Mango fruit quality-oriented water and fertilizer coordinated regulation and control system

Abstract

The invention relates to a mango fruit quality-oriented water and fertilizer coordinated regulation and control system, and belongs to the technical field of intelligent agriculture. In order to solve the problem that the fruit quality is reduced due to the fact that traditional irrigation cannot dynamically meet the requirements of mango growth stages, the system comprises a fruit size monitoring unit, a leaf nitrogen content sensor, a soil humidity sensor, a central controller, a drip irrigation device and a liquid fertilizer applying device. Wherein the fruit size monitoring unit collects images through a canopy multispectral imager and identifies the transverse diameter of a fruit; the leaf nitrogen content sensor detects a leaf reflection spectrum and converts the nitrogen content. The central controller judges the growth stage based on two parameters of the transverse diameter of the fruits and the nitrogen content of the leaves: starting an expansion period strategy when the transverse diameter is 3.0-5.0 cm and the nitrogen content is greater than 2.8% in continuous five days, controlling the drip irrigation device to irrigate twice every day so that the water content of the soil reaches 28-32%, and meanwhile, instructing the liquid fertilizer application device to prepare a fertilizer according to the ratio of nitrogen to phosphorus to potassium being 1.2: 0.8: 1.5. According to the system, the quality of mango fruits is effectively improved through growth stage self-recognition and water and fertilizer cooperative regulation.

Description

Technical Field

[0001] The present invention relates to the technical field of mango planting, and more particularly to a mango fruit quality-oriented water and fertilizer coordinated control system. Background Art

[0002] In mango cultivation, water and fertilizer management requirements differ significantly between the fruit expansion and ripening stages: the expansion stage requires increased irrigation frequency and a higher proportion of specific nutrients, while the ripening stage requires reduced irrigation and an adjusted fertilizer mix. A core limitation of traditional management methods lies in their reliance on a single growth stage criterion, such as physical size or fixed time points, failing to dynamically adjust and manage the tree's nutritional status. This is due to the inherent hysteresis in fruit size changes, making it difficult to accurately capture the critical points of growth stage transitions using this parameter alone. While leaf nutrient levels can reflect the tree's physiological state, existing technologies lack mechanisms for analyzing them in conjunction with fruit development. This leads to two typical problems in actual production: a failure to increase water and fertilizer supply promptly during the early expansion phase, inhibiting fruit cell expansion; and a delayed reduction in nitrogen supply during the ripening phase, impacting sugar accumulation efficiency. Furthermore, when the system is activated, the tree has already entered a specific growth stage, and the traditional pre-set program cannot bypass basic strategies and directly adapt to current needs, further exacerbating the lag in regulation. These inaccurate assessments lead to a misalignment between water and fertilizer supply and physiological needs, ultimately hindering fruit quality. Summary of the Invention

[0003] Another objective of the present invention is to provide a coordinated water and fertilizer control system for mango fruit quality, addressing the inability of traditional mango cultivation to precisely regulate water and fertilizer according to the fruit's growth stage. Traditional methods struggle to accurately track key data such as fruit diameter, leaf nitrogen content, and soil moisture in real time, making it difficult to dynamically adjust irrigation frequency, water volume, and fertilizer ratios. This can easily lead to imbalanced water and fertilizer supply, impacting fruit quality and yield.

[0004] This design addresses the issue of system control accuracy being affected by abnormal sensor data. Sensors can experience data deviations or failures due to various factors. Directly using abnormal data can lead to misjudgments of growth stages and incorrect control strategies, impacting mango growth. This design ensures data reliability.

[0005] This solves the problem of inaccurate proxy data acquisition when sensors are abnormal. When sensors for fruit size, leaf nitrogen content, or soil moisture are abnormal, the lack of clear compensation rules can lead to low reliability of the proxy data, resulting in errors in growth stage determination and control instructions, and affecting the synergistic effect of water and fertilizer.

[0006] This solves the problem of inaccurate nitrogen content detection data from a single leaf. Traditional single-sensor detection is easily affected by factors such as leaf condition and position, failing to reflect the overall nitrogen status and leading to inappropriate fertilization strategies. This design improves detection accuracy through multi-node data collection and data fusion.

[0007] This model addresses the issue of leaf health affecting nitrogen content accuracy. Abnormal spectral data from leaves affected by pests, diseases, or mechanical damage can distort nitrogen content values ​​if included in the calculation, affecting growth stage determination and fertilization control. This model accurately eliminates invalid data.

[0008] This solves the problem of inaccurate identification of blade damage features. Identifying blade damage using a single method can easily lead to misjudgment, resulting in the elimination of valid blade data or the retention of invalid data, affecting the accuracy of nitrogen content detection. This unit improves identification reliability through the collaboration of multiple modules.

[0009] This solves the problem of sudden changes in fertilizer ratios during growth phases, which can affect mango growth. Sudden changes in fertilizer ratios at different stages can easily cause stress in fruit trees, affecting nutrient absorption and fruit development. The phased transition allows for a smooth transition in fertilizer ratios, minimizing adverse effects.

[0010] This solves the problem of uneven water and fertilizer distribution in the root zone. Traditional drip irrigation struggles to address the needs of root zones at varying depths, leading to localized excess or insufficient water or fertilizer, which can affect root absorption. This design improves the uniformity of water and fertilizer distribution through layered layout and pulsed irrigation.

[0011] This module addresses the issue of root zone salinization affecting root function. Salt accumulation can easily occur during irrigation and fertilization, especially during pulse irrigation, which can cause salt to rise, harming the root system and affecting absorption. This module monitors and promptly rinses salt to ensure root health.

[0012] In order to achieve these objects and other advantages according to the present invention, a mango fruit quality-oriented water and fertilizer coordinated control system is provided, characterized in that it includes: a fruit size monitoring unit, a leaf nitrogen content sensor, a soil moisture sensor, a central controller, a drip irrigation device, and a liquid fertilizer application device; The fruit size monitoring unit includes a multispectral imager deployed in the canopy of the fruit tree. The multispectral imager collects the surface image of the mango fruit, calculates the fruit's transverse diameter based on an image recognition algorithm, and transmits the fruit's transverse diameter value to the central controller. The leaf nitrogen content sensor detects the reflectance spectrum of the third fully expanded leaf at the top of the fruit tree in real time, outputs the leaf nitrogen content value through a preset spectrum-nitrogen content conversion model, and transmits the leaf nitrogen content value to the central controller; The soil moisture sensor is buried at the main root zone depth to monitor the soil moisture content in the root zone in real time and transmit it to the central controller; The central controller presets the fruit growth stage judgment logic: when the fruit diameter is between 3.0 cm and 5.0 cm for 5 consecutive days and the leaf nitrogen content is greater than 2.8%, it is judged to have entered the expansion stage; when the fruit diameter is greater than 5.0 cm and the leaf nitrogen content is less than 2.6% or when the fruit diameter is greater than 5.0 cm for 3 consecutive days, it is judged to have entered the maturity stage; if the current data does not meet the expansion or maturity stage judgment conditions, the previous stage control strategy is maintained; if it is the first start or there is no historical stage record, the basic irrigation strategy is implemented, with irrigation once a day, the soil moisture target value is 22%±2%, and nitrogen:phosphorus:potassium = 0.8:0.8:1.0. Among them, when it is first started, if the real-time fruit diameter is >3.0 cm, the expansion stage strategy is directly adopted; if it is >5.0 cm, the maturity stage strategy is adopted; The central controller executes phase-response control: When it determines that the plant has entered the swelling phase, the central controller generates a first control instruction, instructing the drip irrigation device to irrigate twice a day, with each irrigation maintaining the soil moisture content in the root zone at 28-32%. It also instructs the liquid fertilizer applicator to inject irrigation water at a nitrogen:phosphorus:potassium mass ratio of 1.2:0.8:1.5. When it determines that the plant has entered the maturity phase, the central controller generates a second control instruction, instructing the drip irrigation device to irrigate once a day, with each irrigation maintaining the soil moisture content in the root zone at 20%-24%. It also instructs the liquid fertilizer applicator to inject irrigation water at a nitrogen:phosphorus:potassium mass ratio of 0.5:0.5:2.0. The drip irrigation device and the liquid fertilizer dispensing device respond to the first control instruction or the second control instruction.

[0013] Preferably, the central controller further includes a sensor credibility verification module and a dynamic weight allocation unit: The sensor credibility verification module calculates the real-time confidence of each sensor based on historical data. When any sensor data deviates from the historical mean by more than 3 standard deviations for 2 consecutive hours, an abnormal flag is triggered; During the growth phase, the dynamic weight allocation unit automatically reduces the weight of abnormally marked sensor data to 0.3, maintains the weight of unmarked data at 1.0, and activates a multi-source data compensation mechanism: If the fruit size monitoring unit is abnormal, the visible light image of the same fruit tree canopy collected by the multispectral imager is analyzed, and a compensation value is output based on the pre-stored comparison table of leaf projection coverage and fruit size; If the leaf nitrogen content sensor is abnormal, the reflectance data of the multispectral imager in the 700-740nm band is extracted, and the compensation value is output based on the preset reflectance-nitrogen content level mapping table; If the soil moisture sensor is abnormal, the cumulative irrigation amount of the drip irrigation device is integrated with the canopy temperature change trend obtained by the multispectral imager, and the moisture content compensation value is calculated according to the preset temperature-evaporation association rule.

[0014] Preferably, the comparison table between leaf projection coverage and fruit size includes: when the coverage is ≥85%, the fruit transverse diameter compensation value is 5.0±0.2 cm; when the coverage is 70%-85%, the fruit transverse diameter compensation value is 4.2±0.3 cm; when the coverage is <70%, the fruit transverse diameter compensation value is 3.5±0.4 cm; The reflectivity-nitrogen content level mapping table includes: when the average reflectivity in the 700-740nm band is ≥35%, the output nitrogen content compensation value is 2.4%; when the average reflectivity is 30%-35%, the output nitrogen content compensation value is 2.8%; when the average reflectivity is <30%, the output nitrogen content compensation value is 3.1%; The temperature-evaporation association rules include: when the canopy temperature is continuously greater than 32°C for 4 hours, the daily evaporation is defined as 7-9 mm; when the canopy temperature is continuously between 28-32°C for 4 hours, the daily evaporation is defined as 5-6 mm; when the canopy temperature is continuously less than 28°C for 4 hours, the daily evaporation is defined as 3-4 mm.

[0015] Preferably, the leaf nitrogen content sensor comprises a plurality of micro-spectral sensing nodes, which are distributed and deployed at four different positions of the fruit tree canopy, and each sensing node focuses on the third fully expanded leaf at the top of the canopy at the corresponding position; The central controller is equipped with a blade nitrogen content fusion calculation module, which performs the following operations: Synchronously receive the reflection spectrum data detected by all orientation sensor nodes; Based on the preset leaf health assessment model, the spectral data of leaves affected by pests, diseases or mechanical damage are eliminated; The effective spectral data are weighted averaged according to the light exposure time of the corresponding leaves in the canopy, and the calculation results are input into the spectrum-nitrogen content conversion model to output the final leaf nitrogen content value.

[0016] Preferably, the blade health assessment model includes: A multi-band anomaly detection module identifies abnormal chlorophyll degradation or cell structure damage based on the reflectance ratio of the leaf reflectance spectrum at 680nm, 750nm, and 1650nm. The time series change analysis module is used to compare the current detection spectrum with the historical spectrum data of the same leaf within 72 hours. If the reflectance of the 450nm blue light band increases by more than 15% and the reflectance of the 850nm near-infrared band decreases by more than 10%, it is judged as a progressive disease; An auxiliary image verification unit, which includes a micro-camera installed at each sensor node to capture leaf surface images and identify the morphological characteristics of disease spots, insect holes or mechanical damage through a convolutional neural network; Among them, when the multi-band anomaly detection module outputs an abnormal chlorophyll degradation signal or a cell structure damage signal, and the temporal change analysis module outputs a progressive disease judgment, and the auxiliary image verification unit identifies pathological damage characteristics or mechanical damage characteristics, the leaf is judged as an invalid sample and its spectral data is discarded.

[0017] Preferably, the auxiliary image verification unit is further configured with: Multi-spectral narrow-band light source module, which emits narrow-band near-infrared light with center wavelengths of 720nm±5nm and 950nm±10nm respectively; The time-series damage analysis module is used to perform the following operations on continuously captured time-series images of the same leaf: extract the pixel diffusion rate of the diseased area in the 720nm band image, and generate an active disease mark when the diffusion rate is greater than 5 pixels / hour; calculate the texture entropy value change rate of the wormhole edge in the 950nm band image, and determine that mechanical damage is expanding if the entropy value change rate exceeds 0.15 / minute; Dynamic verification logic unit, which is configured as follows: When the following conditions are met at the same time, it is determined that the blade has effective damage characteristics: (i) Convolutional neural network identifies the morphological features of lesions / wormholes / mechanical damage; (ii) the temporal damage analysis module outputs an active disease marker or a mechanical damage extension determination; (iii) The multi-band anomaly detection module detects an anomaly in the reflectivity of one of the 680nm, 750nm, and 1650nm bands.

[0018] Preferably, the liquid fertilizer dispensing device comprises parallel A / B fertilizer liquid tanks and a dynamic fertilizer mixing unit; When the central controller switches control instructions, it performs a phased transition operation: During the pre-flush phase, irrigation water is injected into the irrigation pipe at a volume of 30% of the pipe volume, and no fertilizer is added; During the buffer injection phase, a transition buffer mother solution was mixed with irrigation water in a certain proportion. The transition buffer mother solution contained 60 g / L nitrogen, 50 g / L phosphorus, and 120 g / L potassium. After mixing, the nitrogen, phosphorus, and potassium mass concentration ratios in the irrigation solution were 0.6:0.5:1.2. During the gradient switching phase, after the end of the buffer injection phase, the buffer injection volume decreases day by day by 80% of the total volume of the previous day during daily irrigation operations; the target fertilizer solution injection volume increases accordingly to make up for the reduction in buffer volume; the transition is terminated when the buffer injection volume drops to 5% or less of the initial value.

[0019] Preferably, the drip irrigation device comprises a pressure-compensated drip arrow matrix, which is arranged in two layers of deep and shallow layers according to the three-dimensional topological structure of the main root area of ​​the fruit tree; The central controller is connected to the soil conductivity sensor array to obtain the water and fertilizer distribution map of the root zone profile in real time; When irrigation with a water content of 28-32% is performed during the expansion period, if the difference in water content between the bottom layer and the surface layer of the root zone is greater than 5%, the pulse irrigation strategy is triggered: a single irrigation is divided into three short pulse irrigations with an interval of 10 minutes. After each pulse, the fertilizer solution concentration of the next pulse is dynamically adjusted based on the feedback from the conductivity sensor.

[0020] Preferably, the central controller is further connected to a salt migration monitoring module, which is configured to: The conductivity-salt dynamic modeling unit constructs a root zone profile salt distribution heat map based on real-time data from the soil conductivity sensor array. This model is then combined with historical irrigation parameters from the drip irrigation system and fertilizer injection records from the liquid fertilizer dispenser to establish a salt migration and accumulation model. The salinization risk warning unit, during the pulse irrigation process, if there is a salt upwelling risk mark, the central controller automatically inserts a leaching subroutine between adjacent pulse intervals. The leaching subroutine is: Send a zero fertilizer liquid instruction to the liquid fertilizer dispensing device to generate a pure water irrigation pulse; Control the drip irrigation device to increase the irrigation volume by 20% in the second pulse phase and extend the pulse interval to 15 minutes; The fertilizer solution concentration of the third pulse is dynamically adjusted based on the updated conductivity distribution map to reduce the bottom salt concentration to below the risk threshold.

[0021] The present invention has at least the following beneficial effects: First, the present invention precisely collects key data through a fruit size monitoring unit, leaf nitrogen content sensors, and soil moisture sensors. A central controller determines the fruit growth stage based on preset logic and issues targeted control instructions, achieving intelligent, coordinated regulation of water and fertilizer. This dynamic, fruit-quality-oriented control model precisely adjusts irrigation frequency, soil moisture content, and fertilizer ratios based on the needs of different fruit growth stages. This not only meets the mango's high demand for water, nitrogen, and potassium during the expansion phase, but also accommodates the demand for water control and high-potassium fertilizers during the ripening phase. This effectively avoids the problem of fruit quality decline caused by an imbalance in water and fertilizer supply, significantly improving mango yield and quality.

[0022] Second, the sensor credibility verification module and dynamic weight allocation unit set in the central controller of the present invention further improve the stability and reliability of the system. Sensor credibility verification can promptly detect and mark abnormal data, preventing erroneous data from affecting the determination of the growth stage; dynamic weight allocation, when the data is abnormal, adjusts the weight and activates the multi-source data compensation mechanism, using other relevant data to supplement the abnormal data, thereby ensuring the accuracy of the growth stage determination. This design allows the system to continue to operate normally when the sensor has temporary failures or data fluctuations, reducing control errors caused by sensor problems, ensuring the continuous effectiveness of the water and fertilizer control strategy, and providing a more stable environment for mango growth.

[0023] Third, the present invention clarifies the leaf projection coverage and fruit size comparison table, the reflectivity-nitrogen content level mapping table, and the temperature-evaporation association rules, making the multi-source data compensation mechanism more operational and accurate. When the fruit size monitoring unit, leaf nitrogen content sensor, or soil moisture sensor is abnormal, the compensation value can be quickly obtained based on these specific corresponding relationships, ensuring the reliability of the alternative data and, in turn, the accuracy of the central controller's judgment and control instructions on the growth stage. This detailed compensation rule design enhances the system's ability to respond to sensor anomalies, ensures the consistency of water and fertilizer regulation, and helps mangoes obtain appropriate water and fertilizer supply at different growth stages.

[0024] Fourth, the leaf nitrogen content sensor of the present invention adopts a distributed deployment of multiple micro-spectral sensing nodes, and processes data through the leaf nitrogen content fusion calculation module of the central controller, which effectively improves the accuracy of leaf nitrogen content detection. Collecting data from multiple nodes can avoid the limitations of a single node, eliminate the spectral data of leaves affected by pests and diseases or mechanical damage, and then perform weighted averaging based on the length of leaf light exposure, so that the final output leaf nitrogen content value can more truly reflect the nitrogen nutritional status of the fruit tree. Accurate nitrogen content data provides a reliable basis for the central controller to determine the growth stage and formulate fertilization strategies, which helps to accurately control the proportion of nitrogen in fertilizers, meet the nitrogen needs of mango growth, and promote the improvement of fruit quality.

[0025] Other advantages, objectives and features of the present invention will be reflected in part from the following description and will be understood by those skilled in the art through study and practice of the present invention. DETAILED DESCRIPTION

[0026] The present invention is further described in detail below with reference to the embodiments so that those skilled in the art can implement the invention with reference to the description.

[0027] According to one embodiment of the present invention, a mango fruit quality-oriented water and fertilizer coordinated control system includes: a fruit size monitoring unit, a leaf nitrogen content sensor, a soil moisture sensor, a central controller, a drip irrigation device, and a liquid fertilizer application device; The fruit size monitoring unit includes a multispectral imager deployed in the canopy of the fruit tree. The multispectral imager collects the surface image of the mango fruit, calculates the fruit's transverse diameter based on an image recognition algorithm, and transmits the fruit's transverse diameter value to the central controller. The leaf nitrogen content sensor detects the reflectance spectrum of the third fully expanded leaf at the top of the fruit tree in real time, outputs the leaf nitrogen content value through a preset spectrum-nitrogen content conversion model, and transmits the leaf nitrogen content value to the central controller; The soil moisture sensor is buried at the main root zone depth to monitor the soil moisture content in the root zone in real time and transmit it to the central controller; The central controller presets the fruit growth stage judgment logic: when the fruit diameter is between 3.0 cm and 5.0 cm for 5 consecutive days and the leaf nitrogen content is greater than 2.8%, it is judged to have entered the expansion stage; when the fruit diameter is greater than 5.0 cm and the leaf nitrogen content is less than 2.6%, or when the fruit diameter is greater than 5.0 cm for 3 consecutive days, it is judged to have entered the maturity stage; if the current data does not meet the expansion or maturity stage judgment conditions, the previous stage control strategy is maintained; if it is the first start or there is no historical stage record, the basic irrigation strategy is implemented, with irrigation once a day, the soil moisture content (volume moisture content) target value of 22% ± 2%, and nitrogen: phosphorus: potassium = 0.8:0.8:1.0. Among them, when it is first started, if the real-time fruit diameter is >3.0 cm, the expansion stage strategy is directly adopted; if it is >5.0 cm, the maturity stage strategy is adopted; The central controller executes phase-response control: When it determines that the plant has entered the swelling phase, the central controller generates a first control instruction, instructing the drip irrigation device to irrigate twice a day, with each irrigation maintaining the soil moisture content in the root zone at 28-32%. It also instructs the liquid fertilizer applicator to inject irrigation water at a nitrogen:phosphorus:potassium mass ratio of 1.2:0.8:1.5. When it determines that the plant has entered the maturity phase, the central controller generates a second control instruction, instructing the drip irrigation device to irrigate once a day, with each irrigation maintaining the soil moisture content in the root zone at 20%-24%. It also instructs the liquid fertilizer applicator to inject irrigation water at a nitrogen:phosphorus:potassium mass ratio of 0.5:0.5:2.0. The drip irrigation device and the liquid fertilizer dispensing device respond to the first control instruction or the second control instruction.

[0028] In this technical solution, the fruit size monitoring unit uses a multispectral imager with a wavelength range of 400-1000nm. It can be mounted on an adjustable stand 1.5 meters above the fruit tree canopy. The device collects fruit images every hour from 6:00 AM to 6:00 PM daily, uses the YOLOv5 algorithm to identify the fruit outline and calculate the transverse diameter, with a measurement accuracy of ±0.2 cm. The leaf nitrogen content sensor can use a 650-1100nm miniature fiber optic spectrometer. The probe is fixed 10 cm from the third fully expanded leaf from the top. The spectral data is converted into nitrogen content using a partial least squares regression model. The soil moisture sensor uses a frequency domain reflectometry probe and is buried 30 cm deep in the taproot zone. The central controller has a pre-set growth stage determination program: The expansion phase is determined by fruit diameters between 3.0 and 5.0 cm for five consecutive days and leaf nitrogen content >2.8%. Maturity is determined using a two-conditional logic (fruit diameter >5.0 cm and leaf nitrogen content <2.6%, or diameter >5.0 cm for three consecutive days). The expansion phase control strategy involves daily irrigation at 7:00 AM and 3:00 PM, with each irrigation volume ensuring a soil moisture content of 28-32% at a depth of 30 cm. Water-soluble fertilizer is also administered at a nitrogen:phosphorus:potassium ratio of 1.2:0.8:1.5. During maturity, a single irrigation at 10:00 AM is implemented daily, maintaining a soil moisture content of 20-24%, and adjusting the fertilizer ratio to 0.5:0.5:2.0.

[0029] This technical solution uses a fruit size monitoring unit, leaf nitrogen content sensors, and soil moisture sensors to precisely collect key data. A central controller determines the fruit growth stage based on preset logic and issues targeted control instructions, achieving intelligent, coordinated regulation of water and fertilizer. This dynamic, fruit-quality-oriented control model precisely adjusts irrigation frequency, soil moisture content, and fertilizer ratios based on the needs of different fruit growth stages. This not only meets the mango's high demand for water, nitrogen, and potassium during the expansion phase, but also accommodates the requirements for water control and high-potassium fertilizers during the maturity phase. This effectively avoids the problem of fruit quality decline caused by an imbalance in water and fertilizer supply, significantly improving mango yield and quality.

[0030] According to another embodiment of the present invention, the central controller further includes a sensor credibility verification module and a dynamic weight allocation unit: The sensor credibility verification module calculates the real-time confidence of each sensor based on historical data. When any sensor data deviates from the historical mean by more than 3 standard deviations for 2 consecutive hours, an abnormal flag is triggered; During the growth phase, the dynamic weight allocation unit automatically reduces the weight of abnormally marked sensor data to 0.3, maintains the weight of unmarked data at 1.0, and activates a multi-source data compensation mechanism: If the fruit size monitoring unit is abnormal, the visible light image of the same fruit tree canopy collected by the multispectral imager is analyzed, and a compensation value is output based on the pre-stored comparison table of leaf projection coverage and fruit size; If the leaf nitrogen content sensor is abnormal, the reflectance data of the multispectral imager in the 700-740nm band is extracted, and the compensation value is output based on the preset reflectance-nitrogen content level mapping table; If the soil moisture sensor is abnormal, the system combines the cumulative irrigation volume of the drip irrigation system with the canopy temperature trend obtained by the multispectral imager to calculate a moisture content compensation value based on a preset temperature-evaporation correlation rule. In this technical solution, the central controller has a built-in sensor credibility verification module that establishes a 30-day moving average database. An abnormality flag is triggered when leaf nitrogen content data deviates from the mean by ±3 standard deviations for two consecutive hours (for example, a sudden change from the normal range of 2.5%-3.2% to 1.8% or 3.8%). The dynamic weight allocation unit automatically reduces the weight of abnormal data to 0.3, while the weight of valid data remains at 1.0. The multispectral imager collects canopy images in visible light mode (400-700nm), and the leaf projection coverage is calculated using the OpenCV library. A compensation parameter of 70%-85% coverage corresponds to a fruit diameter of 4.2±0.3 cm. When a leaf sensor detects an anomaly, reflectance data in the 720nm band is extracted, and a compensation value of 2.8% nitrogen content corresponding to a reflectance of 30%-35% can be selected. Soil moisture compensation uses a digital temperature sensor to monitor canopy temperature. When the temperature remains above 32°C for four consecutive hours, daily evaporation is calculated as 8±1 mm according to the FAO-56 model. Combined with the cumulative irrigation data from the drip irrigation system, the root zone moisture content is estimated using a mass balance formula. This technical solution, combined with the sensor credibility verification module and dynamic weight allocation unit incorporated into the central controller of the present invention, further improves the stability and reliability of the system. The sensor credibility verification module promptly detects and flags abnormal data, preventing erroneous data from affecting growth stage determination. The dynamic weight allocation, when abnormal data is detected, adjusts weights and activates a multi-source data compensation mechanism, supplementing the abnormal data with other relevant data to ensure accurate growth stage determination. This design allows the system to function properly even in the event of temporary sensor failures or data fluctuations, reducing control errors caused by sensor issues, ensuring the continued effectiveness of water and fertilizer control strategies, and providing a more stable environment for mango growth.

[0031] According to another embodiment of the present invention, the leaf projection coverage and fruit size comparison table includes: when the coverage is ≥85%, the fruit transverse diameter compensation value is 5.0±0.2 cm (based on the measured data of the Tainong Mang variety); when the coverage is 70%-85%, the fruit transverse diameter compensation value is 4.2±0.3 cm; when the coverage is <70%, the fruit transverse diameter compensation value is 3.5±0.4 cm; The reflectivity-nitrogen content level mapping table includes: when the average reflectivity in the 700-740nm band is ≥35%, the output nitrogen content compensation value is 2.4%; when the average reflectivity is 30%-35%, the output nitrogen content compensation value is 2.8%; when the average reflectivity is <30%, the output nitrogen content compensation value is 3.1%; The temperature-evaporation association rules include: when the canopy temperature is >32°C for four consecutive hours, the daily evaporation is defined as 7-9 mm; when the canopy temperature is 28-32°C for four consecutive hours, the daily evaporation is defined as 5-6 mm; and when the canopy temperature is <28°C for four consecutive hours, the daily evaporation is defined as 3-4 mm. In this technical solution, leaf projected cover is calculated using an RGB image segmentation algorithm; soil moisture compensation is calculated using the formula: current moisture content = value at the end of the previous irrigation - (accumulated evaporation) × 0.7 correction factor. These compensation parameters are established based on three years of field trials. They maintain accurate growth stage determination even under abnormal sensor operating conditions and avoid interruptions in water and fertilizer regulation due to data loss. Using this technical solution, the present invention defines a table comparing leaf projected cover with fruit size, a table mapping reflectance and nitrogen content, and temperature-evaporation association rules, making the multi-source data compensation mechanism more operational and accurate. When the fruit size monitoring unit, leaf nitrogen content sensor, or soil moisture sensor detects an anomaly, compensation values ​​are quickly derived based on these specific corresponding relationships, ensuring the reliability of the replacement data and, in turn, the accuracy of the central controller's growth stage determination and control instructions. This meticulous compensation rule design enhances the system's ability to cope with sensor anomalies, ensures consistent water and fertilizer regulation, and ensures that mangoes receive the appropriate water and fertilizer supply at different growth stages.

[0032] According to another embodiment of the present invention, the leaf nitrogen content sensor comprises a plurality of micro-spectral sensing nodes distributedly deployed at four different locations of the fruit tree canopy, and each sensing node focuses on the third fully expanded leaf at the top of the canopy at the corresponding location; The central controller is equipped with a blade nitrogen content fusion calculation module, which performs the following operations: Synchronously receive the reflection spectrum data detected by all orientation sensor nodes; Based on the preset leaf health assessment model, the spectral data of leaves affected by pests, diseases or mechanical damage are eliminated; A weighted average calculation is performed on the effective spectral data based on the corresponding leaf's light exposure duration in the canopy. The calculated results are input into the spectrum-to-nitrogen content conversion model to output the final leaf nitrogen content value. In this technical solution, the micro-spectral sensing nodes can utilize a three-chip spectral module and be installed 50 cm from the terminal bud in the east, south, west, and north directions of the tree canopy. Each node uses an adjustable focus lens to focus on the third fully expanded leaf from the top layer in its location, and the sampling interval is set to 30 minutes. The light weight coefficient is dynamically calculated based on the sun track data from the meteorological station. The weight of data collected during the midday period (10:00-14:00) is increased to 1.2, while the weight of data collected during cloudy days is reduced to 0.8. Using this technical solution, the leaf nitrogen content sensor of the present invention utilizes a distributed deployment of multiple micro-spectral sensing nodes, and the data is processed by the leaf nitrogen content fusion calculation module of the central controller, effectively improving the accuracy of leaf nitrogen content detection. Collecting data from multiple nodes avoids the limitations of a single node. Spectral data from leaves affected by pests, diseases, or mechanical damage is eliminated, and a weighted average is calculated based on the leaf's light exposure duration. This output ensures that the leaf nitrogen content more accurately reflects the tree's nitrogen nutritional status. Accurate nitrogen content data provides a reliable basis for the central controller to determine growth stage and formulate fertilization strategies. This helps precisely control the nitrogen content in fertilizers, meeting the nitrogen requirements of mangoes and improving fruit quality.

[0033] According to another embodiment of the present invention, the blade health assessment model includes: A multi-band anomaly detection module identifies abnormal chlorophyll degradation or cell structure damage based on the reflectance ratio of the leaf reflectance spectrum at 680nm, 750nm, and 1650nm. The time series change analysis module is used to compare the current detection spectrum with the historical spectrum data of the same leaf within 72 hours. If the reflectance of the 450nm blue light band increases by more than 15% and the reflectance of the 850nm near-infrared band decreases by more than 10%, it is judged as a progressive disease; An auxiliary image verification unit, which includes a micro-camera installed at each sensor node to capture leaf surface images and identify the morphological characteristics of disease spots, insect holes or mechanical damage through a convolutional neural network; When the multi-band anomaly detection module outputs signals of abnormal chlorophyll degradation or cellular structural damage, the temporal change analysis module outputs a progressive disease determination, and the auxiliary image verification unit identifies pathological or mechanical damage, the leaf is deemed invalid and its spectral data is discarded. In this technical solution, multi-band detection sets a 680nm / 750nm reflectance ratio threshold of 1.25, and a 1650nm reflectance greater than 40% is used to identify cellular structural damage. Temporal analysis utilizes a 72-hour sliding window comparison, triggering a disease flag when the 450nm reflectance increases by >15% and the 850nm reflectance decreases by >10%. The micro-camera can be equipped with a sensor and, in conjunction with a ring light, allows for disease spot detection with a minimum detection area of ​​0.5mm², and insect hole detection accuracy of up to 0.2mm. Leaf data is automatically discarded when spectral anomaly, temporal change, and image recognition meet all three criteria. By adopting this technical solution, the leaf health assessment model of the present invention can accurately identify invalid leaf samples and eliminate their spectral data through the collaborative work of the multi-band anomaly detection module, the time series change analysis module and the auxiliary image verification unit. The combination of multi-band reflectance ratio analysis, time series spectrum comparison and leaf surface image recognition can comprehensively and accurately determine whether the leaves have problems such as abnormal chlorophyll degradation, progressive diseases or mechanical damage. After eliminating invalid data, the remaining valid spectral data can more realistically reflect the nitrogen content of the leaves, provide a more reliable basis for the decision-making of the central controller, ensure the accuracy of leaf nitrogen content detection, and thus make the water and fertilizer regulation strategy more in line with the actual nutritional needs of fruit trees.

[0034] According to yet another embodiment of the present invention, the auxiliary image verification unit is further configured with: Multi-spectral narrow-band light source module, which emits narrow-band near-infrared light with center wavelengths of 720nm±5nm and 950nm±10nm respectively; The time-series damage analysis module is used to perform the following operations on continuously captured time-series images of the same leaf: extract the pixel diffusion rate of the diseased area in the 720nm band image, and generate an active disease mark when the diffusion rate is greater than 5 pixels / hour; calculate the texture entropy value change rate of the wormhole edge in the 950nm band image, and determine that mechanical damage is expanding if the entropy value change rate exceeds 0.15 / minute; Dynamic verification logic unit, which is configured as follows: When the following conditions are met at the same time, it is determined that the blade has effective damage characteristics: (i) Convolutional neural network identifies the morphological features of lesions / wormholes / mechanical damage; (ii) the temporal damage analysis module outputs an active disease marker or a mechanical damage extension determination; (iii) The multi-band anomaly detection module detects reflectivity anomalies in one of the 680nm, 750nm, and 1650nm bands. In this technical solution, the narrowband light source can be a dual-wavelength LED array of 720±5nm and 950±10nm, with the illumination intensity set to 1000 lux. The lesion diffusion rate is calculated using the continuous image frame difference method, and when the diffusion rate is greater than 5 pixels / hour, it is marked as an active disease. The entropy analysis of the wormhole texture uses the gray-level co-occurrence matrix algorithm, and the entropy value change rate threshold of 0.15 / minute is determined based on mechanical damage testing. Effective damage determination requires the simultaneous fulfillment of three conditions: the convolutional neural network identifies typical damage morphology, the time series analysis confirms damage extension, and the spectrum detects characteristic band anomalies. Using this technical solution, the auxiliary image verification unit is equipped with a multi-spectral narrowband light source module, a time series damage analysis module, and a dynamic verification logic unit, further improving the accuracy and reliability of leaf damage feature identification. A multi-spectral narrowband light source helps more clearly capture leaf damage signatures. Time-series damage analysis determines whether damage is active or expanding, and a dynamic verification logic unit comprehensively considers various factors to determine whether a leaf has valid damage signatures. This rigorous verification mechanism effectively avoids misjudgments, ensuring that rejected leaf samples are indeed invalid, and thus ensuring the validity of leaf nitrogen content data, providing strong support for precise water and fertilizer regulation.

[0035] According to yet another embodiment of the present invention, the liquid fertilizer dispensing device comprises parallel A / B fertilizer liquid tanks and a dynamic fertilizer mixing unit; When the central controller switches control instructions, it performs a phased transition operation: During the pre-flush phase, irrigation water is injected into the irrigation pipe at a volume of 30% of the pipe volume, and no fertilizer is added; During the buffer injection phase, a transition buffer mother solution was mixed with irrigation water in a certain proportion. The transition buffer mother solution contained 60 g / L nitrogen, 50 g / L phosphorus, and 120 g / L potassium. After mixing, the nitrogen, phosphorus, and potassium mass concentration ratios in the irrigation solution were 0.6:0.5:1.2. During the gradient switching phase, after the buffer injection phase ends, the buffer injection rate decreases daily by 80% of the previous day's total during each irrigation operation. The target fertilizer injection rate increases accordingly to compensate for the buffer reduction. The transition ends when the buffer injection rate drops to 5% or less of the initial value. In this technology, during the pre-flush phase, 30% of the pipe volume (approximately 5-8 liters in a typical system) is injected with clean water. The gradient switching utilizes a PID control algorithm, with the buffer rate decreasing daily by 80% of the previous day's flow rate and the target fertilizer injection rate increasing proportionally to the shortfall. For example, if the initial buffer flow rate is 10 L / day, the buffer rate drops to 8 L after the first day of the switch, and the target fertilizer injection rate is 2 L. The entire transition cycle lasts 3-5 days and ends when the buffer flow rate drops below 5% of the initial value (0.5 L). This technical solution allows the liquid fertilizer distribution device to implement a phased transition through pre-flush, buffer injection, and gradient switching, ensuring a smooth transition in fertilizer ratios when the central controller switches control instructions. Pre-flushing can avoid adverse reactions caused by mixing of fertilizers at different stages in the pipeline. Buffer and gradient switching prevents sudden changes in fertilizer ratios from causing stress on mango growth, allowing fruit trees to gradually adapt to the new fertilizer ratio. This ensures the stability of nutrient supply for mangoes during the transition period of growth stages, is conducive to the continued healthy growth of the fruit, and reduces quality degradation caused by sudden changes in fertilizer ratios.

[0036] According to another embodiment of the present invention, the drip irrigation device includes a pressure-compensated drip arrow matrix, which is arranged in two layers of deep and shallow layers according to the three-dimensional topological structure of the main root area of ​​the fruit tree; The central controller is connected to the soil conductivity sensor array to obtain the water and fertilizer distribution map of the root zone profile in real time; During the bulking phase, when irrigation is performed at a water content of 28-32%, if the difference between the bottom and top layers of the root zone exceeds 5%, a pulse irrigation strategy is triggered: a single irrigation session is split into three short pulses, each 10 minutes apart. After each pulse, the fertilizer concentration is dynamically adjusted based on conductivity sensor feedback. In this technical solution, a dropper array is layered according to a three-dimensional model of the fruit tree's root system: shallow dropper arrays (20 cm depth) are spaced 50 cm apart, while deep dropper arrays (40 cm depth) are spaced 80 cm apart. The conductivity sensor array forms a 10×10 cm grid of monitoring points in the root zone. When the water content difference between the 40 cm and 20 cm depths exceeds 5% (e.g., 28% at the bottom layer vs. 22% at the top layer), a single 30 L irrigation session is split into three 10 L pulses. The 10-minute pulse interval was determined based on infiltration tests in sandy loam soil, and each pulse lasts approximately 15 minutes. After each pulse, the fertilizer concentration is dynamically adjusted via a proportional valve based on the updated conductivity profile, with the adjustment range not exceeding ±20% of the base concentration. Using this technical solution, the drip irrigation system of the present invention utilizes a pressure-compensated drip-arrow matrix arranged in two layers, deep and shallow. This, combined with a soil conductivity sensor array and a pulse irrigation strategy, improves the uniformity and accuracy of irrigation and fertilization. The deep and shallow layers provide more comprehensive coverage of the taproot zone. When there is a significant difference in moisture content between the bottom and surface layers of the root zone, the pulse irrigation strategy splits the irrigation times and dynamically adjusts the fertilizer concentration, achieving a more even distribution of soil moisture and fertilizer within the root zone, avoiding localized excess or deficiency of moisture or fertilizer. This design ensures that the mango roots can evenly absorb water and nutrients, promoting uniform fruit growth and improving fruit quality.

[0037] According to another embodiment of the present invention, the central controller is further connected to a salt migration monitoring module, which is configured to: The conductivity-salt dynamic modeling unit constructs a root zone profile salt distribution heat map based on real-time data from the soil conductivity sensor array. This model is then combined with historical irrigation parameters from the drip irrigation system and fertilizer injection records from the liquid fertilizer dispenser to establish a salt migration and accumulation model. The salinization risk warning unit, during the pulse irrigation process, if there is a salt upwelling risk mark, the central controller automatically inserts a leaching subroutine between adjacent pulse intervals. The leaching subroutine is: Send a zero fertilizer liquid instruction to the liquid fertilizer dispensing device to generate a pure water irrigation pulse; Control the drip irrigation device to increase the irrigation volume by 20% in the second pulse phase and extend the pulse interval to 15 minutes; Based on the updated conductivity profile, the fertilizer concentration in the third pulse is dynamically adjusted to reduce the bottom layer salt concentration to below the risk threshold. In this technical solution, the salt migration monitoring module builds a model based on HYDRUS-2D software, and uses a field-calibrated curve for conductivity-salinity conversion (EC = 1.5 mS / cm corresponds to 0.1% salt). When the salt concentration in the surface layer (0-10 cm) exceeds that in the bottom layer (30-40 cm) by 15%, a salt upwelling warning is triggered. The rinsing process is then executed in the second pulse phase: the pure water injection volume is increased by 20% (for example, from 10 L to 12 L), and the pulse interval is extended to 15 minutes. Simultaneously, the fertilizer injection system is paused to allow irrigation water to fully rinse the salt accumulation above the root zone. After the rinsing phase, the compensation fertilizer concentration for the third pulse phase is automatically calculated based on feedback from the conductivity sensor. Using this technical solution, the salt migration monitoring module, connected to the central controller, effectively monitors salt distribution in the root zone and promptly responds to salinization risks through its conductivity-salinity dynamic modeling unit and salinization risk warning unit. During pulse irrigation, if there's a risk of salt upwelling, the system automatically initiates a rinsing subroutine. This process involves pulsed pure water irrigation, increasing water volume, and adjusting fertilizer concentration to reduce the salt concentration in the bottom layer, preventing salt accumulation from damaging the mango roots. This function ensures a healthy root zone soil environment, promoting proper water and nutrient absorption by the roots, providing favorable conditions for mango growth and ultimately improving fruit quality.

[0038] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the description and implementation methods. They can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and embodiments shown and described herein.

Claims

1. A mango fruit quality-oriented water and fertilizer coordinated control system, characterized in that: include: Fruit size monitoring unit, leaf nitrogen content sensor, soil moisture sensor, central controller, drip irrigation device and liquid fertilizer application device; The fruit size monitoring unit includes a multispectral imager deployed in the canopy of the fruit tree. The multispectral imager collects the surface image of the mango fruit, calculates the fruit's transverse diameter based on an image recognition algorithm, and transmits the fruit's transverse diameter value to the central controller. The leaf nitrogen content sensor detects the reflectance spectrum of the third fully expanded leaf at the top of the fruit tree in real time, outputs the leaf nitrogen content value through a preset spectrum-nitrogen content conversion model, and transmits the leaf nitrogen content value to the central controller; The soil moisture sensor is buried at the main root zone depth to monitor the soil moisture content in the root zone in real time and transmit it to the central controller; The central controller presets the fruit growth stage judgment logic: when the fruit diameter is between 3.0 cm and 5.0 cm for 5 consecutive days and the leaf nitrogen content is greater than 2.8%, it is judged to have entered the expansion stage; when the fruit diameter is greater than 5.0 cm and the leaf nitrogen content is less than 2.6% or when the fruit diameter is greater than 5.0 cm for 3 consecutive days, it is judged to have entered the maturity stage; if the current data does not meet the expansion or maturity stage judgment conditions, the previous stage control strategy is maintained; if it is the first start or there is no historical stage record, the basic irrigation strategy is implemented, with irrigation once a day, the soil moisture target value is 22%±2%, and nitrogen:phosphorus:potassium = 0.8:0.8:1.

0. Among them, when it is first started, if the real-time fruit diameter is >3.0 cm, the expansion stage strategy is directly adopted; if it is >5.0 cm, the maturity stage strategy is adopted; The central controller executes phase-response control: When it determines that the plant has entered the swelling phase, the central controller generates a first control instruction, instructing the drip irrigation device to irrigate twice a day, with each irrigation maintaining the soil moisture content in the root zone at 28-32%. It also instructs the liquid fertilizer applicator to inject irrigation water at a nitrogen:phosphorus:potassium mass ratio of 1.2:0.8:1.

5. When it determines that the plant has entered the maturity phase, the central controller generates a second control instruction, instructing the drip irrigation device to irrigate once a day, with each irrigation maintaining the soil moisture content in the root zone at 20%-24%. It also instructs the liquid fertilizer applicator to inject irrigation water at a nitrogen:phosphorus:potassium mass ratio of 0.5:0.5:2.

0. The drip irrigation device and the liquid fertilizer dispensing device respond to the first control instruction or the second control instruction.

2. The mango fruit quality-oriented water and fertilizer coordinated control system according to claim 1, characterized in that: The central controller further includes a sensor credibility verification module and a dynamic weight allocation unit: The sensor credibility verification module calculates the real-time confidence of each sensor based on historical data. When any sensor data deviates from the historical mean by more than 3 standard deviations for 2 consecutive hours, an abnormal flag is triggered; During the growth phase, the dynamic weight allocation unit automatically reduces the weight of abnormally marked sensor data to 0.3, maintains the weight of unmarked data at 1.0, and activates a multi-source data compensation mechanism: If the fruit size monitoring unit is abnormal, the visible light image of the same fruit tree canopy collected by the multispectral imager is analyzed, and a compensation value is output based on the pre-stored comparison table of leaf projection coverage and fruit size; If the leaf nitrogen content sensor is abnormal, the reflectance data of the multispectral imager in the 700-740nm band is extracted, and the compensation value is output based on the preset reflectance-nitrogen content level mapping table; If the soil moisture sensor is abnormal, the cumulative irrigation amount of the drip irrigation device is integrated with the canopy temperature change trend obtained by the multispectral imager, and the moisture content compensation value is calculated according to the preset temperature-evaporation association rule.

3. The mango fruit quality-oriented water-fertilizer coordinated control system according to claim 2, characterized in that: The leaf projection coverage and fruit size comparison table includes: when the coverage is ≥85%, the fruit transverse diameter compensation value is 5.0±0.2 cm; when the coverage is 70%-85%, the fruit transverse diameter compensation value is 4.2±0.3 cm; when the coverage is <70%, the fruit transverse diameter compensation value is 3.5±0.4 cm; The reflectivity-nitrogen content level mapping table includes: when the average reflectivity in the 700-740nm band is ≥35%, the output nitrogen content compensation value is 2.4%; when the average reflectivity is 30%-35%, the output nitrogen content compensation value is 2.8%; when the average reflectivity is <30%, the output nitrogen content compensation value is 3.1%; The temperature-evaporation association rules include: when the canopy temperature is continuously greater than 32°C for 4 hours, the daily evaporation is defined as 7-9 mm; when the canopy temperature is continuously between 28-32°C for 4 hours, the daily evaporation is defined as 5-6 mm; when the canopy temperature is continuously less than 28°C for 4 hours, the daily evaporation is defined as 3-4 mm.

4. The mango fruit quality-oriented water and fertilizer coordinated control system according to claim 1, characterized in that: The leaf nitrogen content sensor comprises a plurality of micro-spectral sensing nodes, which are distributed and deployed at four different positions of the fruit tree canopy, and each sensing node focuses on the third fully expanded leaf at the top of the canopy at the corresponding position; The central controller is equipped with a blade nitrogen content fusion calculation module, which performs the following operations: Synchronously receive the reflection spectrum data detected by all orientation sensor nodes; Based on the preset leaf health assessment model, the spectral data of leaves affected by pests, diseases or mechanical damage are eliminated; The effective spectral data are weighted averaged according to the light exposure time of the corresponding leaves in the canopy, and the calculation results are input into the spectrum-nitrogen content conversion model to output the final leaf nitrogen content value.

5. The mango fruit quality-oriented water and fertilizer coordinated control system according to claim 4, characterized in that: The blade health assessment model includes: A multi-band anomaly detection module identifies abnormal chlorophyll degradation or cell structure damage based on the reflectance ratio of the leaf reflectance spectrum at 680nm, 750nm, and 1650nm. The time series change analysis module is used to compare the current detection spectrum with the historical spectrum data of the same leaf within 72 hours. If the reflectance of the 450nm blue light band increases by more than 15% and the reflectance of the 850nm near-infrared band decreases by more than 10%, it is judged as a progressive disease; An auxiliary image verification unit, which includes a micro-camera installed at each sensor node to capture leaf surface images and identify the morphological characteristics of disease spots, insect holes or mechanical damage through a convolutional neural network; Among them, when the multi-band anomaly detection module outputs an abnormal chlorophyll degradation signal or a cell structure damage signal, and the temporal change analysis module outputs a progressive disease judgment, and the auxiliary image verification unit identifies pathological damage characteristics or mechanical damage characteristics, the leaf is judged as an invalid sample and its spectral data is discarded.

6. The mango fruit quality-oriented water and fertilizer coordinated control system according to claim 5, characterized in that: The auxiliary image verification unit is further configured with: Multi-spectral narrow-band light source module, which emits narrow-band near-infrared light with center wavelengths of 720nm±5nm and 950nm±10nm respectively; The time-series damage analysis module is used to perform the following operations on continuously captured time-series images of the same leaf: extract the pixel diffusion rate of the diseased area in the 720nm band image, and generate an active disease mark when the diffusion rate is greater than 5 pixels / hour; calculate the texture entropy value change rate of the wormhole edge in the 950nm band image, and determine that mechanical damage is expanding if the entropy value change rate exceeds 0.15 / minute; Dynamic verification logic unit, which is configured as follows: When the following conditions are met at the same time, it is determined that the blade has effective damage characteristics: (i) Convolutional neural network identifies the morphological features of lesions / wormholes / mechanical damage; (ii) the temporal damage analysis module outputs an active disease marker or a mechanical damage extension determination; (iii) The multi-band anomaly detection module detects an anomaly in the reflectivity of one of the 680nm, 750nm, and 1650nm bands.

7. The mango fruit quality-oriented water and fertilizer coordinated control system according to claim 1, characterized in that: The liquid fertilizer dispensing device comprises parallel A / B fertilizer liquid tanks and a dynamic fertilizer mixing unit; When the central controller switches control instructions, it performs a phased transition operation: During the pre-flush phase, irrigation water is injected into the irrigation pipe at a volume of 30% of the pipe volume, and no fertilizer is added; During the buffer injection phase, a transition buffer mother solution was mixed with irrigation water in a certain proportion. The transition buffer mother solution contained 60 g / L nitrogen, 50 g / L phosphorus, and 120 g / L potassium. After mixing, the nitrogen, phosphorus, and potassium mass concentration ratios in the irrigation solution were 0.6:0.5:1.

2. During the gradient switching phase, after the end of the buffer injection phase, the buffer injection volume decreases day by day by 80% of the total volume of the previous day during daily irrigation operations; the target fertilizer solution injection volume increases accordingly to make up for the reduction in buffer volume; the transition is terminated when the buffer injection volume drops to 5% or less of the initial value.

8. The mango fruit quality-oriented water and fertilizer coordinated control system according to claim 1, characterized in that: The drip irrigation device includes a pressure-compensated drip arrow matrix, which is arranged in two layers of deep and shallow layers according to the three-dimensional topological structure of the main root area of ​​the fruit tree; The central controller is connected to the soil conductivity sensor array to obtain the water and fertilizer distribution map of the root zone profile in real time; When irrigation with a water content of 28-32% is performed during the expansion period, if the difference in water content between the bottom layer and the surface layer of the root zone is greater than 5%, the pulse irrigation strategy is triggered: a single irrigation is divided into three short pulse irrigations with an interval of 10 minutes. After each pulse, the fertilizer solution concentration of the next pulse is dynamically adjusted based on the feedback from the conductivity sensor.

9. The mango fruit quality-oriented water and fertilizer coordinated control system according to claim 8, characterized in that: The central controller is further connected to a salt migration monitoring module, which is configured to: The conductivity-salt dynamic modeling unit constructs a root zone profile salt distribution heat map based on real-time data from the soil conductivity sensor array. This model is then combined with historical irrigation parameters from the drip irrigation system and fertilizer injection records from the liquid fertilizer dispenser to establish a salt migration and accumulation model. The salinization risk warning unit, during the pulse irrigation process, if there is a salt upwelling risk mark, the central controller automatically inserts a leaching subroutine between adjacent pulse intervals. The leaching subroutine is: Send a zero fertilizer liquid instruction to the liquid fertilizer dispensing device to generate a pure water irrigation pulse; Control the drip irrigation device to increase the irrigation volume by 20% in the second pulse phase and extend the pulse interval to 15 minutes; The fertilizer solution concentration of the third pulse is dynamically adjusted based on the updated conductivity distribution map to reduce the bottom salt concentration to below the risk threshold.