Underforest crop water and fertilizer intelligent regulation and control method and system based on Internet of Things sensing

Abstract

The invention discloses an intelligent control method and system for water and fertilizer of under-forest crops based on Internet of Things sensing, and relates to the field of agricultural intelligent equipment. The system comprises a sensing layer, a transmission layer, a data processing layer, an execution layer and an interaction layer, the sensing layer collects soil nutrients, humidity and environmental parameters by deploying multi-parameter sensing nodes (avoiding a tree main root area) in a matrix manner, and transmits the soil nutrients, humidity and environmental parameters to a cloud platform through dual-mode communication; the data processing layer calls a water and fertilizer regulation and control model fusing an expert rule base and an improved BP neural network, and generates a regulation and control scheme in combination with a crop growth stage; the execution layer achieves accurate application through zoned drip irrigation and flow feedback, and closed-loop adjustment is formed. The system solves the problems of monitoring distortion, extensive decision, unstable communication and the like in the prior art, realizes intelligent and accurate regulation and control of water and fertilizer of understory crops, improves the utilization efficiency of water and fertilizer, and adapts to a complex understory environment.

Description

Technical Field

[0001] This invention relates to the field of agricultural intelligent equipment technology, specifically to a method and system for precise regulation of crop water and fertilizer in complex forest environments. Background Technology

[0002] Understory planting, as an efficient agroforestry model, can make full use of land resources. However, because trees and crops share water and nutrients and their root systems are intertwined, traditional water and fertilizer management faces many challenges. Existing water and fertilizer regulation devices for understory crops mostly rely on manual experience or simple automated equipment. Their basic structure usually includes a single type of sensor (such as monitoring only soil moisture), a fixed-mode irrigation component (such as a uniform flow drip irrigation pipe), a basic wireless transmission module, and a manual / semi-automatic control unit.

[0003] However, existing technologies have significant shortcomings: First, sensor deployment does not avoid the main root zone of trees, and the collected soil nutrient and moisture data are interfered with by tree roots, failing to reflect the true needs of crops; second, control decisions rely on fixed thresholds, failing to integrate environmental parameters such as understory light, temperature, and humidity, as well as the competitive characteristics of trees and crops, easily leading to excessive or insufficient water and fertilizer; third, communication modules are mostly single-mode transmissions, which are affected by tree shading, resulting in poor signal stability and data real-time performance; fourth, the execution equipment lacks zonal control and dynamic feedback, making it difficult to adapt to the differentiated needs of crops in different areas; fifth, the power supply method is singular (e.g., relying solely on batteries), which is prone to power outages in environments with unstable understory light, and lacks a closed-loop adjustment mechanism, making it impossible to dynamically correct the scheme based on the control effect, leading to problems such as low water and fertilizer utilization efficiency and inhibited crop growth. Summary of the Invention

[0004] To address the problems of inaccurate parameter monitoring, crude decision-making, unstable communication and power supply, low execution accuracy, and lack of dynamic feedback in existing water and fertilizer regulation devices for understory crops, this paper proposes an intelligent water and fertilizer regulation method and system based on Internet of Things (IoT) sensing to achieve precise, dynamic, and intelligent management of water and fertilizer for understory crops.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: This invention constructs an intelligent water and fertilizer regulation system for understory crops through the collaborative work of a sensing layer, a transmission layer, a data processing layer, an execution layer, and an interaction layer, as detailed below: Sensing layer: Multi-parameter sensor nodes are deployed in a matrix in the understory crop planting area, avoiding the main root distribution area of trees (an area with a radius of 5-8 times the diameter at breast height of the tree, centered on the trunk), and each node covers 3-5 crops. The sensor node includes a soil nutrient sensor (detecting the concentration of nitrogen, phosphorus, and potassium at a depth of 10-20 cm), a soil moisture sensor (detecting the root zone volumetric water content), an environmental light sensor (detecting the canopy photosynthetically active radiation), an air temperature and humidity sensor, and a dual power supply module (solar + lithium battery) and a hibernation wake-up function (hibernating when the parameter change rate is below the threshold to save energy).

[0006] Transmission layer: A LoRa and NB-IoT dual-mode low-power wide-area network communication module is used to encrypt and transmit the data collected by the sensor nodes. When the tree blocks the signal in a single mode, it automatically switches to the other mode to ensure real-time data transmission.

[0007] Data processing layer: Includes a cloud platform server, a built-in data preprocessing module (filters and denoises the original data, and removes outliers), a crop growth model library (stores the water and fertilizer demand curves of typical understory crops such as Gastrodia and Rhizoma Polygonati at different growth stages), and a water and fertilizer regulation model. The water and fertilizer regulation model combines an expert rule base (contains the nutrient absorption coefficients of crops under specific light, temperature, and humidity) and an improved BP neural network (input layer: soil nitrogen / phosphorus / potassium concentration, humidity, photosynthetically active radiation, and environmental temperature and humidity; output layer: irrigation volume and nitrogen / phosphorus / potassium application amount; hidden layer nodes optimized by adaptive particle swarm algorithm), based on historical data iteration optimization to adapt to the tree-crop root competition characteristics.

[0008] Execution layer: Includes a local controller and water and fertilizer execution equipment. The water and fertilizer execution equipment includes multiple independent fertilizer storage units (corresponding to nitrogen, phosphorus, and potassium elements or compound fertilizers), a water tank, a mixing chamber (with a stirring device), a flow sensor, and a zoned control drip irrigation pipeline (3-5 outlets per meter, facing the crop root system). The local controller receives cloud platform instructions, adjusts the fertilizer output ratio through electromagnetic valves, controls the water and fertilizer mixing ratio through a variable-speed peristaltic pump, and realizes closed-loop control combined with flow sensor feedback to accurately deliver water and fertilizer to the target area.

[0009] Interaction layer: The user terminal displays real-time monitoring data, regulation schemes, and historical records, supports manual correction of regulation parameters, and feedback of corrected data to the cloud platform for model optimization.

[0010] The regulation method process is as follows: S1. The sensing node collects crop root zone soil nutrients, humidity and environmental parameters; S2. The data is encrypted and transmitted to the cloud platform through the transmission layer, and effective data is obtained after preprocessing; S3. The cloud platform calls the water and fertilizer regulation model, combines the crop type and growth stage threshold, and outputs the target water and fertilizer application amount and timing; S4. The local controller drives the execution equipment to complete the precise water and fertilizer application; S5. Within the preset time after regulation, the sensing node re-collects parameters, and if the threshold is not reached, S3-S4 is repeated to form a closed loop adjustment.

[0011] The one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages compared with the prior art: Monitoring precision: the sensing node avoids the main root zone of the tree and collects multi-dimensional parameters, combined with data preprocessing, to ensure that the real demand of the crop is reflected; Decision intelligence: the regulation model integrates expert rules and machine learning, adapts to the microenvironment under the tree and the competition characteristics of the tree-crop, and makes more accurate decisions; Communication stability: automatic switching of dual-mode communication solves the signal problem caused by tree shading; Efficient execution: zoning control and flow feedback realize differentiated water and fertilizer supply, improving utilization efficiency; Strong self-adaptation: dual power supply mode and hibernation function adapt to the energy conditions under the tree, and the closed-loop feedback mechanism dynamically corrects the scheme to avoid water and fertilizer excess or deficiency; Convenient operation: the user terminal supports real-time monitoring and manual intervention, balancing intelligence and flexibility. Other advantages, objectives and features of the present application will be described to some extent in the subsequent specification, and to some extent, it will be obvious to those skilled in the art based on the study of the following text, or can be taught from the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 The figure is the architecture diagram of the present application based on the Internet of Things sensing of the intelligent regulation method and system of water and fertilizer of crops under trees; Figure 2 The figure is the timing diagram of the intelligent regulation method of water and fertilizer of crops under trees of the present application based on the Internet of Things sensing of the intelligent regulation method and system of water and fertilizer of crops under trees; Figure 3 The figure is a schematic diagram of the deployment of sensing nodes in the crop planting area of the present application based on the Internet of Things sensing of the intelligent regulation method and system of water and fertilizer of crops under trees; Figure 4 The figure is a schematic diagram of the regulation process of the water and fertilizer execution equipment of the present application based on the Internet of Things sensing of the intelligent regulation method and system of water and fertilizer of crops under trees. DETAILED DESCRIPTION

[0013] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all the other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of the present application.

[0014] It should be noted that the terms "vertical", "horizontal", "upper", "lower", "left", "right", and similar expressions used herein are for illustrative purposes only and do not indicate the only implementation.

[0015] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs; the terminology used in the description herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application; the use of the terms "and / or" includes a combination of one or more of the associated listed items.

[0016] As shown in Figures 1-4 The core components of the under-forest crop water and fertilizer intelligent control system based on Internet of Things sensing of the present application include the following: The perception layer includes: a multi-parameter sensing node, the node integrates a soil nutrient sensor (detection depth 10-20 cm, detection parameters are nitrogen, phosphorus, and potassium concentration), a soil humidity sensor (detects the soil volume water content in the root zone), an environmental light sensor (detects the photosynthetically active radiation of the crop canopy), an air temperature and humidity sensor (detects the environmental temperature and relative humidity), a data acquisition module (used for aggregating sensor data), a dual power supply module (a solar power module and a lithium battery pack, supporting adaptive switching of power), and a hibernation wake-up module (triggers hibernation when the parameter change rate is lower than a preset threshold, reducing energy consumption).

[0017] The transmission layer includes: a low-power wide-area network communication module, which adopts a LoRa and NB-IoT dual-mode communication unit, has a built-in encryption chip (supports AES-128 encryption), and has a signal strength detection and mode automatic switching function (switches to another mode when the signal strength of a single mode is lower than -100 dBm).

[0018] The data processing layer includes: a cloud platform server (including a data preprocessing module for Kalman filter denoising and 3σ criterion outlier rejection of original data); a crop growth model library (storing the biomass and water and fertilizer demand relationship curves of typical understory crops such as Gastrodia elata, Polygonatum sibiricum and Panax notoginseng at the seedling stage, growth stage and mature stage); a water and fertilizer regulation model (combining an expert rule base and an improved BP neural network, the expert rule base including the nutrient absorption coefficients of different crops under specific light, temperature and humidity, the input layer of the improved BP neural network being 7 environmental parameter neurons, the number of nodes in the hidden layer being optimized by an adaptive particle swarm algorithm, and the output layer being 4 regulation parameter neurons); and a decision module (converting the model output into executable instructions).

[0019] The execution layer includes: a local controller (using an STM32 series chip to support PWM signal output and data interaction); water and fertilizer execution equipment including at least 3 independent fertilizer storage units (corresponding to nitrogen, phosphorus and potassium elements or compound fertilizers, the outlets being provided with electromagnetic valves to adjust the opening degree), a clean water bucket, a mixing chamber (provided with an internal stirring device to ensure uniform mixing of water and fertilizer), a flow sensor (real-time monitoring of the flow of mixed water and fertilizer and feedback to the controller), and a zoned control drip irrigation pipeline (divided according to the planting rows of crops, 3-5 water outlets per meter, the outlets facing the distribution direction of crop root systems).

[0020] The interaction layer includes: a user terminal (supporting Web and APP access, capable of displaying real-time parameters, regulation schemes and historical curves, and having an interactive interface for manual correction of regulation parameters).

[0021] In the embodiment, the combination of various components realizes a closed-loop link of "precise perception-stable transmission-intelligent decision-making-high efficiency execution-human-machine collaboration" from the implementation points: The multi-parameter sensor of the perception layer and the specific deployment method (avoiding the main root zone of trees and matrix coverage) solve the problem of one-sided monitoring by a single sensor and data distortion caused by interference from tree roots in the prior art - by synchronously collecting soil nutrients, humidity and environmental light, temperature and humidity, and focusing on the real state of the crop root zone, accurate input is provided for regulation and decision-making; the dual power supply and hibernation wake-up module specifically solves the problem of power interruption caused by unstable light in the forest, prolonging the device's endurance time (more than 60% higher than single battery power supply).

[0022] The combination of the dual-mode communication module of the transmission layer solves the problem of signal interruption and data delay caused by tree shading in the prior art - LoRa is suitable for short-distance anti-interference transmission, and NB-IoT relies on operator networks to achieve wide-area coverage, automatic switching of the two ensures that the data transmission success rate is more than 98%, ensuring the real-time nature of the regulation instructions.

[0023] The combination of the "expert rule base + improved BP neural network" model of the data processing layer breaks through the limitations of the existing technology relying on fixed threshold decision and being unable to adapt to the competition characteristics of understory trees and crops: the expert rule base embeds the nutrient absorption rules of understory crops in special environments such as low light and high humidity (for example, when the photosynthetically active radiation is less than 5000 lux, the nitrogen absorption efficiency of crops decreases by 15%-20%), and the improved BP neural network iteratively learns the nutrient competition coefficient between tree roots and crops through historical data, so that the output water and fertilizer scheme not only matches the needs of the growth stage of crops, but also dynamically balances the resource competition of trees, avoiding "fertilizer interception by trees" or "crop nutrient deficiency".

[0024] The combination of the multi-fertilizer storage unit, the zoned drip irrigation pipeline and the flow feedback of the execution layer solves the problem that existing devices cannot achieve differentiated supply and rough flow control: the independent fertilizer unit can accurately control the nitrogen, phosphorus and potassium ratio (error ≤3%), the zoned pipeline adapts to the growth differences of crops in different areas of the forest (such as the demand difference caused by the density of tree shade), and the closed-loop control of the flow sensor and the local controller ensures that the actual application amount deviates from the target value by ≤5%, greatly improving the water and fertilizer utilization efficiency (40%-60% less than traditional flooding irrigation).

[0025] The linkage of the interaction layer and the data processing layer takes into account both intelligence and flexibility - users can intervene in real time to adjust the parameters, correct the data feedback model iteration, and solve the problem of insufficient adaptation to special scenarios (such as emergency adjustment in extreme weather) caused by the lack of human intervention in the fully automated existing system.

[0026] The present application can adapt to various understory planting scenarios, including understory medicinal materials (such as ginseng and rhizoma polygonati), understory vegetables (such as spinach and lettuce), slope understory crops and high-density tree understory crops, etc. The connection, installation and linkage details of the present application with existing technologies / devices and the use state of each structure are as follows: Case one: understory medicinal material planting (taking ginseng as an example, the upper tree is chestnut tree) Scenario characteristics: ginseng is a typical shade-tolerant medicinal material, with shallow roots (mainly distributed in the 10-20 cm soil layer), which needs to avoid water accumulation, and the chestnut tree has developed main roots (diameter about 30 cm, main root distribution area radius 1.5-2.4 m), which needs to avoid competition for water and fertilizer with tree roots.

[0027] Connection and installation details of the present application with existing technologies / devices: Sensing layer: The existing row distance (about 4m) under the chestnut forest is utilized, and the sensor nodes are deployed in the ginseng planting area (1.2m wide) avoiding the main root area of chestnut trees (1.5m outside the trunk). The sensor is installed by inserting: the soil nutrient sensor and humidity sensor are inserted into the 10-15cm soil layer (matching the depth of ginseng root) through a stainless steel probe, and the probe is covered with a corrosion-resistant PVC pipe (compatible with the existing soil environment); the ambient light sensor and air temperature and humidity sensor are fixed on the existing bamboo support (height 50cm, slightly higher than the ginseng canopy). In the dual power supply module, the solar panel (10W) is fixed on the top of the bamboo support (facing southeast, using the existing support to bear weight), and the lithium battery pack (12V / 5Ah) is buried in the 5cm deep soil (waterproof box packaging, to avoid rainwater immersion), and is connected with the sensor through waterproof wire (RVV4x0.5mm²).

[0028] Transmission layer: The LoRa gateway is deployed on the existing power pole at the edge of the chestnut forest (height 3m, using the existing power supply line to take power), and the LoRa module of the sensor node is connected through wireless connection (communication distance 200-300m); the NB-IoT module directly accesses the existing operator base station signal (no need to deploy additional equipment), and the dual-mode switching module is connected with the gateway through the RS485 interface to realize automatic signal switching.

[0029] Execution layer: The existing main water pipe (Φ50mm PE pipe) is used to connect the water inlet pipe of the device through a quick connector (DN20); the fertilizer barrel (3, volume 50L) is fixed on the existing brick platform (height 80cm, higher than the mixing chamber), and the outlet electromagnetic valve is connected to the mixing chamber through a Φ16mm hose, and an existing flow meter is installed in the middle of the hose (compatible with the flow sensor data calibration of the device). The drip irrigation pipe is arranged along the ginseng planting row and connected to the existing drip irrigation main pipe through a tee joint, and the water outlet hole is directed to the distribution side of the ginseng rhizome (each row is independently controlled, and is adapted to the existing planting row distance).

[0030] Data processing layer and interaction layer: The cloud platform is connected to the existing farm management system through the Ethernet interface (API protocol compatible), and the user terminal (mobile phone APP) is connected to the existing farm WiFi network to realize data synchronization.

[0031] The use state of each structure: Sensing layer: The soil humidity threshold is set to 60-65% during the ginseng seedling stage (March-May), and the sensor collects data every 2 hours; when the humidity change rate is less than 3% for 3 consecutive times (such as after rain), the sleep wake-up module triggers sleep (the interval is extended to 6 hours), and the solar panel charges the lithium battery (the charging efficiency is more than 80% on sunny days).

[0032] Transmission layer: During the lush period of chestnut trees (June-August), the LoRa signal is blocked (strength -105 dBm), and the system automatically switches to the NB-IoT mode (signal strength -85 dBm) with a data transmission delay of less than 10 seconds.

[0033] Execution layer: When the soil phosphorus concentration is detected to be lower than 80 mg / kg (peak period of Gastrodia elata phosphorus requirement), the local controller drives the electromagnetic valve of the phosphorus fertilizer storage unit (opening degree 30%), mixes with clean water at a ratio of 1:50 (stirring device speed 300 r / min), and real-time feedback is provided by the flow sensor (accuracy ±2%). The system applies the fertilizer through the 2nd and 3rd row drip irrigation pipes (Gastrodia elata dense area) for 15 minutes and then automatically shuts down.

[0034] Interaction layer: Farmers can view the nutrient trend chart of Gastrodia elata tuber swelling period (September-October) through the APP and manually increase the potassium fertilizer application amount by 10% (based on experience correction). The corrected data is synchronized to the cloud platform for model iteration.

[0035] Scenario two: Slope forest vegetables planting (taking spinach as an example, the upper tree is Pinus massoniana, and the slope is 15°) Scene characteristics: Slope is prone to water and soil loss, spinach has a short growth period (30-40 days), requires high-frequency water and nitrogen supplementation, Pinus massoniana has shallow and widely distributed root systems (diameter at breast height 20 cm, main root zone radius 1-1.6 m), and precise control of water and fertilizer distribution along the slope is required.

[0036] Connection and installation details of existing technology / devices: Sensing layer: Sensing nodes are deployed along the contour direction (avoiding the main root zone of Pinus massoniana), and ground nails are installed at the bottom of the sensors (length 20 cm, suitable for slope soil fixation) to prevent rainwater from washing away. Soil moisture sensors have a detection depth of 8-10 cm (spinach root depth) and are kept 50 cm away from existing slope drainage ditches (to avoid interference from accumulated water). In the dual power supply module, the solar panel is installed at an inclination of 30° (matching the slope angle) and is fixed by existing concrete piers (wind resistance level ≥8).

[0037] Transmission layer: LoRa modules share the LoRa gateway of the existing slope weather station (address code differentiation), and NB-IoT modules enhance the signal through an external antenna (fixed at a height of 2 m on the pine tree trunk) to solve the problem of signal blind area on the slope.

[0038] Execution layer: Drip irrigation pipes are laid along the contour (fixed by existing slope terrace ridges), and check valves are installed on each section of the pipe (compatible with existing anti-backflow design) to prevent water and fertilizer from flowing down the slope. Fertilizer barrels and clean water barrels are fixed on the slope top (1.5 m higher than the planting area, using gravity to assist transportation) and are linked with the existing slope irrigation pump (power 1.5 kW) (the pump start signal is connected to the local controller through a relay).

[0039] Data processing layer: Cloud platform calls spinach growth model (adapt to slope nutrient loss coefficient: 12% higher than flat land), and existing slope soil moisture database (supplement historical data).

[0040] Each structure usage state: Sensing layer: Spinach rapid growth period (15-25 days), large amount of nitrogen required, soil nitrogen concentration sensor collects data every 12 hours, light sensor monitors photosynthetically active radiation (less than 3000 lux, trigger nitrogen fertilizer absorption efficiency correction coefficient).

[0041] Transmission layer: In heavy rain weather (low visibility), LoRa signal attenuation to-110dBm, automatically switch to NB-IoT mode, data encryption transmission (avoid data leakage caused by rain interference).

[0042] Execution layer: When the soil humidity is detected to be lower than 50% (fast evaporation on slope), the local controller starts the irrigation pump and opens the slope middle and slope lower partitioned drip irrigation valves (opening degrees are 80% and 100% respectively, to compensate for the difference in water loss on slope), the flow sensor adjusts the pump speed in real time (to maintain a flow of 0.5m³ / h), to avoid excessive water pressure from damaging spinach seedlings.

[0043] Interaction layer: Farmers set "heavy rain early warning linkage" through the terminal, when the existing weather station pushes heavy rain signal, the system automatically closes all valves and records the current water and fertilizer state, and automatically restores control after the rain.

[0044] Although the present application has been disclosed with the above preferred embodiments, it is not intended to limit the present application, and any person skilled in the art can make various modifications and modifications without departing from the spirit and scope of the present application, therefore the protection scope of the present application should be defined by the claims.

Claims

1. A method for intelligent water and fertilizer regulation of understory crops based on Internet of Things (IoT) sensing, characterized in that, Includes the following steps: S1. Sensor Node Deployment and Parameter Acquisition: Multi-parameter sensor nodes are deployed at a preset density in the understory crop planting area. The sensor nodes include soil nutrient sensors, soil moisture sensors, ambient light sensors, and air temperature and humidity sensors. The soil nutrient sensors are used to collect the concentrations of nitrogen, phosphorus, and potassium elements in the crop root zone soil; the soil moisture sensors are used to collect the volumetric water content of the root zone soil; the ambient light sensors are used to collect the photosynthetically active radiation at the canopy of the understory crops; and the air temperature and humidity sensors are used to collect the temperature and relative humidity of the crop growth environment. The sensor nodes are deployed in a matrix pattern, avoiding the distribution area of ​​the tree taproots. S2. Data transmission and preprocessing: Each sensor node transmits the collected parameter data to the cloud platform in real time through a low-power IoT communication module. The cloud platform performs filtering, noise reduction, and outlier removal on the received raw data to obtain effective monitoring data. S3. Intelligent Decision Generation: The cloud platform calls a preset water and fertilizer regulation model. The water and fertilizer regulation model takes the effective monitoring data obtained in step S2 as input, combines the preset water and fertilizer demand thresholds of the understory crop type and the current growth stage, and outputs the target water and fertilizer application amount and application timing. The water and fertilizer regulation model is constructed by integrating an expert rule base and a machine learning algorithm. The expert rule base contains the nutrient absorption coefficients of different understory crops under specific light, temperature and humidity conditions. The machine learning algorithm iteratively optimizes based on historical regulation data so that the output results are adapted to the root competition characteristics of understory trees and crops. S4. Water and fertilizer application control: The cloud platform sends the target water and fertilizer application amount and application timing instructions generated in step S3 to the local controller. The local controller drives the integrated water and fertilizer application equipment, controls the mixing ratio of liquid fertilizer and irrigation water through an adjustable speed peristaltic pump, and accurately delivers the mixed water and fertilizer to the crop root zone through zoned drip irrigation pipelines. S5. Closed-loop feedback adjustment: Within a preset time after step S4 is executed, soil nutrient concentration and humidity parameters are collected again through the sensor node. If the parameters do not reach the preset water and fertilizer demand threshold, the cloud platform re-executes steps S3-S4 to make dynamic adjustments until the parameters are within the threshold range.

2. The method according to claim 1, characterized in that, In step S1, the detection depth of the soil nutrient sensor is 10-20cm, and the detection range of each sensor node covers 3-5 understory crops; the distribution area of ​​the tree's main root is defined as a circular area with a radius of 5-8 times the tree's diameter at breast height, centered on the tree trunk.

3. The method according to claim 1, characterized in that, In step S3, the machine learning algorithm of the water and fertilizer regulation model adopts an improved BP neural network. Its input layer contains 7 neurons for soil nitrogen concentration, phosphorus concentration, potassium concentration, soil volumetric water content, photosynthetically active radiation, ambient temperature, and ambient relative humidity. The hidden layer optimizes the number of nodes through an adaptive particle swarm optimization algorithm. The output layer contains 4 neurons for irrigation amount, nitrogen fertilizer application amount, phosphorus fertilizer application amount, and potassium fertilizer application amount.

4. The method according to claim 1, characterized in that, In step S4, the integrated water and fertilizer execution device includes a fertilizer tank, a clean water tank, a mixing chamber, and a flow sensor. The mixing chamber has a built-in stirring device, and the flow sensor monitors the flow rate of the mixed water and fertilizer in real time and feeds it back to the local controller to achieve closed-loop control of the target application amount. The density of the water outlet holes in the drip irrigation pipeline is 3-5 per meter, and the water outlet holes face the direction of crop root distribution.

5. A smart water and fertilizer regulation system for understory crops based on Internet of Things (IoT) sensing, characterized in that, include: The perception layer includes multiple distributed sensor nodes, which include soil nutrient sensors, soil moisture sensors, ambient light sensors, air temperature and humidity sensors, and data acquisition modules, used to collect multi-dimensional parameters of the understory crop growth environment. The transport layer, including a low-power wide-area network communication module, is used to encrypt and transmit parameter data collected by the sensing layer to the data processing layer. The data processing layer includes a cloud platform server. The cloud platform server has a built-in data preprocessing module, a water and fertilizer regulation model, and a decision module. The data preprocessing module is used to purify the received data. The water and fertilizer regulation model is used to generate a water and fertilizer regulation scheme based on the purified parameter data and crop information. The decision module is used to convert the regulation scheme into an execution instruction. The execution layer includes a local controller and a water and fertilizer execution device. The local controller receives execution instructions sent by the data processing layer and drives the water and fertilizer execution device to complete precise water and fertilizer application. The water and fertilizer execution device includes a zone-controlled irrigation pipeline, a fertilizer ratio device, and a flow control unit. The interaction layer, including the user terminal, is used to display real-time monitoring data, control schemes, and receive user parameter setting instructions.

6. The system according to claim 5, characterized in that, The sensing nodes of the perception layer adopt a dual power supply mode of solar power module and lithium battery pack, and have a sleep wake-up function. When the detected parameter change rate is lower than the preset threshold, it enters sleep state to reduce energy consumption.

7. The system according to claim 5, characterized in that, The low-power wide-area network communication module of the transmission layer adopts LoRa and NB-IoT dual-mode communication. When the signal of a single communication mode is weak due to the obstruction of trees, it automatically switches to the other communication mode.

8. The system according to claim 5, characterized in that, The cloud platform server of the data processing layer also includes a crop growth model library, which stores the biomass and water and fertilizer demand relationship curves of typical understory crops such as Gastrodia elata, Polygonatum sibiricum, and Panax notoginseng at different growth stages, and is used to provide basic parameters for water and fertilizer regulation models.

9. The system according to claim 5, characterized in that, The fertilizer proportioning device of the execution layer includes at least three independent fertilizer storage units, which correspond to nitrogen, phosphorus, and potassium single-element fertilizers or compound fertilizers, respectively. Each storage unit is equipped with an electromagnetic valve at its outlet, and the fertilizer output ratio is controlled by adjusting the valve opening.

10. The system according to claim 5, characterized in that, The user terminal of the interaction layer can display soil nutrient concentration trend graphs, soil moisture change curves, and historical records of water and fertilizer application. It also supports users to manually correct the output parameters of the water and fertilizer regulation model. The corrected data is fed back to the cloud platform for model iteration and optimization.

Images