Intelligent apple planting management and expert decision-making system

Abstract

The intelligent apple planting management and expert decision-making system comprises a sensor data acquisition module, an image data acquisition module, an edge calculation module, an expert knowledge module, an intelligent analysis module, an expert decision-making module, a wireless communication module, an equipment control module, a growth data tracing analysis module and a user interaction control module. Environmental and growth data are collected through a sensor and an image data acquisition module, and are transmitted to an edge calculation module through wireless communication. The system combines artificial intelligence technology analysis and expert knowledge, evaluates the growth condition of fruit trees, provides a scientific planting strategy, supports automatic control of agricultural operations such as irrigation and fertilization, realizes full-cycle tracing and optimization of growth data, and aims to provide more accurate production management parameters for planters, provide decision basis for scientific and standardized production, and improve the production efficiency. Apple quality is improved, diseases are reduced, planting management is optimized, and orchard production benefits are increased.

Description

Technical Field

[0001] The present invention relates to the technical field of fruit tree planting, and in particular to an intelligent apple planting management and expert decision-making system. Background Art

[0002] Fruit tree planting management typically includes the selection and optimization of the cultivation environment, seedling management, soil and fertilizer management, irrigation and drainage, pruning, and pest and disease control. Traditional fruit tree planting is still primarily manual, requiring full oversight from planters. This relies on planters' experience to conduct real-time orchard inspections to determine whether the current growing environment is normal and, therefore, determine appropriate intervention measures. However, this approach is highly dependent on manual planting experience and is time-consuming and labor-intensive. In recent years, fruit tree management systems have emerged that utilize IoT and sensor technology to acquire data on the fruit tree growing environment and monitor its growth in real time. However, most of these systems only capture limited growth environment data and are unable to accurately determine the quality of the current growing environment. Manual judgment is often still required based on experience to determine whether the data is normal and, ultimately, what intervention measures to take. This results in a lack of timely and scientific guidance on planting strategies, leading to poor orchard production efficiency. Summary of the Invention

[0003] In response to the shortcomings of the above-mentioned prior art, the present invention proposes an intelligent apple planting management and expert decision-making system to provide growers with more precise production management parameters, provide a decision-making basis for scientific and standardized production, improve apple quality, reduce diseases, optimize planting management, and increase orchard production efficiency.

[0004] To solve the above problems, the present invention provides an intelligent apple planting management and expert decision-making system, including a cloud and an edge end, wherein the cloud end is deployed with an expert knowledge module, an expert decision-making module, an extension module, a user interaction control module, and a cloud-edge collaboration module; the edge end is deployed with a data acquisition module, a wireless communication module, and an intelligent analysis module; wherein the cloud-edge collaboration module is used to coordinate the cloud and edge modules to interact with data and work together; the data acquisition module is used to collect real-time images of the current apple tree and the growth environment data during the growth of the apple tree, and send the data to the intelligent analysis module through the wireless communication module; the intelligent analysis module performs intelligent analysis based on the data transmitted by the data acquisition module, and outputs relevant analysis results.

[0005] Furthermore, the data acquisition module includes:

[0006] The sensor data acquisition submodule is used to collect growth environment data of apple trees in different locations of the orchard through different sensors. The growth environment data includes ambient temperature and humidity, light, carbon dioxide, soil temperature and humidity, soil pH, and soil conductivity;

[0007] The data processing submodule is used to convert the Modbus protocol of each sensor into a JSON format that can be recognized by the system to facilitate subsequent data storage;

[0008] The image data acquisition submodule is used to monitor the changes in apple leaves, buds, and fruits in real time through multiple cameras deployed in different areas of the orchard, and store the captured photos in the SSD disk.

[0009] Furthermore, the expert knowledge module is used to convert expert knowledge of fruit tree planting into data that is understandable and quickly retrievable by the system, providing a basis for subsequent intelligent analysis modules and expert decision modules; it includes the following submodules:

[0010] The knowledge base includes an apple tree growth process knowledge base and a pest and disease detection and prevention knowledge base, which respectively store expert knowledge content on the apple tree growth process and expert knowledge content on pest and disease detection and prevention;

[0011] The knowledge retrieval submodule uses a large model to import expert knowledge content into the large model to obtain the corresponding knowledge results.

[0012] Furthermore, the intelligent analysis module includes a growth stage detection submodule, a disease and insect pest detection submodule, a maintenance status detection submodule and a planting status prediction submodule, wherein the growth stage detection submodule,

[0013] The pest and disease detection submodule and the maintenance status detection submodule are constructed based on the growth stage detection model, pest and disease detection model, and maintenance status detection model respectively obtained by training the YOLOv11 target detection algorithm;

[0014] The growth stage detection submodule is used to detect the growth stage of the current fruit tree based on the growth stage detection model, and finally output the growth stage of the current fruit tree in combination with the time period in the knowledge base;

[0015] The pest and disease detection submodule is used to detect whether there are pests and diseases during the growth of fruit trees based on the pest and disease detection model, and store the disease photos and pest and disease detection results in the database;

[0016] The maintenance status detection submodule is used to detect the maintenance operation status during the growth process of the fruit tree based on the maintenance status detection model, and filter out maintenance operations that are not in the current growth stage from the maintenance operation status in combination with the current fruit tree growth stage output by the growth stage detection submodule, and output the final maintenance operation detection result;

[0017] The planting situation prediction submodule is used to output the planting score of the current fruit tree and record the planting events based on the growth environment data, pest and disease detection results, and maintenance operation detection results collected by the sensor data acquisition submodule at different growth stages of the fruit tree.

[0018] Furthermore, the expert decision module is used to construct prompt words based on the growth environment data collected by the sensor data acquisition submodule and the analysis results output by the intelligent analysis module, and input the prompt words into the expert knowledge module, and use the large model to output planting decision suggestions and problem solutions.

[0019] Furthermore, the expansion module includes:

[0020] The device control submodule is used to control external devices connected through relay switches and automatically start or shut down corresponding external devices based on the intelligent suggestions obtained by the expert decision module;

[0021] The alarm notification submodule is used to monitor and record abnormal information transmitted by the growth stage detection module, disease detection module, sensor data module, maintenance detection module, and equipment control module in real time, and send alarm information to the user when abnormal information is received;

[0022] The growth data tracing and analysis submodule is used to trace and analyze the detailed historical data of fruit tree growth.

[0023] Furthermore, the growth data retrospective analysis submodule includes:

[0024] Historical data storage unit, used to store all collected growth data in the database, and supports long-term storage and fast retrieval to ensure data integrity and traceability;

[0025] A data query function unit is used for users to quickly retrieve historical data stored in the historical data storage unit according to different parameters;

[0026] Data visualization unit, used to display complex data through charts, line charts, and bar charts to help users understand data changes and trends;

[0027] Growth trend analysis unit, which analyzes historical data to obtain trends and patterns in fruit tree growth, providing reference for future planting activities;

[0028] Environmental impact assessment unit, used to analyze the impact of environmental factors on fruit tree growth and evaluate growth performance under different environmental conditions;

[0029] The maintenance effect evaluation unit evaluates the effects of different maintenance measures by comparing growth data before and after maintenance, providing data support for formulating more effective maintenance plans;

[0030] User-customizable report unit, users can generate customized reports according to their needs, including growth cycle reports, environmental change reports, and maintenance effect reports.

[0031] Furthermore, the user interaction control module includes:

[0032] Real-time growth status display unit, used to display real-time environmental factor data, various situation forecasts, growth stages, maintenance conditions, disease conditions, and core data of equipment operation status;

[0033] The expert decision suggestion viewing unit is used to view the detailed expert decision suggestion list output by the expert decision module and search for relevant suggestions based on conditions;

[0034] The knowledge base module unit is used to provide an interface for uploading and managing relevant documents of the expert knowledge base, thereby enhancing the intelligent capabilities of the knowledge base;

[0035] The visual module control unit is used to set the parameters of the small target detection optimization method in the intelligent analysis module, including ROI, confidence level, time interval and other parameters;

[0036] The historical data analysis unit is used to provide an interface for searching and querying all historical data of fruit tree growth according to conditions, and to display various generated reports;

[0037] Equipment control record setting unit, used to set the parameters of automatically controlled equipment and query the historical data of equipment control;

[0038] The alarm module setting and data query unit is used to query all historical alarm records, set the alarm media, alarm format, and alarm content.

[0039] Furthermore, the wireless communication module includes 4G and LORA wireless communication modules, wherein the 4G wireless communication module is used to realize communication between edge devices and the cloud; the LORA wireless communication module is used to remotely obtain sensor data.

[0040] Therefore, the present invention adopts the above-mentioned intelligent apple planting management and expert decision-making system, which has the following beneficial effects:

[0041] First, the present invention uses sensors to collect data related to environmental growth, and uses monitoring cameras to obtain real-time image data of fruit tree growth, monitors the growth status of apple trees in real time, and uses AI deep learning methods through intelligent analysis modules to monitor the growth, disease status, and maintenance status of fruit trees in real time, and comprehensively judges the rating of fruit tree planting;

[0042] Second, the present invention constructs expert advice, knowledge and other content required in the process of fruit tree growth through the expert knowledge module, supports the ability to quickly retrieve and analyze knowledge, and uses the expert decision module to comprehensively judge the current fruit tree planting situation based on sensor data and intelligent analysis results. Combined with the knowledge base module, it gives the latest planting decision suggestions and problem-solving solutions to help fruit farmers quickly discover and solve problems.

[0043] Third, the present invention is equipped with an expansion module that can reversely manually or automatically control some equipment in fruit tree planting (such as irrigation and fertilization systems), and use the alarm notification module to promptly notify users through SMS or email;

[0044] Fourth, the present invention traces and optimizes the entire cycle of fruit tree growth through the growth data tracing and analysis module, thereby providing growers with a more accurate variety of data in the production management parameter statistical analysis system, and supports viewing historical data, providing users with more dimensional data references for judging fruit tree planting.

[0045] Fifth, the present invention utilizes the user interaction control module to view the real-time data of the fruit tree environment, the fruit tree growth report, the fruit tree growth suggestion, the fruit tree growth historical data and the like.

[0046] In summary, the present invention can monitor the growth status of apple trees in real time, identify problems in the growth process of fruit trees, and provide planting suggestions and intervention measures based on expert knowledge for the problems that arise. At the same time, it can also realize the digitization and traceability of the apple tree growth process through UI interaction, ultimately improving apple yield and fruit quality and reducing diseases.

[0047] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a module architecture diagram of the present invention.

[0049] Figure 2 This is the structural diagram of the optimized YOLOv11 target detection model.

[0050] Figure 3a-3c This is the brown spot disease form of apple trees.

[0051] Figure 4 This is the form of anthracnose leaf blight on apple trees.

[0052] Figure 5 This is the ring rot morphology of apple trees.

[0053] Figure 6a-6c This is the form of apple rust on apple trees.

[0054] Figure 7 This is a workflow diagram of the expert decision-making module of the present invention.

[0055] Figure 8 This is a diagram of the implementation and deployment architecture of the system proposed in the present invention. DETAILED DESCRIPTION

[0056] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art will make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims of the application.

[0057] The present invention proposes an intelligent apple planting management and expert decision-making system, including cloud and edge terminals, such as Figure 1 、 Figure 8 As shown in the figure, the cloud is deployed with an expert knowledge module, an expert decision module, an extension module, a user interaction control module, and a cloud-edge collaboration module; the edge is deployed with a data acquisition module, a wireless communication module, and an intelligent analysis module; the cloud-edge collaboration module is used to coordinate data interaction and collaborative work between the cloud and edge modules;

[0058] The data acquisition module is used to collect real-time images of the current apple tree and the growth environment data of the apple tree during its growth process, and send the data to the intelligent analysis module through the wireless communication module; the intelligent analysis module performs intelligent analysis based on the data transmitted by the data acquisition module and outputs relevant analysis results;

[0059] The following is an introduction to each of the above modules:

[0060] 1. Data acquisition module

[0061] This module includes the following two sub-modules:

[0062] (1) Sensor data acquisition submodule

[0063] This module is used to collect growth environment data of apple trees at different points in the orchard through different sensors. The growth environment data includes ambient temperature and humidity, light, carbon dioxide, soil temperature and humidity, pH value, and soil conductivity. These data are crucial to the growth of fruit trees (apples grow by absorbing nutrients through their roots. A good root system is a high-speed channel for nutrients, and the soil environment of the root system determines its growth state; soil factors such as soil pH, conductivity, air permeability, humidity, and ground temperature affect root growth, thereby affecting the effectiveness of nutrients).

[0064] The sensors include temperature and humidity sensors, light sensors, carbon dioxide concentration sensors, soil pH sensors, and soil conductivity sensors. Multiple sets of sensor equipment are deployed at different locations in the orchard to prevent inaccurate data detection due to local environmental changes. Subsequent data processing and storage can be judged based on sensor location attributes.

[0065] (2) Data processing submodule

[0066] This module is used to convert the Modbus protocol of each sensor into a JSON format that can be recognized by the system to facilitate subsequent data storage. The JSON format needs to be defined during the protocol conversion process, for example: {'humidity_v':56.6,'temperature_v':24.3,'conductivity_v':146}.

[0067] The Modbus protocol is a query-response mode. The query protocol structure is as follows: 08 03 00 0000 0704 91, which respectively correspond to the communication address (0803), query content (00000007), and CRC check code (0491). The response protocol example is as follows: 08 03 0e 00b5 01 55 00 00d5 98 00 00 00 0002 22 33ae, which respectively correspond to the communication address (0803), data length (0e), data content (00b5 01550000d598 0000 0000 0222), and CRC check code (33ae). 55 is the hexadecimal data of temperature, which is converted into actual temperature data as follows: (1*16*16+5*16+5) / 10=34.1℃; 00b5 is the humidity data, and the same calculation is: (11*16+5) / 10=18.1%; d598 is the light data, and the calculation is as follows: (13*16*16*16+5*16*16+9*16+8)=54680LUX, that is, the light is 54680LUX.

[0068] (3) Image data acquisition submodule

[0069] This module is used to monitor the changes in apple leaves, buds, and fruits in real time using multiple cameras deployed in different areas of the orchard. It includes the following units:

[0070] The image acquisition unit is configured to capture an image of the apple using a camera, wherein the camera is a high-definition surveillance camera with a resolution of 1080P or higher; the image acquisition frequency is once every 10 minutes, and the acquisition time period is from 8:00 a.m. to 6:00 p.m. During this time period, most areas have good lighting, and the camera can capture clear images, which is conducive to detection by the target detection method;

[0071] Image encoding and decoding unit, used to encode and decode the real-time data stream of the image collected by the camera through the H.264 / H.265 algorithm;

[0072] The image preprocessing unit is used to filter out abnormal data in the image processed by the image encoding and decoding unit, so as to ensure that the images entering the intelligent analysis module are clear and reliable data, thereby improving the detection and decision-making capabilities of the system.

[0073] 2. Expert knowledge module

[0074] This module converts expert knowledge about fruit tree planting into system-understandable and quickly retrievable data, providing a basis for subsequent intelligent analysis and decision-making modules. This expert knowledge module, built on a large language model system, is simpler, more scalable, and more flexible than traditional knowledge modules that require structured processing.

[0075] The expert knowledge module includes a knowledge base and a knowledge retrieval submodule. The knowledge base is constructed based on the content retrieved by the knowledge retrieval submodule, specifically including a knowledge base on the apple tree growth process and a knowledge base on pest and disease detection and prevention. The knowledge retrieval submodule uses a large model.

[0076] Traditional knowledge base construction requires categorizing and storing fruit tree planting knowledge according to specific attributes. However, since fruit trees may have different attributes at different growth stages, this makes it difficult to achieve uniformity in data abstraction. Furthermore, when expanding the knowledge system, it is also necessary to consider whether it will impact existing systems. Constructing a knowledge base based on a large model, on the other hand, simply requires loading expert knowledge content through MaxKB and accessing the desired data by calling Ollama's interface (using the Qwen2-7B model on the backend).

[0077] For example, the prompt words for obtaining data through the Ollama interface can be referred to as follows:

[0078] Please provide environmental factor data for apple trees during the budding stage, including temperature, soil moisture, light duration, pH value, etc., in a format similar to the following: {"temperature":5,"soilhumidity":30,"illumination":36000,"PH":5.2}

[0079] The large model summarizes and outputs the following information based on the knowledge base: {“temperature”: “7-10”, “soilhumidity”: “60-80”, “illumination”: “6-8”, “PH”: “5-8”}.

[0080] The following is an introduction to the main contents of the above two knowledge bases (Table 1 and Table 2):

[0081] Table 1 Knowledge base of apple tree growth process

[0082]

[0083]

[0084]

[0085]

[0086]

[0087]

[0088] Table 2 Knowledge base of pest and disease detection and control

[0089]

[0090]

[0091]

[0092] 3. Intelligent analysis module

[0093] This module is used to determine the current growth status of fruit trees based on the image data collected by the YOLOv11 target detection algorithm and the image data acquisition submodule. Specifically, it includes growth stage detection, pest and disease detection, maintenance status detection, and planting status prediction.

[0094] Among them, growth stage detection, pest and disease detection, and maintenance status detection all use the optimized YOLOv11 target detection algorithm to accurately detect the classification of each feature. YOLOv11 has high performance advantages and is very suitable for edge computing scenarios. However, in this system, due to the small size of the camera imaging targets (such as leaf buds), the small target detection capability of the original YOLOv11 algorithm was optimized. Based on the YOLOv11 model, the growth stage detection model, pest and disease detection model, and maintenance status detection model were trained separately. During the training process of each model, at least 100 pictures need to be collected as a training data set for each model feature. For example, there are 11 features in growth stage detection, 100 pictures are collected for each feature, and a total of 1,100 pictures are collected. The training set and test set are allocated at a ratio of 9:1, and the training accuracy is at least 98%. The training processes for the above three detection models are similar. They all obtain fruit tree growth images through the data acquisition module. Due to limited computing resources, there is no high requirement for detection time. Therefore, the image data is serially inferred through the three models in sequence. The inference results of each model are updated to the database in the edge device. Through real-time automatic monitoring and recording of fruit tree growth stages, disease conditions, and maintenance conditions, the intelligent expert decision-making module provides a basis for judgment.

[0095] This paper is based on the YOLOv11 target detection algorithm. By modifying the small target detection head and reducing the minimum detectable target size, the YOLOv11 target detection algorithm is optimized. The optimized YOLOv11 model is as follows: Figure 2 shown.

[0096] from Figure 2 It can be seen that the input vector sizes of the three detection heads in the red box are reduced by half compared to the original network. The C3K2 of the 16th layer network is changed from the original 64x80x80 to 32x80x80, the C3K2 of the 19th layer network is changed from the original 128x40x40 to 64x40x40, and the C3K2 of the 22nd layer network is changed from the original 256x20x20 to 128x20x20. Due to the reduced detection size, the improved network has higher detection accuracy in small target detection, while also improving the detection speed and effectively reducing the computing power required for detection.

[0097] The intelligent analysis module includes the following sub-modules:

[0098] ①Growth stage detection submodule

[0099] The growth stages of fruit trees are divided into dormancy, budding, critical period for flower bud differentiation, fruit expansion before harvest, and maturity. In addition to using the time period in the expert knowledge module to determine the growth stage, it is also necessary to use appearance characteristics to comprehensively judge whether the growth of the fruit tree meets expectations. Therefore, it is necessary to define appearance characteristic factors for different growth stages:

[0100] Dormant period: yellow leaves, fallen leaves

[0101] Bud stage: leaf bud

[0102] Critical period of flower bud differentiation: flower buds, sepals, petals, and stamens

[0103] Fruit expansion period before harvest: young fruit, bagging

[0104] Ripening period: bag removal, fruit ripening

[0105] When training the growth phase detection model, 100 images were collected for each of the 11 different features mentioned above.

[0106] Based on the growth stage detection model, the growth stage of the current fruit tree is detected. The detection output result is one of the five growth stages. The current growth stage is comprehensively judged by combining the multi-day average temperature (data obtained from the temperature sensor) and the date (current time of the system), and the final growth stage is output.

[0107] ② Pest and disease detection submodule

[0108] This module is used to detect whether there are diseases and pests during the growth of fruit trees. It obtains high-definition image data of leaves and branches during the growth process of fruit trees through the image data acquisition submodule. Through the disease and pest detection model, it calculates in real time whether there are diseases and pests in fruit trees, stores the detection results in the database, and finally uses the expert decision module to output cause analysis and prevention and control suggestions.

[0109] Key design of the pest and disease detection submodule:

[0110] i) To reduce the false positive rate of disease, it is required to detect disease characteristics in at least three different locations, and further confirm the disease situation based on the judgment results of the growth stage;

[0111] ii) The same disease may have multiple different characteristics, and it is necessary to distinguish these characteristics to make the detection algorithm more accurate, such as brown spot disease No. 1, brown spot disease No. 2, and brown spot disease No. 3;

[0112] iii) The detection output is the specific disease type and disease image, which are saved in the database. In the expert decision-making module, disease prevention and control suggestions are given in combination with the disease knowledge base, and the disease recovery status is continuously monitored.

[0113] ③Maintenance status detection submodule

[0114] Fruit tree cultivation requires specialized maintenance procedures at different stages, such as whitewashing, trellising, bagging, bag removal, spraying, irrigation, tillage, fertilization, flower and fruit thinning, and pruning. Different growth stages and external environmental factors may correspond to different maintenance procedures, and expert decision-making regarding the appropriate maintenance procedures at the right time is crucial. This module monitors the maintenance status of fruit trees during their growth. Based on high-definition trunk image data acquired by the image data acquisition submodule, it utilizes a maintenance detection model to calculate the tree's maintenance status in real time and stores the results in a database.

[0115] Afterwards, the expert decision-making module will comprehensively judge whether and what maintenance operations should be performed based on data such as environmental factors, growth stages, and expert knowledge base, and give users the most reasonable maintenance recommendations.

[0116] The image features of the maintenance operation of the maintenance condition detection model are as follows:

[0117] Dormant period: whether whitewashed (branch characteristics), whether plowed (ground characteristics), whether pruned, whether irrigated (ground characteristics)

[0118] Bud stage: whether to thin out the buds, whether the thinning meets the standards, and whether the branches are evenly distributed.

[0119] Flower bud differentiation period: whether to thin out flowers, whether the flower bud distribution meets the standards, and whether the flower bud growth is healthy.

[0120] Fruit expansion period before harvest: whether to build racks, whether to put bags on the fruit, whether the fruit is evenly distributed

[0121] Maturity: whether the branches and leaves are healthy, whether the bags have been removed, and whether the fruits are healthy

[0122] When training the maintenance condition detection model, 100 image datasets are collected for each of the above 16 features to train the model.

[0123] The image data must first pass through the growth stage detection model to determine the current growth stage, and then pass through the maintenance detection model to determine whether the maintenance operations in the current growth stage are reasonable. The specific method is to perform judgment and filtering during the post-processing process of the detection. For example, if the current growth stage is the budding stage, then only the thinning of buds and the distribution of branches need to be paid attention to, and other detected results need to be filtered out.

[0124] ④Planting situation prediction submodule

[0125] This module is used to combine sensor data from each stage, pest and disease detection results, and maintenance test results to make a comprehensive judgment and output a planting score with a total score of 5 points. At the same time, it is necessary to record planting events (such as when and what data did not meet the standards, what diseases occurred, and which maintenance did not meet the standards). The specific evaluation criteria are shown in Table 1:

[0126] Table 3 Planting situation scoring table

[0127] Score name Judging Criteria 1 point Difference The important environmental data required for fruit tree growth do not meet the standards, diseases seriously affect the yield, and there is no maintenance 2 points generally Some environmental data do not meet the standards, diseases occur, and maintenance actions are not in place 3 points ordinary Some environmental data do not meet the standards and diseases occur. Timely maintenance should be carried out to avoid serious diseases. 4 points good Environmental data at each stage meet the standards, no diseases occur, and maintenance conditions basically meet the standards 5 points excellent Environmental data at all stages meet standards, disease prevention and control are timely, no diseases occur, and maintenance is excellent

[0128] This relatively intuitive scoring method can quickly understand the health of the current planting management. In addition, it can also provide more detailed information to understand which problems have not been solved in a timely manner at what stage, and provide more professional advice to help users accumulate experience in fruit tree planting.

[0129] 4. Expert decision-making module

[0130] The expert decision-making module is used to comprehensively judge the current fruit tree planting situation based on sensor data and the analysis results of the intelligent analysis module, and combined with the expert knowledge module, it provides the latest planting decision suggestions and problem-solving solutions to help fruit farmers quickly identify and solve problems.

[0131] This module is used to integrate sensor data, AI detection results and knowledge base data to make comprehensive decisions. The specific process is as follows: Figure 7 As shown:

[0132] Step 1: Determine the current growth stage of the fruit tree by using the 7-day average temperature data, the current date, and the detection results of the growth stage detection module;

[0133] Step 2: Real-time acquisition of sensor data, maintenance and disease detection results, and then retrieval through the expert knowledge module, with the large model providing a comprehensive conclusion;

[0134] The data input into the expert knowledge module in step 2 includes sensor data, maintenance detection results (such as detection of unpainted or unpruned branches), disease detection results (such as detection of brown spot disease), and the growth stage obtained in step 1. The expert knowledge module constructs prompt words for the large model based on these data and sends the constructed prompt words to the large model. Finally, the large model outputs the planting situation prediction results and expert decision recommendations.

[0135] The prompt word template is as follows:

[0136]

[0137] The content in {} in the above prompt word template is the current environmental variable data and the current growth stage;

[0138] For example, if the input values ​​are: air temperature 8°C, light intensity 2000 LUX, soil moisture 40%, budding stage, pH 6.8, maintenance test result: uneven branch distribution due to lack of bud thinning, and disease test result: no disease found, the expert knowledge base module will process these values ​​based on the large model and output the following results:

[0139]

[0140] The system will list detailed operating specifications, matters that need to be paid attention to during the budding stage, and expert advice in the output results.

[0141] 5. Extension Module

[0142] (1) Equipment control module

[0143] This module is used to control external devices connected via relay switches, such as irrigation equipment and fertilization equipment. In addition to manual control of each external device, this system also supports automatic control. By combining intelligent suggestions from the expert decision-making module, it automatically starts or shuts down the corresponding equipment, thereby achieving unmanned and intelligent automatic control. In addition, this module can monitor the status of external devices in real time (the device supports status data collection). Users can view the operating status of the device at any time, and if the device fails, they can also be notified through the alarm module. This module can also view the historical execution history of the device, such as irrigation duration, irrigation volume, and other data. This data also provides a basis for growth data tracing.

[0144] (2) Alarm notification module

[0145] This module is a key component of the system, used to monitor and record negative events in the orchard that may affect fruit tree health and yield, namely abnormal data transmitted from the growth stage detection module, disease detection module, sensor data module, maintenance detection module, and equipment control module. Through an event processing mechanism, it captures and records various key information in real time, ensuring that users can respond quickly and take necessary measures. The following are the main functions and features of this module:

[0146] Comprehensive event recording: The module can record all key negative events during the growth process of fruit trees, including but not limited to changes in growth stages, the occurrence of diseases, abnormal fluctuations in environmental data, and events related to maintenance and automated equipment control.

[0147] Multi-channel notification system: To ensure users receive important information in a timely manner, this module supports sending notifications via multiple communication media such as SMS, WeChat, and email. Users can set the method of receiving notifications according to their preferences to ensure timely delivery of information.

[0148] Detailed Recording: Messages are recorded in detail, including not only the type and time of the incident, but also specific data and recommendations, such as the type of disease, specific values ​​of environmental parameters, and specific maintenance measures. These details help users fully understand the severity and urgency of the incident, allowing them to make more accurate decisions.

[0149] Through these features, the module not only improves the efficiency and responsiveness of fruit tree management, but also enhances users' understanding and control over fruit tree growth, thereby helping to improve the overall yield and quality of the orchard.

[0150] (3) Growth data tracing and analysis module

[0151] The growth data tracing and analysis module is used to trace and analyze detailed historical data on fruit tree growth, including:

[0152] Historical data storage unit, used to store all collected growth data in the database, and supports long-term storage and fast retrieval to ensure data integrity and traceability;

[0153] A data query unit is used for users to quickly retrieve historical data stored in the historical data storage unit according to different parameters (such as date, growth stage, etc.);

[0154] Data visualization unit, used to display complex data through charts, line charts, and bar charts to help users understand data changes and trends;

[0155] Growth trend analysis unit, which analyzes historical data to obtain trends and patterns in fruit tree growth, providing reference for future planting activities;

[0156] Environmental impact assessment unit, used to analyze the impact of environmental factors (such as temperature, humidity, and light) on fruit tree growth and evaluate growth performance under different environmental conditions;

[0157] The maintenance effect evaluation unit evaluates the effects of different maintenance measures by comparing growth data before and after maintenance, providing data support for formulating more effective maintenance plans;

[0158] User-customizable report unit, users can generate customized reports according to their needs, including growth cycle reports, environmental change reports, and maintenance effect reports.

[0159] The growth data traceability analysis module not only allows users to gain an in-depth understanding of the growth history of fruit trees, but also enables them to make more accurate planting decisions based on the data, thereby improving the scientific nature and efficiency of orchard management.

[0160] 6. User interaction control module

[0161] The user interaction control module is the bridge of communication between the system and the user. It is responsible for receiving user instructions, displaying system status, and providing an operation interface so that users can effectively manage and control the entire system. The interactive function modules provided by this module mainly include:

[0162] The real-time growth status display unit is used to display real-time environmental factor data, planting situation forecast, growth stage, maintenance status, disease status, and core data of equipment operation status.

[0163] The expert decision suggestion viewing unit is used to view the detailed expert decision suggestion list output by the expert decision module and search for relevant suggestions based on conditions.

[0164] The knowledge base module unit is used to provide an interface for uploading and managing relevant documents of the expert knowledge base, thereby enhancing the intelligent capabilities of the knowledge base.

[0165] The visual module control unit is used to set the parameters of the small target detection optimization method in the intelligent analysis module, including ROI, confidence, and time interval parameters.

[0166] The historical data analysis unit is used to provide an interface for searching and querying all historical data of fruit tree growth according to conditions, and to display various generated reports.

[0167] The device control record setting unit is used to set the parameters of the automatically controlled equipment and query the historical data of the equipment control.

[0168] The alarm module setting and data query unit is used to query all historical alarm records, set the alarm media, alarm format, and alarm content.

[0169] 7. Wireless communication equipment

[0170] Wireless communication equipment mainly includes 4G and LORA wireless communication modules, which are used to provide support for system data communication.

[0171] Since most orchards do not have Wi-Fi or Ethernet, data needs to be uploaded via 4G. Therefore, a 4G wireless communication module is set up in this system. The 4G wireless communication module in this system is mainly used for communication between the edge computing box and the cloud management platform.

[0172] Sensor devices are deployed in various corners of the orchard. Wired wiring increases the difficulty of wiring. Therefore, this system uses the LoRa wireless communication module to remotely obtain sensor data. To achieve this goal, it is necessary to modify the ID and communication channel of different sensors in the LoRa communication module to avoid mutual interference when LoRa modules in different areas communicate.

[0173] 8. Cloud-edge collaboration module

[0174] The cloud-edge collaboration module is mainly used for collaborative work between the cloud and the edge. The cloud operates the management system, and the edge provides AI computing units and data collection functions to support deep learning inference calculations.

[0175] Edge devices are mainly responsible for the operation of target detection algorithms and data collection functions. The entire platform is deployed in the cloud, so that the platform can access the data of multiple users at the same time, isolate data through account permissions, and ensure data security.

[0176] Edge devices use computing devices based on the RK3576 chip, supporting at least eight cameras and four AI algorithms. Because the algorithms run at relatively long intervals (minutes), the computing power requirements for edge devices are low; 6TB of computing power is sufficient. The results of the object detection algorithm execution are sent to the cloud for storage.

[0177] In addition, data is also collected to the edge device through the LORA communication module. The box device performs simple data processing and then sends it to the cloud for storage.

[0178] When communication between the edge and the cloud fails, data can be temporarily stored locally on the edge. Due to the limited storage capacity of the edge, the temporary storage period does not exceed one month. When communication with the cloud is restored, all data will be sent to the cloud.

[0179] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. An intelligent apple planting management and expert decision-making system, characterized by: It includes the cloud and edge, among which the cloud is deployed with expert knowledge module, expert decision module, extension module, user interaction control module, and cloud-edge collaboration module; the edge is deployed with data acquisition module, wireless communication module and intelligent analysis module; among which, the cloud-edge collaboration module is used to coordinate the cloud and edge modules to interact with data and work together; the data acquisition module is used to collect real-time images of the current apple tree and the growth environment data during the growth process of the apple tree, and send the data to the intelligent analysis module through the wireless communication module; the intelligent analysis module performs intelligent analysis based on the data transmitted by the data acquisition module, and outputs relevant analysis results.

2. The intelligent apple planting management and expert decision-making system according to claim 1, characterized in that: The data acquisition module includes: The sensor data acquisition submodule is used to collect growth environment data of apple trees in different locations of the orchard through different sensors. The growth environment data includes ambient temperature and humidity, light, carbon dioxide, soil temperature and humidity, soil pH, and soil conductivity; The data processing submodule is used to convert the Modbus protocol of each sensor into a JSON format that can be recognized by the system to facilitate subsequent data storage; The image data acquisition submodule is used to monitor the changes in apple leaves, buds, and fruits in real time through multiple cameras deployed in different areas of the orchard, and store the captured photos in the SSD disk.

3. The intelligent apple planting management and expert decision-making system according to claim 2, characterized in that: The expert knowledge module is used to convert the expert knowledge of fruit tree planting into data that can be understood by the system and quickly retrieved, providing a basis for the subsequent intelligent analysis module and expert decision-making module; That Includes the following submodules: The knowledge base includes an apple tree growth process knowledge base and a pest and disease detection and prevention knowledge base, which respectively store expert knowledge content on the apple tree growth process and expert knowledge content on pest and disease detection and prevention; The knowledge retrieval submodule uses a large model to import expert knowledge content into the large model to obtain the corresponding knowledge results.

4. The intelligent apple planting management and expert decision-making system according to claim 3, characterized in that: The intelligent analysis module includes a growth stage detection submodule, a disease and pest detection submodule, a maintenance status detection submodule, and a planting status prediction submodule, wherein the growth stage detection submodule, the disease and pest detection submodule, and the maintenance status detection submodule are respectively constructed based on the growth stage detection model, the disease and pest detection model, and the maintenance status detection model obtained by training the YOLOv11 target detection algorithm; The growth stage detection submodule is used to detect the growth stage of the current fruit tree based on the growth stage detection model, and finally output the growth stage of the current fruit tree in combination with the time period in the knowledge base; The pest and disease detection submodule is used to detect whether there are pests and diseases during the growth of fruit trees based on the pest and disease detection model, and store the disease photos and pest and disease detection results in the database; The maintenance status detection submodule is used to detect the maintenance operation status during the growth process of the fruit tree based on the maintenance status detection model, and filter out maintenance operations that are not in the current growth stage from the maintenance operation status in combination with the current fruit tree growth stage output by the growth stage detection submodule, and output the final maintenance operation detection result; The planting situation prediction submodule is used to output the planting score of the current fruit tree and record the planting events based on the growth environment data, pest and disease detection results, and maintenance operation detection results collected by the sensor data acquisition submodule at different growth stages of the fruit tree.

5. The intelligent apple planting management and expert decision-making system according to claim 4, characterized in that: The expert decision module is used to construct prompt words based on the growth environment data collected by the sensor data acquisition submodule and the analysis results output by the intelligent analysis module, and input the prompt words into the expert knowledge module, and use the large model to output planting decision suggestions and problem solutions.

6. The intelligent apple planting management and expert decision-making system according to claim 5, characterized in that: The expansion module includes: The device control submodule is used to control external devices connected through relay switches and automatically start or shut down corresponding external devices based on the intelligent suggestions obtained by the expert decision module; The alarm notification submodule is used to monitor and record abnormal information transmitted by the growth stage detection module, disease detection module, sensor data module, maintenance detection module, and equipment control module in real time, and send alarm information to the user when abnormal information is received; The growth data tracing and analysis submodule is used to trace and analyze the detailed historical data of fruit tree growth.

7. The intelligent apple planting management and expert decision-making system according to claim 6, characterized in that: The growth data retrospective analysis submodule includes: Historical data storage unit, used to store all collected growth data in the database, and supports long-term storage and fast retrieval to ensure data integrity and traceability; A data query function unit is used for users to quickly retrieve historical data stored in the historical data storage unit according to different parameters; Data visualization unit, used to display complex data through charts, line charts, and bar charts to help users understand data changes and trends; Growth trend analysis unit, which analyzes historical data to obtain trends and patterns in fruit tree growth, providing reference for future planting activities; Environmental impact assessment unit, used to analyze the impact of environmental factors on fruit tree growth and evaluate growth performance under different environmental conditions; The maintenance effect evaluation unit evaluates the effects of different maintenance measures by comparing growth data before and after maintenance, providing data support for formulating more effective maintenance plans; User-customizable report unit, users can generate customized reports according to their needs, including growth cycle reports, environmental change reports, and maintenance effect reports.

8. The intelligent apple planting management and expert decision-making system according to claim 7, characterized in that: The user interaction control module includes: Real-time growth status display unit, used to display real-time environmental factor data, various situation forecasts, growth stages, maintenance conditions, disease conditions, and core data of equipment operation status; The expert decision suggestion viewing unit is used to view the detailed expert decision suggestion list output by the expert decision module and search for relevant suggestions based on conditions; The knowledge base module unit is used to provide an interface for uploading and managing relevant documents of the expert knowledge base, thereby enhancing the intelligent capabilities of the knowledge base; The visual module control unit is used to set the parameters of the small target detection optimization method in the intelligent analysis module, including ROI, confidence level, time interval and other parameters; The historical data analysis unit is used to provide an interface for searching and querying all historical data of fruit tree growth according to conditions, and to display various generated reports; Equipment control record setting unit, used to set the parameters of automatically controlled equipment and query the historical data of equipment control; The alarm module setting and data query unit is used to query all historical alarm records, set the alarm media, alarm format, and alarm content.

9. The intelligent apple planting management and expert decision-making system according to claim 8, characterized in that: The wireless communication module includes 4G and LORA wireless communication modules, wherein the 4G wireless communication module is used to realize communication between edge devices and the cloud; the LORA wireless communication module is used to remotely obtain sensor data.

Images