A landscaping intelligent irrigation control system and method

Abstract

This invention relates to the field of intelligent irrigation technology, and discloses an intelligent irrigation control system and method for landscaping, including a weather station, an infrared temperature sensor, a solenoid valve, and a controller. The controller establishes a global three-dimensional spatial coordinate system containing two-dimensional ground and facade grid cells, generates a shadow projection matrix based on global solar shortwave radiation, and calculates the effective cumulative radiation on the ground and facade respectively. The effective original canopy temperature is extracted based on the two-dimensional temperature matrix. The lateral longwave thermal radiation warming equivalent is calculated using the effective cumulative radiation on the facade and the equivalent thermal hysteresis coefficient, and this equivalent is subtracted from the original canopy temperature to obtain the corrected true canopy temperature. An adaptive judgment threshold is generated based on the effective cumulative radiation on the ground, the difference between the true canopy temperature and the ambient air temperature is calculated, and the difference is compared with the threshold to generate a command to drive the solenoid valve. This invention helps to eliminate interference from surrounding environmental thermal radiation and improves the accuracy of irrigation control.

Description

Technical Field

[0001] This invention relates to the field of intelligent irrigation technology, specifically to an intelligent irrigation control system and method for landscaping. Background Technology

[0002] Intelligent irrigation systems for landscaping typically utilize infrared temperature sensors to collect canopy temperature and combine it with environmental parameters to assess the vegetation's water deficit status, thereby driving solenoid valves in the pipeline network to execute irrigation. However, existing control schemes have shortcomings in practical applications. In complex garden spaces, surrounding building facades and hard paving absorb short-wave solar radiation, generating environmental heat and continuously radiating lateral long-wave thermal radiation to the surrounding vegetation. Existing systems often directly use the raw measured canopy temperature as the basis for calculations, failing to effectively quantify and deduct the interference from this external thermal radiation. This results in the acquired temperature data deviating from the actual physiological state of the vegetation, reducing the accuracy of water deficit assessment.

[0003] Existing irrigation decisions often use preset fixed thresholds when setting drought criteria, which cannot be dynamically adjusted according to actual accumulated sunlight. When encountering local airflow changes or short-term cloud cover, infrared temperature data will fluctuate in the short term. The existing fixed threshold judgment logic is easily affected by this transient interference and will produce misjudgments, resulting in unstable irrigation decision command output.

[0004] Furthermore, existing systems mostly use open-loop control for driving solenoid valves at the end of the pipeline network, lacking a closed-loop monitoring mechanism for the electrical status of the underlying actuators. Garden irrigation nodes are constantly exposed to outdoor humid conditions, making solenoid valves prone to leakage due to insulation aging or moisture intrusion. Existing solutions cannot promptly interrupt the output level when leakage or abnormal current drops occur, easily leading to damage to control nodes and even upstream power supply lines, resulting in low overall system safety. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent irrigation control system and method for landscaping, which solves the problem that existing landscaping irrigation systems rely on a single environmental parameter or uncorrected infrared surface temperature when determining the water deficit status of vegetation. This fails to eliminate radiation obstruction and long-wave thermal radiation interference caused by surrounding hard buildings or paving, resulting in deviations in the calculation of crown temperature difference and affecting the accuracy of irrigation control.

[0006] The first aspect of this invention provides an intelligent irrigation control system for landscaping, comprising: Weather stations are used to collect ambient air temperature and global shortwave solar radiation. Infrared temperature sensor, used to collect a two-dimensional temperature matrix of the target garden green area; Solenoid valves are installed on the water supply network of the target garden and green area; The controller establishes a global three-dimensional spatial coordinate system containing two-dimensional ground grid units and two-dimensional facade grid units. It generates a shadow projection matrix based on global solar shortwave radiation and uses this matrix to calculate the effective cumulative ground radiation of the two-dimensional ground grid units and the effective cumulative facade radiation of the two-dimensional facade grid units. Based on a two-dimensional temperature matrix, it extracts the effective original canopy temperature of the two-dimensional ground grid units. Based on the effective cumulative facade radiation and a preset equivalent thermal hysteresis coefficient, it calculates the lateral longwave thermal radiation warming equivalent. Subtracting the lateral longwave thermal radiation warming equivalent from the effective original canopy temperature yields the corrected true canopy temperature. Based on the effective cumulative ground radiation, it generates an adaptive judgment threshold, calculates the difference between the corrected true canopy temperature and the ambient air temperature to obtain the real-time compensated canopy temperature difference, and compares the real-time compensated canopy temperature difference with the adaptive judgment threshold to generate control commands and execute intelligent irrigation operations.

[0007] Preferably, when establishing a global three-dimensional spatial coordinate system, the controller discretizes the ground area with vegetation into two-dimensional ground grid units, obtains the three-dimensional geometric contours of surrounding stationary obstacles, and discretizes the illuminated facade into two-dimensional facade grid units, assigning an initial equivalent thermal hysteresis coefficient to the two-dimensional facade grid units; the controller calculates the solar altitude angle and solar azimuth angle according to a set time step, performs ray projection simulation calculation to generate a shadow projection matrix, extracts the global solar shortwave radiation collected by the meteorological station and multiplies it with the shadow projection matrix, and integrates and outputs the effective cumulative radiation of the ground of the two-dimensional ground grid units and the effective cumulative radiation of the facade of the two-dimensional facade grid units within a sliding time window.

[0008] Preferably, the process by which the controller extracts the effective original canopy temperature of the two-dimensional ground grid cells based on the two-dimensional temperature matrix includes: The bimodal histogram threshold segmentation algorithm is used to divide the two-dimensional temperature matrix into high-temperature background pixel clusters and low-temperature vegetation pixel clusters. Data in the high-temperature background pixel clusters are excluded and the average temperature in the low-temperature vegetation pixel clusters is calculated as the effective original canopy temperature of the two-dimensional ground grid unit.

[0009] Preferably, after obtaining the effective original canopy temperature, the controller constructs a spatially correlated neighborhood of the target grid, calculates the local temperature mean and local temperature standard deviation within the spatially correlated neighborhood, and constructs a confidence interval based on the local temperature mean and local temperature standard deviation; the controller filters out abnormal pixels that do not fall within the confidence interval and calculates the effective average canopy temperature weighted by spatial distance.

[0010] Preferably, under the nighttime parameter self-calibration condition, the controller uses the temperature residual to inversely solve and update the equivalent thermal hysteresis coefficient of the two-dimensional facade grid unit; the controller constructs a discrete difference equation for thermodynamic dissipation, builds a historical observation matrix, solves for the fitting slope that minimizes the sum of squared observation errors, and takes the reciprocal of the absolute value of the fitting slope as the updated equivalent thermal hysteresis coefficient.

[0011] Preferably, the controller extracts the background equivalent emission temperature of the high-temperature background heat source under real-time compensation conditions during the day, and solves the approximate viewing angle factor based on the relative spatial position of the two-dimensional ground grid unit and the two-dimensional facade grid unit in the global three-dimensional spatial coordinate system; the controller combines the approximate viewing angle factor and the Stefan-Boltzmann law to solve the net absorption of background longwave radiation by the canopy, and uses the thermal impedance network model to convert the net absorption of background longwave radiation into the equivalent of lateral longwave thermal radiation warming.

[0012] Preferably, the process of generating an adaptive judgment threshold based on the effective cumulative ground radiation includes: The controller calculates the ratio of the effective cumulative radiation on the ground to the preset standard cumulative light intensity to obtain the light intensity normalization coefficient; the controller multiplies the preset baseline residual threshold used to characterize the degree of water deficit by the light intensity normalization coefficient to generate an adaptive judgment threshold.

[0013] Preferably, the controller calculates the algebraic difference between the crown temperature difference after real-time compensation and the adaptive judgment threshold, and generates a transient drought flag. The controller counts the proportion of time-series values ​​where the transient drought flag is set to logic high within a sliding statistical window. When the proportion of time-series values ​​exceeds the lower confidence limit, a synthetic irrigation enable signal is sent to the solenoid valve as a control command.

[0014] Preferably, the controller synchronously collects the real-time feedback current flowing through the coil during the driving of the solenoid valve, and performs equivalent impedance calculation based on the set rated driving voltage and the real-time feedback current to obtain the equivalent operating impedance; when the real-time feedback current is lower than the preset minimum leakage current limit, the controller marks the equivalent operating impedance as infinite and cuts off the output level.

[0015] A second aspect of the present invention provides a control method for an intelligent irrigation control system for landscaping, comprising the following steps: Establish a global three-dimensional spatial coordinate system, discretize the ground vegetation area and the light-receiving facade of the surrounding static obstacles into two-dimensional ground grid units and two-dimensional facade grid units respectively, and assign an initial equivalent thermal hysteresis coefficient. Perform a ray projection simulation to generate a shadow projection matrix. Multiply the acquired global solar shortwave radiation by the shadow projection matrix and integrate it to calculate the effective cumulative radiation of the ground for the two-dimensional ground grid cell and the effective cumulative radiation of the facade for the two-dimensional facade grid cell. The high-temperature background data in the two-dimensional temperature matrix was excluded by using the bimodal histogram threshold segmentation algorithm, and the average temperature of low-temperature vegetation pixels was calculated as the effective original canopy temperature of the two-dimensional ground grid unit. The equivalent thermal hysteresis coefficient is updated based on the working conditions, and the lateral long-wave thermal radiation warming equivalent is calculated by combining the approximate viewing angle factor, the effective cumulative radiation of the facade and the equivalent thermal hysteresis coefficient. The lateral long-wave thermal radiation warming equivalent is then subtracted from the effective original canopy temperature to obtain the corrected real canopy temperature. An adaptive judgment threshold is generated based on the effective cumulative radiation on the ground. The difference between the corrected real canopy temperature and the ambient air temperature is calculated to obtain the real-time compensated canopy temperature difference. The real-time compensated canopy temperature difference is compared with the adaptive judgment threshold to generate control commands, which drive the solenoid valve to perform intelligent irrigation operations.

[0016] This invention provides an intelligent irrigation control system and method for landscaping. It has the following beneficial effects: 1. This invention establishes a global three-dimensional spatial coordinate system including ground and facade grids, and uses the shadow projection matrix and equivalent thermal hysteresis coefficient to calculate the lateral long-wave thermal radiation warming equivalent. The external thermal radiation generated by the surrounding building facades and hard paving is deducted from the original canopy temperature, eliminating the environmental heat storage deviation in complex garden spaces, making the obtained real canopy temperature consistent with the actual physiological state of vegetation, and improving the accuracy of water deficit determination.

[0017] 2. This invention extracts the effective cumulative radiation on the ground to generate an adaptive judgment threshold, and combines it with the real-time compensation crown temperature difference to generate a transient drought flag. The irrigation operation is triggered by calculating the temporal proportion of this flag within a sliding statistical window. The drought judgment criteria are dynamically adjusted based on the actual cumulative sunlight, and short-term temperature data fluctuations caused by local airflow or cloud cover are filtered out, ensuring the stability of irrigation decision command output.

[0018] 3. This invention synchronously collects the real-time feedback current of the coil when driving the solenoid valve to calculate the equivalent operating impedance, and cuts off the output level when the current drops to the minimum leakage current limit. It performs closed-loop monitoring of the electrical connection status of the solenoid valve at the end of the pipeline network, and can promptly block the output when the equipment experiences insulation aging or water vapor intrusion leading to leakage, thereby improving the operational safety of electrical control nodes under outdoor humid conditions. Attached Figure Description

[0019] Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a flowchart of the method of the present invention; Figure 3 This is a comparison diagram of canopy temperatures for thermodynamic compensation according to the present invention. Detailed Implementation

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see the appendix Figure 1 - Appendix Figure 2 This invention provides an intelligent irrigation control system for landscaping, comprising: a weather station, an infrared temperature sensor, a solenoid valve, and a controller.

[0022] The weather station is deployed in an open, unobstructed area to collect ambient air temperature and global shortwave solar radiation in real time.

[0023] The infrared temperature sensor is installed on a high-pole device, and its detection field covers the target garden green area to collect a two-dimensional temperature matrix of the target garden green area.

[0024] The solenoid valve is installed on the water supply network of the target garden green area, and the physical pipeline is divided according to the irrigation micro-zone.

[0025] The weather station, infrared temperature sensor, and solenoid valve establish communication connections with the controller; the controller receives ambient air temperature, global solar shortwave radiation, and a two-dimensional temperature matrix, and sends control commands to the solenoid valve.

[0026] This invention provides an intelligent irrigation control method for landscaping, comprising the following steps: S1. Establish a global three-dimensional spatial coordinate system for the target garden greening area and its surrounding environment. Discretize the ground area with vegetation into horizontal two-dimensional ground grid units. Obtain the three-dimensional geometric contours of the surrounding stationary obstacles and discretize their illuminated facades into vertical two-dimensional facade grid units. Assign an initial equivalent thermal hysteresis coefficient to the two-dimensional facade grid units. S2, calculate the solar altitude angle and solar azimuth angle according to the set time step, perform ray projection simulation calculation to generate a shadow projection matrix, extract the global solar shortwave radiation collected by the weather station and multiply it with the shadow projection matrix, integrate within the sliding time window and output the effective cumulative radiation of the ground of the two-dimensional ground grid unit and the effective cumulative radiation of the facade of the two-dimensional facade grid unit. S3 receives the two-dimensional temperature matrix collected by the infrared temperature sensor, executes the bimodal histogram threshold segmentation algorithm to divide the two-dimensional temperature matrix into high-temperature background pixel clusters and low-temperature vegetation pixel clusters, excludes the data in the high-temperature background pixel clusters and calculates the average temperature in the low-temperature vegetation pixel clusters as the effective original canopy temperature of the two-dimensional ground grid unit. S4. Based on the global solar shortwave radiation, the environmental conditions are determined. Under the nighttime parameter self-calibration condition, the equivalent thermal hysteresis coefficient of the two-dimensional facade grid unit is solved and updated in reverse using the temperature residual. Under the daytime real-time compensation condition, the surface dynamic temperature value of the two-dimensional facade grid unit is calculated based on the effective cumulative radiation of the facade and the equivalent thermal hysteresis coefficient. The lateral longwave thermal radiation warming equivalent generated by the stationary obstacle on the vegetation is calculated in combination with the viewing angle factor. The corrected real canopy temperature is obtained by subtracting the lateral longwave thermal radiation warming equivalent from the effective original canopy temperature. S5 inputs the effective cumulative radiation of the ground into a monotonically decreasing mapping function to generate an adaptive judgment threshold, calculates the difference between the corrected real canopy temperature and the ambient air temperature to obtain the real-time compensated canopy temperature difference, compares the real-time compensated canopy temperature difference with the adaptive judgment threshold, and sends an opening command or a closing command to the corresponding solenoid valve according to the comparison result.

[0027] When implementing intelligent irrigation control methods for landscaping, the controller needs to model and discretize the physical space to calculate local light differences and thermal radiation distribution. By transforming the continuous physical space into a discrete mesh matrix, geometric data support is provided for subsequent light occlusion determination and the solution of the viewing angle factor for long-wave thermal radiation.

[0028] In establishing a global three-dimensional spatial coordinate system, in this embodiment, the controller uses a preset fixed reference point within the target garden greening area as the origin of the coordinate system, sets the due east direction as the positive X-axis, the due north direction as the positive Y-axis, and the vertical upward direction as the positive Z-axis, and acquires the terrain elevation data and physical boundary coordinate data of the target garden greening area.

[0029] The ground area with vegetation is discretized into horizontal two-dimensional ground grid units: Based on physical boundary coordinate data, the controller divides the vegetation-containing areas of the target garden greening area into multiple adjacent two-dimensional ground grid units. Preferably, the two-dimensional ground grid units are set as squares or rectangles. The grid side length is generally set between 0.5m and 2.0m, and its specific value needs to be determined in conjunction with the actual size of the irrigation micro-zone and the computing power of the controller's built-in processor. If the grid side length is too large, it will lead to a decrease in the recognition accuracy of local shadows; if the grid side length is too small, it will cause a significant increase in the spatial matrix dimension, exceeding the computing power limit of the edge computing device. In addition, for incomplete grids generated at irregular physical boundaries, the controller makes a judgment based on the area ratio. When the ratio of the vegetation coverage area to the total area of ​​the grid in the incomplete grid is greater than or equal to a preset area ratio threshold, the grid is retained; otherwise, it is discarded to avoid data anomalies caused by the area approaching zero in subsequent spatial area integration calculations. A two-dimensional ground grid unit is denoted as ,in Indexed by natural numbers. The controller records each two-dimensional ground grid cell. 3D coordinates of the center point .

[0030] Acquiring the 3D geometric contours of surrounding stationary obstacles: In real-world urban garden scenarios, stationary obstacles include, but are not limited to, buildings, solid walls, or bridge piers. The controller extracts the 3D geometric contours of stationary obstacles by reading pre-stored building information model files or 3D laser point cloud data. After extraction, the controller analyzes the normal orientation of the 3D geometric contours and, combined with the orientation information of the global 3D spatial coordinate system, identifies the surfaces facing the target garden green area and designates them as illuminated facades.

[0031] The illuminated facade is discretized into vertical two-dimensional facade grid units: Based on the set facade grid resolution, the controller divides the illuminated facade into multiple two-dimensional facade grid units. (Setting the first...) Each two-dimensional facade grid unit is denoted as ,in Indexed by natural numbers. The controller records each two-dimensional facade grid cell. 3D coordinates of the center point Simultaneously, record the normal vector of the plane containing the two-dimensional facade mesh element. Normal vector Defined as a unit vector perpendicular to the plane containing the two-dimensional elevation grid cells and pointing outwards from the target landscaped area. Normal vector. This data will serve as the foundation for subsequent determination of the solar radiation incident state and calculation of the spatial perspective factor. Through the above discretization process, the continuous three-dimensional space is mapped into a discrete matrix structure, thereby supporting subsequent ray projection simulations and thermodynamic calculations.

[0032] After spatial discretization, to quantify the physical process by which surrounding stationary obstacles absorb short-wave solar radiation during the day and release long-wave thermal radiation to vegetation at night or in shadow, the controller needs to assign thermophysical property parameters to the established two-dimensional facade mesh units. Conventional building thermodynamics modeling typically relies on inputting multi-dimensional physical parameters such as wall density, material specific heat capacity, and surface emissivity. However, in open urban garden environments, manually acquiring the physical parameters of various complex building materials is too costly and easily introduces static errors. Therefore, a comprehensive dimensionless parameter, the equivalent thermal hysteresis coefficient, is introduced. This parameter macroscopically characterizes the temperature rise sensitivity of the two-dimensional facade mesh units after absorbing solar radiation and the heat decay inertia after losing the radiation source. The initial equivalent thermal hysteresis coefficient is set according to the appearance and material type of the stationary obstacles. In this embodiment, the controller acquires material classification information from on-site survey records and matches empirically preset initial parameter values ​​to light-receiving facades with different material properties. The specific sub-characteristics of the material categories of stationary obstacles include high-heat-storage solid materials and low-heat-storage lightweight materials. High-heat-storage solid materials typically include cast-in-place concrete walls, solid brick and tile walls, and stone cladding; low-heat-storage lightweight materials typically include glass curtain walls and metal decorative panels.

[0033] As a preferred approach, the controller is a two-dimensional facade mesh unit. The initial equivalent thermal hysteresis coefficient is denoted as . Since this coefficient is a relative value after normalization and dimensionless processing, The value range is strictly constrained to between 0.01 and 0.99. The specific basis for this value is: the closer the value is to 1, the greater the thermal inertia of the material, meaning the temperature change lags behind the change in solar radiation intensity; the closer the value is to 0, the smaller the thermal inertia. Based on this physical correlation, for high-heat-storage solid materials, the controller assigns a higher initial range value, for example, a range of 0.6 to 0.85; for low-heat-storage lightweight materials, the controller assigns a lower initial range value, for example, a range of 0.15 to 0.35. If clear material classification information cannot be obtained in actual deployment, the controller will default to assigning an initial mean value in the middle range to all two-dimensional facade mesh elements, for example, setting it to 0.5. This initial mean value setting ensures that the system has a steady-state response benchmark in the initial startup phase, avoiding calculation anomalies caused by missing parameters.

[0034] The initial thermal hysteresis coefficient matrix is ​​established and verified. For all the previously discretized two-dimensional facade mesh elements, the controller constructs an initial one-dimensional array of thermal parameters with a length equal to the total number of facade meshes, or constructs a corresponding diagonal matrix structure to accommodate subsequent matrix multiplication for accelerated computation. Assume the illuminated facade is divided into... If there are two-dimensional facade grid units, then the controller constructs a dimension of diagonal matrix The elements on the main diagonal are the initial equivalent thermal hysteresis coefficients of the corresponding mesh. All other off-diagonal elements are zero. Meanwhile, to ensure the completeness and robustness of subsequent algorithm logic, the controller will check each element written into the matrix. Perform extreme value verification. If the coefficient is incorrectly set to absolute zero due to abnormal data input, the controller will forcibly assign it a very small positive lower limit compensation value, for example, setting the compensation value to 0.01. The technical purpose of this is to avoid program crashes or singularity errors in matrix operations caused by the denominator approaching zero when the controller subsequently performs inverse division under the night parameter self-calibration condition.

[0035] By pre-configuring the initial equivalent thermal hysteresis coefficient, the system does not need to rely on static and easily misaligned manual calibration parameters for extended periods. This initial value serves only as the calculation benchmark for the control system on the first day of initial power-on operation. To ensure the alignment of multi-source data across time, the controller records the current system time as the initial timestamp and uses this as the starting point for initiating the timing of the sliding time window. After the first natural day-night cycle, the system can adaptively iterate and cover the nighttime temperature residual mechanism, thereby reducing the difficulty of engineering deployment while ensuring the convergence accuracy of long-wave thermal radiation assessment.

[0036] After discretizing the 3D mesh of the physical space, the controller needs to dynamically evaluate the real-time state of the light and shadow distribution within the physical space. The obstruction of sunlight by stationary obstacles within the space is the core physical cause of temperature differences in the garden's microclimate. Based on celestial motion, by introducing local latitude and longitude and time variables, the light and shadow boundaries at each moment can be accurately reconstructed using analytical geometry. To transform the dynamic physical occlusion process into a mathematical matrix that can be processed by a computer, the controller needs to perform a light projection simulation, the specific process of which includes the following steps: The solar altitude angle and solar azimuth angle are calculated according to a set time step. In this embodiment, the controller obtains the current standard time through a built-in real-time clock module and performs astronomical parameter calculations by combining the preset latitude and longitude data of the target garden green area. To balance the controller's computational load with the accuracy of shadow tracking, the set time step is generally set to a range of 5 to 15 minutes. Simultaneously, to ensure temporal consistency during multi-source data fusion, the controller strictly aligns the set time step with the sampling period of the global solar shortwave radiation from the meteorological station to avoid timestamp mismatches in subsequent integration calculations. Each time a new time step is entered, the controller calculates the solar altitude angle for the current moment. and solar azimuth .

[0037] A solar ray direction vector is constructed in a global three-dimensional spatial coordinate system. Based on the geometric relationships set when establishing the global three-dimensional spatial coordinate system—the X-axis (eastward), the Y-axis (northward), and the Z-axis (opposite to gravity)—the controller determines the solar altitude angle in the spherical coordinate system. With solar azimuth Convert to absolute direction in Cartesian coordinates. Define the direction vector of sunlight as... The calculation formula is as follows: ; In the formula, Characterizes the angle between sunlight and the horizontal ground; This represents the clockwise angle between the projection of sunlight onto the horizontal plane and the direction of true north. The physical purpose of this direction vector is to provide an absolute reference ray path for subsequent space ray tracing algorithms. To ensure the completeness and efficiency of the algorithm logic, the controller introduces pre-condition judgment logic here: the controller comprehensively judges the solar altitude angle. And the global solar shortwave radiation collected synchronously by the weather station. When detected If the global solar shortwave radiation is lower than the preset minimum threshold, it indicates that the sun is below the horizon or is blocked by extremely strong clouds. The system confirms that it has entered a state of no effective direct sunlight. Under this multi-dimensional judgment condition, the controller will skip all subsequent ray intersection calculations and directly force the occlusion status of all grid cells within the current time step to be assigned the value representing shadow.

[0038] Perform ray projection simulation calculations on two-dimensional ground grid cells, and determine... Furthermore, under daytime operating conditions with effective solar radiation, the controller operates with each two-dimensional ground grid unit. 3D coordinates of the center point Starting from the direction of the sun's rays, the vector... A virtual test ray is emitted in a specific direction. Using a 3D spatial analytical geometry intersection algorithm, the controller determines whether the virtual test ray intersects with the 3D geometric contours of surrounding stationary obstacles during its extension. To avoid misjudgments of geometric coplanarity or edge tangency points due to the precision limitations of computer floating-point numbers (i.e., the pseudo-intersection singularity problem), the controller introduces a value of 10 during the intersection determination. -6 The geometric tolerance threshold. If the intersection distance is greater than this tolerance threshold, it indicates that the direct sunlight path of the grid at the current moment has been substantially truncated, and the controller will change the two-dimensional ground grid cell. Current occlusion status Marked as 0; if no intersection occurs or the intersection penetration distance is within the tolerance threshold, it indicates that the grid is completely exposed to direct sunlight, and the controller will shade it. Marked as 1. The binarization of the state parameters aims to achieve mathematical isolation of complex light and shadow boundaries in the local microclimate, thereby providing discrete switching variables for subsequent radiative energy integration.

[0039] When performing ray projection simulation calculations on two-dimensional facade mesh elements, the stationary obstacle faces a self-occlusion effect due to the difference in normals between its illuminated facade and the surrounding surface. As a preferred approach, the controller extracts the two-dimensional facade mesh elements before emitting the virtual test ray. Normal vector of the plane The controller calculates the negative vector of the direction vector of sunlight. With the normal vector The inner product is used to assess whether incident light can illuminate the front of the facade. Considering that the radiant energy transmitted under grazing conditions is weak and prone to integration errors, the system presets a positive cosine threshold close to zero. If the calculated inner product is less than or equal to this cosine threshold, it indicates that the sunlight is illuminating from the back of the facade or grazing at the facade at a very small angle. Due to self-shading or an excessively large incident angle, the facade cannot obtain effective direct sunlight energy, and the controller directly converts the two-dimensional facade mesh unit to the appropriate value. occlusion state Marked as 0. If the inner product is greater than the cosine threshold, the controller then uses the two-dimensional facade mesh unit. A virtual test ray is emitted from the center point of the system to determine whether it is secondary occluded by other taller, stationary obstacles within the garden, and the occlusion status is assessed based on the 3D intersection results. Assign a value of 1 or 0.

[0040] The controller will generate a shadow casting matrix and then determine all ground occlusion states. and all facade obstruction status The data is integrated to construct a shadow projection matrix corresponding to the current time step. This matrix, as a discrete mask composed of Boolean values, fully records the microscopic light-receiving attributes of the target garden area at the current moment. Based on this shadow projection matrix, the controller can perform spatial dimensionality reduction and mapping on the single global radiation data collected by the macroscopic meteorological station, thereby providing geometric weighting coefficients for calculating the cumulative amount of solar radiation actually absorbed locally in different grids.

[0041] The global solar shortwave radiation collected by the weather station reflects the macroscopic energy input in an open environment. In a spatially heterogeneous garden setting, the actual heat energy absorbed by the vegetation canopy and surrounding building facades is directly related to their local light history. Due to the thermal inertia exhibited by plant transpiration and the heating process of building materials, instantaneous radiation data at a single moment cannot accurately characterize the true thermodynamic state under the microclimate. Based on the aforementioned thermodynamic hysteresis characteristics, the controller needs to combine the previously generated shadow projection matrix to perform radiation integral flow calculations within a set sliding time window. This data fusion process includes the following steps: In this embodiment, a sliding time window is set and multi-source time series are aligned. The controller uses the current system time as the baseline cutoff point and traces back a fixed-length period of time along the historical timeline to construct the sliding time window. The length of the sliding time window is generally set to 2 to 6 hours. The physical basis for this value is that if the time window is too short, it cannot fully cover the heat transfer hysteresis cycle of the building walls, resulting in insufficient heat storage assessment; if the time window is too long, it will introduce invalid meteorological features remaining from yesterday, overshadowing the temperature response characteristics of the current environment. To ensure the completeness of data calculation, the controller extracts the global solar shortwave radiation sequence recorded by the meteorological station within the sliding time window and verifies whether its timestamp is strictly aligned with the time series of the shadow projection matrix. If data loss due to sensor communication delay is detected, the controller uses linear interpolation to smoothly fill in the missing radiation data. To avoid interpolation distortion caused by long-term disconnection, the controller synchronously calculates the time series proportion of missing data. When the percentage of missing data exceeds the preset fault tolerance threshold (e.g., set to 15%), the controller will stop interpolation and call the valid historical radiation data of the same period of the previous natural day as the degradation compensation benchmark, so as to ensure the continuity and physical rationality of the discrete integration step size.

[0042] The effective cumulative ground radiation of a two-dimensional ground grid cell is calculated. For ground areas exposed to the natural environment, even if they are in the physical shadow zone cast by stationary obstacles at a specific moment, they will still receive scattered radiation from atmospheric molecules and clouds. To ensure the physical completeness of the heat assessment logic and avoid singularities where the energy in the shadow zone reaches absolute zero due to simple binarization, the controller introduces a scattered radiation reduction factor. The controller calculates the effective cumulative ground radiation of each two-dimensional ground grid cell based on a discrete summation algorithm, and the corresponding calculation formula is as follows: ; In the formula, This represents the effective cumulative radiation on the ground. This represents the total number of discrete time steps contained within the sliding time window; Indicates a time point; Indicates in Global shortwave solar radiation collected by the weather station at all times; Indicates in The time frame mapped by the shadow projection matrix The ground occlusion status of a two-dimensional ground grid cell, which takes the value of 1 or 0; Indicates the reduction factor for scattered radiation; This indicates the set time step. As a preferred method, the controller reads multi-dimensional data such as air humidity and particulate matter concentration synchronously collected by the weather station, inputs this multi-dimensional meteorological data into a preset empirical mapping table, and dynamically outputs the current diffuse radiation reduction coefficient. Its value range is usually constrained to between 0.10 and 0.25.

[0043] Through the above calculation logic, the controller effectively maps macroscopic meteorological observation data to grid microscopic physical absorption, quantifying the historical light and heat load of the vegetation area.

[0044] The calculation of the effective cumulative radiation of a two-dimensional facade grid cell differs from that of a horizontal ground grid. The efficiency of a vertical building facade in absorbing shortwave solar radiation is strictly limited by the angle of incidence of sunlight. Based on Lambert's cosine law, the controller extracts the negative vector of the previously calculated solar ray direction when calculating the facade heat source input. Normal vector of the plane containing the two-dimensional facade mesh element The inner product is used to characterize the cosine of the angle between the direction of the direct ray and the wall normal vector. The physical purpose of this inner product calculation is to quantify the energy projection density of oblique rays on the facade. When the inner product result is less than or equal to zero, it indicates a back-shading or parallel grazing state. To avoid introducing inverse invalid algebraic superposition into the energy integral, the system uses a nonlinear extremum function to forcibly truncate it. The controller calculates the effective cumulative radiation of the facade for each two-dimensional facade grid cell, using the following formula: ; In the formula, Indicates the effective cumulative radiation of the facade; Indicates in The current state of facade obstruction; This is a function to find the maximum value, used to output the larger of the two values ​​within the parentheses; Indicates in Global shortwave solar radiation collected by the weather station at all times; This indicates the set time step.

[0045] The effective cumulative radiation output by the above formula accurately reflects the energy loss caused by shading and unfavorable incident angles over time. This calculation result serves as a core dynamic intermediate variable, providing a reliable external heat source driving force for the subsequent first-order thermal inertia hysteresis model to accurately estimate the wall surface temperature.

[0046] After acquiring the light and shadow distribution and radiation energy of the target green area, the controller needs to obtain the real physical temperature of the vegetation canopy in real time as a closed-loop feedback parameter for irrigation decisions. The two-dimensional temperature matrix obtained by the infrared temperature sensor through the detection field of view not only includes the vegetation canopy but also inevitably mixes in background materials such as bare soil and paved paths. Because plant leaves consume a large amount of latent heat through transpiration via stomata, their surface temperature is usually lower than that of hard background surfaces without transpiration during the day. Based on the above thermodynamic differences, the two-dimensional temperature matrix statistically exhibits a typical bimodal distribution of high and low temperatures. In order to accurately remove background thermal radiation noise and extract pure vegetation temperature, the controller needs to perform adaptive threshold segmentation calculation of the bimodal histogram based on the principle of maximizing inter-class variance. The data processing process includes the following steps: The controller receives the two-dimensional temperature matrix from the infrared temperature sensor at a set sampling period via a communication interface and performs spatial coordinate mapping. To ensure time alignment of multi-source data, the controller strictly compares the received timestamp of this data with the reference time of the aforementioned sliding time window, with the allowable time deviation tolerance typically set to within 1 minute. If the timestamp deviation exceeds this tolerance range, the controller will reject the current frame data and maintain the valid temperature array of the previous period, thereby avoiding time misalignment between meteorological radiation and surface temperature under transient cloud cover conditions. Subsequently, the controller, combining the installation height, pitch angle, and lens field of view parameters of the infrared temperature sensor, establishes a perspective projection transformation relationship between the image pixel coordinate system and the global three-dimensional spatial coordinate system. Based on this transformation relationship, the controller assigns each pixel in the two-dimensional temperature matrix to the corresponding two-dimensional ground grid unit. Inside.

[0047] Construct a histogram for the temperature pixels within a single two-dimensional ground grid cell, targeting the first... Two-dimensional ground grid unit The controller extracts all valid temperature pixels falling within the physical boundaries of the grid. Considering that sensor lenses are easily obstructed by flying insects or subjected to occasional high-frequency electromagnetic interference in outdoor environments, the controller performs median filtering on the extracted pixel sequence before statistical analysis to remove extreme values ​​(e.g., distorted data above 80°C or below -20°C) that exceed physical norms. After filtering, the controller determines the lowest temperature value within the current grid. With the highest temperature value Divide it into A series of equidistant temperature ranges. As a preferred method, however, it is limited by the 8-bit quantization depth calculation efficiency of conventional microprocessors. The default value is 256. The controller counts the number of pixels falling within each temperature range and divides this number by the total number of pixels in that grid to obtain the probability density for each temperature range. When performing the above division operation, if the total number of valid pixels in the grid is zero due to a blind spot in the field of view, it will directly cause the program to crash due to the invalid denominator. Before performing the probability density calculation, the controller checks the total number of pixels. If the total number is equal to zero, the grid is directly marked as invisible, and the subsequent segmentation calculation of the current grid is stopped and skipped.

[0048] An adaptive segmentation threshold is calculated using an inter-class variance maximization algorithm. The physical objective of this algorithm is to find an optimal temperature segmentation point that maximizes the thermodynamic difference between the segmented high-temperature background cluster and the low-temperature vegetation cluster, while minimizing the internal temperature fluctuations within each cluster. The controller iterates through all candidate temperature ranges. ( ), calculate the inter-class variance for each corresponding segmentation state. The calculation formula is as follows: ; ; ; ; ; In the formula, Indicates the candidate temperature range index; This represents the index of the internal interval traversed during the summation process; and These represent the pixel accumulation probabilities of low-temperature vegetation clusters and high-temperature background clusters, respectively. and Let represent the expected temperature values ​​for both intervals. When performing the above expected value division operation, if... or A value approaching zero will cause the denominator to become invalid, leading to calculation errors. Therefore, the controller is set to a value of 10. -4 The controller sets a minimum probability protection threshold. When the denominator is less than this threshold, the controller forcibly skips the variance calculation of the current candidate threshold. After traversal, the controller selects the threshold that minimizes the inter-class variance. The one that gets the maximum value The value is then inversely mapped back to the actual physical temperature scale to obtain the current adaptive segmentation threshold. .

[0049] Extract and output the effective raw canopy temperature based on the calculated adaptive segmentation threshold. The controller controls the two-dimensional ground grid cells. All temperature pixels within the range are binarized. Values ​​greater than or equal to... Pixels with temperatures below a certain value are divided into high-temperature background pixel clusters and removed; pixels with temperatures below a certain value are removed. The pixels are divided into low-temperature vegetation pixel clusters. The controller calculates the arithmetic mean of all physical temperature values ​​within each low-temperature vegetation pixel cluster and uses it as the unit of the two-dimensional ground grid. The effective original canopy temperature is denoted as . As a preferred approach, if, under specific operating conditions (such as the winter leaf fall period or the newly planted seedling stage), the total number of pixels within a low-temperature vegetation pixel cluster is less than a preset effective baseline (e.g., less than 5% of the total pixels), directly averaging will not represent the true canopy state. The basis for setting this preset effective baseline is that when the proportion of vegetation pixels is too low, the system is prone to misidentifying residual sensor noise or blurred object edges as vegetation itself, thus introducing observational quiescent errors that are difficult to converge. In this case, the controller will trigger data degradation compensation logic, directly extracting the ambient air temperature collected in real-time by the weather station as the current temperature of the grid. This ensures the completeness and anti-interference capability of the subsequent thermodynamic compensation calculation chain.

[0050] After initially obtaining the vegetation body temperature of the target garden green area through adaptive threshold segmentation, the two-dimensional ground grid unit... Effective raw canopy temperature recorded internally It is still prone to local physical disturbances. In open, real natural environments, plant leaves are inevitably affected by dynamic wind loads causing leaf flipping, transient light spots transmitted through canopy gaps (i.e., the light spot effect), and the superposition of thermal noise from the underlying infrared sensor. These physical phenomena cause the raw temperature data to exhibit high-frequency discrete fluctuations in spatial distribution. Based on the above-mentioned thermal disturbance characteristics, if the controller directly relies on the instantaneous raw temperature of a single grid for microclimate assessment, it is easy to cause logical oscillations in subsequent irrigation decisions. In order to filter out spatial discrete noise and extract steady-state thermodynamic parameters that characterize the true transpiration state of vegetation, the controller needs to perform mean domain extraction of effective raw canopy temperature. The specific process includes the following steps: In this embodiment, the controller constructs a spatially related neighborhood of the target mesh. Two-dimensional ground grid unit Using the geometric center as the axis, a dynamic search radius is defined on the global two-dimensional physical plane. The search radius is generally constrained to be between 1.0 meter and 3.0 meter. The physical basis for determining this range is that it must match the typical canopy radius of vegetation within the target area, thus ensuring that grids included in the neighborhood belong to the same plant or adjacent plant clusters with similar microclimate characteristics in terms of plant physiology. The controller traverses the two-dimensional matrix, selecting grids with Euclidean distances less than or equal to... All two-dimensional ground grid cells are extracted, and a spatially related neighborhood set of that grid is constructed. Assume that this set contains a total of... Each grid cell is indexed as _ . ,in The value range is 1 to .

[0051] To quantify the degree of local thermal heterogeneity of the vegetation canopy, the controller calculates the temperature statistical distribution parameters within the spatially correlated neighborhood. This involves analyzing the original canopy temperatures of all elements included in the spatially correlated neighborhood set. Statistical analysis is performed. The controller sequentially calculates the local temperature mean within the neighborhood. and local temperature standard deviation The corresponding calculation formula is as follows: ; ; In the formula, Represents the spatially related neighborhood of the first The effective original canopy temperature of each grid cell; This represents the total number of grid cells contained in the neighborhood. When performing the standard deviation division, since calculating the unbiased estimate of the sample requires one degree of freedom, if there is only one isolated valid grid cell in the neighborhood (i.e., ... This will directly cause the program to crash when the denominator is zero. To ensure the integrity of the algorithm's underlying logic, the controller incorporates extreme value protection logic here: when an extreme value is detected... At that time, the local temperature standard deviation was forcibly adjusted. The value is assigned to 0 to indicate that the current grid lacks a spatial fluctuation reference.

[0052] After obtaining the statistical distribution parameters for outlier pixel removal based on dynamic confidence intervals, the system needs to physically strip away extreme temperature values ​​within the neighborhood. As a preferred approach, the controller constructs a temperature domain boundary representing high confidence based on the local temperature mean and standard deviation. ; In the formula, This represents the local average temperature. Indicates the local temperature standard deviation; This represents the confidence scaling factor. Considering that the edge leaves of garden vegetation exhibit a certain degree of gradient broadening due to localized self-shading or long-wave radiation heating from the soil, the actual physical temperature of these leaves is affected by this gradient. A strict Gaussian distribution is not typically used. The standard was instead moderately relaxed to between 1.5 and 2.0 to preserve the normal canopy thermal gradient. The controller compares each element in its neighborhood one by one. If the temperature value falls strictly within the above confidence interval, it indicates that it belongs to a stable blade body temperature, and the controller assigns its effectiveness weight. Marked as 1; if the temperature value exceeds the limit, it indicates that it is likely a discrete outlier caused by a transient light spot directly hitting the bottom dead leaves or misjudging the background at the edge of the field of view, and the controller will assign it an effectiveness weight. Marked as 0.

[0053] To calculate the effective average canopy temperature weighted by spatial distance, after outlier removal, the controller introduces attenuation-weighted logic based on spatial distance to enhance the contribution of core region leaves to the final temperature characterization. The controller calculates the Euclidean distance from the center anchor point to each preserved grid within the spatially correlated neighborhood. Based on this, spatial weight coefficients are generated. The closer the distance, the greater the weighting coefficient. The controller calculates the final effective average canopy temperature of the target grid using a weighted summation formula. The specific formula is as follows: ; In the formula, Indicates the spatial weighting coefficient; Indicates the validity weight; Represents the spatially related neighborhood of the first The effective original canopy temperature of each grid cell. Considering that under extreme disturbance conditions, large-area, drastic temperature jumps can easily occur within the neighborhood, leading to changes in the effective original canopy temperature of all grid cells. All are judged as 0. Under this singular boundary condition, the weighted denominator of the above formula will approach zero.

[0054] To address this computational risk, the controller verifies the denominator state before performing fusion. If the sum of the denominators is detected to be zero, the controller will abandon the current spatial weighted calculation and directly use the historical smoothed temperature latched in the previous time step for a downgraded replacement. Through this mean domain extraction mechanism, the system not only effectively filters out high-frequency thermal disturbances in the microenvironment but also provides a high signal-to-noise ratio temperature state variable for subsequent quantification of water loss due to transpiration in vegetation.

[0055] After obtaining the steady-state effective average canopy temperature, the system needs to further assess the actual water stress state of the vegetation. Plant leaf temperature is not only affected by cooling from internal stomatal transpiration but also strongly driven by external air temperature, solar radiation, and meteorological quantities such as wind speed. To accurately isolate external environmental thermal radiation noise and extract thermodynamic characteristics that solely characterize the physiological water shortage state of the vegetation, the controller needs to perform a process of condition determination and extraction of pure thermal radiation residuals. This data processing includes the following steps: In this embodiment, the controller performs synchronization of multi-source environmental parameters and determines microclimate conditions. It receives external driving data such as global solar shortwave radiation, ambient air temperature, relative humidity, and wind speed, synchronously collected from a meteorological station. To ensure the validity and consistency of the data during multi-source fusion, the controller forcibly compares the timestamps of the meteorological data with the timestamps of the current grid canopy temperature, with an allowable time alignment tolerance of less than one minute. If the time deviation exceeds this tolerance, the controller triggers linear interpolation compensation logic to forcibly align the time axis. Based on this time alignment logic, the controller further determines the current plant physiological conditions. Specifically, the controller comprehensively determines whether the current solar shortwave radiation is higher than the set effective radiation threshold and verifies whether the solar altitude angle is greater than a preset elevation angle threshold (e.g., 15°). The effective radiation threshold is typically set to 200 W / m². 2 Up to 300W / m 2 The basis for this determination is that when the effective photosynthetic radiation is below this range, the stomata of vegetation are often in a physiologically closed state due to insufficient light. Under this non-transpiration condition, calculating the leaf surface temperature difference residual lacks real physiological significance. If the above radiation and light conditions are met, the system confirms that it has entered the effective transpiration condition and triggers subsequent residual calculations; if not, the controller will suspend the residual calculation for the current time step, directly call and maintain the historical residual state of the previous cycle, thereby avoiding the introduction of invalid disturbances at night or under heavy cloudy conditions.

[0056] To calculate the theoretical temperature difference benchmark under sufficient moisture conditions, after confirming effective transpiration, the controller needs to calculate a reference frame for the canopy-air temperature difference under ideal water supply conditions. To avoid complex calculus derivations, the controller performs theoretical value estimation based on a simplified Penman-Montis energy balance principle. The controller obtains the air saturated vapor pressure difference and a preset aerodynamic drag under the current environment, calculates their physical product to form an aerodynamic driving term; simultaneously, the controller extracts net radiation and the theoretical minimum stomatal resistance of vegetation, calculates their physical product to form a radiation driving term. The controller subtracts the radiation driving term from the aerodynamic driving term and divides the resulting difference by the comprehensive impedance coefficient, which is weighted and corrected by aerodynamic drag and thermodynamic wet / dry constants, to output the theoretical temperature difference benchmark. During the above algebraic division operation, if the external environment is in an extremely calm state with wind speeds approaching zero, the inversely proportional aerodynamic drag will approach infinity, leading to numerical overflow in the division operation and crashes of the underlying program. To ensure the physical integrity of the algorithm, the controller sets a minimum protection limit of 0.5 m / s for the wind speed input port. When the actual detected wind speed is lower than this limit, the relevant resistance is calculated based on this limit value. The theoretical temperature difference benchmark output by this step aims to provide a dynamic baseline that is free from meteorological fluctuations for actual temperature assessment.

[0057] To calculate the actual canopy temperature difference under real-world conditions, the controller extracts the effective average canopy temperature output from the target two-dimensional ground grid cells in the preceding steps and subtracts the time-aligned ambient air temperature. As a preferred method, considering the nonlinear calibration drift that infrared temperature sensors and contact air temperature sensors are prone to in different ambient temperature zones, the controller uses a built-in calibration compensation matrix to perform benchmark alignment of the two sets of temperature data before performing the subtraction operation. This calibration compensation matrix is ​​generated by fitting multi-point calibration data from a blackbody radiation furnace and a constant temperature and humidity chamber before the equipment leaves the factory. The row and column indices of the matrix correspond to the discretized ambient temperature and humidity ranges, respectively, and the internally stored values ​​are dynamic temperature correction biases. The actual ambient canopy temperature difference obtained through this subtraction operation macroscopically characterizes the final thermodynamic performance of the vegetation at the current moment after the coupling effect of its own water transpiration cooling and ambient thermal radiation heating.

[0058] The controller extracts and outputs the pure thermal radiation residual. After obtaining the actual environmental canopy temperature difference and the theoretical temperature difference benchmark, it calculates the algebraic difference between the two, which is used as the current pure thermal radiation residual for the grid cell. The physical essence of this subtraction operation is the common-mode suppression of environmental meteorological driving forces. Through this operation, the system successfully isolates homogeneous temperature fluctuations caused by sudden temperature rises or abrupt changes in solar radiation, while retaining the additional temperature rise caused by abnormal increases in stomatal resistance due to soil water shortage. In terms of the physical meaning of the output result, if the residual approaches zero or is negative, it indicates that the current local soil moisture supply is sufficient; if the residual deviates positively, it reveals that the vegetation in the grid is facing physiological drought stress. To avoid biased decision-making due to reliance on a single extreme value or transient noise, the controller compares the residual with a preset stress temperature rise threshold in real time. The threshold is usually set between 1.0℃ and 2.5℃, and its specific value is determined based on the drought resistance physiological characteristics of the target plant (e.g., higher values ​​for drought-resistant shrubs and lower values ​​for water-loving lawns). The controller uses a set time sliding window to count the percentage of time series where the residual continuously exceeds the aforementioned threshold. Only when this percentage exceeds a set weight lower limit (e.g., 60%) does the system finally confirm the current drought anomaly. The extracted pure thermal radiation residual will serve as the core physical state variable and be imported into the subsequent irrigation decision-making logic chain.

[0059] During a complete day-night meteorological cycle, there is no shortwave solar radiation input at night, vegetation stomata are basically closed, and transpiration heat exchange tends to stagnate. At this time, the thermodynamic evolution of the garden microenvironment is mainly dominated by longwave radiative cooling and sensible heat convection of the surface and canopy. Due to differences in canopy density, soil moisture content, and the material of surrounding hard paving, each grid area exhibits distinctly different thermal inertia characteristics at night. Based on the above-mentioned purely thermodynamic dissipation characteristics at night, the controller needs to move beyond the daytime photosynthetic transpiration assessment framework and enter the nighttime in-depth operating condition, extracting the equivalent thermal hysteresis coefficient of each grid through unsupervised back-calculation, which serves as an auxiliary decision dimension for assessing the water storage heat capacity of the underlying soil. This data processing process includes the following steps: In this embodiment, the controller synchronizes and determines the nighttime deep-dive operating conditions by monitoring whether the global solar shortwave radiation is continuously below the nighttime radiation threshold and checking whether the solar altitude angle is less than zero. The nighttime radiation threshold is typically set to 0 W / m². 2 Up to 5W / m 2 The verification window for continuous time is set to 1 to 2 hours. The physical purpose of establishing this verification window is to completely filter out transient interferences from residual shortwave radiation and atmospheric scattering during dusk, ensuring that the microclimate system truly enters the pure longwave radiation and sensible heat exchange dissipation stage. After the conditions are met, the controller extracts the effective average canopy temperature sequence and ambient air temperature sequence within this time window. To ensure the consistency of multi-source heterogeneous data, the controller strictly registers the timestamps of the two, with an allowable time tolerance set to within 30 seconds. If there is a time axis gap due to sensor communication packet loss, the controller will call the cubic spline interpolation algorithm to smoothly reconstruct the missing sequence, avoiding step noise caused by the crude zero-order hold, thereby ensuring the continuity of subsequent microcalculation calculations. As a preferred protection logic, if the system detects that the proportion of missing effective temperature data within the verification window exceeds the preset safety limit (e.g., exceeding 30% of the total sampling points), spline interpolation will introduce uncontrollable overfitting distortion. In this case, the controller will directly abandon the back-calculation of the current night cycle and call the effective historical coefficients of the previous night as a substitute.

[0060] A discrete difference equation for thermodynamic dissipation is constructed. After entering the nighttime steady state, the heat loss rate of vegetation and the surrounding surface mainly depends on the temperature gradient between the vegetation and the surrounding air, as well as its own equivalent heat capacity. To transform the continuous physical cooling process into a discrete model solvable by a microprocessor, the controller constructs a discrete cooling equation based on the principle of lumped-parameter thermodynamics. The specific expression of this discrete cooling equation is as follows: ; In the formula, and These represent the current time step. Compared to the previous time step The effective average canopy temperature; This represents the discrete time interval between two moments; This indicates the ambient air temperature at the current time step. This is the equivalent thermal hysteresis coefficient to be determined. The controller uses this to calculate the temperature derivative term on the left side of the equation, representing the cooling rate, and the temperature potential difference term within square brackets on the right side, representing the heat dissipation driving force. During the division calculation of the above derivative term, if the system clock jumps or the sampling frequency abnormally increases, causing the discrete time interval to approach zero, it will trigger a computational overflow where the derivative term exceeds infinity. To avoid this underlying risk, the controller enforces a timer interrupt at the system level, limiting the minimum allowable time interval parameter (e.g., 5 minutes). High-frequency redundant sampling data with intervals shorter than this will be directly discarded.

[0061] To construct a historical observation matrix and verify algebraic singularity, and to achieve unsupervised inverse calculation of the equivalent thermal hysteresis coefficient, the system does not rely on externally labeled prior information but instead uses the accumulated monitoring sequences within the historical time window for autoregressive data fitting. The controller concatenates temperature potential difference terms across multiple consecutive time steps to construct an independent variable vector, and concatenates the corresponding temperature derivative terms to construct a dependent variable vector. During conventional least-squares matrix fitting, if extreme isothermal weather conditions such as dense fog, high humidity, or strong calm winds occur at night, the canopy temperature and air temperature remain highly consistent and unfluctuated for extended periods, causing the internal variance of the independent variable vector to approach zero infinitely. Based on this thermodynamic lockout phenomenon, directly calculating the inverse of the covariance matrix by the controller will lead to matrix singularity and iterative divergence. To ensure computational completeness, the controller introduces a matrix condition number pre-check mechanism before inverse calculation. When the condition number of the independent variable matrix exceeds a preset safety threshold (e.g., 10), a pre-check mechanism is implemented. 3 When approaching the singular boundary, the controller automatically applies a value, typically 10, to the diagonal. -4 The minimal regularization penalty term is used to forcibly truncate divergent paths through the ridge regression compensation mechanism, ensuring that the linear equation system has a unique stable solution.

[0062] After solving for the equivalent thermal hysteresis coefficient and performing multi-dimensional physical mapping to eliminate matrix singularity issues, the controller uses algebraic inner product operations to find the optimal linear regression slope that minimizes the sum of squared observation errors across all time steps. This fitted slope, mathematically derived, corresponds to the equation described above. The controller takes the reciprocal of the absolute value of the fitted slope, and uses this as the equivalent thermal hysteresis coefficient of the extracted two-dimensional ground grid cell. In a physical sense, this coefficient represents the thermal inertial resistance of the target area to changes in ambient temperature. Water, as a substance with a high specific heat capacity in nature, directly determines the thermal inertia of the medium. The larger the equivalent thermal hysteresis coefficient, the more abundant the moisture in the underlying soil, and the extremely slow the cooling process; the smaller the coefficient, the more arid the soil, and the more drastic the cooling response.

[0063] After performing a joint evaluation of day and night characteristics and obtaining the nighttime equivalent thermal hysteresis coefficient, the controller did not directly use it as a single switching quantity to trigger the vortex. As a preferred approach, to avoid biased decision-making errors caused by relying solely on nighttime extreme cooling values ​​(such as false rapid cooling caused by abnormal cold air advection at night), the controller converts the extracted nighttime equivalent thermal hysteresis coefficient into a water sufficiency score and weights it with the pure thermal radiation residual output during the day. The controller assigns a higher decision weight to the pure thermal radiation residual during the day (e.g., set to 0.6 to 0.7), as it directly reflects the true physiological stress of plant stomata; at the same time, it assigns a lower auxiliary weight to the nighttime equivalent thermal hysteresis coefficient (e.g., set to 0.3 to 0.4), using it as background evidence of the underlying physical water content. Through the above multi-dimensional day-night weighted joint evaluation logic, the system effectively overcomes the blind spot of microclimate monitoring where there is no transpiration signal at night, and achieves all-weather closed-loop perception of the drought status of the garden grid without increasing the hardware cost of the contact soil moisture sensor.

[0064] In the complex garden environment during the daytime, target vegetation is not only directly driven by solar shortwave radiation, but also heated by additional longwave thermal radiation from surrounding high-temperature background materials such as exposed soil, hard paving, or building walls. This external thermal coupling caused by the spatial heterogeneity of the microenvironment leads to non-physiological temperature increases in the vegetation canopy. If the system directly relies on uncompensated pure thermal radiation residuals for moisture assessment, it is easy to misjudge the external heating of this physical environment as drought stress on the vegetation itself, thereby triggering ineffective over-irrigation. Based on the above thermodynamic crosstalk phenomenon, the controller needs to enter daytime compensation mode, calculate and remove the corresponding warming equivalent through perspective factor coupling. This data processing includes the following steps: In this embodiment, the controller synchronously extracts the high-temperature background heat source features by calling the high-temperature background pixel clusters extracted in the aforementioned adaptive threshold segmentation stage. To quantify the radiation intensity of the background heat source, the controller extracts all valid raw temperature data belonging to the high-temperature background within a set radiation influence radius, centered on the target two-dimensional ground grid cell. This radiation influence radius is typically set to 2.0 meters to 5.0 meters. The basis for this determination is that, since long-wave radiation energy decreases sharply with the square of distance, the background radiation contribution beyond this radius has been dissipated by natural air, and its remaining magnitude is usually below the noise floor limit of the infrared temperature sensor, thus requiring no computational overhead. After extraction, the controller calculates the arithmetic mean temperature of the aforementioned background pixel clusters, using it as the local equivalent background emission temperature. To prevent time-series misalignment of multi-source heterogeneous data under rapidly changing meteorological conditions, the controller forcibly checks the timestamp sequence of the current operating condition at this point, ensuring that the alignment deviation of the acquisition time of the background temperature matrix, target canopy temperature, and environmental meteorological parameters is strictly constrained within a preset 30-second tolerance. If communication congestion causes timestamp mismatch, the controller will trigger historical state latching and reject the high-temperature background data of the current frame.

[0065] The solution is based on the spatially distributed geometric perspective factor, which characterizes the effective proportion of the spatial solid angle of the target canopy receiving thermal radiation from the surrounding background. Considering the enormous computational resources required for real-time reconstruction of a high-precision 3D surface model and performance of dual-vehicle area integration in a conventional embedded microprocessor, the system employs a dimensionality-reduced approximation using a distance attenuation model based on 2D mesh projection. The controller traverses each background mesh within the radiation influence radius, calculating the physical plane Euclidean distance between its geometric center and the geometric center of the target canopy mesh. The controller will determine the physical area of ​​each background grid. Divide by the corresponding distance The square of the result is then summed up, and the summation is multiplied by an empirical canopy morphology correction factor. Thus, the approximate viewpoint factor after dimensionality reduction is obtained. As a preferred method, this canopy morphology correction factor The value range is typically set from 0.1 to 0.3, with the specific value determined based on the geometric crown morphology of the target plant. For example, for spherical shrubs with a large surface area, the cross-section receiving lateral radiation is larger, so the system uses a higher upper limit value (e.g., 0.3), while for vertically spreading conical trees, the system uses a lower lower limit value (e.g., 0.1). When performing the above division calculation of area and distance squared, if the background mesh and target mesh physically overlap due to coordinate system quantization errors or mesh polarization, the distance... This will approach zero. This not only deviates from the logic of real physical space occupancy, but will also directly lead to the failure of the division denominator and overflow of the underlying program. To address this, the controller is forcibly equipped with a minimum distance protection function, which forcibly assigns a value of 0.1 meters to all calculation distances less than 0.1 meters, thereby characterizing the extreme strong radiation coupling condition where the plant roots are close to the edge of the hard paving.

[0066] The coupled calculation of the net absorption of background longwave radiation involves obtaining the background equivalent emission temperature and the approximate viewing angle factor. The controller then uses the Stefan-Boltzmann law to solve for the additional thermal radiation flux absorbed by the canopy. Specifically, the controller calculates the background equivalent emission temperature... The fourth power of the effective average canopy temperature The fourth power is calculated, and the algebraic difference between the two is obtained. Then, the controller multiplies this difference by the Stefan-Boltzmann constant in sequence. Average longwave absorption rate of vegetation canopy (Generally set between 0.95 and 0.98 depending on the plant species), and the approximate viewing factor calculated above. The product of the above physical quantities yields the net absorption of background longwave radiation. The technical objective is to quantify the additional heat passively absorbed by the target plant from the surrounding high-temperature background.

[0067] Thermal impedance network transformation and warming equivalent extraction. After absorbing the additional longwave radiation flux, the target vegetation will exchange sensible heat with the surrounding flowing air through surface convection, thereby re-establishing a local thermal dynamic equilibrium. To map the energy-dimensional heat radiation flux into an intuitive temperature-dimensional state variable, the controller uses a thermal impedance network model for transformation. The controller converts the net absorption of background longwave radiation... Multiplied by the aerodynamic drag of the vegetation canopy Then divide by the physical density of ambient air Specific heat capacity of air at constant pressure The product of these two values ​​yields the quotient, which is the equivalent of the lateral long-wave thermal radiation warming. .

[0068] When performing the division and multiplication conversion logic here, although the density and specific heat capacity of the ambient atmosphere fluctuate very little within the normal meteorological range, and the denominator will not fail to approach zero; however, when calculating aerodynamic drag... Since this resistance value is usually inversely proportional to the ambient wind speed, under extreme calm wind conditions, the calculated resistance value will tend to infinity. To avoid an infinite temperature rise in the mathematical output of the heating equivalent at this time, the controller calls the wind speed lower limit protection logic, and at the same time sets the physical extreme upper limit of the heating equivalent at the final output port (usually set to 3.0℃ to 4.5℃), forcibly truncating divergent calculations that exceed the heat tolerance limit of conventional plants.

[0069] After the correction of the pure thermal radiation residual is integrated with multi-dimensional decision-making, and the temperature increase equivalent is extracted, the system enters the state closed-loop correction stage. The controller performs an algebraic subtraction operation on the preliminary pure thermal radiation residual calculated in the previous stage, subtracting the temperature increase equivalent at the current moment. This allows us to obtain the true compensated residual after eliminating the influence of long-wave radiation crosstalk from the environment. As a preferred protection mechanism, if the residual after deduction is negative due to high-frequency wind speed disturbances or a sudden drop in background temperature, the controller will not accept this reverse anomaly but will force it to zero, thus indicating that the current vegetation is not only free from physiological drought stress but also that the microenvironment is in a state of abundant sensible heat cooling.

[0070] When deciding whether to trigger the bottom-level solenoid valve to start irrigation, the controller's output judgment does not solely rely on the extreme parameter of the compensated thermal radiation residual. To avoid biased decisions caused by localized contamination of the infrared sensor lens or abnormal hot air, the controller imports the compensated residual, the effective precipitation probability for the next 24 hours pushed by the weather station, and the soil texture water-holding coefficient matrix corresponding to the bottom-level target grid into the comprehensive weighted judgment module. Only when the compensated residual exceeds the warning threshold for three consecutive cycles, the probability of future precipitation is less than 20%, and the soil texture allows for free gravity water infiltration, does the controller synthesize and issue the final irrigation control value. This joint judgment logic based on multi-dimensional physical quantities effectively suppresses decision fluctuations caused by single-point measurement errors and improves the long-term operational robustness of the garden water-saving system.

[0071] In garden environments, due to building obstruction and the canopy projection of tall trees, different grid areas exhibit light heterogeneity. Vegetation in shaded micro-zones typically exhibits lower stomatal conductance and thermal response sensitivity compared to vegetation in full-sun areas. If the system uses a static and singular drought threshold across the entire area, weak drought signals in shaded areas will be masked, or physiological temperature rises in full-sun areas will be misjudged as severe water shortages. Based on these physical differences in micro-habitats, the controller needs to incorporate micro-zone light history as a regulating variable, achieving adaptive generation of the threshold through data extrapolation. The core technical principle of this calculation process lies in quantifying the long-term radiant energy received by vegetation and using this as a physical scale to dynamically scale the baseline thermal residual. This data processing process includes the following steps: In this embodiment, the controller retrieves the global solar shortwave radiation sequence collected by a local weather station and simultaneously receives the target grid shading coefficient matrix output by the shadow analysis module. To ensure the spatiotemporal consistency of the micro-area illumination data, the controller forcibly compares the timestamps of the radiation sequence and the shading coefficient matrix. The allowable time alignment tolerance is set to within 1 minute. If a time deviation exceeding this tolerance is detected, the controller triggers linear interpolation compensation logic to forcibly align the time axis to prevent miscalculation of radiation caused by asynchronous data. After alignment, the controller multiplies the global solar shortwave radiation by the corresponding shading coefficient to obtain the actual received radiance of the target two-dimensional ground grid cell in the micro-area at a specific time step. .

[0072] After obtaining transient radiance, the controller needs to assess the long-term impact of historical light accumulation on vegetation physiological characteristics by discretely integrating the effective light accumulation in a micro-area. Specifically, the controller sets a historical assessment time window. This window typically spans 3 to 7 days, determined to fully cover the short-term acclimatization cycle of common garden plants' stomatal morphology to the microclimate environment. Within this window, the controller applies all light exceeding the lower limit of physiologically effective radiation (e.g., set at 50 W / m²). 2 The actual received radiance of the micro-area With the set time step Perform multiplication and discrete summation to calculate the effective light accumulation in the micro-area. When performing the multiplication operation of the above discrete time intervals, if the calculated time difference is negative or close to zero due to an abnormal reset of the underlying timer, the controller will force the use of a preset rated sampling period (e.g., 10 minutes) to replace it, ensuring the physical continuity of the integral components.

[0073] Based on the adaptive generation of a threshold using a normalized ratio, after obtaining the effective cumulative ground radiation, the controller dynamically adjusts the residual threshold for determining drought stress. The controller also extracts the system's preset standard cumulative solar radiation. This standard value is provided by the system's built-in local meteorological historical database and represents the theoretical radiation integral benchmark for an unobstructed, full-sunlight environment at the same geographical latitude. The controller will calculate the effective cumulative radiation at the ground. Divide by standard cumulative light intensity Calculate the illumination normalization coefficient Subsequently, the controller sets the system baseline residual threshold. Multiply by the illumination normalization coefficient to generate the adaptive decision threshold for the current mesh. When performing the above normalized division operation, if the standard cumulative illumination is reduced due to lost system initialization parameters or bad memory blocks, the error may occur. A value approaching zero will directly trigger a division overflow fault in the microprocessor. To ensure the integrity of the underlying algorithm, the controller incorporates a minimum standard threshold protection mechanism for the denominator, forcibly limiting the denominator to no less than a set minimum self-radiation constant (e.g., 1.0 MJ / m²). 2 Through the aforementioned ratio-based adjustment mechanism, the adaptive threshold generated in the long-term shaded area will be reduced proportionally, effectively improving the system's physical sensitivity to capturing weak transpiration stress signals.

[0074] The adaptive threshold's boundary constraints and multi-dimensional fusion judgment directly determine the trigger boundary for the underlying irrigation decision. As a preferred approach, to avoid excessive collapse of the adaptive threshold caused by extreme weather events (such as continuous heavy rain leading to extremely low light accumulation), which could trigger malfunctions of the solenoid valve even with slight temperature fluctuations, the controller must impose boundary constraints on the generated threshold. The controller extracts the physiological tolerance limit corresponding to the target plant species and sets a hard physical upper limit (e.g., 3.5℃) and lower limit (e.g., 0.5℃) for the adaptive threshold. Any calculation results exceeding this constraint range are forcibly truncated to the boundary value. In the final decision stage, the output drought judgment result does not simply rely on the extreme value of the comparison between the pure thermal radiation residual and the adaptive threshold. The controller correlates and superimposes the above threshold comparison results, the measured soil surface moisture curve, and the recent irrigation calendar. Only when the residual exceeds the adaptive threshold for multiple consecutive judgment cycles, and the soil physical moisture content falls below the preset warning line, does the controller synthesize and output the final drought confirmation signal. This logical design, which superimposes physical boundaries and multi-dimensional causality, effectively suppresses control system jitter caused by single meteorological distortions.

[0075] After completing the derivation of the micro-area adaptive threshold, the system needs to transform the abstract thermodynamic state parameters into underlying mechanical irrigation actions. Due to high-frequency thermal disturbances caused by sudden changes in wind speed or transient cloud cover, if the controller directly closes the relay based solely on the temperature extreme value at a single moment, it will lead to frequent opening and closing of the solenoid valve. This will not only shorten the hardware lifespan but also fail to meet the agronomical requirements for deep soil immersion. Based on these control engineering challenges, the controller introduces decoupling comparison logic in the time series dimension and combines it with multi-dimensional data hysteresis intervals to execute robust solenoid valve drive control. The core technical principle of this control process lies in transforming the easily disturbed temperature parameters into physical drive signals with disturbance resistance through time-domain sliding filtering and impedance closed-loop feedback in the underlying hardware. This data processing and control process includes the following steps: In this embodiment, the controller retrieves the real-time compensated crown temperature difference (CATD value, hereinafter referred to as CATD) extracted by the preceding logic for synchronous extraction and state locking of real-time feature quantities. ) and the corresponding adaptive decision threshold for mesh generation To ensure physical consistency of the comparison benchmark, the controller forcibly compares the timestamps of the compensated crown temperature difference sequence and the adaptive judgment threshold sequence, setting an allowable time alignment tolerance of within 30 seconds. Because the underlying communication bus is susceptible to electromagnetic interference from high-power water pumps during startup in real-world garden environments, resulting in data packet loss, if three consecutive sampling periods are detected... If data is missing, the controller will trigger a state lockout mechanism. In this state, the controller refuses to execute subsequent numerical comparison logic and forcibly maintains the drive output level of the previous valid cycle, thereby avoiding malfunctions of the solenoid valve caused by zero-value input or discrete random codes.

[0076] After performing decoupling comparison and transient drought marker location, and obtaining synchronization baseline data, the controller calls the microprocessor's low-level arithmetic logic unit to calculate the crown temperature difference after real-time compensation. With adaptive decision threshold The algebraic difference. The technical purpose of this calculation is to quantify the physical degree to which the current vegetation deviates from a healthy water supply baseline. When real-time compensation is applied to the crown temperature difference... Greater than the adaptive judgment threshold When the threshold is reached, it indicates that the stomatal closure effect caused by water scarcity within the vegetation has overwhelmed the sensible heat dissipation capacity of the external microenvironment, resulting in a true physiological water shortage in the local area. At this point, the controller sets the transient drought flag for the current time step to logic high; otherwise, if the threshold is not exceeded, the flag is set to 0.

[0077] Based on a sliding time window-based hysteresis filtering and multi-dimensional decision-making, to filter out high-frequency noise pulses caused by birds blocking the infrared field of view or occasional temperature drift of sensors, the controller does not directly map transient flags to control words of the underlying actuators. As a preferred approach, the controller allocates a sliding statistical window in memory with a time span of 15 to 30 minutes. The width of this window is determined based on the physiological inertial cycle of common garden plants' stomata responding extensively to soil moisture changes and causing macroscopic fluctuations in canopy temperature, which is typically within this time range. The controller calculates the cumulative percentage of time the transient drought flag is set to 1 within this window. Only when this percentage exceeds a set confidence threshold (e.g., 80%), and the physical moisture content returned by the underlying soil moisture sensor is simultaneously below the wilting coefficient warning line for the target plant, does the controller synthesize and issue the final physical irrigation enable signal. Through the aforementioned hysteresis filtering and multi-source weighted logic, the system effectively blocks the transmission of non-stationary meteorological noise to the downstream mechanical layer.

[0078] The solenoid valve drive employs isolation and current feedback closed-loop control. Upon receiving the irrigation enable signal, the controller outputs a high logic level through its general-purpose input / output port. To prevent the back electromotive force generated by the external pressurized water pump and the 24V AC solenoid valve during shutdown from damaging the main control chip, the controller utilizes an optocoupler for opto-isolation. This drives the high-power solid-state relay at the back end to close, directing the pressurized irrigation water from the pipeline network to the target garden grid. During the drive execution, the controller synchronously acquires the real-time feedback current flowing through the solenoid valve coil. To achieve real-time monitoring of the underlying hardware status, the controller performs equivalent impedance calculation based on Ohm's law, and sets the rated drive voltage accordingly. (Typically 24V) divided by the real-time feedback current Calculate the equivalent operating impedance of the coil. When performing the above division operation, if the real-time feedback current is transmitted back due to communication disconnection or physical disconnection... If the denominator approaches zero, it will directly introduce an underlying program overflow in the division operation due to the denominator being zero. To ensure the physical integrity of the algorithm, the controller incorporates an overflow protection mechanism here: when an overflow is detected... When the current is below the preset minimum leakage current limit (e.g., 5mA), the controller skips the impedance division operation and uses its equivalent impedance. If the value is forcibly marked as infinite, it indicates a physical circuit open circuit or poor terminal contact; if the current is fed back in real time... If the current exceeds the upper limit of the rated operating current, it indicates that there is an inter-turn short circuit in the coil or mechanical jamming of the valve core. For such underlying hardware failure boundaries, the controller will immediately cut off the output level for hardware protection and report the abnormal alarm through the IoT gateway to ensure the stable operation of the microclimate water-saving system.

[0079] Specific application examples: In this application embodiment, the assessment of the vegetation canopy temperature in the target landscaped area is subject to strict constraints of light and shadow variations and thermodynamic dissipation. At 14:00 (corresponding to the high radiation period in the chart), the controller detects data entry from the infrared temperature sensor and the weather station. At this time, the microenvironment is in a period of high radiation fluctuation, and the surrounding tall concrete landscape walls absorb a large amount of short-wave radiation and release strong lateral long-wave heat energy. The controller extracts the spatial grid set corresponding to the currently entered data, confirms the existence of thermal radiation crosstalk caused by the landscape walls, sets the original canopy temperature state to a state to be compensated, and writes an absolute timestamp. When the two-dimensional temperature matrix is ​​subsequently entered and meets the operating condition alignment tolerance threshold, the system accurately triggers joint calculation, packages the integrated multi-source environmental parameters and sends them to the bimodal histogram threshold segmentation module, according to the formula... The optimal segmentation point was selected by iteratively calculating the inter-class variance. After confirming the removal of high-temperature background pixels, the multi-dimensional compensation channel was activated in parallel. Due to the objective influence of the intense local baking at the time, the original compensation canopy temperature showed a significant distortion exceeding the ambient temperature (corresponding to the dark gray dashed line bulge), and the thermodynamic deviation output a certain temperature rise value accordingly. Based on the perspective factor and thermal impedance network conversion logic of the daytime real-time compensation solver, the temperature increase equivalent of this local grid was calculated to reach 2.1℃. If a traditional timed and quantitative or fixed temperature threshold control system is used, the external temperature rise of up to 2.1℃ will directly trigger a false drought judgment, causing the environmental thermal crosstalk to be mistakenly judged as plant water shortage and resulting in invalid false spraying.

[0080] The dynamic threshold generation and thermodynamic compensation module of this invention adaptively calculates the true canopy temperature after deducting the warming equivalent by solving the shadow projection matrix. The system compares and judges, accurately identifying that the temperature rise anomaly belongs to the normal light spot and lateral baking effect under the microclimate background, thereby generating a status code to safely suppress the malfunction of the bottom solenoid valve.

[0081] When the system operates smoothly until the nighttime period when there is no shortwave radiation input, the ambient longwave radiation cools and falls back to the nighttime baseline. The weather station records another batch of environmental data, in which the global shortwave radiation is consistently below the nighttime radiation threshold. The thermodynamic dissipation analysis module immediately captures this purely sensible heat exchange dissipation process and calls the discrete cooling equation. The equivalent thermal hysteresis coefficient of the target two-dimensional ground grid is obtained by reverse deduction. At this point, the micro-area illumination history configuration unit detected that the shrubs near the landscape wall were in a long-term physical shaded area, posing a risk of weakened transpiration signals. It immediately invoked the normalization ratio rule to adaptively lower its residual threshold and triggered the extreme value extraction channel to directly obtain the actual canopy temperature difference within the daytime historical sliding window. Subsequently, the system forcibly output the final transient drought flag bit through decoupled comparison logic. Under the actual condition of gradually depleting soil moisture, the dynamic judgment threshold of this invention, derived from the effective cumulative illumination, was significantly lower than that of the full-sunlight area. The system comparison revealed that the actual canopy temperature difference continuously exceeded this adaptive threshold, thus triggering a high-level water shortage warning.

[0082] Upon receiving the irrigation enable status code, the solenoid valve-driven isolation and closed-loop execution unit immediately closes the high-power solid-state relay in the underlying hardware circuit via an optocoupler, and simultaneously collects feedback current to confirm that the coil's equivalent impedance is normal, thus completely opening the pressurized water supply path from the pipeline network to the target garden micro-area. At the same time, the system generates a visual comparison curve showing the smooth trend of the corrected actual canopy temperature (black solid line) and the ambient air temperature (light gray dotted line), and pushes it to the monitoring terminal of the garden management personnel. Just as... Figure 3 And the results of the 30-day comparative experiment revealed: Traditional irrigation control mechanisms suffer from a rigid, fixed temperature boundary and a lack of micro-habitat heat radiation stripping channels, resulting in a false spraying rate (false drought warning rate) as high as 27% and a total water consumption of 145.6 cubic meters. In contrast, the intelligent irrigation system of this invention, through the fusion calculation of the aforementioned light and shadow tracking and thermodynamic compensation dual-track system, significantly reduces the drought false spraying rate to below 1.5%, while converging the total water consumption limit to 89.2 cubic meters (achieving approximately 38.7% water resource savings), thus completely solving the ineffective irrigation blind spots under complex garden heterogeneous microclimates from a thermodynamic perspective.

Claims

1. A landscaping intelligent irrigation control system, characterized in that, include: Weather stations are used to collect ambient air temperature and global shortwave solar radiation. Infrared temperature sensor, used to collect a two-dimensional temperature matrix of the target garden green area; Solenoid valves are installed on the water supply network of the target garden greening area; The controller establishes a global three-dimensional spatial coordinate system containing two-dimensional ground grid units and two-dimensional facade grid units, performs ray projection simulation to generate a shadow projection matrix, multiplies and integrates the global solar shortwave radiation with the shadow projection matrix, and calculates the effective cumulative radiation of the ground of the two-dimensional ground grid unit and the effective cumulative radiation of the facade of the two-dimensional facade grid unit respectively. The effective original canopy temperature of the two-dimensional ground grid unit is extracted based on the two-dimensional temperature matrix. The lateral long-wave thermal radiation warming equivalent is calculated based on the effective cumulative radiation of the facade and the preset equivalent thermal hysteresis coefficient. The corrected real canopy temperature is obtained by subtracting the lateral long-wave thermal radiation warming equivalent from the effective original canopy temperature. An adaptive judgment threshold is generated based on the effective cumulative radiation on the ground. The difference between the corrected real canopy temperature and the ambient air temperature is calculated to obtain the real-time compensated canopy temperature difference. The real-time compensated canopy temperature difference is compared with the adaptive judgment threshold to generate a control command and execute intelligent irrigation operation. 2.The intelligent irrigation control system for landscaping according to claim 1, wherein, The global three-dimensional spatial coordinate system is established, the ground area with vegetation is discretized into the two-dimensional ground grid unit, the three-dimensional geometric contour of the surrounding stationary obstacles is obtained, the light-receiving facade is discretized into the two-dimensional facade grid unit, and the initial equivalent thermal hysteresis coefficient is assigned to the two-dimensional facade grid unit. The controller calculates the solar altitude angle and solar azimuth angle according to a set time step, performs ray projection simulation calculation to generate the shadow projection matrix, extracts the global solar shortwave radiation collected by the weather station and multiplies it with the shadow projection matrix, integrates within a sliding time window and outputs the effective cumulative radiation of the ground of the two-dimensional ground grid unit and the effective cumulative radiation of the facade of the two-dimensional facade grid unit.

3. The intelligent irrigation control system for landscaping according to claim 1, characterized in that, The controller extracts the effective original canopy temperature of the two-dimensional ground grid cell based on a two-dimensional temperature matrix, including: The bimodal histogram threshold segmentation algorithm is used to divide the data into a high-temperature background pixel cluster and a low-temperature vegetation pixel cluster. Data in the high-temperature background pixel cluster is excluded, and the average temperature in the low-temperature vegetation pixel cluster is calculated as the effective original canopy temperature of the two-dimensional ground grid unit.

4. The intelligent irrigation control system for landscaping according to claim 3, characterized in that, After obtaining the effective original canopy temperature, the controller constructs a spatially correlated neighborhood of the target grid, calculates the local temperature mean and local temperature standard deviation within the spatially correlated neighborhood, and constructs a confidence interval based on the local temperature mean and local temperature standard deviation. The controller filters out abnormal pixels that do not fall within the confidence interval and calculates the effective average canopy temperature weighted by spatial distance. In subsequent calculations, the effective average canopy temperature replaces the effective original canopy temperature in the calculation.

5. The intelligent irrigation control system for landscaping according to claim 1, characterized in that, The controller uses the temperature residual to inversely solve and update the equivalent thermal hysteresis coefficient of the two-dimensional facade mesh unit under the night parameter self-calibration condition; The controller constructs a discrete difference equation for thermodynamic dissipation, builds a historical observation matrix, solves for the fitting slope that minimizes the sum of squared observation errors, and takes the reciprocal of the absolute value of the fitting slope as the updated equivalent thermal hysteresis coefficient.

6. The intelligent irrigation control system for landscaping according to claim 3, characterized in that, The controller extracts the background equivalent emission temperature of the high-temperature background heat source under the real-time compensation condition during the day, and solves the approximate viewing angle factor based on the relative spatial position of the two-dimensional ground grid unit and the two-dimensional facade grid unit in the global three-dimensional spatial coordinate system. The controller combines the approximate viewing angle factor with the Stefan-Boltzmann law to solve for the net absorption of background longwave radiation by the canopy, and uses a thermal impedance network model to convert the net absorption of background longwave radiation into the equivalent of lateral longwave thermal radiation warming.

7. The intelligent irrigation control system for landscaping according to claim 1, characterized in that, An adaptive threshold for determining radiation is generated based on the effective cumulative ground radiation, including: The controller calculates the ratio of the effective cumulative radiation on the ground to the preset standard cumulative illumination to obtain the illumination normalization coefficient; The controller multiplies a preset baseline residual threshold used to characterize the degree of water deficit by the light normalization coefficient to generate the adaptive judgment threshold.

8. The intelligent irrigation control system for landscaping according to claim 1, characterized in that, The controller calculates the algebraic difference between the real-time compensated crown temperature difference and the adaptive judgment threshold to generate a transient drought flag. The controller counts the percentage of time periods in which the transient drought flag is set to logic high within a sliding statistical window. When the percentage of time periods exceeds the lower confidence limit, a synthetic irrigation enable signal is sent to the solenoid valve as the control command.

9. The intelligent irrigation control system for landscaping according to claim 1, characterized in that, The controller synchronously acquires the real-time feedback current flowing through the coil during the driving of the solenoid valve, and performs equivalent impedance calculation based on the set rated driving voltage and the real-time feedback current to obtain the equivalent operating impedance. When the real-time feedback current is lower than the preset minimum leakage current limit, the controller marks the equivalent operating impedance as infinite and cuts off the output level.

10. A control method for a smart irrigation control system for landscaping according to any one of claims 1-9, characterized in that, Includes the following steps: Establish a global three-dimensional spatial coordinate system, discretize the ground vegetation area and the light-receiving facade of the surrounding static obstacles into two-dimensional ground grid units and two-dimensional facade grid units respectively, and assign an initial equivalent thermal hysteresis coefficient. A ray projection simulation is performed to generate a shadow projection matrix. The obtained global solar shortwave radiation is multiplied by the shadow projection matrix and integrated to calculate the effective cumulative radiation of the ground of the two-dimensional ground grid unit and the effective cumulative radiation of the facade of the two-dimensional facade grid unit, respectively. The high-temperature background data in the two-dimensional temperature matrix is ​​excluded by using a bimodal histogram threshold segmentation algorithm, and the average temperature of the low-temperature vegetation pixels is calculated as the effective original canopy temperature of the two-dimensional ground grid unit. The equivalent thermal hysteresis coefficient is updated according to the working conditions, and the lateral long-wave thermal radiation warming equivalent is calculated by combining the approximate viewing angle factor, the effective cumulative radiation of the facade, and the equivalent thermal hysteresis coefficient. The lateral long-wave thermal radiation warming equivalent is subtracted from the effective original canopy temperature to obtain the corrected real canopy temperature. An adaptive judgment threshold is generated based on the effective cumulative radiation of the ground. The difference between the corrected real canopy temperature and the ambient air temperature is calculated to obtain the real-time compensated canopy temperature difference. The real-time compensated canopy temperature difference is compared with the adaptive judgment threshold to generate a control command, which drives the solenoid valve to perform intelligent irrigation operation.

Images