuous spatial distribution map covering the entire grid through Kriging interpolation.
[0072] Data fusion: Cross-validate and fuse data on the same indicator from different sources (such as precipitation data from weather stations and satellites) to generate the optimal estimate.
[0073] Step 1.33: Calculation of key derived metrics (calculated as needed) The platform automatically calculates the input metrics required for the game theory model: Total resource budget =Estimated daily photosynthetic assimilation - daily respiratory consumption.
[0074] The environmental stress index D(t) = f(moisture stress index, pest and disease stress index, nutrient stress index, temperature stress index). Each sub-index is calculated from the raw data.
[0075] Participant Capital Status C ᵢThe proxy index for (t) is: for example, the fruit enlargement rate can be used to proxy C1, and the soil moisture absorption efficiency can be used to proxy C2.
[0076] Step 2: Divide the pear tree physiological system into fruit growth subsystem P1, root development subsystem P2, disease and pest resistance subsystem P3, and symbiotic support subsystem P4, and construct a resource allocation game optimization model with each subsystem as a game participant.
[0077] The construction process of the resource allocation game optimization model includes: Define the resource allocation ratio of each subsystem within each time period t, and the sum of the ratios is 1; Establish the payoff function of the i-th subsystem at stage t. ; A resource allocation game optimization model is constructed with the goal of maximizing the discounted present value of the total revenue of each subsystem within growth stage T. The optimization objective is defined as follows: ,in The time discount factor, Let be the time-varying weighting coefficient of the i-th subsystem.
[0078] Payoff function Depending on the amount of resources allocated, the resource allocation of other subsystems, and the level of environmental stress, it includes at least: Based on the logarithmic revenue term of resource allocation, its revenue decreases as resources increase; The magnitude of the competition loss term, which is related to competition with other functional subsystems, is determined by the competition intensity matrix. An environmental stress penalty is imposed, and the penalty is triggered when the capital of a functional subsystem falls below a maintenance threshold. Payoff function Defined as: ; in, For stage-related priority coefficients, The resource utilization efficiency coefficient. This is the competition sensitivity coefficient. For elements of the competition intensity matrix, This represents the environmental stress penalty coefficient. To maintain resource thresholds for the functionality of each participant. To take 0 and Maximum value operation in the context of .
[0079] Competition intensity matrix elements The settings are dynamically configured based on the collaborative or competitive relationships between the subsystems, and are dynamically adjusted according to the following rules: ,in, The benchmark competition intensity is given by k, which is an adjustment coefficient. Capital status scaling factor; Specifically: Based on the synergistic relationship between the fruit growth subsystem and the symbiotic support subsystem, the following is set: ; Based on the strong competition between the root development subsystem and the disease and pest resistance subsystem, the following is set .
[0080] Time-varying weighting coefficients The adjustments are made dynamically based on the degree of environmental stress, and the adjustment rules are as follows: ; in, Let i be the base weight of the i-th game participant. and For stress response parameters, For indicator functions, Let be the coercion response threshold for the i-th game participant.
[0081] Step 3: Based on the collected multi-dimensional data, dynamically calculate the current capital status of each game participant. and environmental stress index .
[0082] Among them, the ecological coordination index of pear trees The calculation is performed using the following formula: ,in, Represents the Euclidean norm, and the measured assigned vector. Obtained by carbon isotope tracing or by inferring the growth of each organ; Specifically: when When this is determined to be an imbalance in ecological coordination, a corresponding ecological management strategy is generated and output.
[0083] In one possible implementation, the environmental stress index The calculation can be performed using the following procedure: Step 3.1: Define and calculate the four basic stress factor indices First, based on the collected multi-dimensional data, four basic stress factor indices that affect all game participants are calculated. Each index has a value range of [0,1], including: 1) Water Stress Index (WSI) Data sources: soil volumetric water content (θ, multi-layer sensor), atmospheric vapor pressure deficit (VPD, calculated from temperature and humidity), and trunk flow rate (SF); calculation formulas are as follows: ; in The water content at the wilting point. It refers to field water holding capacity. Vapor pressure deficit, which is the difference between the saturated water vapor pressure of the air and the actual water vapor pressure, reflects the dryness of the atmosphere and the ability of plants to "pump water" from their leaves. When the critical vapor pressure is deficient, pear trees begin to actively close their stomata to reduce excessive water loss. If this value is exceeded, the stomatal conductance will decrease significantly, and photosynthesis will be inhibited. The real-time stem flow rate is the amount of water transported upward through the xylem of the trunk per unit time, which directly reflects the actual transpiration water consumption intensity of the tree. The maximum potential stem flow rate is the maximum transpiration rate that a pear tree can reach during this phenological period under conditions of sufficient water and optimal atmospheric conditions (light and VPD), representing the theoretical capacity of the tree's water transport system. Based on the basal / nighttime stem flow rate, the weak sap flow is mainly caused by root pressure and trunk tissue water rebalancing when transpiration almost stops at night; All are weighting coefficients.
[0084] 2) Biological Stress Index (BSI) Data sources: Automatic pest and disease monitoring image recognition results (number of pests, area of lesions), pest activity frequency identified by insect voiceprint sensors, and spore trap data (optional). Calculation method: The ratio of the occurrence degree of major pests (pear psyllid, pear fruit moth) and diseases (black spot disease, rust disease) to the preset economic threshold is calculated, and the maximum value is taken as the biological stress index.
[0085] 3) Temperature Stress Index (TSI) Data source: Air temperature (T) air Fruit surface temperature (T) fruit Thermal infrared image inversion), soil temperature (T) soil ).
[0086] Calculation method: ; in, These are the optimal, minimum, and maximum tolerable temperatures for pear tree growth (which vary with phenological stages).
[0087] 4) Nutritional Stress Index (NSI) Data sources: Leaf nutrient diagnostic data (N, P, K content, etc.), soil EC value (reflecting salt and total ion content), Soil solution ion concentration (optional) Calculation method: ; in, Real-time soil electrical conductivity (EC) reflects the total concentration of soluble salt ions in the soil solution and is a general indicator of the degree of salinization. Excessively high EC values can lead to osmotic stress and ion poisoning (such as...). It affects root water absorption and disrupts nutrient balance. It is monitored in real time by soil EC sensors buried in the root layer (e.g., at a depth of 20cm), which are usually integrated with soil moisture and temperature sensors. The salt stress threshold for pear trees is the critical soil EC value at which pear tree growth begins to be significantly inhibited. Exceeding this value, salt becomes one of the main stress factors. The value was determined by consulting the literature: pear trees are moderately salt-tolerant fruit trees, with a typical threshold range of 1.5 - 2.5 dS / m (saturated extract method). For any nutrient element X (such as N, P, K, Ca, Mg, Fe, Zn, B, etc.), the deficiency response function is defined as follows: , The real-time concentration of element X in the leaves is the actual content of element X in the functional leaves of the pear tree at time t. It reflects the tree's absorption and internal inventory of the element and is the most direct indicator of nutritional status. It is obtained through periodic leaf sampling and laboratory analysis. The critical concentration is the minimum leaf element concentration required to ensure the normal growth and development of pear trees under a specific phenological stage Φ(t). Below this value, visible deficiency symptoms will appear and physiological functions will be affected. It can usually be determined by field test calibration: set different fertilization level gradients, establish the relationship curve between yield / growth and leaf concentration, and set the concentration corresponding to a 5-10% decrease in yield as the critical value, or by consulting reference values in the literature.
[0088] Step 3.2: Determine the sensitivity weight matrix of each participant to coercion. In one possible implementation, a 4 (participant) x 4 (stress factor) sensitivity weight matrix W can be established, as shown in the table below:
[0089] It should be noted that: Weight assignment basis: P1 (Fruit): Extremely sensitive to moisture, pests and diseases, and extreme temperatures (sunburn, frost damage), with high weighting.
[0090] P2 (root system): Directly affected by soil moisture and nutrient availability, with a high weight; not directly sensitive to above-ground pests and diseases and temperature.
[0091] P3 (Defense): Biological stress is its "trigger" and requires special handling (see below). Other stresses indirectly affect it by consuming tree resources.
[0092] P4 (symbiotic): Mycorrhizal fungi and the like are sensitive to soil moisture and nutrient conditions, but not directly sensitive to stress on the aboveground parts.
[0093] Note: Some fungicides can harm symbiotic microorganisms. This is reflected in P4 as an independent input through the "Agricultural Operation Record".
[0094] Weight normalization: The weights of each row need to be normalized so that their sum is 1, representing the relative importance of each coercion to the participant.
[0095] Step 3.3: Calculate the participant-specific stress index Dᵢ(t) 1) For P1, P2, P4: Linear weighted model: ; in, Represent It is the corresponding element in the sensitivity weight matrix W.
[0096] 2) For P3 (disease and pest resistance subsystem): Nonlinear triggering and amplification model P3 itself is designed to respond to biological stress, therefore its stress index is... It should reflect "the load that the defense system needs to withstand in response to all current threats," and the calculation method is as follows: ; It should be noted that: For the baseline load, the biological stress index (BSI)(t) is directly used. The more severe the biological stress, the more resources the defense system needs to mobilize.
[0097] For indirect pressure, the coercions (D1, D2, D4) experienced by other subsystems will compete with the defense system for resources, thereby aggravating the "internal pressure" of the defense system; η is an amplification factor (e.g., 0.2-0.3).
[0098] Step 3.4: Dynamic Adjustment and Phenological Period Correction 1) Phenological period sensitivity correction: During the flowering period, the sensitivity weight of P1 (at this time, the floral organ) to low temperature stress increases sharply.
[0099] During the fruit enlargement period, the sensitivity weight of P1 to water stress reaches its peak.
[0100] During the later stages of mining, P2 The importance of (root growth) should be increased to prepare for overwintering.
[0101] Implementation: Introduce the phenological period correction factor matrix M(Φ(t)) and perform a Hadamard product (element-by-element multiplication) with the basic weight matrix W to obtain the dynamic weights. 2) Stress memory effect: Continuous stress will cause cumulative damage. Therefore, D ᵢ The actual input value of (t) can be filtered using a first-order hysteresis filter: ; in, The memory decay factor (e.g., 0.7) enables the model to "remember" the stress experienced in the previous days.
[0102] Step 4: Budgeting the total resources for pear trees As constraints, a reinforcement learning algorithm is used to solve the resource allocation game optimization model to obtain the optimal resource allocation vector. .
[0103] Among them, the resource allocation game optimization model is solved using reinforcement learning algorithms, including: The upper-level central coordinator takes the capital status, environmental stress index, phenological period, and total resources of each subsystem as inputs, and its state space... Encoding phenological periods using resource allocation vectors For output; the reward function of the upper-level central coordinator. The definition is based on a comprehensive consideration of the current revenue, expected future revenue, and stability of each capital state of each functional subsystem. The lower-level subsystem strategy network takes the capital status, environmental stress, previous stage allocation and benefits of each subsystem as input, and outputs the expected resource ratio. Reinforcement learning algorithms iteratively update the parameters of the upper-level central coordinator and the lower-level subsystem policy network by alternately optimizing a fixed policy. Reward function of the upper-level central coordinator Defined as: ; in, Weights for expected future earnings. This is the penalty coefficient for capital volatility. It is a function of standard deviation. Let be the value function, representing the estimated cumulative expected return that the system can obtain from the current time step to a future period.
[0104] Step 5: Compare the optimal resource allocation vector Compared with the measured assigned vector Calculate the ecological coordination index of pear trees .
[0105] pear tree ecological coordination index The calculation is performed using the following formula: ,in, Represents the Euclidean norm, and the measured assigned vector. Obtained through carbon isotope tracing or by inferring growth from individual organs; when When this is determined to be an imbalance in ecological coordination, a corresponding ecological management strategy is generated and output.
[0106] Step 6: Based on ecological coordination indicators Optimal resource allocation vector and the current capital status of each game participant It generates and outputs corresponding ecological management strategies.
[0107] The steps involved in generating an ecological management strategy include: Calculate the allocation bias of the i-th game participant. ; Pre-defined prescription rules are matched based on the allocation deviation; The prescription is adjusted and quantified in a personalized manner, taking into account tree age, variety characteristics, and agricultural feasibility. The final prescription is converted into augmented reality guidance instructions to guide field operations; The prescription rules include at least the following: Rule 1: When and At that time, a root-promoting management prescription is generated. The target capital state for the second game participant; Rule 2: When When the disease index in D(t) is greater than 0.6, a defense enhancement prescription is generated; Rule 3: When When the value is less than 0.1 for three consecutive stages, a symbiotic system repair prescription is generated.
[0108] Example 2: This invention also provides a pear tree ecological indicator assessment system based on comprehensive data analysis. This system is used to implement the aforementioned pear tree ecological indicator assessment method based on comprehensive data analysis. (See also...) Figure 2 The system architecture diagram shown includes: A multi-source data acquisition module is used to acquire tree physiological, environmental stress, and resource input data in real time; The game modeling and solving module includes: This is a construction unit for building a resource allocation game optimization model with the goal of maximizing the discounted value of the total revenue of each subsystem within growth stage T. This is used to solve a resource allocation game optimization model using reinforcement learning algorithms, in order to obtain the optimal resource allocation vector. Hierarchical reinforcement learning to solve units; The ecological index calculation module is used to compare the optimal resource allocation vector. Compared with the measured assigned vector Calculate the ecological coordination index of pear trees ; The prescription generation module is used to manage prescriptions based on ecological harmony indicators. Optimal resource allocation vector and the current capital status of each game participant Generate and output corresponding ecological management strategies; The visualization and interaction module is used to show users the assessment results of pear tree ecological indicators and provide augmented reality task guidance.
[0109] Example 3: The present invention also provides a computer-readable storage medium, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the above-mentioned method for evaluating pear tree ecological indicators based on resource allocation game theory and dynamic optimization is implemented.
[0110] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.
Claims
1. A method for evaluating ecological indicators of pear trees based on comprehensive data analysis, characterized in that: Includes the following steps: Establish a multi-dimensional data collection system for pear trees to acquire tree physiological data, environmental stress data, and resource input data in real time; The physiological system of pear trees is divided into a fruit growth subsystem. Root development subsystem Disease and pest defense subsystem Symbiotic Support Subsystem A resource allocation game optimization model is constructed with each subsystem as a game participant. Based on the collected multi-dimensional data, the current capital status of each game participant is dynamically calculated. and environmental stress index ; Based on the total resource budget of pear trees ; Using the total resource budget of the pear tree as a constraint, a reinforcement learning algorithm is employed to solve the resource allocation game optimization model, thereby obtaining the optimal resource allocation vector. ; By comparing the optimal resource allocation vector Compared with the measured assigned vector Calculate the ecological coordination index of pear trees ; Based on the aforementioned ecological coordination indicators Optimal resource allocation vector and the current capital status of each of the game participants. It generates and outputs corresponding ecological management strategies.
2. The method for evaluating pear tree ecological indicators based on comprehensive data analysis according to claim 1, characterized in that, The construction process of the resource allocation game optimization model includes: Define the resource allocation ratio of each subsystem within each time period t, and the sum of the ratios is 1; Establish the payoff function of the i-th subsystem at stage t. ; The resource allocation game optimization model is constructed with the objective of maximizing the discounted present value of the total revenue of each subsystem within growth stage T. The optimization objective is defined as follows: ,in The time discount factor, Let be the time-varying weighting coefficient of the i-th subsystem.
3. The method for evaluating pear tree ecological indicators based on comprehensive data analysis according to claim 2, characterized in that, The profit function Depending on the amount of resources allocated, the resource allocation of other subsystems, and the level of environmental stress, it includes at least: Based on the logarithmic revenue term of resource allocation, its revenue decreases as resources increase; The magnitude of the competition loss term, which is related to competition with other functional subsystems, is determined by the competition intensity matrix. An environmental stress penalty is imposed, and the penalty is triggered when the capital of a functional subsystem falls below a maintenance threshold. The profit function Defined as: ; in, For stage-related priority coefficients, The resource utilization efficiency coefficient. This is the competition sensitivity coefficient. For elements of the competition intensity matrix, This represents the environmental stress penalty coefficient. To maintain resource thresholds for the functionality of each participant. To take 0 and Maximum value operation in the context of .
4. The method for evaluating pear tree ecological indicators based on comprehensive data analysis according to claim 3, characterized in that, The elements of the competition intensity matrix The settings are dynamically configured based on the collaborative or competitive relationships between the subsystems, and are dynamically adjusted according to the following rules: ,in, Let k be the baseline competitive strength. Capital status scaling factor; Specifically: Based on the synergistic relationship between the fruit growth subsystem and the symbiotic support subsystem, the following is set: ; Based on the strong competitive relationship between the root development subsystem and the disease and pest resistance subsystem, the following is set: .
5. The method for evaluating pear tree ecological indicators based on comprehensive data analysis according to claim 2, characterized in that, The time-varying weighting coefficient The adjustments are made dynamically based on the degree of environmental stress, and the adjustment rules are as follows: ; in, Let i be the base weight of the i-th game participant. and For stress response parameters, For indicator functions, Let be the coercion response threshold for the i-th game participant.
6. The method for evaluating pear tree ecological indicators based on comprehensive data analysis according to claim 1, characterized in that, The method of solving the resource allocation game optimization model using reinforcement learning algorithms includes: The upper-level central coordinator takes the capital status, environmental stress index, phenological period, and total resources of each subsystem as inputs, and its state space... Encoding phenological periods using resource allocation vectors For output; the reward function of the upper-level central coordinator. The definition is based on a comprehensive consideration of the current revenue, expected future revenue, and stability of each capital state of each functional subsystem. The lower-level subsystem strategy network takes the capital status, environmental stress, previous stage allocation and benefits of each subsystem as input, and outputs the expected resource ratio. The reinforcement learning algorithm iteratively updates the parameters of the upper-level central coordinator and the lower-level subsystem policy network through a fixed-policy alternating optimization method. The reward function of the upper-level central coordinator Defined as: ; in, Weights for expected future earnings. This is the penalty coefficient for capital volatility. It is a function of standard deviation. Let be the value function, representing the estimated cumulative expected return that the system can obtain from the current time step to a future period.
7. The method for evaluating pear tree ecological indicators based on comprehensive data analysis according to claim 1, characterized in that, The pear tree ecological coordination index The calculation is performed using the following formula: ,in, The measured assigned vector represents the Euclidean norm. Obtained by carbon isotope tracing or by inferring the growth of each organ; Specifically: when When this is determined to be an imbalance in ecological coordination, a corresponding ecological management strategy is generated and output.
8. The method for evaluating pear tree ecological indicators based on comprehensive data analysis according to claim 1, characterized in that, The steps for generating the ecological management strategy include: Calculate the allocation bias of the i-th game participant. ; The pre-defined prescription rules are matched according to the allocation deviation; The prescription is adjusted and quantified in a personalized manner, taking into account tree age, variety characteristics, and agricultural feasibility. The final prescription is converted into augmented reality guidance instructions to guide field operations; The prescription rules include at least the following: Rule 1: When and At that time, a root-promoting management prescription is generated. The target capital state for the second game participant; Rule 2: When When the disease index in D(t) is greater than 0.6, a defense enhancement prescription is generated; Rule 3: When When the value is less than 0.1 for three consecutive stages, a symbiotic system repair prescription is generated.
9. A pear tree ecological index assessment system based on comprehensive data analysis, characterized in that, The system is used to implement the pear tree ecological index assessment method based on comprehensive data analysis as described in any one of claims 1-8, including: A multi-source data acquisition module is used to acquire tree physiological, environmental stress, and resource input data in real time; The game modeling and solving module includes: This is a construction unit for building a resource allocation game optimization model with the goal of maximizing the discounted value of the total revenue of each subsystem within growth stage T. This is used to solve a resource allocation game optimization model using reinforcement learning algorithms, in order to obtain the optimal resource allocation vector. Hierarchical reinforcement learning to solve units; The ecological index calculation module is used to compare the optimal resource allocation vector. Compared with the measured assigned vector Calculate the ecological coordination index of pear trees ; The prescription generation module is used to manage prescriptions based on ecological harmony indicators. Optimal resource allocation vector and the current capital status of each game participant Generate and output corresponding ecological management strategies; The visualization and interaction module is used to show users the assessment results of pear tree ecological indicators and provide augmented reality task guidance.
10. A computer-readable storage medium, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements the pear tree ecological index evaluation method based on resource allocation game theory and dynamic optimization as described in any one of claims 1-8.