Ecological three-dimensional greening intelligent maintenance water-saving system for garden forest seedling culture

Abstract

The invention discloses an ecological three-dimensional greening intelligent maintenance water-saving system for garden forest seedling culture, and belongs to the technical field of garden forest seedling culture. The system comprises an environment sensing module, a data preprocessing module, a growth stage identification module, an intelligent diagnosis module, an intelligent execution module, a user interaction module and a water resource recovery module. The environment sensing module collects environment data of a seedling raising area and forest tree seedling pictures, after data preprocessing, the growth stage recognition module determines the growth stage of forest tree seedlings through image recognition, the intelligent diagnosis module generates a maintenance regulation and control strategy according to the growth stage, and the intelligent execution module controls equipment to execute irrigation and temperature and light adjustment operation. The user interaction module realizes data visualization and strategy parameter adjustment, and the water resource recovery module realizes water saving by collecting rainwater and irrigation water and recycling the rainwater and the irrigation water after treatment. Through multi-module cooperation, intelligent identification and precise maintenance regulation and control of the forest tree seedling growth stage are achieved, and the seedling raising efficiency and the water resource utilization rate are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of garden tree seedling cultivation, and in particular to an ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedling cultivation. Background Art

[0002] In the field of garden tree seedling cultivation, the maintenance of seedlings for ecological three-dimensional greening places high demands on seedling survival rate, growth quality and water resource utilization efficiency. Traditional maintenance methods rely on manual monitoring of environmental parameters such as soil moisture, temperature, light intensity and experience-based operations, and have technical defects such as extensive environmental monitoring, lagging maintenance and regulation, low water resource utilization and insufficient human-computer interaction.

[0003] With the development of intelligent sensing, image recognition, and automated control technologies, there is an urgent need for an integrated system that combines environmental perception, intelligent diagnosis, and water conservation and recycling. This system can address the low accuracy and resource waste in traditional seedling cultivation and maintenance, achieve accurate identification and maintenance control of seedling growth status, and improve seedling efficiency and ecological benefits. To this end, we propose an ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedling cultivation. Summary of the Invention

[0004] The purpose of the present invention is to provide an ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedling cultivation to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: An ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedlings, including an environmental perception module, a data preprocessing module, a growth stage recognition module, an intelligent diagnosis module, an intelligent execution module and a user interaction module; The environmental perception module is used to collect environmental data of the nursery area and pictures of the tree seedlings; The data preprocessing module is used to preprocess the seedling area environment data and the forest seedling pictures, and store the preprocessed seedling area environment data and the forest seedling pictures in the database; The growth stage recognition module is used to perform image recognition processing on the pre-processed forest seedling pictures to obtain growth status indicators, and based on the growth status indicators, obtain the growth stage of the forest seedlings through a growth stage discrimination algorithm; The intelligent diagnosis module is used to construct a maintenance and control model, input the growth stage of the forest seedlings and the pre-processed seedling area environment data into the maintenance and control model for processing to obtain a maintenance and control strategy; The intelligent execution module is used to control the environmental control equipment to perform maintenance operations based on the maintenance control strategy; The user interaction module is used to realize the visualization display of the growth stages of forest seedlings and the pre-processed seedling area environmental data, as well as to realize the adjustment operation of the maintenance control strategy and the weight parameters of the growth measurement weighted formula.

[0006] Preferably, in the environmental perception module, the process of collecting the environmental data of the nursery area and the pictures of the tree seedlings is: Soil moisture sensors and soil temperature sensors were buried in the soil of the nursery area in a 5×5 array layout. The row and column spacing between adjacent sensors was 1 meter, and the burial depth was 15 to 20 centimeters. This enabled grid-based real-time collection of soil moisture and temperature in the nursery area. At the same time, a light intensity sensor was fixed 2 meters above the ground directly above the nursery area to continuously collect light intensity data, thereby obtaining environmental data of the nursery area including soil moisture, soil temperature, and light intensity. By carrying the level Rotation and vertical to The 24-megapixel visible light camera on the electric cloud platform with pitch adjustment function evenly selects 3 to 5 standard sample plants for every 100 square meters of the nursery area, and sets the shooting program to automatically trigger at 9 am and 3 pm every day to capture multi-angle high-definition images of the standard sample plants. The built-in image preprocessing unit of the visible light camera can simultaneously complete the resolution standardization of the captured images to ensure that the acquired forest seedling images are representative and consistent.

[0007] Preferably, in the data preprocessing module, the process of preprocessing the nursery area environmental data and the forest plant pictures is as follows: For the environmental data of the nursery area, the mean standard deviation algorithm based on the sliding window is used to identify outliers. Calculate the outlier threshold, automatically mark and delete the sampling points that exceed the outlier threshold, where: is the mean of the data in the window, is the standard deviation, and the nursery area environmental data processed with data outliers are interpolated and filled using the linear interpolation method using the adjacent valid data points before and after to obtain the pre-processed nursery area environmental data; For the seedling images, salt and pepper noise was first eliminated using a median filter algorithm with a 5×5 window. Then, Gaussian filtering was performed using a 3×3 Gaussian kernel with a standard deviation of 1.5 to remove Gaussian noise. The denoised seedling images were then processed using a contrast-limited adaptive histogram equalization algorithm. The images were divided into 8×8 pixel blocks with a contrast threshold of 40. The dynamic range of pixel values ​​was expanded to [0, 255] using histogram statistics to obtain the preprocessed seedling images.

[0008] Preferably, in the growth stage identification module, the process of obtaining the growth status indicator is: The pre-processed tree seedling images were input into a pre-trained YOLOv8 object detection model. The YOLOv8 object detection model was pre-trained on the COCO plant dataset and fine-tuned on a local tree seedling leaf dataset. The pre-processed tree seedling images were normalized and then features were extracted using the C2f backbone network and PANet feature fusion structure of the YOLOv8 object detection model. DIOU-NMS was used to output leaf target boxes. The actual number of leaves N was obtained by counting the number of leaf target boxes and combining it with an image distortion correction algorithm. The COCO plant dataset is a special dataset built for the plant field based on the COCO dataset format; The DIOU-NMS is a key post-processing module of the YOLOv8 target detection model.

[0009] To calculate the leaf area index (LAI), a U-Net semantic segmentation model with a ResNet34 backbone encoder and a decoder containing a four-layer upsampling module was used. The U-Net semantic segmentation model was jointly trained using the cross-entropy loss function and the Dice loss function. The preprocessed tree seedling image was input and resized to 512×512 using bilinear interpolation. The leaf pixel mask was then output. The actual physical size conversion coefficient obtained by visible light camera calibration was used to count the number of mask pixels and convert it into actual leaf area. The mask was then divided by the land area of ​​the sampling area to obtain the leaf area index (LAI). Based on this, growth status indicators including leaf number and leaf area index were obtained.

[0010] Preferably, in the growth stage identification module, the process of obtaining the growth stage of the forest seedlings by the growth stage discrimination algorithm is as follows: The growth stage discrimination algorithm sets the number of leaves and leaf area index The growth status measurement value is calculated by the growth measurement weighted formula, based on the growth status measurement value , the growth stages of forest seedlings are divided into seedling stage, growth stage, lignification stage or maturity stage; like , then the growth stage of forest seedlings is the seedling stage; like , then the growth stage of forest seedlings is the growing period; like , then the growth stage of forest seedlings is the lignification stage; like , then the growth stage of forest seedlings is the maturity stage; The growth measurement weighted formula is: ; in, is the number of leaves, is the leaf area index, is the leaf quantity weight, is the foliage index weight.

[0011] Preferably, the growth stage step thresholds of coniferous tree species, broadleaved tree species or shrub species are determined by a K-means clustering algorithm: Construct a multi-species growth characteristic dataset and collect the number of leaves of different tree species throughout their growth cycle and leaf area index Data, including at least 200 groups of coniferous tree samples, at least 250 groups of broad-leaved tree samples, and at least 180 groups of shrub samples, with each group of data corresponding to manually annotated growth stages including seedling stage, growing stage, lignification stage, and maturity stage; K-means clustering was performed on coniferous tree samples, broad-leaved tree samples, and shrub samples, and the optimal cluster number K for each tree species was independently determined by the elbow rule. After clustering the three types of tree species samples, the centers of each cluster center of the three types of tree species samples were obtained and restored to the actual growth state measurement value through denormalization processing. The distribution interval of the conifer sample is set as the center of the smallest cluster center among the cluster centers of the conifer sample. , the midpoint between the second smallest and third cluster centers is set to , the midpoint between the third and largest cluster centers is set to , the maximum cluster center is set to Based on this, the growth stage step thresholds of leafy tree species are obtained, and broad-leaved tree samples and shrub samples are processed according to the same logic.

[0012] Preferably, in the intelligent diagnosis module, the process of inputting the growth stage of the forest seedlings and the pre-processed seedling area environment data into the maintenance and control model for processing is: Pre-collect environmental data for the entire life cycle of three tree species: conifers, broadleaf trees, and shrubs, during the seedling, growth, lignification, and maturity stages. This data set, including soil moisture, soil temperature, and light intensity, is constructed and stored in a database. At the same time, database query statements are invoked to match the growth stages of the forest seedlings to the corresponding reference data set for the nursery area environment in real time. The pre-processed nursery area environmental data is compared item by item with the nursery area environmental reference data set corresponding to the real-time matching of the forest seedling growth stage, and the environmental parameter deviation value is calculated using the environmental parameter deviation value formula. The environmental parameter deviation value includes soil moisture deviation, soil temperature deviation and light intensity deviation; The environmental parameter deviation value formula is: ; ; ; in, 、 and They are soil moisture deviation, soil temperature deviation and light intensity deviation, 、 and is the pre-processed environmental data of the nursery area. 、 and are the reference values ​​of soil moisture, soil temperature and light intensity in the nursery area environmental reference dataset; The target irrigation amount, target temperature and target illumination are generated based on the environmental parameter deviation values. The soil moisture deviation, soil temperature deviation and light intensity deviation have positive and negative differences. If the soil moisture deviation, soil temperature deviation or light intensity deviation is positive, it indicates that the corresponding reference value has been reached or exceeded. At this time, the corresponding target irrigation amount, target temperature or target illumination is zero. If the soil moisture deviation, soil temperature deviation or light intensity deviation is negative, it indicates that the corresponding reference value has not been reached. At this time, the soil moisture deviation is the target irrigation amount, the soil temperature deviation is the target temperature, and the light intensity deviation is the light intensity deviation. Output the maintenance control strategy including target irrigation amount, target temperature and target illumination to the intelligent execution module.

[0013] Preferably, in the intelligent execution module, based on the maintenance control strategy, the process of controlling the environmental control equipment to perform the maintenance operation is as follows: Extracting target irrigation volume in maintenance and regulation strategies , target temperature and target illuminance ; The target irrigation amount The irrigation duration is calculated using the flow-time formula, and a solenoid valve opening instruction is generated based on the irrigation duration to control the opening of the solenoid valve of the irrigation equipment for water replenishment. The flow time formula is: ; in, is the irrigation duration, is the irrigation equipment flow rate, is the amount of water corresponding to 1% soil moisture; The target temperature Calculate the power of the temperature control device using the temperature control power formula, and control the heating device or ventilation device based on the power of the temperature control device to cool down or heat up; The temperature regulation power formula is: ; in, is the power of the temperature control equipment, is the temperature regulation coefficient; The target illuminance The power of the light control device is calculated using the light control power formula, and the fill light intensity is adjusted based on the power of the light control device to provide fill light. The light adjustment power formula is: ; in, It is the power of the lighting control device. is the light compensation coefficient.

[0014] Preferably, in the user interaction module, a process for realizing a visual display of the growth stages of the forest seedlings and the pre-processed seedling area environmental data is implemented: Color-coded progress bars show the growth stages of tree seedlings, and overlay historical growth curves to show growth status measurements Trends over time; A multi-axis line chart is used to synchronously display the real-time monitoring values ​​and reference values ​​of the pre-processed soil moisture, soil temperature and light intensity. The reference values ​​are the reference values ​​of soil moisture, soil temperature and light intensity that match the growth stage of the forest seedlings in the preset nursery area environmental reference data set.

[0015] Preferably, in the user interaction module, the process of adjusting the weight parameters of the maintenance control strategy and the growth measurement weighted formula is implemented: Provides a drag-and-drop rule editor, allowing users to modify the target irrigation amount in the maintenance and control strategy through the drag-and-drop rule editor , target temperature and target illuminance , and call the API to update the maintenance and control strategy; Provides a tree species selection interface, where users can select coniferous, broadleaved, or shrub species based on the tree species of the seedlings in the nursery area. Based on the selected tree species, the weight parameters of the growth measurement weighting formula corresponding to the tree species type are automatically matched and loaded; The weight parameters of the growth measurement weighted formula include the weight of the number of leaves and leaf area index weight , Corresponding coniferous tree species, broad-leaved tree species and shrub species The corresponding values ​​for coniferous trees, broadleaved trees and shrubs are 0.4, 0.5 and 0.6 respectively. They are 0.6, 0.5 and 0.4 respectively.

[0016] Preferably, an ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedlings further includes a water resource recovery module, wherein the water resource recovery module includes a rainwater collection unit and an irrigation water recovery unit; The rainwater collection unit is used to collect rainwater through the HDPE siphon drainage trough laid on the roof of the seedling shed, and after filtering the rainwater through the automatic backwash filter, it is transported to the water reservoir for irrigation; The irrigation water recovery unit is used to collect excess irrigation water through a perforated drainage pipe laid on the ground, and after filtering the excess irrigation water with quartz sand and disinfecting it with ultraviolet light, transport it to a reservoir for use as irrigation water. BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 It is a schematic flow chart of the system function modules of the present invention.

Claims

1. An ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedlings, characterized in that: include: Environmental perception module, used to collect environmental data of the nursery area and pictures of forest seedlings; A data preprocessing module is used to preprocess the seedling area environmental data and forest seedling pictures, and store the preprocessed seedling area environmental data and forest seedling pictures in a database; The growth stage recognition module is used to perform image recognition processing on the pre-processed forest seedling pictures to obtain growth status indicators, and based on the growth status indicators, obtain the growth stage of the forest seedlings through the growth stage discrimination algorithm; The intelligent diagnosis module is used to build a maintenance and control model, inputting the growth stage of forest seedlings and pre-processed seedling area environmental data into the maintenance and control model for processing to obtain a maintenance and control strategy; Intelligent execution module, used to control environmental control equipment to perform maintenance operations based on maintenance control strategies; The user interaction module is used to realize the visualization of the growth stages of forest seedlings and the pre-processed environmental data of the nursery area, as well as to realize the adjustment of the maintenance control strategy and the weight parameters of the growth measurement weighted formula.

2. The ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedling cultivation according to claim 1 is characterized in that: In the environmental perception module, the process of collecting the environmental data of the nursery area and the pictures of the tree seedlings is as follows: Soil moisture sensors and soil temperature sensors are buried in the soil of the nursery area in an array to collect soil moisture and soil temperature. A light intensity sensor is deployed directly above the nursery area to collect light intensity. Based on this, environmental data of the nursery area including soil moisture, soil temperature and light intensity are obtained. Pictures of standard sample tree seedlings in the nursery area were taken using a visible light camera equipped with an electric cloud platform.

3. The ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedling cultivation according to claim 2 is characterized in that: In the data preprocessing module, the process of preprocessing the seedling area environmental data and forest plant pictures: Data outlier processing was performed on the nursery area environmental data. Outliers that exceeded the mean ± 3 times the standard deviation were identified and deleted. At the same time, the nursery area environmental data that had undergone data outlier processing were filled with missing values ​​using linear interpolation to obtain pre-processed nursery area environmental data. The forest seedling pictures were denoised using the median filter algorithm and the Gaussian filter algorithm in turn, and the denoised forest seedling pictures were enhanced using the histogram equalization method to obtain the preprocessed forest seedling pictures.

4. The ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedling cultivation according to claim 3 is characterized in that: In the growth stage identification module, the process of obtaining the growth status indicator is as follows: The pre-processed tree seedling images were subjected to the pre-trained YOLOv8 target detection model and U-Net semantic segmentation model to obtain the leaf number and leaf area index, based on which the growth status indicators including the leaf number and leaf area index were obtained.

5. The ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedling cultivation according to claim 4 is characterized in that: In the growth stage identification module, the process of obtaining the growth stage of the forest seedlings through the growth stage discrimination algorithm is as follows: The growth stage discrimination algorithm uses the number of leaves and leaf area index The growth status measurement value is calculated by the growth measurement weighted formula, based on the growth status measurement value , the growth stages of forest seedlings are divided into seedling stage, growth stage, lignification stage or maturity stage; like , then the growth stage of forest seedlings is the seedling stage; like , then the growth stage of forest seedlings is the growing period; like , then the growth stage of forest seedlings is the lignification stage; like , the growth stage of forest seedlings is the maturity stage.

6. The ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedling cultivation according to claim 5 is characterized in that: described 、 、 and The step thresholds of the growth stages of coniferous tree species, broad-leaved tree species or shrub species are determined by a K-means clustering algorithm.

7. The ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedling cultivation according to claim 6 is characterized in that: In the intelligent diagnosis module, the growth stage of the forest seedlings and the pre-processed seedling area environment data are input into the maintenance and control model for processing, including the following steps: Step 1: matching a preset nursery area environmental reference dataset based on the growth stage of the tree seedlings; Step 2: Compare the pre-processed seedling raising area environmental data with the seedling raising area environmental reference data set item by item to calculate the environmental parameter deviation value; Step 3: Generate target irrigation amount, target temperature and target illumination based on the environmental parameter deviation value; Step 4: Output the maintenance control strategy including target irrigation amount, target temperature and target illumination to the intelligent execution module.

8. The ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedling cultivation according to claim 7 is characterized in that: In the intelligent execution module, based on the maintenance control strategy, the environmental control equipment is controlled to perform the maintenance operation process: Extracting target irrigation volume in maintenance and regulation strategies , target temperature and target illuminance ; The target irrigation amount The irrigation duration is calculated using the flow-time formula, and a solenoid valve opening instruction is generated based on the irrigation duration to control the opening of the solenoid valve of the irrigation equipment for water replenishment. The target temperature Calculate the power of the temperature control device using the temperature control power formula, and control the heating device or ventilation device based on the power of the temperature control device to cool down or heat up; The target illuminance The power of the light control device is calculated using the light control power formula, and the fill light intensity is adjusted based on the power of the light control device to perform fill light.

9. The ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedling cultivation according to claim 8, characterized in that: In the user interaction module, the process of realizing the visualization of the growth stages of forest seedlings and the pre-processed seedling area environmental data, and the adjustment of the maintenance and control strategy and the weight parameters of the growth measurement weighting formula is implemented: Color-coded progress bars show the growth stages of tree seedlings, and overlay historical growth curves to show growth status measurements The changing trend over time is displayed simultaneously using a multi-axis line chart to display the real-time monitoring values ​​and reference values ​​of pre-processed soil moisture, soil temperature and light intensity; Provides a drag-and-drop rule editor, allowing users to modify the target irrigation amount in the maintenance and control strategy through the drag-and-drop rule editor , target temperature and target illuminance , and call the API to update the maintenance and control strategy; Provides a tree species selection interface, where users can select coniferous, broadleaved, or shrub species based on the tree species of the seedlings in the nursery area. Based on the selected tree species, the weight parameters of the growth measurement weighting formula corresponding to the tree species type are automatically matched and loaded; The reference values ​​are the reference values ​​of soil moisture, soil temperature and light intensity that match the growth stage of the forest seedlings in the preset nursery area environmental reference data set; The weight parameters of the growth measurement weighted formula include the weight of the number of leaves and leaf area index weight , Corresponding coniferous tree species, broad-leaved tree species and shrub species The corresponding values ​​for coniferous trees, broadleaved trees and shrubs are 0.4, 0.5 and 0.6 respectively. They are 0.6, 0.5 and 0.4 respectively.

10. The ecological three-dimensional greening intelligent maintenance and water-saving system for garden tree seedling cultivation according to claim 9, characterized in that: It also includes a water resource recovery module, which includes: The rainwater collection unit is used to collect rainwater through the HDPE siphon drainage trough laid on the roof of the seedling shed. After filtering the rainwater through the automatic backwash filter, the rainwater is transported to the reservoir for irrigation water; The irrigation water recovery unit is used to collect excess irrigation water through perforated drainage pipes laid on the ground, and after filtering the excess irrigation water with quartz sand and disinfecting it with ultraviolet light, it is transported to the reservoir for use as irrigation water.

Images