Intelligent service system and method for garden management

Abstract

The invention discloses an intelligent service system and method for garden management. The intelligent service system comprises an Internet of Things sensing module, an edge computing module, a cloud service module and an application service module. And comprehensive intelligentization and refinement of garden management are realized. Through file recording, multi-module collaborative operation and deep analysis and intelligent decision of big data, the efficiency and quality of garden maintenance are improved, and the tourist experience and the garden safety management level are also significantly enhanced. The system can monitor the health state of plants in real time, accurately predict and prevent diseases and pests, optimize the irrigation strategy and guarantee the optimal growth conditions of the plants. Meanwhile, through the digital twinborn platform, a manager can deduce different maintenance schemes in a virtual environment, scientifically evaluate a long-term effect and make an optimal decision.

Description

Technical Field

[0001] The present invention relates to the technical field of garden management, and in particular to an intelligent service system and method for garden management. Background Art

[0002] Traditional garden management relies on manual inspections and empirical judgment, resulting in low efficiency, waste of resources, and delayed decision-making. While existing technologies incorporate the Internet of Things and data analytics, they lack multi-module collaboration capabilities and lack the dynamic simulation and predictive management capabilities for complex scenarios.

[0003] Traditional garden management relies primarily on manual inspections, with managers relying on their own experience to assess various garden conditions. This approach is extremely inefficient. For example, in large gardens, a manual inspection often takes a significant amount of time and is difficult to achieve in a comprehensive and detailed manner. Furthermore, due to the lack of precise resource allocation, resource waste is common, such as excessive pesticide spraying during pest control. Decision-making is also delayed, with issues addressed only after they become apparent. While existing technologies have attempted to integrate the Internet of Things (IoT) to interconnect devices and leverage data analysis to uncover potential insights, in practice, the interoperability between various modules is poor, preventing the formation of an efficient linkage mechanism. Furthermore, dynamic simulation of complex and ever-changing garden scenarios, such as the growth of vegetation under different seasons and weather conditions, is difficult. Furthermore, there is a lack of forward-looking predictive management, making it difficult to formulate effective response strategies in advance. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent service system and method for garden management to solve the problems raised in the above background technology.

[0005] To achieve the above-mentioned object, the present invention provides the following technical solutions: an intelligent service system for garden management, comprising an Internet of Things perception module, an edge computing module, a cloud service module, and an application service module;

[0006] The Internet of Things perception module includes:

[0007] The environmental monitoring submodule is used to detect soil moisture, environmental temperature and humidity, light intensity, and CO2 concentration, as well as water quality and weather conditions, to achieve multi-dimensional data collection.

[0008] The plant health submodule is used to monitor leaf chlorophyll content, signs of pests and diseases, and tree stem deformation to predict drought or root problems;

[0009] The facility operation and maintenance submodule is used to monitor the operating status of equipment and the water level of the reservoir in real time to achieve dynamic management of water resources;

[0010] The edge computing module includes:

[0011] The data preprocessing node performs real-time noise reduction and target detection on camera images, and implements outlier filtering and feature extraction through FPGA acceleration;

[0012] The localized decision engine performs emergency actions based on a rules engine and uses lightweight machine learning models to predict the probability of equipment failure;

[0013] The cloud service module includes:

[0014] The data center is used to build a distributed database to store historical data, support time series data query and spatial analysis, and use blockchain technology to record equipment maintenance logs and material traceability information to ensure that data cannot be tampered with.

[0015] The intelligent decision-making center uses deep learning models to identify pest and disease types, generate prevention and control plans, use reinforcement learning to optimize irrigation strategies, and dynamically adjust water volume based on weather forecasts and plant water demand models;

[0016] The digital twin platform builds a three-dimensional garden model, maps the state of the physical world in real time, simulates the impact of extreme weather on plants, supports the deduction of maintenance plans in virtual scenarios, and evaluates the long-term effects of different decisions;

[0017] The application service module includes:

[0018] The intelligent maintenance unit can automatically generate maintenance work orders, dispatch robots to perform pruning and spraying tasks, and provide health records of ancient and famous trees, recording growth ring data and maintenance history;

[0019] The visitor service unit uses AR guides to display plant information, push flowering forecasts and event notifications, and integrates smart trash cans and charging stations to enhance the visitor experience;

[0020] Emergency management units can combine GIS maps and real-time data to plan fire evacuation routes, coordinate firefighting equipment, predict the risk of flooding caused by heavy rain, activate drainage systems and issue early warning information.

[0021] Preferably, the environmental monitoring submodule is specifically configured to deploy a soil moisture sensor, which is a capacitive or TDR type; deploy a temperature and humidity sensor, which uses a thermocouple or RTD technology to accurately measure ambient temperature and humidity; deploy a light sensor, which is based on the principle of a photodiode and can keenly sense light intensity, helping garden plants to photosynthesize reasonably; deploy a CO2 sensor, which uses NDIR technology to monitor the CO2 concentration in the air in real time, providing data support for plant respiration and carbon cycle; integrate a water quality monitor to measure the key indicators of pH and dissolved oxygen in the water, control the water quality in the garden in real time, ensure the safety of aquatic plants and irrigation water, and be equipped with a professional weather station that can accurately measure meteorological parameters such as wind speed and precipitation;

[0022] Specifically, the plant health submodule uses a multispectral camera covering near-infrared and visible light bands. Through continuous photography and analysis of plant leaves, it can accurately monitor the chlorophyll content of leaves in real time, promptly detect changes in the photosynthesis capacity of plants, and sensitively capture early signs of pests and diseases, thereby buying valuable time for pest and disease control. Pressure sensors are deployed at specific parts of tree stems to accurately monitor the deformation of tree stems. When trees encounter drought stress or root problems, the stems will produce slight deformations. The pressure sensors can sense and transmit data in a timely manner to predict plant health risks and formulate response strategies in advance.

[0023] Preferably, the facility operation and maintenance submodule is specifically to embed RFID tags in various types of equipment, whether it is an irrigation pump, an unmanned lawn mowing robot, or other garden tools, to track their positions in real time through RFID tags, install vibration sensors at key parts of the equipment, and use them to collect and analyze the vibration signals generated when the equipment is running, so as to detect the operating status of the irrigation pump, unmanned lawn mowing robot or other garden tools, and install liquid level sensors on the water tank, which can monitor the water level of the water tank 24 hours a day, and realize dynamic management of water resources according to the garden water demand and real-time water level data, while ensuring the water demand for garden irrigation, etc., to maximize the rational and efficient use of water resources.

[0024] Preferably, the data preprocessing node specifically uses an Nvidia Jetson edge computing box and applies a Gaussian filtering algorithm to perform real-time noise reduction on the camera image. The algorithm formula is:

[0025]

[0026] Where G(x,y) represents the filtered image at the (x,y) coordinate, and σ is the standard deviation, which determines the width of the Gaussian function and controls the strength of the filter. The larger the standard deviation, the blurrier the filtered image and the stronger the noise reduction effect.

[0027] The YOLO target detection algorithm is used to identify pest targets. Its core formula is:

[0028]

[0029] Among them, S 2 is the number of grids into which the image is divided, and B is the number of bounding boxes predicted for each grid; and is an indicator function, indicating whether the j-th bounding box in the i-th grid contains the target or not; are the center coordinates and width and height of the predicted bounding box, (t x , t y , t w , t h ) is the corresponding value of the ground-truth bounding box; λ noobj is a parameter that controls the weight of the no-object bounding box loss, C i is the confidence of the i-th bounding box, and p i (c) are the predicted and true probabilities of category c in the i-th grid, respectively. This loss function comprehensively considers the bounding box position, confidence, and category prediction error to achieve accurate object detection;

[0030] Fusion sensor data, with the help of FPGA acceleration, through the median filter algorithm formula:

[0031]

[0032] Where X is the data sequence, N is the sequence length, and outlier filtering is achieved using the principal component analysis algorithm formula:

[0033] X new =U T (X-μ)

[0034] Among them, U is the eigenvector matrix, μ is the mean vector, and feature extraction is performed.

[0035] Preferably, the localized decision engine specifically relies on a rule engine mechanism, which can quickly respond to and perform emergency operations. When the sensor feedback shows that the soil moisture is lower than the preset threshold, the system will immediately start the irrigation equipment to ensure that the garden plants have sufficient water for growth. In addition, the decision engine also uses a lightweight machine learning model to conduct in-depth analysis of the garden equipment operation data, so as to accurately predict the probability of equipment failure, make maintenance arrangements in advance, and ensure the stable operation of garden management work.

[0036] Preferably, the data center is specifically built on Apache Hadoop and Apache Spark TMIt is a distributed database with a core architecture. The database has powerful storage capabilities and can efficiently store massive amounts of historical garden data, including plant growth cycle records, meteorological and environmental data, etc. It can not only smoothly support complex time-series data queries and quickly trace back the status of the garden at different time points, but also has excellent spatial analysis capabilities, accurately analyzing the ecological connections and resource distribution in different areas of the garden. The Hyperledger Fabric blockchain technology is introduced to create a safe and reliable recording system for equipment maintenance logs and material traceability information. The time, personnel, operation content of each equipment maintenance, as well as the procurement source, transportation route, and use link information of the materials are all stored in encrypted form in the blockchain to ensure that the data cannot be tampered with throughout the life cycle, greatly improving the credibility and traceability of the data;

[0037] Specifically, the intelligent decision-making center uses the advanced YOLOv5 model in the field of deep learning to intelligently identify plant diseases and pests. By learning from a large number of sample images of pests and diseases, it can accurately determine the type of pest and disease and quickly generate targeted, scientific and effective prevention and control plans to protect the health of garden plants. It uses reinforcement learning algorithms to deeply optimize irrigation strategies. Combining weather forecast data with a refined plant water requirement model, it dynamically and accurately adjusts irrigation water volume based on real-time weather changes, soil moisture conditions, and the water requirements of plants at different growth stages, achieving efficient use of water resources and precise regulation of the plant growth environment.

[0038] The digital twin platform specifically uses modeling technology to construct a highly realistic three-dimensional garden model, which can map the real state of the physical world of the garden in real time and accurately, and can present everything from the growth status of plants to the operation status of facilities. It can simulate the impact of various extreme weather conditions, such as heavy rain, drought, and strong winds on garden plants, estimate risks in advance, and support comprehensive deduction of different maintenance plans in virtual scenarios. Through multi-dimensional data analysis, it can intuitively evaluate the comprehensive effects of different decisions on garden ecology and landscape effects in the long term, helping managers make the best decisions.

[0039] Preferably, the intelligent maintenance unit automatically generates maintenance work orders, accurately formulates task contents according to plant growth cycles and pest and disease monitoring results, and then dispatches professional garden robots to perform maintenance tasks such as pruning branches and leaves and precisely spraying pesticides, ensuring that garden plants always maintain the best growth state, providing detailed health records of ancient and famous trees, using non-destructive testing technology to record annual ring data, accurately tracing their growth history, and fully retaining historical information on the time, method, and personnel of each maintenance operation, providing solid data support for the long-term protection of ancient and famous trees;

[0040] Specifically, the visitor service unit utilizes AR navigation technology, powered by the Unity engine, to vividly and intuitively display information about various plants. This not only showcases plant morphology and distribution patterns, but also accurately predicts flowering periods based on big data analysis, ensuring visitors don't miss out on any spectacular blooms. It also provides timely notifications of various park events to enrich the visitor experience. Smart trash cans with built-in overflow detection sensors are integrated into the park. Once a bin is nearly full, an overflow alarm is sent to cleaning staff, ensuring a clean and tidy environment. Charging stations are strategically located to meet the charging needs of visitors' electronic devices, comprehensively enhancing the visitor experience within the park.

[0041] Specifically, the emergency management unit combines the precise positioning function of GIS maps with real-time collected meteorological and environmental data to quickly and accurately plan fire evacuation routes, ensuring that tourists and staff can evacuate safely in the shortest possible time. It also seamlessly links with the fire-fighting equipment in the park, automatically activates fire-fighting devices, and improves fire response efficiency. It uses big data models and meteorological monitoring technology to predict the risk of flooding caused by heavy rain in advance. Once the risk approaches, it automatically activates the efficient drainage system to ensure smooth drainage in the park. It also promptly issues early warning information through the park's broadcasts and electronic display screens to remind tourists and staff to take preventive measures.

[0042] A method for using an intelligent service system for garden management comprises the following steps:

[0043] Step 1: Real-time monitoring and precise identification: Video surveillance equipment and sensors deployed at key locations in the garden capture real-time dynamic information about plants, facilities, and visitors. The video data is then transmitted to the data preprocessing node of the edge computing module. This node uses a Gaussian filter algorithm to reduce noise on the image to improve image quality. It then uses the YOLO target detection algorithm to accurately identify pests and diseases, as well as abnormal behaviors or states, such as signs of equipment failure, to achieve preliminary intelligent analysis and early warning.

[0044] Step 2: Localized Decision-Making and Preventive Maintenance: The localized decision-making engine responds quickly and executes emergency actions based on preset rules, such as automatically activating irrigation equipment to address low soil moisture. Simultaneously, it uses lightweight machine learning models to conduct in-depth analysis of equipment operating data, predict potential failure risks, and provide decision support for preventive maintenance.

[0045] Step 3: Cloud-based data processing and intelligent decision-making: In the cloud service module, the data center builds a distributed database to efficiently store and manage historical garden data, including plant growth records and meteorological and environmental data, supporting complex data queries and spatial analysis. The intelligent decision-making center uses deep learning models and reinforcement learning algorithms to intelligently identify pests and diseases, generate prevention and control plans, and optimize irrigation strategies to achieve precise regulation of water resources. The digital twin platform constructs a three-dimensional garden model, mapping the state of the physical world in real time, supporting maintenance plan deduction and decision-making evaluation in virtual scenarios.

[0046] Step 4. Application service optimization and emergency response: In the application service module, the smart maintenance unit automatically generates maintenance work orders and dispatches robots to perform pruning and spraying tasks. At the same time, it provides health records of ancient and famous trees, records annual ring data and maintenance history; the visitor service unit uses AR guidance technology to display plant information, push flowering forecasts and event notifications, and integrates smart trash cans and charging stations to enhance the visitor experience; the emergency management unit combines GIS maps with real-time data to plan fire evacuation routes, link fire-fighting equipment, predict and respond to the risk of heavy rain and waterlogging, and ensure the safety of the park.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] It has achieved comprehensive intelligence and refinement in garden management. Through archival records, multi-module collaborative operations, and in-depth analysis and intelligent decision-making of big data, it has not only improved the efficiency and quality of garden maintenance, but also significantly enhanced the visitor experience and park safety management level. The system can monitor the health status of plants in real time, accurately predict and prevent pests and diseases, optimize irrigation strategies, and ensure the best growth conditions for plants. At the same time, through the digital twin platform, managers can deduce different maintenance plans in a virtual environment, scientifically evaluate long-term effects, and make the best decisions. In addition, the intelligent upgrade of visitor service units, such as AR guides, smart trash cans and charging piles, allows tourists to enjoy the natural beauty while also feeling convenient and comfortable. In terms of emergency management, the system's rapid response and accurate prediction capabilities provide a solid guarantee for park safety. In summary, the present invention has brought revolutionary changes to garden management and promoted the intelligent development of the garden industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a system principle diagram of the present invention;

[0050] Figure 2 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0051] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0052] See also Figure 1-2 , the present invention provides an intelligent service system for garden management, including an Internet of Things perception module, an edge computing module, a cloud service module and an application service module;

[0053] The IoT perception module includes:

[0054] The environmental monitoring submodule is used to detect soil moisture, environmental temperature and humidity, light intensity and CO2 concentration, and to detect water quality and weather to achieve multi-dimensional data collection. Specifically, the environmental monitoring submodule includes: deploying soil moisture sensors, which are capacitive or TDR types; deploying temperature and humidity sensors, which use thermocouple or RTD technology to accurately measure ambient temperature and humidity; deploying light sensors, which are based on the principle of photodiodes and can keenly sense light intensity to help garden plants carry out photosynthesis reasonably; deploying CO2 sensors, which use NDIR technology to monitor the CO2 concentration in the air in real time, providing data support for plant respiration and carbon cycle; integrating water quality monitors to measure key indicators such as pH value and dissolved oxygen in water, to control the water quality in the garden in real time, to ensure the safety of aquatic plants and irrigation water; and being equipped with professional weather stations that can accurately measure meteorological parameters such as wind speed and precipitation.

[0055] The plant health submodule is used to monitor leaf chlorophyll content, signs of pests and diseases, and tree stem deformation to predict drought or root problems. Specifically, the plant health submodule uses a multispectral camera covering near-infrared and visible light bands. Through continuous photography and analysis of plant leaves, it can accurately monitor leaf chlorophyll content in real time, promptly detect changes in plant photosynthesis capacity, and sensitively capture early signs of pests and diseases, buying valuable time for pest and disease prevention. Pressure sensors are deployed at specific locations on tree stems to accurately monitor stem deformation. When trees encounter drought stress or root problems, the stems will undergo subtle deformation. The pressure sensors can promptly sense and transmit data to predict plant health risks and formulate response strategies in advance.

[0056] The facility operation and maintenance submodule is used to monitor the operating status of equipment and the water level of the reservoir in real time to achieve dynamic management of water resources. Specifically, the facility operation and maintenance submodule embeds RFID tags in various types of equipment, whether it is an irrigation pump, an unmanned lawn mower robot, or other garden tools, and tracks their location in real time through RFID tags. Vibration sensors are installed at key parts of the equipment to collect and analyze the vibration signals generated by the equipment during operation to detect the operating status of the irrigation pump, unmanned lawn mower robot or other garden tools. Liquid level sensors are installed in the reservoir to monitor the water level of the reservoir 24 hours a day. According to the garden water demand and real-time water level data, dynamic management of water resources is achieved, while ensuring the water demand for garden irrigation and other water needs, while maximizing the rational and efficient use of water resources.

[0057] The edge computing module includes:

[0058] The data preprocessing node performs real-time noise reduction and target detection on camera images, and implements outlier filtering and feature extraction through FPGA acceleration. Specifically, the data preprocessing node uses the Nvidia Jetson edge computing box and the Gaussian filtering algorithm to perform real-time noise reduction on camera images. The algorithm formula is:

[0059]

[0060] Where G(x,y) represents the filtered image at the (x,y) coordinate, and σ is the standard deviation, which determines the width of the Gaussian function and controls the strength of the filter. The larger the standard deviation, the blurrier the filtered image and the stronger the noise reduction effect.

[0061] The YOLO target detection algorithm is used to identify pest targets. Its core formula is:

[0062]

[0063] Among them, S 2 is the number of grids into which the image is divided, and B is the number of bounding boxes predicted for each grid; and is an indicator function, indicating whether the j-th bounding box in the i-th grid contains the target or not; are the center coordinates and width and height of the predicted bounding box, (t x , t y , t w , t h ) is the corresponding value of the ground-truth bounding box; λ noobj is a parameter that controls the weight of the no-object bounding box loss, C i is the confidence of the i-th bounding box, and p i(c) are the predicted and true probabilities of category c in the i-th grid, respectively. This loss function comprehensively considers the bounding box position, confidence, and category prediction error to achieve accurate object detection;

[0064] Fusion sensor data, with the help of FPGA acceleration, through the median filter algorithm formula:

[0065]

[0066] Where X is the data sequence, N is the sequence length, and outlier filtering is achieved using the principal component analysis algorithm formula:

[0067] X new =U T (X-μ)

[0068] Among them, U is the eigenvector matrix, μ is the mean vector, and feature extraction is performed;

[0069] The localized decision engine executes emergency actions based on a rules engine and uses lightweight machine learning models to predict the probability of equipment failure. Relying on a rules engine mechanism, the localized decision engine can quickly respond to and execute emergency actions. When sensors indicate that soil moisture is below a preset threshold, the system immediately activates irrigation equipment to ensure sufficient water for garden plants. Furthermore, the decision engine uses lightweight machine learning models to conduct in-depth analysis of garden equipment operating data, accurately predicting the probability of equipment failure and making advance maintenance arrangements to ensure the stable operation of garden management.

[0070] Cloud service modules include:

[0071] The data center is used to build a distributed database to store historical data, support time series data query and spatial analysis, and use blockchain technology to record equipment maintenance logs and material traceability information to ensure that data cannot be tampered with. The data center is specifically built on Apache Hadoop and Apache Spark. TM It is a distributed database with a core architecture. The database has powerful storage capabilities and can efficiently store massive amounts of garden historical data, including plant growth cycle records, meteorological and environmental data, etc. It can not only smoothly support complex time series data queries and quickly trace back the status of the garden at different time points, but also has excellent spatial analysis capabilities, accurately analyzing the ecological connections and resource distribution in different areas of the garden. The Hyperledger Fabric blockchain technology is introduced to create a safe and reliable recording system for equipment maintenance logs and material traceability information. The time, personnel, operation content of each equipment maintenance, as well as the procurement source, transportation route, and use link information of the materials are all stored in encrypted form in the blockchain to ensure that the data cannot be tampered with throughout the life cycle, greatly improving the credibility and traceability of the data;

[0072] The Intelligent Decision Center uses deep learning models to identify pest and disease types, generate prevention and control plans, and optimize irrigation strategies using reinforcement learning. It also dynamically adjusts water volumes based on weather forecasts and plant water requirement models. Specifically, the Intelligent Decision Center leverages the advanced YOLOv5 model in deep learning to intelligently identify pests and diseases in garden plants. By studying a large number of pest and disease sample images, it can accurately determine pest and disease types and quickly generate targeted, scientifically effective prevention and control plans to safeguard the health of garden plants. It uses reinforcement learning algorithms to deeply optimize irrigation strategies, combining weather forecast data with a refined plant water requirement model. Based on real-time weather changes, soil moisture conditions, and the water requirement characteristics of plants at different growth stages, it dynamically and accurately adjusts irrigation water volumes, achieving efficient use of water resources and precise regulation of the plant growth environment.

[0073] The digital twin platform constructs a three-dimensional garden model, mapping the state of the physical world in real time, simulating the impact of extreme weather on plants, supporting the deduction of maintenance plans in virtual scenarios, and evaluating the long-term effects of different decisions. Specifically, the digital twin platform uses modeling technology to build a highly realistic three-dimensional garden model. It can accurately map the true state of the garden's physical world in real time, presenting everything from plant growth to facility operation. It can simulate the impact of various extreme weather conditions, such as heavy rain, drought, and strong winds, on garden plants, estimate risks in advance, and support comprehensive deduction of different maintenance plans in virtual scenarios. Through multi-dimensional data analysis, it can intuitively evaluate the comprehensive effects of different decisions on the garden ecology and landscape effects from a long-term perspective, helping managers make the best decisions.

[0074] Application service modules include:

[0075] The smart maintenance unit can automatically generate maintenance work orders, dispatch robots to perform pruning and spraying tasks, provide health records of ancient and famous trees, and record annual ring data and maintenance history. Specifically, the smart maintenance unit automatically generates maintenance work orders, accurately formulates task content based on plant growth cycles and pest and disease monitoring results, and then dispatches professional garden robots to perform maintenance tasks such as pruning branches and leaves and precisely spraying pesticides to ensure that garden plants always maintain optimal growth conditions. It also provides detailed health records of ancient and famous trees, uses non-destructive testing technology to record annual ring data, accurately traces their growth history, and completely retains historical information on the time, method, and personnel of each maintenance operation, providing solid data support for the long-term protection of ancient and famous trees.

[0076] The visitor service unit uses AR guides to display plant information, push flowering forecasts and event notifications, and integrate smart trash cans and charging piles to enhance the visitor experience. Specifically, the visitor service unit uses AR guide technology and the Unity engine as a carrier to vividly and intuitively display various plant information to tourists. It can not only present the morphological characteristics and habit distribution of plants, but also accurately push flowering forecasts based on big data analysis, so that tourists do not miss every blooming flower scene. At the same time, various event notifications in the park are released in a timely manner to enrich the tourists' experience. Smart trash cans are integrated in the park with built-in overflow detection sensors. Once the trash can is almost full, an overflow alarm signal will be immediately sent to the cleaning staff to ensure that the park environment is always clean and tidy. Charging piles are reasonably arranged to meet the charging needs of tourists' electronic devices, thereby comprehensively improving the tourists' experience in the park.

[0077] The emergency management unit can combine GIS maps and real-time data to plan fire evacuation routes, link fire-fighting equipment, predict the risk of heavy rain and waterlogging, activate the drainage system and issue early warning information; the emergency management unit specifically combines the precise positioning function of the GIS map with the meteorological and environmental data collected in real time to quickly and accurately plan fire evacuation routes, ensuring that tourists and staff can evacuate safely in the shortest time possible, and achieve seamless linkage with the fire-fighting equipment in the park, automatically activate the fire-fighting device, improve the efficiency of fire response, use big data models and meteorological monitoring technology to predict the risk of heavy rain and waterlogging in advance, and automatically start the efficient drainage system once the risk is approaching to ensure smooth drainage of the park, and promptly issue early warning information through the park's broadcasts and electronic display screens to remind tourists and staff to take preventive measures.

[0078] A method for using an intelligent service system for garden management comprises the following steps:

[0079] Step 1: Real-time monitoring and precise identification: Video surveillance equipment and sensors deployed at key locations in the garden capture real-time dynamic information about plants, facilities, and visitors. The video data is then transmitted to the data preprocessing node of the edge computing module. This node uses a Gaussian filter algorithm to reduce noise on the image to improve image quality. It then uses the YOLO target detection algorithm to accurately identify pests and diseases, as well as abnormal behaviors or states, such as signs of equipment failure, to achieve preliminary intelligent analysis and early warning.

[0080] Step 2: Localized Decision-Making and Preventive Maintenance: The localized decision-making engine responds quickly and executes emergency actions based on preset rules, such as automatically activating irrigation equipment to address low soil moisture. Simultaneously, it uses lightweight machine learning models to conduct in-depth analysis of equipment operating data, predict potential failure risks, and provide decision support for preventive maintenance.

[0081] Step 3: Cloud-based data processing and intelligent decision-making: In the cloud service module, the data center builds a distributed database to efficiently store and manage historical garden data, including plant growth records and meteorological and environmental data, supporting complex data queries and spatial analysis. The intelligent decision-making center uses deep learning models and reinforcement learning algorithms to intelligently identify pests and diseases, generate prevention and control plans, and optimize irrigation strategies to achieve precise regulation of water resources. The digital twin platform constructs a three-dimensional garden model, mapping the state of the physical world in real time, supporting maintenance plan deduction and decision-making evaluation in virtual scenarios.

[0082] Step 4. Application service optimization and emergency response: In the application service module, the smart maintenance unit automatically generates maintenance work orders and dispatches robots to perform pruning and spraying tasks. At the same time, it provides health records of ancient and famous trees, records annual ring data and maintenance history; the visitor service unit uses AR guidance technology to display plant information, push flowering forecasts and event notifications, and integrates smart trash cans and charging stations to enhance the visitor experience; the emergency management unit combines GIS maps with real-time data to plan fire evacuation routes, link fire-fighting equipment, predict and respond to the risk of heavy rain and waterlogging, and ensure the safety of the park.

[0083] Example:

[0084] Hardware deployment: Install sensor nodes (50 meters apart) in key areas of the garden and deploy edge computing boxes (one per 1,000 square meters).

[0085] Data collection: The sensor uploads data every 5 minutes, and the camera captures images every hour.

[0086] Model training: Update pest and disease identification models monthly and optimize irrigation strategies based on historical data.

[0087] User interaction: Administrators view the digital twin interface through the web, and tourists use the WeChat mini-program to obtain AR guides.

[0088] A city's landscape management bureau introduced this intelligent service system to improve the quality and efficiency of landscape management. Initially, the system deployed video surveillance equipment to comprehensively cover key landscape areas, capturing real-time activity within the garden. Edge computing modules rapidly processed video data, accurately identifying pests and diseases and abnormal conditions, such as irrigation pump failure warnings, and promptly notifying management personnel for intervention. A localized decision engine automatically responded to low soil moisture alerts, activating irrigation equipment to ensure adequate watering for plants. Simultaneously, a lightweight machine learning model predicted potential irrigation pump failures, allowing maintenance to be scheduled in advance to avoid disrupting normal irrigation operations.

[0089] In the cloud-based service module, the data center efficiently stores historical garden data, supports time-series data query and spatial analysis, and provides managers with detailed plant growth records and meteorological and environmental data. The intelligent decision-making center uses deep learning models to intelligently identify pest and disease types and generate scientifically effective prevention and control plans. It also incorporates reinforcement learning algorithms to optimize irrigation strategies and achieve precise regulation of water resources. The three-dimensional garden model constructed on the digital twin platform reflects the physical world in real time. Managers can simulate different maintenance plans in a virtual setting and evaluate their long-term effects, providing strong support for decision-making.

[0090] At the application service level, the smart maintenance unit automatically generates maintenance work orders and dispatches garden robots to perform tasks such as pruning and spraying, effectively improving maintenance efficiency. The establishment of health records for ancient and famous trees provides solid data support for long-term conservation. The visitor service unit uses AR navigation technology to provide visitors with rich plant information and event notifications. The integration of smart trash cans and charging stations further enhances the visitor experience. The emergency management unit accurately predicts flooding risks during heavy rain seasons, promptly activates drainage systems, and issues early warnings, effectively ensuring park safety.

[0091] After a period of operation, the intelligent service system has significantly improved the work efficiency and management level of the Garden Management Bureau, reduced the occurrence of pests and diseases, optimized water resource utilization, enhanced the visitor experience, and made important contributions to the city's green development.

[0092] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An intelligent service system for garden management, characterized by: Including IoT perception module, edge computing module, cloud service module and application service module; The Internet of Things perception module includes: The environmental monitoring submodule is used to detect soil moisture, environmental temperature and humidity, light intensity, and CO2 concentration, as well as water quality and weather conditions, to achieve multi-dimensional data collection. The plant health submodule is used to monitor leaf chlorophyll content, signs of pests and diseases, and tree stem deformation to predict drought or root problems; The facility operation and maintenance submodule is used to monitor the operating status of equipment and the water level of the reservoir in real time to achieve dynamic management of water resources; The edge computing module includes: The data preprocessing node performs real-time noise reduction and target detection on camera images, and implements outlier filtering and feature extraction through FPGA acceleration; The localized decision engine performs emergency actions based on a rules engine and uses lightweight machine learning models to predict the probability of equipment failure; The cloud service module includes: The data center is used to build a distributed database to store historical data, support time series data query and spatial analysis, and use blockchain technology to record equipment maintenance logs and material traceability information to ensure that data cannot be tampered with. The intelligent decision-making center uses deep learning models to identify pest and disease types, generate prevention and control plans, use reinforcement learning to optimize irrigation strategies, and dynamically adjust water volume based on weather forecasts and plant water demand models; The digital twin platform builds a three-dimensional garden model, maps the state of the physical world in real time, simulates the impact of extreme weather on plants, supports the deduction of maintenance plans in virtual scenarios, and evaluates the long-term effects of different decisions; The application service module includes: The intelligent maintenance unit can automatically generate maintenance work orders, dispatch robots to perform pruning and spraying tasks, and provide health records of ancient and famous trees, recording growth ring data and maintenance history; The visitor service unit uses AR guides to display plant information, push flowering forecasts and event notifications, and integrates smart trash cans and charging stations to enhance the visitor experience; Emergency management units can combine GIS maps and real-time data to plan fire evacuation routes, coordinate firefighting equipment, predict the risk of flooding caused by heavy rain, activate drainage systems and issue early warning information.

2. The intelligent service system for garden management according to claim 1, characterized in that: The environmental monitoring submodule specifically includes: deploying a soil moisture sensor, which is a capacitive or TDR type; deploying a temperature and humidity sensor, which uses thermocouple or RTD technology to accurately measure ambient temperature and humidity; deploying a light sensor, which is based on the principle of photodiode and can keenly sense light intensity to help garden plants to photosynthesize reasonably; deploying a CO2 sensor, which uses NDIR technology to monitor the CO2 concentration in the air in real time, providing data support for plant respiration and carbon cycle; integrating a water quality monitor to measure the key indicators of pH value and dissolved oxygen in water, to control the water quality in the garden in real time, to ensure the safety of aquatic plants and irrigation water; and being equipped with a professional weather station that can accurately measure meteorological parameters such as wind speed and precipitation; Specifically, the plant health submodule uses a multispectral camera covering near-infrared and visible light bands. Through continuous photography and analysis of plant leaves, it can accurately monitor the chlorophyll content of leaves in real time, promptly detect changes in the photosynthesis capacity of plants, and sensitively capture early signs of pests and diseases, thereby buying valuable time for pest and disease control. Pressure sensors are deployed at specific parts of tree stems to accurately monitor the deformation of tree stems. When trees encounter drought stress or root problems, the stems will produce slight deformations. The pressure sensors can sense and transmit data in a timely manner to predict plant health risks and formulate response strategies in advance.

3. The intelligent service system for garden management according to claim 1, characterized in that: Specifically, the facility operation and maintenance submodule embeds RFID tags in various types of equipment, whether it is an irrigation pump, an unmanned lawn mower robot, or other garden tools, and tracks their positions in real time through RFID tags. Vibration sensors are installed at key parts of the equipment to collect and analyze the vibration signals generated by the equipment during operation, so as to detect the operating status of the irrigation pump, unmanned lawn mower robot or other garden tools. Liquid level sensors are installed in the reservoir to monitor the water level of the reservoir 24 hours a day. According to the garden water demand and real-time water level data, dynamic management of water resources is achieved, while ensuring the water demand for garden irrigation and other water needs, the rational and efficient use of water resources is maximized.

4. The intelligent service system for garden management according to claim 1, characterized in that: Specifically, the data preprocessing node uses the Nvidia Jetson edge computing box and applies the Gaussian filtering algorithm to perform real-time noise reduction on the camera image. The algorithm formula is: Where G(x,y) represents the filtered image at the (x,y) coordinate, and σ is the standard deviation, which determines the width of the Gaussian function and controls the strength of the filter. The larger the standard deviation, the blurrier the filtered image and the stronger the noise reduction effect. The YOLO target detection algorithm is used to identify pest targets. Its core formula is: Among them, S 2 is the number of grids into which the image is divided, and B is the number of bounding boxes predicted for each grid; and is an indicator function, indicating whether the j-th bounding box in the i-th grid contains the target or not; are the center coordinates and width and height of the predicted bounding box, (t x , t y , t w , t h ) is the corresponding value of the ground-truth bounding box; λ noobj is a parameter that controls the weight of the no-object bounding box loss, C i is the confidence of the i-th bounding box, and p i (c) are the predicted and true probabilities of category c in the i-th grid, respectively. This loss function comprehensively considers the bounding box position, confidence, and category prediction error to achieve accurate object detection; Fusion sensor data, with the help of FPGA acceleration, through the median filter algorithm formula: Where x is the data sequence and N is the sequence length. To filter outliers, the principal component analysis algorithm formula is used: X new =U T (X-μ) Among them, U is the eigenvector matrix, μ is the mean vector, and feature extraction is performed.

5. The intelligent service system for garden management according to claim 1, characterized in that: The localized decision engine specifically relies on a rule engine mechanism that can quickly respond to and execute emergency operations. When the sensor feedback indicates that the soil moisture is lower than the preset threshold, the system will immediately start the irrigation equipment to ensure that the garden plants have sufficient water for growth. In addition, the decision engine also uses a lightweight machine learning model to conduct in-depth analysis of garden equipment operation data, thereby accurately predicting the probability of equipment failure, making maintenance arrangements in advance, and ensuring the stable operation of garden management work.

6. The intelligent service system for garden management according to claim 1, characterized in that: The data center is specifically built on Apache Hadoop and Apache Spark TM The distributed database with Hyperledger Fabric blockchain technology as its core architecture creates a secure and reliable record system for equipment maintenance logs and material traceability information. The time, personnel, operation content of each equipment maintenance, as well as the procurement source, transportation route, and usage information of the materials are all stored in encrypted form on the blockchain, ensuring that the data cannot be tampered with throughout its life cycle, greatly improving the credibility and traceability of the data. Specifically, the intelligent decision-making center uses the advanced YOLOv5 model in the field of deep learning to intelligently identify plant diseases and pests. By learning from a large number of sample images of pests and diseases, it can accurately determine the type of pest and disease and quickly generate targeted, scientific and effective prevention and control plans to protect the health of garden plants. It uses reinforcement learning algorithms to deeply optimize irrigation strategies. Combining weather forecast data with a refined plant water requirement model, it dynamically and accurately adjusts irrigation water volume based on real-time weather changes, soil moisture conditions, and the water requirements of plants at different growth stages, achieving efficient use of water resources and precise regulation of the plant growth environment. The digital twin platform specifically uses modeling technology to construct a highly realistic three-dimensional garden model, which can map the real state of the physical world of the garden in real time and accurately, and can present everything from the growth status of plants to the operation status of facilities. It can simulate various extreme weather conditions, estimate risks in advance, and support comprehensive deduction of different maintenance plans in virtual scenarios. Through multi-dimensional data analysis, it can intuitively evaluate the comprehensive effects of different decisions on the garden ecology and landscape effects in the long term, helping managers make the best decisions.

7. The intelligent service system for garden management according to claim 1, characterized in that: Specifically, the intelligent maintenance unit automatically generates maintenance work orders, accurately formulates task content based on plant growth cycles and pest and disease monitoring results, and then dispatches professional garden robots to perform maintenance tasks such as pruning branches and leaves and precisely spraying pesticides, ensuring that garden plants always maintain optimal growth conditions. It also provides detailed health records of ancient and famous trees, uses non-destructive testing technology to record annual ring data, accurately traces their growth history, and fully retains historical information on the time, method, and personnel of each maintenance operation, providing solid data support for the long-term protection of ancient and famous trees. Specifically, the visitor service unit uses AR guide technology and the Unity engine as a carrier to vividly and intuitively display various plant information to tourists. Smart trash cans with built-in overflow detection sensors are integrated in the park. Once the trash can is almost full, an overflow alarm signal is immediately sent to the cleaning staff to ensure that the park environment is always clean and tidy. Charging piles are reasonably arranged to meet the charging needs of tourists' electronic devices. Specifically, the emergency management unit combines the precise positioning function of GIS maps with real-time collected meteorological and environmental data to quickly and accurately plan fire evacuation routes, ensuring that tourists and staff can evacuate safely in the shortest possible time. It also seamlessly links with the fire-fighting equipment in the park, automatically activates fire-fighting devices, and improves fire response efficiency. It uses big data models and meteorological monitoring technology to predict the risk of flooding caused by heavy rain in advance. Once the risk approaches, it automatically activates the efficient drainage system to ensure smooth drainage in the park. It also promptly issues early warning information through the park's broadcasts and electronic display screens to remind tourists and staff to take preventive measures.

8. A method for using an intelligent service system for garden management according to any one of claims 1 to 7, characterized in that: The following steps are involved: Step 1: Real-time monitoring and precise identification: Video surveillance equipment and sensors deployed at key locations in the garden capture dynamic information about plants, facilities, and visitors in real time. The video data is then transmitted to the data preprocessing node of the edge computing module. This node uses a Gaussian filter algorithm to reduce noise on the image to improve image quality. It then uses the YOLO object detection algorithm to accurately identify pests and diseases, as well as abnormal behavior or status. Step 2: Localized Decision-Making and Preventive Maintenance: The localized decision-making engine responds quickly and executes emergency actions based on preset rules. Simultaneously, it uses lightweight machine learning models to conduct in-depth analysis of equipment operating data, predict potential failure risks, and provide decision support for preventive maintenance. Step 3: Cloud-based data processing and intelligent decision-making: In the cloud service module, the data center builds a distributed database to efficiently store and manage historical garden data, including plant growth records and meteorological and environmental data, supporting complex data queries and spatial analysis. The intelligent decision-making center uses deep learning models and reinforcement learning algorithms to intelligently identify pests and diseases, generate prevention and control plans, and optimize irrigation strategies to achieve precise regulation of water resources. The digital twin platform constructs a three-dimensional garden model, mapping the state of the physical world in real time, supporting maintenance plan deduction and decision-making evaluation in virtual scenarios. Step 4. Application service optimization and emergency response: In the application service module, the smart maintenance unit automatically generates maintenance work orders and dispatches robots to perform pruning and spraying tasks. At the same time, it provides health records of ancient and famous trees, records annual ring data and maintenance history; the visitor service unit uses AR guidance technology to display plant information, push flowering forecasts and event notifications, and integrates smart trash cans and charging stations to enhance the visitor experience; the emergency management unit combines GIS maps with real-time data to plan fire evacuation routes, link fire-fighting equipment, predict and respond to the risk of heavy rain and waterlogging, and ensure the safety of the park.

Images