Intelligent fruit tree planting monitoring method and system

Abstract

The invention relates to an intelligent fruit tree planting monitoring method and system, and belongs to the technical field of intelligent agriculture. According to the scheme, a multi-source sensor network is deployed to collect soil, crop and environment data, a salinity-moisture collaborative stress decision-making model is constructed, and parameters such as soil salinity, canopy temperature and weather forecast are deeply fused; dynamically generating a non-fixed precise irrigation instruction; meanwhile, a low-visibility robust vision algorithm is integrated, and it is ensured that fruits and diseases and pests can still be reliably recognized in severe weather; by establishing a collaborative scheduling mechanism, efficient linkage of water and fertilizer management, plant protection operation and farming operation is realized. According to the method, the technical span from single threshold judgment to multi-factor intelligent decision is realized, the utilization efficiency of water resources and fertilizers is effectively improved, the production cost is remarkably reduced, and the mode of fruit tree planting is promoted to be converted from traditional experience management to data-driven intelligent management.

Description

Technical Field

[0001] This invention relates to the field of smart agriculture technology, specifically to an intelligent fruit tree planting monitoring method and system. Background Technology

[0002] As a crucial component of agricultural production, the level of intelligence in fruit tree cultivation management directly impacts fruit quality, resource utilization efficiency, and overall economic benefits. Traditional fruit tree cultivation management relies heavily on growers' personal experience for agricultural decisions, inherently suffering from high subjectivity, low standardization, and inability to be replicated on a large scale. With the development of modern agricultural technology, particularly the application of the Internet of Things (IoT) in agriculture, automated monitoring systems based on environmental sensors have emerged. These systems, by deploying single or a small number of sensors for soil moisture, temperature, and other parameters, collect data on specific environmental parameters and can automatically trigger operations such as irrigation based on preset threshold values.

[0003] However, existing technological solutions still have significant limitations. First, their decision-making logic is usually based on a single or a few isolated environmental parameters (e.g., soil moisture alone), failing to comprehensively consider the synergistic effects of multiple factors such as soil salinity, crop physiological state, and short-term weather changes. This piecemeal, single-dimensional decision-making model cannot address complex coupled environmental problems such as salt and water stress, and may even exacerbate secondary soil salinization due to improper irrigation. Second, the system's perception capabilities perform poorly in complex and variable field environments. Conventional visual recognition technologies experience a sharp decline in accuracy under low visibility conditions such as fog, rain, and low light, leading to missing or unreliable monitoring data. Third, automated systems for different agricultural processes such as water and fertilizer management, pest and disease control, and weeding are often independent of each other, forming "information silos" and "execution silos," lacking efficient cross-subsystem collaboration mechanisms and limiting further improvements in overall management efficiency.

[0004] Therefore, there is an urgent need in this field for an intelligent fruit tree planting monitoring solution that can deeply integrate multi-source information, simulate expert decision-making logic, possess forward-looking judgment capabilities, and coordinate the scheduling of various agricultural equipment, so as to achieve a fundamental shift from traditional experience-based management to data-driven intelligent management. Summary of the Invention

[0005] To address the problems of existing technologies, this invention provides an intelligent fruit tree planting monitoring method and system.

[0006] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: Firstly, an intelligent fruit tree planting monitoring method, the method comprising the following steps: Data acquisition steps: Collect multi-source data on the fruit tree's growth environment using deployed sensors. This multi-source data includes at least soil volumetric water content from the root zone of the fruit trees. Soil soluble salt content Temperature of fruit tree canopy and air temperature from meteorological data sources and short-term forecast rainfall ; Data fusion and decision-making steps: The multi-source data is input into the salinity-moisture synergistic stress decision-making model, which performs the following calculation process: Based on soil soluble salt content With crop salt tolerance threshold The difference is used to calculate the required rinsing water volume. : in, The rinsing efficiency coefficient is... For irrigated area, The depth of the root layer; Based on the temperature of the fruit tree canopy With air temperature The difference is used to calculate the crop water stress index. And based on this, determine the physiological water requirement. : in, and These are the baseline values ​​for canopy-air temperature difference under conditions of full irrigation and severe drought, respectively. The total water required for rinsing Physiological water requirement and short-term forecast rainfall Generate the final non-fixed-volume irrigation control instructions. : in, Rainfall utilization coefficient; Collaborative execution steps: Drive the integrated water and fertilizer irrigation equipment to execute the irrigation control commands. .

[0007] In one specific implementation of the first aspect, the data acquisition step further includes: Image data of fruit trees is collected through cameras deployed in the orchard and equipped with a dynamic convolutional neural network model that has been enhanced by conditions of haze and low light. The dynamic convolutional neural network model improves feature extraction capabilities in low-visibility environments by introducing attention-based dynamic convolutional kernels into the encoder. The intelligent decision-making process also includes: Based on the image data, early disease and pest diagnosis is performed using a trained disease and pest identification model. When early symptoms of pests and diseases are diagnosed, precise pesticide application instructions and regional weed control instructions are generated simultaneously.

[0008] In one specific implementation of the first aspect, the intelligent decision-making step further includes a predictive hierarchical step: Obtain time-series image sequences during the fruit growth cycle ; The time-series image sequence is input into a fruit growth prediction model built on a convolutional neural network and a long short-term memory network, and the model outputs a future specified time point. Predicted fruit diameter ; Based on the predicted fruit diameter Compared with the preset grading standard threshold , The comparison results generate harvesting decision instructions that include real-time grading information.

[0009] Secondly, an intelligent fruit tree planting monitoring system, the system comprising: The monitoring module is used to collect multi-source data on the fruit tree's growth environment, and it includes at least: The underground sensing unit includes soil moisture sensors and soil soluble salt content sensors deployed in the main root layer of fruit trees. The ground-based sensing unit includes an infrared temperature sensor and a multispectral camera, which are used to collect the temperature of the fruit tree canopy. and canopy images; Environmental information acquisition unit, used to acquire air temperature and short-term forecast rainfall ; The decision control module is communicatively connected to the monitoring module and integrates a data fusion unit and an AI decision unit. The AI ​​decision-making unit is embedded with a salt-moisture coordinated stress decision-making model. The execution module, communicatively connected to the decision control module, includes an integrated water and fertilizer irrigation device for receiving and executing the non-fixed-quantity irrigation control commands. .

[0010] In one specific embodiment of the first aspect, the monitoring module further includes a machine vision unit, which includes a camera and a processor; The processor runs a dynamic convolutional neural network model as defined in claim 2, used to maintain the accuracy of identifying fruit and pest targets under adverse weather conditions, with an average target detection accuracy of [missing information]. The efficiency should be no less than 80% under simulated smog conditions.

[0011] In one specific implementation of the first aspect, the execution module further includes: Plant protection execution equipment is used to receive precise application instructions generated based on pest and disease identification results; The agricultural collaborative scheduling unit is configured to automatically generate and send instructions to the intelligent mowing equipment when the plant protection execution equipment is triggered, so as to collaboratively perform grass clearing operations.

[0012] Thirdly, an intelligent fruit tree planting monitoring device, the device comprising: One or more processors; Memory; and one or more computer programs; The one or more computer programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for performing an intelligent fruit tree planting monitoring method as described in any one of claims 1 to 3. Fourthly, a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of an intelligent fruit tree planting monitoring method. Fifthly, a computer program product comprising computer instructions, which, when executed by a processor, are used to implement steps of an intelligent fruit tree planting monitoring method.

[0013] The beneficial effects of this invention are as follows: 1. This invention upgrades traditional simple control relying on a single threshold judgment to collaborative intelligent decision-making based on the full-dimensional growth status of crops by constructing a core architecture of multi-source data fusion and intelligent decision-making. For the first time, the system deeply integrates soil environment, crop physiology, and meteorological factors, utilizing advanced sensing and algorithm models to achieve a leap from passive response to proactive prediction. This effectively overcomes the technical bottlenecks of traditional agricultural monitoring systems, such as single decision-making dimensions, poor environmental adaptability, and insufficient predictability, forming a complete technical closed loop integrating perception, decision-making, and execution. 2. This invention achieves efficient utilization of water and fertilizer resources through precise resource allocation and intelligent collaborative management, significantly reducing production costs. Simultaneously, it effectively improves fruit yield and quality through proactive pest and disease early warning and growth regulation. More importantly, this invention transforms the traditional decentralized management model relying on manual experience into a data-driven, intensive, and intelligent production system, comprehensively enhancing the precision, reliability, and automation level of agricultural production, and providing strong technical support for the high-quality development of modern fruit industry. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the calculation process of the salt-water synergistic stress decision-making model of the present invention. Detailed Implementation

[0015] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0016] like Figure 1 This paper presents an intelligent fruit tree planting monitoring method and system.

[0017] Example 1: System Architecture The system of the present invention includes a monitoring module, a decision control module, and an execution module.

[0018] The monitoring module is responsible for comprehensive data collection, including: The underground sensing unit consists of soil temperature and humidity sensors and soil soluble salt content (EC value) sensors. These sensors should be deployed in a grid pattern within the main root system layer of the fruit tree (usually 20-40 cm below the surface) to ensure that the collected data accurately reflects the environmental conditions of the crop root zone.

[0019] Ground-based sensing unit: Includes an infrared temperature sensor for non-contact measurement of fruit tree canopy temperature. ; and multispectral cameras, used to acquire canopy image data reflecting crop growth and nutrients.

[0020] Environmental information acquisition unit: typically a small weather station used to collect air temperature data. It can collect data such as humidity, wind speed, light intensity, and rainfall, and obtain short-term (e.g., 24-72 hours) weather forecast data, especially rainfall forecasts, via an internet interface. .

[0021] The machine vision unit consists of high-definition cameras deployed at key nodes in the orchard and their built-in processors. These processors run specially trained, low-visibility robust vision algorithms.

[0022] The decision control module is the brain of the system, typically deployed on a cloud server or local industrial control computer, and includes: Data fusion unit: Cleans, aligns (spatiotemporal registration) and standardizes data from different units and at different spatiotemporal scales.

[0023] AI Decision Unit: Embedded with the core models of this invention, including a salt-water synergistic stress decision model, a pest and disease identification model, and a fruit growth prediction model.

[0024] The execution module performs specific operations according to instructions, including: Integrated water and fertilizer irrigation equipment: It receives irrigation instructions and can be linked with the fertilizer tank to achieve simultaneous application of water and fertilizer.

[0025] Plant protection equipment: such as precision spraying vehicles or drones based on Beidou navigation.

[0026] Agricultural Coordination and Scheduling Unit: A logic control unit used to automatically schedule other equipment (such as intelligent lawn mowing equipment) to work together when a specific decision is triggered (such as when plant protection is started).

[0027] Example 2: Method Flow and Core Model The overall process of this method includes data acquisition, fusion decision-making, and collaborative execution. The core parts are described in detail below.

[0028] 1. Salinity-Water Coordinated Stress Decision-Making Model The specific steps for this model are as follows: a) Rinse water requirement calculate This section addresses the problem of soil salinization. The calculation formula is as follows: : The measured soluble salt content of soil, the unit is dS / m (decimeters per meter), which is the standard unit for measuring soil salinity.

[0029] Crop salt tolerance threshold. This is a key agronomic parameter, and the threshold varies for different tree species and even different varieties. For example, the salt tolerance threshold for apple trees is approximately 0.35 dS / m, while pear trees may tolerate up to 0.51 dS / m. This parameter needs to be set by the user during system initialization based on the type of crop being planted.

[0030] The planned irrigated area, in units of... .

[0031] Effective root depth of fruit trees, in units of _____. The value is usually taken as 0.3 m. 0.5 m.

[0032] Leaching efficiency coefficient. This is an empirical coefficient related to soil texture, reflecting the efficiency of irrigation water in leaching salts. Typical values ​​for sandy soils, loam, and clay are 0.8–1.0, 0.5–0.8, and 0.3–0.5, respectively. Users can calibrate this coefficient based on soil survey data or field trials.

[0033] b) Physiological water requirement calculate This section uses crop canopy temperature, a physiological indicator, to determine if the crop is "thirsty." First, the crop water stress index is calculated. : The canopy temperature measured by an infrared temperature sensor, in units of... .

[0034] The air temperature measured by the weather station, in units of... .

[0035] The lower limit of the baseline represents the temperature difference between the canopy temperature and the air temperature when there is sufficient moisture and high transpiration. It can be obtained by measurement under fully irrigated conditions or by consulting literature; a typical value is approximately [missing value]. .

[0036] The upper limit of the baseline represents the temperature difference between the canopy temperature and the air temperature when there is severe water stress and transpiration has essentially ceased. It can be obtained through measurements under drought conditions, with a typical value of approximately [value missing]. .

[0037] get Then (its value is between 0 and 1), it can be determined through a predefined relationship. For example, a simple linear relationship can be established: when At that time, the system determines that irrigation is needed. It can be calculated based on the difference between the soil field capacity and the current humidity; when hour, A better solution is to establish... A lookup table or nonlinear mapping function for irrigation compensation.

[0038] c) Final irrigation volume calculate This step integrates crop needs and natural rainfall to achieve optimal resource utilization.

[0039] Forecast rainfall in the near future (e.g., 24 hours), in mm.

[0040] Rainfall utilization factor. This factor takes into account runoff loss and evaporation loss from rainfall, and its value is typically in the range of 0.6. 0.9. Higher values ​​can be used for fields with heavy clay soil and gentle slopes, and lower values ​​can be used for the opposite.

[0041] This function ensures that the irrigation amount is not negative. If the effective rainfall is sufficient to meet the needs, no additional irrigation is required.

[0042] 2. Low-visibility robust vision algorithm This algorithm is based on a Dynamic Convolutional Network (CNN). Unlike standard CNNs, its innovation lies in its dynamic encoder. This encoder dynamically generates a set of convolutional kernel weights based on global contextual information of the input image (such as fog density and illumination level) through an attention mechanism, instead of using fixed convolutional kernels. This means that the network actually uses different feature extractors for clear and foggy images, significantly improving the model's robustness in low-visibility conditions.

[0043] In practical implementation, the network can be jointly trained using public datasets (such as ImageNet) and self-built datasets of orchard haze and low-light images. Testing showed that the model achieves a stable average accuracy (mAP) of over 80% for fruit detection in simulated haze orchard scenarios, significantly higher than traditional models.

[0044] 3. Fruit growth prediction and grading model This model is a CNN-BiLSTM-Attention hybrid model.

[0045] Input: A time-series image sequence during the fruit growth cycle For example, images of the same fruit taken once a week.

[0046] Feature extraction and sequence modeling: First, a CNN (such as ResNet) is used to extract features from each image, resulting in feature vectors representing the size, color, and texture of the fruit.

[0047] The temporal feature vector sequence is then input into a bidirectional long short-term memory (BiLSTM) network. BiLSTM can learn the temporal patterns and dependencies of fruit growth from both forward and backward directions.

[0048] An attention mechanism is introduced into the output of BiLSTM, allowing the model to automatically focus on the growth stages most critical to predicting future size (e.g., the rapid expansion period).

[0049] Output: The model outputs the value at a specified future time. Predicted fruit diameter .

[0050] Hierarchical decision-making: The system will predict the diameter Compared with the preset grading standard threshold (such as , By comparing the results, grading instructions can be generated during harvesting, realizing "harvesting and sorting at the same time".

[0051] Example 3: Complete Workflow Example Assuming an apple orchard, the system initialization parameters are as follows: dS / m, , , , .

[0052] Data acquisition: measured by sensors . area ,depth .

[0053] Decision calculation: ; according to =0.78, from the table we get ; (Note: 1 mm of precipitation = 10 m³ on 1 hectare). ; Collaborative execution: The system issues instructions to the integrated water and fertilizer equipment to irrigate 165 cubic meters. At the same time, the machine vision unit identifies early-stage leaf spot disease, and the system simultaneously launches a plant protection drone to precisely apply pesticides to the affected area, and dispatches an intelligent lawnmower to clear weeds around the affected area.

[0054] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent monitoring method for fruit tree planting, characterized in that, The method includes the following steps: Data acquisition steps: Collect multi-source data on the fruit tree's growth environment using deployed sensors. This multi-source data includes at least soil volumetric water content from the root zone of the fruit trees. Soil soluble salt content Temperature of fruit tree canopy and air temperature from meteorological data sources and short-term forecast rainfall ; Data fusion and decision-making steps: The multi-source data is input into the salinity-moisture synergistic stress decision-making model, which performs the following calculation process: Based on soil soluble salt content With crop salt tolerance threshold The difference is used to calculate the required rinsing water volume. : in, The rinsing efficiency coefficient is... For irrigated area, The depth of the root layer; Based on the temperature of the fruit tree canopy With air temperature The difference is used to calculate the crop water stress index. And based on this, determine the physiological water requirement. : in, and These are the baseline values ​​for canopy-air temperature difference under conditions of full irrigation and severe drought, respectively. The total water required for rinsing Physiological water requirement and short-term forecast rainfall Generate the final non-fixed-volume irrigation control instructions. : in, Rainfall utilization coefficient; Collaborative execution steps: Drive the integrated water and fertilizer irrigation equipment to execute the irrigation control commands. .

2. The intelligent fruit tree planting monitoring method according to claim 1, characterized in that: The data acquisition step also includes: Image data of fruit trees is collected through cameras deployed in the orchard and equipped with a dynamic convolutional neural network model that has been enhanced by conditions of haze and low light. The dynamic convolutional neural network model improves feature extraction capabilities in low-visibility environments by introducing attention-based dynamic convolutional kernels into the encoder. The intelligent decision-making process also includes: Based on the image data, early disease and pest diagnosis is performed using a trained disease and pest identification model. When early symptoms of pests and diseases are diagnosed, precise pesticide application instructions and regional weed control instructions are generated simultaneously.

3. The intelligent fruit tree planting monitoring method according to claim 2, characterized in that: The intelligent decision-making process also includes a predictive grading step: Obtain time-series image sequences during the fruit growth cycle ; The time-series image sequence is input into a fruit growth prediction model built on a convolutional neural network and a long short-term memory network, and the model outputs a future specified time point. Predicted fruit diameter ; Based on the predicted fruit diameter Compared with the preset grading standard threshold , The comparison results generate harvesting decision instructions that include real-time grading information.

4. An intelligent fruit tree planting monitoring system, characterized in that, The system includes: The monitoring module is used to collect multi-source data on the fruit tree's growth environment, and it includes at least: The underground sensing unit includes soil moisture sensors and soil soluble salt content sensors deployed in the main root layer of fruit trees. The ground-based sensing unit includes an infrared temperature sensor and a multispectral camera, which are used to collect the temperature of the fruit tree canopy. and canopy images; Environmental information acquisition unit, used to acquire air temperature and short-term forecast rainfall ; The decision control module is communicatively connected to the monitoring module and integrates a data fusion unit and an AI decision unit. The AI ​​decision-making unit is embedded with a salt-moisture coordinated stress decision-making model. The execution module, communicatively connected to the decision control module, includes an integrated water and fertilizer irrigation device for receiving and executing the non-fixed-quantity irrigation control commands. .

5. The intelligent fruit tree planting monitoring system according to claim 4, characterized in that: The monitoring module also includes a machine vision unit, which includes a camera and a processor; The processor runs a dynamic convolutional neural network model as defined in claim 2, used to maintain the accuracy of identifying fruit and pest targets under adverse weather conditions, with an average target detection accuracy of [missing information]. The efficiency should be no less than 80% under simulated smog conditions.

6. The intelligent fruit tree planting monitoring system according to claim 4, characterized in that: The execution module further includes: Plant protection execution equipment is used to receive precise application instructions generated based on pest and disease identification results; The agricultural collaborative scheduling unit is configured to automatically generate and send instructions to the intelligent mowing equipment when the plant protection execution equipment is triggered, so as to collaboratively perform grass clearing operations.

7. An intelligent fruit tree planting monitoring device, characterized in that, The device includes: One or more processors; Memory; and one or more computer programs; The one or more computer programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for performing an intelligent fruit tree planting monitoring method as described in any one of claims 1 to 3.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent fruit tree planting monitoring method as described in any one of claims 1 to 3.

9. A computer program product, characterized in that, The computer program product includes computer instructions, which, when executed by a processor, are used to implement the steps of an intelligent fruit tree planting monitoring method as described in any one of claims 1 to 3.

Images