Intelligent precision spraying control method and system for smart garden

Abstract

The application discloses a kind of intelligent precision spraying control method and system for smart garden, its method includes the following steps: S1: multi-source environment and equipment state perception: collection garden environment parameter, vegetation state parameter and spraying equipment operating parameter, and constructs environment state vector.The application realizes the dynamic precision control of different space unit spraying water volume by multi-source environment and equipment state fusion perception, garden area spatial differentiation modeling, intelligent prediction of water demand and wind field and evaporation loss compensation mechanism, and forms closed-loop control system in combination with spraying effect feedback and self-learning optimization, so that spraying decision is changed from static experience rule to data-driven adaptive strategy, still can maintain stable, accurate irrigation effect under complex and changeable meteorology and growth environment, significantly improve water resource utilization efficiency and intelligent level of garden maintenance.

Description

Technical Field

[0001] This invention relates to the field of smart garden technology, and in particular to an intelligent and precise spraying control method and system for smart gardens. Background Technology

[0002] With the continuous advancement of smart garden and digital city construction, garden greening maintenance is gradually developing from traditional manual experience-driven methods to informatization and automation. Automatic sprinkler systems such as sprinkler irrigation and micro-sprinklers, as an important part of water-saving irrigation in gardens, have been widely used in urban parks, scenic green spaces and municipal green areas. Most existing smart garden sprinkler control methods are based on single or limited environmental parameters. They mainly use fixed thresholds or preset rules to determine the spraying duration and water volume. While these methods can achieve basic automated sprinkler control, they still have the following significant shortcomings: Existing technologies have limited sensing dimensions and weak environmental characterization capabilities: most systems only collect a few indicators such as soil moisture or ambient temperature, lacking comprehensive perception of multi-source information such as light intensity, wind speed and direction, vegetation growth status and spraying equipment operation status, making it difficult to fully reflect the real state of the garden environment. Spraying decision rules are static and have poor adaptability: Existing spraying control strategies are usually based on experience thresholds or fixed models, which cannot be adaptively adjusted according to dynamic environmental changes. This results in spraying strategies lagging behind during seasonal changes, extreme weather, or changes in vegetation growth stages, and unstable spraying effects. Lack of spatial differentiation modeling mechanism: Existing systems mostly spray uniformly on the whole park or large area, without spatial fine-grained modeling of different vegetation types, soil conditions and shading conditions, making it difficult to achieve zoned fine irrigation, which can easily lead to over-irrigation or under-irrigation in local areas. Insufficient spraying precision and inadequate consideration of wind field and evaporation loss: Existing technologies usually ignore the influence of wind speed, wind direction and evaporation factors on the effective infiltration rate of water. Under conditions of strong wind, high temperature or strong light, a large amount of sprayed water drifts or evaporates, resulting in serious waste of water resources. Existing systems generally lack feedback and self-learning mechanisms for spraying effects: after spraying, there is a lack of assessment of changes in soil moisture and vegetation growth response, making it impossible to continuously optimize water demand models and spraying strategies based on historical effects, and making it difficult to form an intelligent, self-evolving, and precise spraying closed-loop control system. In light of the above, there is an urgent need to propose an intelligent and precise spraying control method and system for smart gardens. This system should enable closed-loop control that incorporates multi-source sensing, spatially differentiated modeling, water demand prediction, wind field and evaporation loss compensation, and self-learning optimization of spraying effects. This would improve spraying accuracy, reduce water waste, and enhance the intelligence and precision of garden maintenance. Summary of the Invention

[0003] Based on the technical problems existing in the background technology, this invention proposes an intelligent and precise spraying control method and system for smart gardens, which solves the problems of single perception dimension, weak environmental characterization ability, static spraying decision rules, poor adaptability, lack of spatial differentiation modeling mechanism, insufficient spraying accuracy, insufficient consideration of wind field and evaporation loss, and lack of spraying effect feedback and self-learning mechanism in the existing technology.

[0004] This invention proposes an intelligent and precise spraying control method for smart gardens, comprising the following steps: S1: Multi-source environment and equipment status perception: Collect garden environment parameters, vegetation status parameters and spraying equipment operation parameters, and construct an environmental status vector; S2: Spatial Differentiation Modeling: Divide the garden area into multiple spatial units and establish an independent environmental state model and vegetation water requirement model for each spatial unit; S3: Water demand prediction: Based on historical data and current environmental conditions, the water demand of each spatial unit within a preset time window is predicted to obtain the target spraying water volume; S4: Wind field and evaporation loss compensation: Based on real-time wind speed, wind direction, temperature and light intensity, the predicted spraying water volume is effectively corrected for infiltration, and a compensated spraying control volume is generated. S5: Dynamic spraying decision and execution: Based on the compensated spraying control volume, dynamically adjust the nozzle flow rate, spraying angle and spraying duration to achieve spatially differentiated precision spraying; S6: Spraying effect feedback and evaluation: Collect data on changes in soil moisture and vegetation growth status after spraying to evaluate the spraying effect and water use efficiency. S7: Self-learning optimization: Based on the spraying effect evaluation results, the water demand prediction model and spraying decision parameters are updated to form a self-learning closed loop.

[0005] Preferably, the specific logical steps of S1 are as follows: S101: In various spatial units of the garden Multiple types of sensor nodes are deployed internally to periodically collect soil parameters, meteorological parameters, vegetation parameters, and equipment status. Soil parameters include soil volumetric moisture content. Soil temperature Soil electrical conductivity EC, meteorological parameters including air temperature Air humidity (H) and wind speed ,wind direction Light intensity Vegetation parameters include Normalized Difference Vegetation Index (NDVI), Leaf Area Index (LAI), and Canopy Temperature. Equipment status includes nozzle flow rate. Valve opening and spray pressure ; S102: Construct the original observation vector based on the parameters collected in S101: ; in This represents the original observation vector, with the superscript T indicating the transpose operation, used to transform the vector into its original form. Transform row vectors into column vectors; S103: Use sliding window mid-value filtering to remove outliers: ; The data for each dimension are then normalized. ; in This represents the result of median filtering on the k-th dimension observation data at time t. This represents the original observation data in the k-th dimension. This indicates the current time, and L represents the length of the sliding window. ( () represents the mean value function. This indicates that the k-th dimension observation data is at "time". At the time "The set of all observations within this time window, This represents the result of normalizing the k-th dimension observation data at time t. This represents the historical minimum value of the k-th dimension observation data. This represents the historical maximum value of the Kth dimension observation data, where K represents the dimension index of the observation data; S104: Based on the fusion environment state of the previous moment The predicted state is obtained by predicting the current environmental state: Then the predicted state is mapped to the predicted observation. and compared with the current measured values Calculate the residuals: ; in This indicates that at time t, only the previous time period is used. Based on the observation data at and before time t, the predicted estimate of the system state X(t) is given. Indicates the previous moment, ( ) represents the state transition function. This represents the observation residual at time t. ( ) represents the observation mapping function, This represents the actual observation vector at time t; S105: Construct an adaptive weight matrix based on the historical stability and real-time fluctuation of each sensor. Finally, the predicted state is corrected to obtain the current fused environment state vector: ; in This represents the final state estimate at time t obtained after incorporating the observation residuals. The gain matrix represents time t; S106: The final environmental state vector of the garden space unit is obtained: ; in This represents the environmental state vector of a garden space unit. This represents the state estimation vector of a spatial element at time t. These represent the estimated values ​​for soil volumetric moisture content, soil temperature, air temperature, air humidity, wind speed, light intensity, normalized difference vegetation index, leaf area index, canopy temperature, sprinkler flow rate, valve opening, and spray pressure, respectively.

[0006] Preferably, the specific logical steps of S2 are as follows: S201: Divide the entire garden area into several spatial units according to geographical coordinates, vegetation distribution, and sprinkler network structure: ; in This represents the garden space grid cell in row i and column j. This refers to the collection of areas within the entire garden. This indicates the total number of grids that divide the garden area along the "row" direction. This represents the total number of grid cells in the garden area along the "column" direction, i This represents the spatial cell coordinate index, where i represents the row and j represents the column; S202: Merge the environment state vector obtained in S105 Mapped to the corresponding spatial unit: ; in , This represents the environmental state vector of the spatial grid cell in the i-th row and j-th column at time t. This represents the state estimation vector of a spatial element at time t. Let represent the soil volumetric water content, soil temperature, air humidity, wind speed, light intensity, normalized vegetation index, leaf area index, and canopy temperature of grid (i,j), respectively. S203: Based on vegetation type, soil type, and functional zone attributes, assign weighting coefficients to each spatial unit: ,in This represents the overall weighting coefficient. These represent the weighting coefficients for vegetation type, soil type, and functional zone attributes, respectively. S204: Constructing the basic water demand model for each spatial unit: The reference evaporation rate is as follows: ,in represents the reference evapotranspiration, Indicates the basic water demand. , and All represent empirical fit coefficients. Indicates light intensity. Indicates the soil reference temperature; S205: Adjust water requirement based on current soil moisture content and vegetation status: ; in This indicates the revised water requirement. Weighting coefficients representing vegetation growth The weighting coefficient representing soil moisture content. The optimal normalized vegetation index represents the vegetation. This indicates the optimum volumetric moisture content of the soil.

[0007] Preferably, the specific logical steps of S3 are as follows: S301: For each spatial unit Obtain historical environment state sequence: and the environmental state vector of garden space unit ; in This indicates that the grid (i,j) is at time "time". arrive "The historical state sequence during this period, This indicates that the grid (i,j) is in the " , until "State vectors at each time step, k is the length of the historical time window, t represents the current time step, i..." This represents the spatial cell coordinate index, where i represents the row and j represents the column; S302: Establish a predictive model for each spatial unit based on historical data and the current environmental state: ,in Indicates a prediction of the future time window Water demand within the area Represents the prediction function. These parameters represent vegetation type and soil type, respectively, reflecting the water requirements of plants and the water-holding capacity of soil. This indicates that the grid (i,j) is at time "time". arrive "The historical state sequence during this period;" S303: Apply a weighted correction to the forecast results based on historical trends: ; in This indicates that the grid (i,j) is in the " "Water demand after real-time fusion correction" This indicates that the fusion coefficient ranges from 0 to 1, and is used to control the weighting of historical data and model predictions. This indicates that the grid (i,j) is in the " "Historical experience at every moment requires water." Indicates time interval, This indicates that the grid (i,j) is in the " "Current calculated water demand at any given moment;" S304: Use the revised water demand as the target spray water volume for the space unit. ,in This indicates the target spray volume of water.

[0008] Preferably, the specific logical steps of S4 are as follows: S401: For each spatial unit Obtain the predicted target spray volume and real-time environmental parameters, including wind speed. ,wind direction air temperature (t), light intensity (t) and air humidity (t); S402: Correct spray deviation based on wind speed and direction, defining wind field correction coefficients. : ; in Indicates the wind sensitivity coefficient. Indicates the direction of the spray from the nozzle. ( ) represents an exponential function. Represents the cosine function; S403: Calculate the evaporation loss correction factor based on ambient temperature, light intensity, and air humidity. : ; in Indicates the evaporation sensitivity coefficient. Indicates the air reference temperature. This represents the normalized maximum light intensity. S404: Applying wind field and evaporation correction factors to predict spray volume yields the compensated spray volume for each spatial unit. ; in Indicates the amount of compensation sprayed; S405: Output This informs the next steps in dynamic spraying decisions and execution.

[0009] Preferably, the specific logical steps of S5 are as follows: S501: For each spatial unit Obtain the compensated spray water volume And the parameters of the spraying equipment, including the maximum flow rate of the nozzles. Spraying angle range and spray pressure range ; S502: Based on the target spray volume and spray duration Calculate the nozzle flow rate: ; in For the area of ​​a spatial unit, Indicates the nozzle flow rate, if Then the restriction is And adjust the spraying time accordingly; S503: Adjust the spraying angle based on vegetation distribution and wind direction within the spatial unit. : ; in , This represents the wind direction correction factor. Indicates the reference direction for spraying. Indicates the spray reference angle of the nozzle. This indicates the amount of spray angle correction caused by wind direction. Indicates wind speed. Indicates wind direction, t represents the current time, and i This represents the spatial cell coordinate index, where i represents the row and j represents the column; S504: Based on the calculated nozzle flow rate and target spray volume The spraying time is recalculated to ensure that the sprayed water volume accurately matches the compensated target volume. The formula used is as follows: ; in Indicates the spraying duration; S505: Adjust nozzle flow rate Spraying angle and spraying duration Send the message to the spray controller to complete the dynamic spraying execution.

[0010] Preferably, the specific logical steps of S6 are as follows: S601: Preset evaluation time after spraying is completed Collect data from each spatial unit. The feedback data includes soil moisture content before spraying. Soil moisture content after spraying Vegetation growth indicators and the actual amount of water sprayed ; S602: Calculate the increase in soil moisture resulting from spraying: ; in This represents the increase in soil moisture in spatial unit i. This indicates that spatial unit i is at the moment of spraying effect evaluation. Soil volumetric water content, This indicates that spatial unit i is at the moment of spraying start. The soil volumetric water content, where i represents the index of the spatial cell. Indicates the moment for evaluating the spraying effect; S603: Convert soil moisture increase into effective infiltration volume: ; in This represents the effective depth of the soil layer in spatial unit i. Represents the area of ​​a spatial unit. Indicates the effective infiltration volume; S604: Calculate water use efficiency: ; in Indicates water use efficiency; S605: Constructing a spraying effect scoring function: ; in This indicates the actual effect score. , This represents a weighting coefficient used to balance water-saving efficiency and vegetation growth. Indicates that spatial unit i is at the evaluation time Vegetation growth indicators This indicates that spatial unit i is at the moment of spraying start. Vegetation growth indicators; S606: Rate the actual effect With the target effect Comparison: ,in Indicates error; And based on error The parameters of the water demand model are adaptively updated to provide feedback for the next cycle S2-S5, thereby achieving closed-loop optimization control.

[0011] Preferably, the specific logical steps of S7 are as follows: S701: Spraying effect error obtained from S606 In each spatial unit Constructing learning samples: ,in This represents the environmental feature vector of the unit. Representing spatial units The learning sample set This indicates the actual amount of water sprayed. Indicates the actual effect score; S702: With error As a loss function, the parameters of the water demand prediction model Perform gradient updates: ,in For learning rate, , Represents the loss function. express Model parameters after the next iteration This represents the model parameters at the t-th iteration. Represents the loss function For parameters The gradient; S703: Based on water use efficiency Dynamically adjust spray control parameters: ; in , , This represents the adjustment coefficient, F represents the flow rate, and T represents the duration. Indicates angle, Indicates the first Spray flow control parameters for the wheel, Indicates the first Spray flow control parameters for the wheel, Indicates the first Spraying duration control parameters for the wheel, Indicates the first Spraying duration control parameters for the wheel, Indicates the first Spray angle control parameters of the wheel, Indicates the first The spray angle control parameter of the wheel, where t represents the iteration cycle; S704: When multiple consecutive cycles satisfy: If the historical spraying strategy of the spatial unit is marked as a failed strategy, its sample weight is reduced to avoid amplifying erroneous experience. Indicates the error threshold. This represents the absolute value of the spraying effect error of spatial unit i; S705: The updated water demand model parameters and spraying decision parameters are written back to the S3-S5 process, so that the water demand forecast and dynamic spraying decision in subsequent cycles can be continuously optimized based on historical effect feedback, and a closed-loop control mechanism of "prediction-execution-evaluation-relearning" is constructed.

[0012] This invention also proposes an intelligent precision spraying control system for smart gardens, including a multi-source sensing module, a spatial modeling module, a water demand prediction module, a spraying compensation and decision-making module, a spraying execution module, and a feedback evaluation and self-learning module. The multi-source sensing module is used to collect garden environmental parameters, vegetation growth status parameters and spraying equipment operation status parameters, and to perform time synchronization and noise suppression processing on the multi-source heterogeneous data to construct an environmental status observation vector. The spatial modeling module is used to divide the garden area into multiple spatial units and to establish an independent environmental state model and vegetation water requirement model for each spatial unit, so as to realize the spatial differentiation digital modeling of the garden area. The water demand prediction module is used to predict the water demand of each spatial unit within a preset time window based on historical spraying data, vegetation water demand model and current integrated environmental status, and generate the target spraying water volume for each spatial unit. The spraying compensation and decision module is used to combine real-time wind speed, wind direction, temperature and light intensity to compensate for wind field drift and evaporation loss of the target spraying water volume, and generate the spraying control quantity corresponding to each spatial unit based on the compensation result. The spraying execution module is used to dynamically adjust the spraying flow rate, spraying angle and spraying duration of each nozzle according to the spraying control quantity, so as to achieve differentiated and precise spraying control for different spatial units. The feedback evaluation and self-learning module is used to collect data on changes in soil moisture and vegetation growth status after spraying, evaluate the spraying effect and water use efficiency, and adaptively update the water demand prediction model and spraying decision parameters based on the evaluation results, forming a closed-loop optimization control mechanism of prediction-execution-evaluation-self-learning.

[0013] Compared with existing technologies, the beneficial effects of this invention are: By constructing a multi-source environment and equipment status perception mechanism, integrating multi-dimensional data such as temperature, humidity, light, wind speed and direction, soil moisture, vegetation growth status and spraying equipment operation status, a highly reliable environmental status vector is formed, which significantly improves the characterization ability of the garden environment and avoids the adverse effects of single sensor distortion on spraying control. By dividing the garden area into multiple spatial units and establishing environmental state models and vegetation water requirement models for each unit, spatial differentiation modeling of the garden area is achieved, effectively solving the problem of uneven spraying caused by differences in different vegetation types, soil conditions and shading conditions, and significantly improving the precision of spraying control. By using a water demand prediction mechanism based on historical data and current environmental conditions, dynamic water demand prediction for each spatial unit within a preset time window can be achieved, transforming spraying decisions from being driven by experience rules to being data-driven, and improving the adaptability of spraying strategies to seasonal changes and meteorological fluctuations. By introducing a wind field drift and evaporation loss compensation model, the target spraying water volume is corrected in real time. Even under adverse conditions such as strong winds, high temperatures, or strong sunlight, the effective infiltration water volume can still be guaranteed, significantly reducing water waste and improving water use efficiency. By incorporating changes in soil moisture and vegetation growth status after spraying into the control closed loop through a spraying effect feedback evaluation and self-learning optimization mechanism, the water demand prediction model and spraying decision parameters are continuously updated to achieve adaptive evolution of the spraying control strategy and construct an intelligent closed-loop control system of prediction-execution-evaluation-self-learning. This invention achieves dynamic and precise control of spraying water volume in different spatial units through multi-source environmental and equipment status fusion perception, spatial differentiation modeling of garden areas, intelligent prediction of water demand, and compensation for wind field and evaporation loss. It also forms a closed-loop control system by combining spraying effect feedback and self-learning optimization, transforming spraying decisions from static experience rules to data-driven adaptive strategies. Even in complex and ever-changing weather and growth environments, it can maintain stable and precise irrigation effects, significantly improving water resource utilization efficiency and the level of intelligent garden maintenance. Attached Figure Description

[0014] Figure 1 This is a flowchart of an intelligent and precise spraying control method for smart gardens proposed in this invention; Figure 2 This is a schematic diagram illustrating the division of a garden area into multiple spatial units in an intelligent and precise spraying control method for smart gardens proposed in this invention. Figure 3 This is a schematic diagram illustrating the principle of wind field and evaporation loss compensation in an intelligent and precise spraying control method for smart gardens proposed in this invention. Figure 4 This is a block diagram of an intelligent precision spraying control system for smart gardens proposed in this invention. Detailed Implementation

[0015] The present invention will be further explained below with reference to specific embodiments.

[0016] Example 1 Reference Figure 1-3 This embodiment proposes an intelligent and precise spraying control method for smart gardens, including the following steps: S1: Multi-source environment and equipment status perception: Collect garden environment parameters, vegetation status parameters and spraying equipment operation parameters, and construct an environmental status vector; The specific logical steps are as follows: S101: In various spatial units of the garden Multiple types of sensor nodes are deployed internally to periodically collect soil parameters, meteorological parameters, vegetation parameters, and equipment status. Soil parameters include soil volumetric moisture content. Soil temperature Soil electrical conductivity EC, meteorological parameters including air temperature Air humidity (H) and wind speed ,wind direction Light intensity Vegetation parameters include Normalized Difference Vegetation Index (NDVI), Leaf Area Index (LAI), and Canopy Temperature. Equipment status includes nozzle flow rate. Valve opening and spray pressure ; S102: Construct the original observation vector based on the parameters collected in S101: ; in This represents the original observation vector, with the superscript T indicating the transpose operation, used to transform the vector into its original form. Transform row vectors into column vectors; S103: Use sliding window mid-value filtering to remove outliers: ; The data for each dimension are then normalized. ; in This represents the result of median filtering on the k-th dimension observation data at time t. This represents the original observation data in the k-th dimension. This indicates the current time, and L represents the length of the sliding window. ( () represents the mean value function. This indicates that the k-th dimension observation data is at "time". At the time "The set of all observations within this time window, This represents the result of normalizing the k-th dimension observation data at time t. This represents the historical minimum value of the k-th dimension observation data. This represents the historical maximum value of the Kth dimension observation data, where K represents the dimension index of the observation data; Among them, the sliding window midpoint filter is used to suppress the impact of sudden abnormal data on spraying decisions and avoid the spraying strategy from deviating drastically due to the short-term failure of a single sensor; S104: Based on the fusion environment state of the previous moment The predicted state is obtained by predicting the current environmental state: Then the predicted state is mapped to the predicted observation. and compared with the current measured values Calculate the residuals: ; in This indicates that at time t, only the previous time period is used. Based on the observation data at and before time t, the predicted estimate of the system state X(t) is given. Indicates the previous moment, ( ) represents the state transition function. This represents the observation residual at time t. ( ) represents the observation mapping function, This represents the actual observation vector at time t; S105: Construct an adaptive weight matrix based on the historical stability and real-time fluctuation of each sensor. Finally, the predicted state is corrected to obtain the current fused environment state vector: ; in This represents the final state estimate at time t obtained after incorporating the observation residuals. The gain matrix represents time t; S106: The final environmental state vector of the garden space unit is obtained: ; in This represents the environmental state vector of a garden space unit. This represents the state estimation vector of a spatial element at time t. These represent the estimated values ​​for soil volumetric moisture content, soil temperature, air temperature, air humidity, wind speed, light intensity, normalized vegetation index, leaf area index, canopy temperature, sprinkler flow rate, valve opening, and spray pressure, respectively. S2: Spatial Differentiation Modeling: Divide the garden area into multiple spatial units and establish an independent environmental state model and vegetation water requirement model for each spatial unit; The specific logical steps are as follows: S201: Divide the entire garden area into several spatial units according to geographical coordinates, vegetation distribution, and sprinkler network structure: ; in This represents the garden space grid cell in row i and column j. This refers to the collection of areas within the entire garden. This indicates the total number of grids that divide the garden area along the "row" direction. This represents the total number of grid cells in the garden area along the "column" direction, i This represents the spatial cell coordinate index, where i represents the row and j represents the column; S202: Merge the environment state vector obtained in S105 Mapped to the corresponding spatial unit: ; in , This represents the environmental state vector of the spatial grid cell in the i-th row and j-th column at time t. This represents the state estimation vector of a spatial element at time t. Let represent the soil volumetric water content, soil temperature, air humidity, wind speed, light intensity, normalized vegetation index, leaf area index, and canopy temperature of grid (i,j), respectively. S203: Based on vegetation type, soil type, and functional zone attributes, assign weighting coefficients to each spatial unit: ,in This represents the overall weighting coefficient. These represent the weighting coefficients for vegetation type, soil type, and functional zone attributes, respectively. The vegetation type weight coefficient, soil type weight coefficient, and functional zone attribute weight coefficient are obtained through historical irrigation data statistics or manual calibration, and their specific value ranges are not considered as limiting conditions of this invention. S204: Constructing the basic water demand model for each spatial unit: The reference evaporation rate is as follows: ,in represents the reference evapotranspiration, Indicates the basic water demand. , and All represent empirical fit coefficients. Indicates light intensity. Indicates the soil reference temperature; S205: Adjust water requirement based on current soil moisture content and vegetation status: ; in This indicates the revised water requirement. Weighting coefficients representing vegetation growth The weighting coefficient representing soil moisture content. The optimal normalized vegetation index represents the vegetation. This indicates the optimum volumetric moisture content of the soil. S3: Water demand prediction: Based on historical data and current environmental conditions, the water demand of each spatial unit within a preset time window is predicted to obtain the target spraying water volume; The specific logical steps are as follows: S301: For each spatial unit Obtain historical environment state sequence: and the environmental state vector of garden space unit ; in This indicates that the grid (i,j) is at time "time". arrive "The historical state sequence during this period, This indicates that the grid (i,j) is in the " , until "State vectors at each time step, k is the length of the historical time window, t represents the current time step, i..." This represents the spatial cell coordinate index, where i represents the row and j represents the column; S302: Establish a predictive model for each spatial unit based on historical data and the current environmental state: ,in Indicates a prediction of the future time window Water demand within the area Represents the prediction function. These parameters represent vegetation type and soil type, respectively, reflecting the water requirements of plants and the water-holding capacity of soil. This indicates that the grid (i,j) is at time "time". arrive "The historical state sequence during this period;" Where the prediction function This can be achieved using linear regression, neural networks, or other machine learning models; S303: Apply a weighted correction to the forecast results based on historical trends: ; in This indicates that the grid (i,j) is in the " "Water demand after real-time fusion correction" This indicates that the fusion coefficient ranges from 0 to 1, and is used to control the weighting of historical data and model predictions. This indicates that the grid (i,j) is in the " "Historical experience at every moment requires water." Indicates time interval, This indicates that the grid (i,j) is in the " "Current calculated water demand at any given moment;" S304: Use the revised water demand as the target spray water volume for the space unit. ,in Indicates the target spray volume; S4: Wind field and evaporation loss compensation: Based on real-time wind speed, wind direction, temperature and light intensity, the predicted spraying water volume is effectively corrected for infiltration, and a compensated spraying control volume is generated. The specific logical steps are as follows: S401: For each spatial unit Obtain the predicted target spray volume and real-time environmental parameters, including wind speed. ,wind direction air temperature (t), light intensity (t) and air humidity (t); S402: Correct spray deviation based on wind speed and direction, defining wind field correction coefficients. : ; in Indicates the wind sensitivity coefficient. Indicates the direction of the spray from the nozzle. ( ) represents an exponential function. Represents the cosine function; It should also be noted that the exponential function is used to describe the nonlinear decay relationship between wind speed and spray drift. S403: Calculate the evaporation loss correction factor based on ambient temperature, light intensity, and air humidity. : ; in Indicates the evaporation sensitivity coefficient. Indicates the air reference temperature. The maximum light intensity normalization is represented by the wind field correction coefficient and the evaporation correction coefficient, which are used to describe the combined effects of wind speed, temperature and light on the effective infiltration rate of sprayed water. Their functional forms can be adjusted according to engineering needs and are not intended to limit the scope of protection of this invention. S404: Applying wind field and evaporation correction factors to predict spray volume yields the compensated spray volume for each spatial unit. ; in Indicates the amount of compensation sprayed; S405: Output To inform the next steps in dynamic spraying decisions and execution; S5: Dynamic spraying decision and execution: Based on the compensated spraying control volume, dynamically adjust the nozzle flow rate, spraying angle and spraying duration to achieve spatially differentiated precision spraying; The specific logical steps are as follows: S501: For each spatial unit Obtain the compensated spray water volume And the parameters of the spraying equipment, including the maximum flow rate of the nozzles. Spraying angle range and spray pressure range ; S502: Based on the target spray volume and spray duration Calculate the nozzle flow rate: ; in For the area of ​​a spatial unit, Indicates the nozzle flow rate, if Then the restriction is And adjust the spraying time accordingly; S503: Adjust the spraying angle based on vegetation distribution and wind direction within the spatial unit. : ; in , This represents the wind direction correction factor. Indicates the reference direction for spraying. Indicates the spray reference angle of the nozzle. This indicates the amount of spray angle correction caused by wind direction. Indicates wind speed. Indicates wind direction, t represents the current time, and i This represents the spatial cell coordinate index, where i represents the row and j represents the column; S504: Based on the calculated nozzle flow rate and target spray volume The spraying time is recalculated to ensure that the sprayed water volume accurately matches the compensated target volume. The formula used is as follows: ; in Indicates the spraying duration; S505: Adjust nozzle flow rate Spraying angle and spraying duration Send the message to the spray controller to complete the dynamic spraying execution; S6: Spraying effect feedback and evaluation: Collect data on changes in soil moisture and vegetation growth status after spraying to evaluate the spraying effect and water use efficiency. The specific logical steps are as follows: S601: Preset evaluation time after spraying is completed Collect data from each spatial unit. The feedback data includes soil moisture content before spraying. Soil moisture content after spraying Vegetation growth indicators and the actual amount of water sprayed ; S602: Calculate the increase in soil moisture resulting from spraying: ; in This represents the increase in soil moisture in spatial unit i. This indicates that spatial unit i is at the moment of spraying effect evaluation. Soil volumetric water content, This indicates that spatial unit i is at the moment of spraying start. The soil volumetric water content, where i represents the index of the spatial cell. Indicates the moment for evaluating the spraying effect; S603: Convert soil moisture increase into effective infiltration volume: ; in This represents the effective depth of the soil layer in spatial unit i. Represents the area of ​​a spatial unit. Indicates the effective infiltration volume; S604: Calculate water use efficiency: ; in Indicates water use efficiency; S605: Constructing a spraying effect scoring function: ; in This indicates the actual effect score. , This represents a weighting coefficient used to balance water-saving efficiency and vegetation growth. Indicates that spatial unit i is at the evaluation time Vegetation growth indicators This indicates that spatial unit i is at the moment of spraying start. Vegetation growth indicators; S606: Rate the actual effect With the target effect Comparison: ,in Indicates error; And based on error The parameters of the water demand model are adaptively updated to provide feedback for the next cycle S2-S5, thereby achieving closed-loop optimization control. S7: Self-learning optimization: Based on the spraying effect evaluation results, the water demand prediction model and spraying decision parameters are updated to form a self-learning closed loop; The specific logical steps are as follows: S701: Spraying effect error obtained from S606 In each spatial unit Constructing learning samples: ,in This represents the environmental feature vector of the unit. Representing spatial units The learning sample set This indicates the actual amount of water sprayed. Indicates the actual effect score; S702: With error As a loss function, the parameters of the water demand prediction model Perform gradient updates: ,in For learning rate, , Represents the loss function. express Model parameters after the next iteration This represents the model parameters at the t-th iteration. Represents the loss function For parameters The gradient; In addition, gradient update methods can be gradient descent, approximate gradient, or heuristic update; S703: Based on water use efficiency Dynamically adjust spray control parameters: ; in , , This represents the adjustment coefficient, F represents the flow rate, and T represents the duration. Indicates angle, Indicates the first Spray flow control parameters for the wheel, Indicates the first Spray flow control parameters for the wheel, Indicates the first Spraying duration control parameters for the wheel, Indicates the first Spraying duration control parameters for the wheel, Indicates the first Spray angle control parameters of the wheel, Indicates the first The spray angle control parameter of the wheel, where t represents the iteration cycle; S704: When multiple consecutive cycles satisfy: If the historical spraying strategy of the spatial unit is marked as a failed strategy, its sample weight is reduced to avoid amplifying erroneous experience. Indicates the error threshold. This represents the absolute value of the spraying effect error of spatial unit i; S705: Write the updated water demand model parameters and spraying decision parameters back to the S3-S5 process, so that the water demand forecast and dynamic spraying decision in subsequent cycles can be continuously optimized based on historical effect feedback, and a closed-loop control mechanism of "prediction-execution-evaluation-relearning" can be constructed. Through the aforementioned self-learning closed-loop mechanism, this embodiment can continuously correct the water demand prediction model and spraying control parameters during long-term operation, so that the spraying strategy gradually tends to the optimal level, effectively overcoming the problem of water waste caused by the long-term reliance on fixed rules in traditional garden spraying systems.

[0017] Example 2 Based on the same inventive concept as the intelligent precision spraying control method for smart gardens described in the foregoing embodiments, such as Figure 4 As shown, this embodiment proposes an intelligent precision spraying control system for smart gardens, including a multi-source sensing module, a spatial modeling module, a water demand prediction module, a spraying compensation and decision-making module, a spraying execution module, and a feedback evaluation and self-learning module. The multi-source sensing module is used to collect garden environmental parameters, vegetation growth status parameters, and spraying equipment operation status parameters, and performs time synchronization and noise suppression processing on multi-source heterogeneous data to construct an environmental status observation vector. The spatial modeling module is used to divide the garden area into multiple spatial units and to establish an independent environmental state model and vegetation water requirement model for each spatial unit, so as to realize the spatial differentiation digital modeling of the garden area. The water demand prediction module is used to predict the water demand of each spatial unit within a preset time window based on historical spraying data, vegetation water demand model and current integrated environmental status, and generate the target spraying water volume for each spatial unit. The spraying compensation and decision module is used to combine real-time wind speed, wind direction, temperature and light intensity to compensate for wind field drift and evaporation loss of the target spraying water volume, and generate the spraying control quantity corresponding to each spatial unit based on the compensation result. The spray execution module is used to dynamically adjust the spray flow rate, spray angle and spray duration of each nozzle according to the spray control quantity, so as to achieve differentiated and precise spray control for different spatial units; The feedback evaluation and self-learning module is used to collect data on changes in soil moisture and vegetation growth status after spraying, evaluate the spraying effect and water use efficiency, and adaptively update the water demand prediction model and spraying decision parameters based on the evaluation results, forming a closed-loop optimization control mechanism of prediction-execution-evaluation-self-learning. The intelligent precision spraying control system for smart gardens provided in this embodiment can execute the intelligent precision spraying control method for smart gardens provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method.

[0018] Example 3 Based on the intelligent precision spraying control method for smart gardens in Embodiment 1 and the intelligent precision spraying control system for smart gardens in Embodiment 2, taking a lawn area in a city park as an example: The garden area, which is 40m×40m in size, is divided into 8×8 spatial units. Each unit is equipped with sensors for soil moisture, air temperature and humidity, wind speed and direction, and light intensity. During a particular spraying cycle, the system collected data from spatial units. The soil volumetric moisture content at (3,5) is 0.18, the wind speed is 3.1 m / s, and the light intensity is 720 W / m². 2The target spraying water volume was calculated to be 15.2L by the S3 water demand prediction module, and then corrected to 18.6L after S4 wind evaporation compensation. After the spraying was completed, the soil moisture content was measured to be 0.27 by the S6 feedback evaluation module, and the water use efficiency reached 0.83, which is about 27% higher than the original timed spraying method. Subsequently, the system self-learned and updated the parameters of the water demand prediction model based on the spraying effect error.

[0019] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for intelligent and precise spraying control in smart gardens, characterized in that, Includes the following steps: S1: Multi-source environment and equipment status perception: Collect garden environment parameters, vegetation status parameters and spraying equipment operation parameters, and construct an environmental status vector; S2: Spatial Differentiation Modeling: Divide the garden area into multiple spatial units and establish an independent environmental state model and vegetation water requirement model for each spatial unit; S3: Water demand prediction: Based on historical data and current environmental conditions, the water demand of each spatial unit within a preset time window is predicted to obtain the target spraying water volume; S4: Wind field and evaporation loss compensation: Based on real-time wind speed, wind direction, temperature and light intensity, the predicted spraying water volume is effectively corrected for infiltration, and a compensated spraying control volume is generated. S5: Dynamic spraying decision and execution: Based on the compensated spraying control volume, dynamically adjust the nozzle flow rate, spraying angle and spraying duration to achieve spatially differentiated precision spraying; S6: Spraying effect feedback and evaluation: Collect data on changes in soil moisture and vegetation growth status after spraying to evaluate the spraying effect and water use efficiency. S7: Self-learning optimization: Based on the spraying effect evaluation results, the water demand prediction model and spraying decision parameters are updated to form a self-learning closed loop.

2. The intelligent and precise spraying control method for smart gardens according to claim 1, characterized in that, The specific logical steps of S1 are as follows: S101: In various spatial units of the garden Multiple types of sensor nodes are deployed internally to periodically collect soil parameters, meteorological parameters, vegetation parameters, and equipment status. Soil parameters include soil volumetric moisture content. Soil temperature Soil electrical conductivity EC, meteorological parameters including air temperature Air humidity (H) and wind speed ,wind direction Light intensity Vegetation parameters include Normalized Difference Vegetation Index (NDVI), Leaf Area Index (LAI), and Canopy Temperature. Equipment status includes nozzle flow rate. Valve opening and spraying pressure ; S102: Construct the original observation vector based on the parameters collected in S101: ； in This represents the original observation vector, with the superscript T indicating the transpose operation, used to transform the vector into its original form. Transform row vectors into column vectors; S103: Use sliding window mid-value filtering to remove outliers: ; The data for each dimension are then normalized. ; in This represents the result of median filtering on the k-th dimension observation data at time t. This represents the original observation data in the k-th dimension. This indicates the current time, and L represents the length of the sliding window. ( () represents the mean value function. This indicates that the k-th dimension observation data is at time "time". At the time "The set of all observations within this time window, This represents the result of normalizing the k-th dimension observation data at time t. This represents the historical minimum value of the k-th dimension observation data. This represents the historical maximum value of the Kth dimension observation data, where K represents the dimension index of the observation data; S104: Based on the fusion environment state of the previous moment The predicted state is obtained by predicting the current environmental state: Then the predicted state is mapped to the predicted observation. and compared with the current measured values Calculate the residuals: ; in This indicates that at time t, only the previous time period is used. Based on the observation data at and before time t, the predicted estimate of the system state X(t) is given. Indicates the previous moment, ( ) represents the state transition function. This represents the observation residual at time t. ( ) represents the observation mapping function, This represents the actual observation vector at time t; S105: Construct an adaptive weight matrix based on the historical stability and real-time fluctuation of each sensor. Finally, the predicted state is corrected to obtain the current fused environment state vector: ； in This represents the final state estimate at time t obtained after incorporating the observation residuals. The gain matrix at time t; S106: The final environmental state vector of the garden space unit is obtained: ； in This represents the environmental state vector of a garden space unit. This represents the state estimation vector of a spatial element at time t. These represent the estimated values ​​for soil volumetric moisture content, soil temperature, air temperature, air humidity, wind speed, light intensity, normalized difference vegetation index, leaf area index, canopy temperature, sprinkler flow rate, valve opening, and spray pressure, respectively.

3. The intelligent and precise spraying control method for smart gardens according to claim 2, characterized in that, The specific logical steps of S2 are as follows: S201: Divide the entire garden area into several spatial units according to geographical coordinates, vegetation distribution, and sprinkler network structure: ; in This represents the garden space grid cell in row i and column j. This refers to the collection of areas within the entire garden. This indicates the total number of grids that divide the garden area along the "row" direction. This represents the total number of grid cells in the garden area along the "column" direction, i This represents the spatial cell coordinate index, where i represents the row and j represents the column; S202: Merge the environment state vector obtained in S105 Mapped to the corresponding spatial unit: ; in , This represents the environmental state vector of the spatial grid cell in the i-th row and j-th column at time t. This represents the state estimation vector of a spatial element at time t. Let represent the soil volumetric water content, soil temperature, air humidity, wind speed, light intensity, normalized vegetation index, leaf area index, and canopy temperature of grid (i,j), respectively. S203: Based on vegetation type, soil type, and functional zone attributes, assign weighting coefficients to each spatial unit: ,in This represents the overall weighting coefficient. These represent the weighting coefficients for vegetation type, soil type, and functional zone attributes, respectively. S204: Constructing the basic water demand model for each spatial unit: The reference evaporation rate is as follows: ,in represents the reference evapotranspiration, Indicates the basic water demand. , and All represent empirical fit coefficients. Indicates light intensity. Indicates the soil reference temperature; S205: Adjust water requirement based on current soil moisture content and vegetation status: ； in This indicates the revised water requirement. Weighting coefficients representing vegetation growth The weighting coefficient representing soil moisture content. The optimal normalized vegetation index represents the vegetation. This indicates the optimum volumetric moisture content of the soil.

4. The intelligent and precise spraying control method for smart gardens according to claim 3, characterized in that, The specific logical steps of S3 are as follows: S301: For each spatial unit Obtain historical environment state sequence: and the environmental state vector of garden space unit ; in This indicates that the grid (i,j) is at time " arrive "The historical state sequence during this period, This indicates that the grid (i,j) is in " , until "State vectors at each time step, k is the length of the historical time window, t represents the current time step, i..." This represents the spatial cell coordinate index, where i represents the row and j represents the column; S302: Establish a predictive model for each spatial unit based on historical data and the current environmental state: ,in Indicates a prediction of the future time window Water demand within the area Represents the prediction function. These parameters represent vegetation type and soil type, respectively, reflecting the water requirements of plants and the water-holding capacity of soil. This indicates that the grid (i,j) is at time " arrive "The historical state sequence during this period;" S303: Apply a weighted correction to the forecast results based on historical trends: ； in This indicates that the grid (i,j) is in " "Water demand after real-time fusion correction" This indicates that the fusion coefficient ranges from 0 to 1, and is used to control the weighting of historical data and model predictions. This indicates that the grid (i,j) is in " "Historical experience at every moment requires water." Indicates time interval, This indicates that the grid (i,j) is in " "Current calculated water demand at any given moment;" S304: Use the revised water demand as the target spray water volume for the space unit. ,in This indicates the target spray volume of water.

5. The intelligent precision spraying control method for smart gardens according to claim 4, characterized in that, The specific logical steps of S4 are as follows: S401: For each spatial unit Obtain the predicted target spray volume and real-time environmental parameters, including wind speed. ,wind direction air temperature (t), light intensity (t) and air humidity (t); S402: Correct spray deviation based on wind speed and direction, defining wind field correction coefficients. : ； in Indicates the wind sensitivity coefficient. Indicates the direction of the spray from the nozzle. ( ) represents an exponential function. Represents the cosine function; S403: Calculate the evaporation loss correction factor based on ambient temperature, light intensity, and air humidity. : ； in Indicates the evaporation sensitivity coefficient. Indicates the air reference temperature. This represents the normalized maximum light intensity. S404: Applying wind field and evaporation correction factors to predict spray volume yields the compensated spray volume for each spatial unit. ; in Indicates the amount of compensation sprayed; S405: Output This informs the next steps in dynamic spraying decisions and execution.

6. The intelligent precision spraying control method for smart gardens according to claim 5, characterized in that, The specific logical steps of S5 are as follows: S501: For each spatial unit Obtain the compensated spray water volume And the parameters of the spraying equipment, including the maximum flow rate of the nozzles. Spraying angle range and spray pressure range ; S502: Based on the target spray volume and spray duration Calculate the nozzle flow rate: ； in For the area of ​​a spatial unit, Indicates the nozzle flow rate, if Then the restriction is And adjust the spraying time accordingly; S503: Adjust the spraying angle based on vegetation distribution and wind direction within the spatial unit. : ； in , This represents the wind direction correction factor. Indicates the reference direction for spraying. Indicates the spray reference angle of the nozzle. This indicates the amount of spray angle correction caused by wind direction. Indicates wind speed. Indicates wind direction, t represents the current time, and i This represents the spatial cell coordinate index, where i represents the row and j represents the column; S504: Based on the calculated nozzle flow rate and target spray volume The spraying time is recalculated to ensure that the sprayed water volume accurately matches the compensated target volume. The formula used is as follows: ； in Indicates the spraying duration; S505: Adjust nozzle flow rate Spraying angle and spraying duration Send the message to the spray controller to complete the dynamic spraying execution.

7. The intelligent precision spraying control method for smart gardens according to claim 6, characterized in that, The specific logical steps of S6 are as follows: S601: Preset evaluation time after spraying is completed Collect data from each spatial unit. The feedback data includes soil moisture content before spraying. Soil moisture content after spraying Vegetation growth indicators and the actual amount of water sprayed ; S602: Calculate the increase in soil moisture resulting from spraying: ; in This represents the increase in soil moisture in spatial unit i. This indicates that spatial unit i is at the moment of spraying effect evaluation. Soil volumetric water content, This indicates that spatial unit i is at the moment of spraying start. The soil volumetric water content, where i represents the index of the spatial cell. Indicates the moment for evaluating the spraying effect; S603: Convert soil moisture increase into effective infiltration volume: ; in This represents the effective depth of the soil layer in spatial unit i. Represents the area of ​​a spatial unit. Indicates the effective infiltration volume; S604: Calculate water use efficiency: ; in Indicates water use efficiency; S605: Constructing a spraying effect scoring function: ; in This indicates the actual effect score. , This represents a weighting coefficient used to balance water-saving efficiency and vegetation growth. Indicates that spatial unit i is at the evaluation time Vegetation growth indicators This indicates that spatial unit i is at the moment of spraying start. Vegetation growth indicators; S606: Rate the actual effect With the target effect Comparison: ,in Indicates error; And based on error The parameters of the water demand model are adaptively updated to provide feedback for the next cycle S2-S5, thereby achieving closed-loop optimization control.

8. The intelligent and precise spraying control method for smart gardens according to claim 7, characterized in that, The specific logical steps of S7 are as follows: S701: Spraying effect error obtained from S606 In each spatial unit Constructing learning samples: ,in This represents the environmental feature vector of the unit. Representing spatial units The learning sample set This indicates the actual amount of water sprayed. Indicates the actual effect score; S702: With error As a loss function, the parameters of the water demand prediction model Perform gradient updates: ,in For learning rate, , Represents the loss function. express Model parameters after the next iteration This represents the model parameters at the t-th iteration. Represents the loss function For parameters The gradient; S703: Based on water use efficiency Dynamically adjust spray control parameters: ； in , , This represents the adjustment coefficient, F represents the flow rate, and T represents the duration. Indicates angle, Indicates the first Spray flow control parameters for the wheel, Indicates the first Spray flow control parameters for the wheel, Indicates the first Spraying duration control parameters for the wheel, Indicates the first Spraying duration control parameters for the wheel, Indicates the first Spray angle control parameters of the wheel, Indicates the first The spray angle control parameter of the wheel, where t represents the iteration cycle; S704: When multiple consecutive cycles satisfy: If the historical spraying strategy of the spatial unit is marked as a failed strategy, its sample weight is reduced to avoid amplifying erroneous experience. Indicates the error threshold. This represents the absolute value of the spraying effect error of spatial unit i; S705: The updated water demand model parameters and spraying decision parameters are written back to the S3-S5 process, so that the water demand forecast and dynamic spraying decision in subsequent cycles can be continuously optimized based on historical effect feedback, and a closed-loop control mechanism of "prediction-execution-evaluation-relearning" is constructed.

9. An intelligent precision spraying control system for smart gardens, used to implement the method described in any one of claims 1-8, characterized in that, It includes a multi-source sensing module, a spatial modeling module, a water demand prediction module, a spraying compensation and decision-making module, a spraying execution module, and a feedback evaluation and self-learning module; The multi-source sensing module is used to collect garden environmental parameters, vegetation growth status parameters and spraying equipment operation status parameters, and to perform time synchronization and noise suppression processing on the multi-source heterogeneous data to construct an environmental status observation vector. The spatial modeling module is used to divide the garden area into multiple spatial units and to establish an independent environmental state model and vegetation water requirement model for each spatial unit, so as to realize the spatial differentiation digital modeling of the garden area. The water demand prediction module is used to predict the water demand of each spatial unit within a preset time window based on historical spraying data, vegetation water demand model and current integrated environmental status, and generate the target spraying water volume for each spatial unit. The spraying compensation and decision module is used to combine real-time wind speed, wind direction, temperature and light intensity to compensate for wind field drift and evaporation loss of the target spraying water volume, and generate the spraying control quantity corresponding to each spatial unit based on the compensation result. The spraying execution module is used to dynamically adjust the spraying flow rate, spraying angle and spraying duration of each nozzle according to the spraying control quantity, so as to achieve differentiated and precise spraying control for different spatial units. The feedback evaluation and self-learning module is used to collect data on changes in soil moisture and vegetation growth status after spraying, evaluate the spraying effect and water use efficiency, and adaptively update the water demand prediction model and spraying decision parameters based on the evaluation results, forming a closed-loop optimization control mechanism of prediction-execution-evaluation-self-learning.

Images