Self-adaptive management system for picking in orchard

Abstract

The invention, which relates to the technical field of orchard picking management, discloses an orchard picking adaptive management system comprising an environment monitoring module, a data analysis and prediction module, an automatic control module, an intelligent picking module, a pest and disease monitoring module and a pest and disease control module. The environment monitoring module is used for collecting orchard environment data and transmitting the orchard environment data to the data analysis and prediction module, and the data analysis and prediction module is used for preparing and preprocessing the data, determining a model order and establishing, checking and optimizing an ARIMA model for yield prediction. According to the invention, through the integrated ARIMA model, the orchard picking adaptive management system can carry out accurate yield prediction based on the historical time sequence data of the orchard; the problems of inaccurate orchard yield prediction, complex data processing and model establishment, difficult model optimization and selection, lack of basis for picking plan making and low orchard management efficiency can be solved.

Description

Technical Field

[0001] The invention relates to the technical field of orchard picking management systems, and in particular to an orchard picking adaptive management system. Background Art

[0002] Traditional orchards often lack scientific basis when formulating picking plans, resulting in improper selection of picking timing, affecting fruit quality and yield. The orchard picking management system of this application can solve the problems of inaccurate orchard yield prediction, complex data processing and model establishment, difficult model optimization and selection, lack of basis for picking plan formulation, and inefficient orchard management.

[0003] The defects of the existing orchard picking management system are:

[0004] 1. Patent document CN108782145B discloses an intelligent orchard management system. The document mainly considers how to accurately and effectively identify apples in the crown and remove suspicious obstacles in the early stage of picking. It does not consider how to solve the problems of inaccurate orchard yield prediction, complex data processing and model establishment, difficult model optimization and selection, lack of basis for picking plan formulation, and low orchard management efficiency.

[0005] 2. Patent document CN109573571B discloses an orchard management system, which mainly considers how to transport the picked fruits to the ground quickly and conveniently, but does not consider how to solve the problems of low efficiency of manual picking, insufficient accuracy of fruit recognition, low efficiency of fruit sorting and packaging, low level of intelligent orchard management, and fruit damage during picking;

[0006] 3. Patent document CN111428834B discloses an orchard spraying information management system and method based on blockchain and information fusion. The document mainly considers how to provide an orchard spraying information management system and method with high detection accuracy, low detection cost and high transparency of detection results, but does not consider how to solve the problems of inaccurate identification of pests and diseases, untimely warning, poor transmission of warning information and low level of intelligent management of orchard pests and diseases;

[0007] 4. Patent document CN116050699B discloses an orchard management data platform and its control method. The document mainly considers how to solve the problem of single function of systems and devices used for orchard crop planting and farmland management, but does not consider how to solve the problems of lack of systematicness and accuracy in pest control knowledge, improper selection of control methods, poor transmission and understanding of control recommendations, and lack of scientific basis for orchard pest control decisions. Summary of the invention

[0008] The object of the present invention is to provide an orchard picking adaptive management system to solve the problems raised in the above-mentioned background technology.

[0009] To achieve the above-mentioned object, the present invention provides the following technical solutions: an orchard picking adaptive management system, comprising an environmental monitoring module, a data analysis and prediction module, an automatic control module, an intelligent picking module, a pest monitoring module and a pest control module;

[0010] The environmental monitoring module is used to collect orchard environmental data and transmit it to the data analysis and prediction module. The data analysis and prediction module is used to prepare and preprocess data, determine the model order, establish and test the ARIMA model for yield prediction. The automation control module is used to generate and send picking equipment control instructions and monitor the working status of the equipment according to the analysis results. The intelligent picking module is used for image recognition, robotic arm control picking, and fruit sorting and packaging. The pest and disease monitoring module is used to shoot, preprocess, extract features and classify pests and diseases on fruit trees. The pest and disease control module is used to set early warning rules, generate and send early warning information, and generate and output prevention and control suggestions;

[0011] The environmental monitoring module includes a sensor unit, a data acquisition unit and a communication unit;

[0012] The data analysis and prediction module includes a data preparation and preprocessing unit, a model order determination unit, an ARIMA model establishment unit, a model testing and tuning unit, and a prediction and result analysis unit.

[0013] Preferably, the sensor unit includes an ambient temperature and humidity sensor, an ambient light sensor, a soil moisture sensor and a soil pH sensor, wherein the temperature and humidity sensor and the ambient light sensor are used to monitor the temperature, humidity and light intensity of the orchard, respectively, and the soil moisture sensor and the soil pH sensor are used to monitor the moisture content and pH value in the soil, respectively;

[0014] The data acquisition unit is used to receive the data collected by the sensor unit and convert it into digital signals for processing and storage;

[0015] The communication unit is used to transmit the processed data to the data analysis and prediction module.

[0016] Preferably, the data preparation and preprocessing unit includes a data preparation subunit, a data visualization subunit and a stationarity test subunit;

[0017] The data preparation subunit is used to collect historical time series data of the orchard;

[0018] The data visualization subunit is used to display data using charts and graphs to preliminarily determine the trend and periodicity of the data;

[0019] The stationarity test subunit is used to determine whether the data is stationary using ADF and other test methods. If it is not stationary, differential processing is performed;

[0020] The model order determination unit is used to preliminarily determine the values ​​of p and q by calculating ACF and PACF, and to determine the value of d by observing the difference results;

[0021] The ARIMA model building unit is used to build an ARIMA model for orchard yield prediction using the ARIMA function according to the determined model order;

[0022] The model testing and tuning unit is used to evaluate the model using residual test and goodness of fit test. If the residual test and goodness of fit test results show that the model is not suitable, the model order is re-determined to establish the model and the model test is re-performed to select the optimal model;

[0023] The forecasting and result analysis unit is used to use the verified ARIMA model to perform time series forecasting, calculate the confidence interval and forecast error of the forecast value, and formulate an orchard picking plan based on the forecast results.

[0024] Preferably, the expression of the ARIMA model is: ;

[0025] Among them, B is the backshift operator, , is the autoregressive coefficient, and p is the autoregressive order, which means that the model considers the impact of the observations at the past p time points on the current observations;

[0026] is the difference operator, d is the difference order, which is used to transform the non-stationary time series into a stationary time series;

[0027] , is the moving average coefficient, q is the moving average order, and the moving average part is used to deal with random fluctuations in the time series, indicating that the current forecast error is the white noise of the past q time points Linear combinations of

[0028] is a white noise sequence with a mean of 0 , the variance is constant , and the white noise at different time points are independent of each other , in the orchard yield prediction model, represents the random factors that the model cannot explain through the autoregression and moving average parts.

[0029] Preferably, the automation control module includes a control instruction generating unit, a device control unit and a status monitoring unit;

[0030] The control instruction generating unit is used to generate control instructions for the picking equipment according to the analysis results, and the control instructions include picking equipment start instructions, picking equipment stop instructions, picking path planning instructions, equipment parameter adjustment instructions, pesticide spraying instructions and fertilizer spraying instructions;

[0031] The equipment control unit is used to send control instructions to the picking equipment;

[0032] The status monitoring unit is used to monitor the working status of orchard picking equipment in real time.

[0033] Preferably, the intelligent picking module includes an image recognition unit, a robotic arm control unit and a fruit sorting unit;

[0034] The image recognition unit includes an image acquisition subunit, an image preprocessing subunit, a fruit feature extraction subunit and a fruit recognition and classification subunit;

[0035] The robot arm control unit is used to control the robot arm to pick according to the recognition result;

[0036] The fruit sorting unit is used to sort and pack the picked fruits.

[0037] Preferably, the image acquisition subunit includes a high-resolution color camera and a light compensation device, the high-resolution color camera is used to obtain images of fruit trees and fruits in the orchard, and the light compensation device is used to automatically adjust the brightness and color temperature according to the data of the ambient light sensor;

[0038] The image preprocessing subunit is used to perform image filtering and noise reduction, image enhancement processing and image geometric transformation processing on the image acquired by the image acquisition subunit;

[0039] The fruit feature extraction subunit is used to extract the geometric features and color features of the fruit using computer vision technology, and to extract the high-level abstract features of the fruit based on the CNN deep learning algorithm;

[0040] The fruit recognition and classification subunit is used to collect images of fruits of different varieties and different degrees of maturity as training data sets to train the classification model, establish a fruit feature database, and input the images after preprocessing and feature extraction into the trained classification model for real-time recognition and classification.

[0041] Preferably, the pest monitoring module includes a pest identification unit and an early warning unit;

[0042] The pest identification unit includes an image acquisition device, an image preprocessing subunit, a pest feature extraction subunit and a pest classification model.

[0043] The image acquisition device is used to capture the characteristics of pests and diseases on fruit trees. The image preprocessing subunit is used to crop, scale and enhance the acquired images. The pest and disease feature extraction subunit is used to extract the morphological characteristics of pests and diseases using computer vision technology, and automatically extract abstract features in the image based on the CNN deep learning algorithm. The pest and disease classification model is used to accurately classify pests and diseases into different categories based on the extracted features.

[0044] Preferably, the warning unit includes a warning rule setting subunit, a warning information generating subunit and a warning information sending subunit;

[0045] The early warning rule setting subunit is used to set the classification standards according to the types of pests and diseases and their potential harm to fruit tree growth, fruit yield and quality, set the early warning threshold of the pest and disease infection range in combination with the area of ​​the orchard and the planting density of fruit trees, and determine the occurrence time of pests and diseases according to the growth cycle of fruit trees;

[0046] The warning information generation subunit is used to generate warning messages according to the warning rules and the pest identification results. The formats of the warning messages include SMS warning, mobile application push and email;

[0047] The warning information sending subunit is used to determine the recipients of the warning information. Different recipients are set with different receiving permissions and notification methods. Orchard field workers receive SMS warnings, and managers and technicians receive mobile application push and emails through mobile applications and emails.

[0048] Preferably, the pest control module includes a pest knowledge base, a control suggestion generating unit and a control suggestion output unit;

[0049] The pest knowledge base is used to store the biological information of various pests and diseases, including the life habits of pests and the pathogenesis of diseases, as well as the prevention and control methods of different pests and diseases, including chemical control, biological control and physical control;

[0050] The control suggestion generation unit is used to retrieve the most suitable combination of control methods from the pest knowledge base according to the pest types determined by the pest identification unit;

[0051] The prevention and control suggestion output unit is used to output the prevention and control suggestions in the format of text reports, operation guides and charts, and send them to orchard on-site workers, managers and technicians through the early warning information sending unit.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] 1. The invention uses an integrated ARIMA model, and the orchard picking adaptive management system can accurately predict the yield based on the historical time series data of the orchard. The ARIMA model takes into account the time series characteristics of the data, including trend, seasonality and periodicity, thereby improving the accuracy of the prediction. The model testing and tuning unit evaluates the established ARIMA model through residual test and goodness of fit test. If the model is not suitable, the model order is re-determined and the model is re-established until the optimal model is found, thereby ensuring the accuracy and reliability of the model. The prediction and result analysis unit uses the ARIMA model that has passed the test to perform time series prediction. It can predict the time series and calculate the confidence interval and prediction error of the predicted value. These prediction results can provide a scientific basis for orchard managers, help them formulate reasonable picking plans, and avoid losses caused by over-ripe or premature picking of fruits. Through automated data analysis and prediction, the system can update the orchard's yield forecast information in real time, help orchard managers respond quickly to market changes, adjust picking strategies, and thus improve the orchard's overall management efficiency and profitability. Therefore, it can solve the problems of inaccurate orchard yield forecasts, complex data processing and model building, difficult model optimization and selection, lack of basis for picking plan formulation, and inefficient orchard management.

[0054] 2. The present invention can significantly improve picking efficiency and reduce labor costs by introducing an intelligent picking module. The robotic arm control unit can automatically pick fruits according to the recognition results of the image recognition unit, greatly reducing the labor burden. During the orchard picking process, the image recognition unit obtains high-quality images through a high-resolution color camera and a light compensation device, and uses computer vision technology and CNN deep learning algorithms to extract multiple features of fruits, thereby achieving high-precision fruit recognition and classification, which helps to avoid picking unripe or overripe fruits and improve the quality and commercial value of fruits. Traditional fruit sorting and packaging processes often rely on manual labor, which is inefficient and prone to errors. The fruit sorting unit can automatically sort and package the picked fruits according to the variety, size, and maturity of the fruits. At the same time, traditional orchard management lacks intelligent means, and it is difficult to grasp the production situation of the orchard in real time. By introducing the intelligent picking module, orchard managers can obtain information such as the picking situation, quality and quantity of fruits in real time, providing strong support for the precise management and decision-making of the orchard. In addition, during manual picking, improper operation or lack of experience may cause damage to the fruits. The intelligent picking module can minimize fruit damage and improve the integrity and aesthetics of the fruits by precisely controlling the robotic arm for picking. Therefore, it can solve the problems of low efficiency of manual picking, insufficient accuracy of fruit recognition, low efficiency of fruit sorting and packaging, low level of intelligent orchard management and fruit damage during picking.

[0055] 3. The present invention integrates image acquisition equipment, image preprocessing subunit, pest feature extraction subunit and pest classification model, and the orchard picking adaptive management system can automatically, accurately and quickly identify pests and diseases in the orchard, greatly improving the accuracy and efficiency of identification. The occurrence of orchard pests and diseases is often sudden and unpredictable, and traditional early warning methods are often difficult to achieve timely and accurate early warning. Through the early warning rule setting subunit, the orchard picking adaptive management system can set reasonable early warning thresholds and early warning rules according to information such as the type of pests and diseases, the degree of damage, the orchard area, the fruit tree planting density and the fruit tree growth cycle, so as to achieve graded early warning and targeted early warning of pests and diseases. Traditional early warning information transmission methods also have the problems of poor transmission and reception. The problem of untimely collection makes it impossible for orchard managers and workers to take effective countermeasures in time. Through the early warning information generation subunit and the early warning information sending subunit, the orchard picking adaptive management system can set different receiving permissions and notification methods according to different receiving objects to ensure that the early warning information can be accurately and timely delivered to relevant personnel, providing strong support for the timely prevention and control of orchard pests and diseases. By introducing the pest and disease monitoring module, orchard managers can grasp the occurrence of orchard pests and diseases in real time, providing strong support for the precise management and decision-making of the orchard, and at the same time improving the intelligent level of orchard pest and disease management. Therefore, it can solve the problems of inaccurate pest and disease identification, untimely early warning, poor transmission of early warning information and low level of intelligent orchard pest and disease management.

[0056] 4. The present invention provides comprehensive and detailed biological information and control methods of pests and diseases by setting up a pest control module and integrating a pest and disease knowledge base, thereby providing orchard managers and workers with a scientific and reliable basis for control. Different pests and diseases may require different control methods, while traditional control methods often lack pertinence, which easily leads to poor control effects or unnecessary environmental pollution. Through the control suggestion generation unit, the most suitable combination of control methods can be retrieved from the set knowledge base according to the identified pest and disease types, so as to provide orchard managers and workers with personalized control suggestions. In addition, the traditional control suggestion transmission method has the problems of poor information transmission and difficulty in understanding, which makes it impossible for orchard managers and workers to accurately understand and implement the control suggestions. Through the control suggestion output unit, the control suggestions can be output in the form of text reports and operation guides. The prevention and control suggestions are output in the format of tables and graphs, making them more intuitive and easy to understand. At the same time, the early warning information sending unit can ensure that the prevention and control suggestions are delivered to the orchard workers, managers and technicians in a timely and accurate manner, thereby improving the prevention and control efficiency. In addition, orchard managers often lack scientific basis when making decisions on pest and disease control, and are prone to make decisions based on experience or subjective judgment, resulting in poor prevention and control effects or waste of resources. By integrating the pest and disease control module, scientific and systematic pest and disease control decision-making support can be provided to orchard managers, helping them to formulate prevention and control plans more accurately, improve prevention and control effects and resource utilization efficiency, and thus solve the problems of lack of systematicness and accuracy in pest and disease control knowledge, improper selection of prevention and control methods, poor transmission and understanding of prevention and control suggestions, and lack of scientific basis for orchard pest and disease control decisions. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the present invention are described clearly and completely. Obviously, the described embodiments 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 creative work are within the scope of protection of the present invention.

[0058] Embodiment 1:

[0059] An orchard picking adaptive management system, comprising an environmental monitoring module, a data analysis and prediction module, an automatic control module, an intelligent picking module, a pest monitoring module and a pest control module;

[0060] The environmental monitoring module is used to collect orchard environmental data and transmit it to the data analysis and prediction module. The data analysis and prediction module is used to prepare and preprocess data, determine the model order, establish and test the ARIMA model for yield prediction. The automation control module is used to generate and send picking equipment control instructions and monitor the working status of the equipment according to the analysis results. The intelligent picking module is used for image recognition, robotic arm control picking, and fruit sorting and packaging. The pest and disease monitoring module is used to shoot, preprocess, extract features and classify pests and diseases on fruit trees. The pest and disease control module is used to set early warning rules, generate and send early warning information, and generate and output prevention and control suggestions;

[0061] The environmental monitoring module includes a sensor unit, a data acquisition unit and a communication unit;

[0062] The sensor unit includes an ambient temperature and humidity sensor, an ambient light sensor, a soil moisture sensor, and a soil pH sensor. The temperature and humidity sensor and the ambient light sensor are used to monitor the temperature, humidity, and light intensity of the orchard, respectively. The soil moisture sensor and the soil pH sensor are used to monitor the moisture content and pH value in the soil, respectively.

[0063] The data acquisition unit is used to receive the data collected by the sensor unit and convert it into digital signals for processing and storage;

[0064] The communication unit is used to transmit the processed data to the data analysis and prediction module.

[0065] Furthermore, the environmental temperature and humidity sensor monitors the temperature and humidity in the orchard in real time, providing key environmental parameters for crop growth. The environmental light sensor is used to measure the light intensity in the orchard, reflecting the exposure of natural light and artificial light. The soil moisture sensor monitors the moisture content in the soil, providing a basis for irrigation management. The soil pH sensor is used to measure the acidity and alkalinity of the soil, reflecting the soil fertility.

[0066] The data acquisition unit receives data from the sensor unit and performs preprocessing, including signal amplification, filtering, and analog-to-digital conversion. The data acquisition unit supports real-time data processing and long-term storage, realizes real-time data collection and preliminary processing, and provides a reliable data source for subsequent data analysis and prediction. The communication unit transmits the processed data to the data analysis and prediction module to realize remote access and sharing of data, thereby realizing remote monitoring and real-time early warning of the orchard environment and improving management efficiency.

[0067] The data collected by the sensor unit is preprocessed by the data acquisition unit and then transmitted to the data analysis and prediction module by the communication unit. The data analysis and prediction module receives the data from the environmental monitoring module for further analysis and prediction.

[0068] Embodiment 2:

[0069] An orchard picking adaptive management system, wherein a data analysis and prediction module comprises a data preparation and preprocessing unit, a model order determination unit, an ARIMA model establishment unit, a model testing and tuning unit, and a prediction and result analysis unit.

[0070] The data preparation and preprocessing unit includes a data preparation subunit, a data visualization subunit, and a stationarity test subunit;

[0071] The data preparation subunit is used to collect historical time series data of the orchard;

[0072] The data visualization subunit is used to display data using charts and graphs to preliminarily determine the trend and periodicity of the data;

[0073] The stationarity test subunit is used to determine whether the data is stationary using ADF and other test methods. If it is not stationary, differential processing is performed;

[0074] The model order determination unit is used to preliminarily determine the values ​​of p and q by calculating ACF and PACF, and to determine the value of d by observing the difference results;

[0075] The ARIMA model building unit is used to build an ARIMA model for orchard yield prediction using the ARIMA function according to the determined model order;

[0076] The model testing and tuning unit is used to evaluate the model using residual test and goodness of fit test. If the residual test and goodness of fit test results show that the model is not suitable, the model order is re-determined to establish the model and the model test is re-performed to select the optimal model;

[0077] The forecasting and result analysis unit is used to use the verified ARIMA model to perform time series forecasting, calculate the confidence interval and forecast error of the forecast value, and formulate an orchard picking plan based on the forecast results.

[0078] Furthermore, the data preparation subunit is used to collect historical time series data of the orchard, including annual yield, temperature, humidity, rainfall, soil data, etc. These data are the basis for subsequent analysis and prediction, and are crucial to the accuracy and reliability of the model. The data visualization subunit uses charts to display data, including line charts, bar charts, and scatter plots, etc., to preliminarily determine the trend and periodicity of the data. Visualization helps to intuitively understand the data characteristics and provide clues for subsequent analysis. The stationarity test subunit uses the ADF test method to determine whether the data is stationary. If the data is not stationary, differential processing is performed until the data meets the stationarity requirements. Stationarity is the premise assumption of the ARIMA model. Data that does not meet stationarity may lead to inaccurate model prediction results.

[0079] The model order determination unit preliminarily determines the values ​​of p and q by calculating ACF and PACF, and determines the value of d by observing the difference results. ACF is the autocorrelation coefficient, and PACF is the partial autocorrelation coefficient. By observing the tailing and truncation characteristics of the ACF and PACF graphs, the possible values ​​of p and q are preliminarily determined. According to the stationarity test results of the data, the difference order d is determined;

[0080] The ARIMA model building unit uses the ARIMA function to build the ARIMA model for orchard yield prediction according to the determined model order. The model testing and tuning unit uses the residual test and the goodness of fit test to evaluate the model. If the residual test and the goodness of fit test results show that the model is not suitable, the model order is re-determined to build the model and the model test is re-performed until the optimal model is found.

[0081] The forecasting and result analysis unit uses the verified ARIMA model to perform time series forecasting, calculates the confidence interval and forecast error of the forecast value, formulates orchard picking plans based on the forecast results, such as determining the picking time and picking amount, analyzes the accuracy and reliability of the forecast results, and provides decision support for orchard management.

[0082] Embodiment three:

[0083] An adaptive management system for orchard picking, the expression of the ARIMA model is: ;

[0084] Among them, B is the backshift operator, , is the autoregressive coefficient, and p is the autoregressive order, which means that the model considers the impact of the observations at the past p time points on the current observations;

[0085] is the difference operator, d is the difference order, which is used to transform the non-stationary time series into a stationary time series;

[0086] , is the moving average coefficient, q is the moving average order, and the moving average part is used to deal with random fluctuations in the time series, indicating that the current forecast error is the white noise of the past q time points Linear combinations of

[0087] is a white noise sequence with a mean of 0 , the variance is constant , and the white noise at different time points are independent of each other , in the orchard yield prediction model, represents the random factors that the model cannot explain through the autoregression and moving average parts.

[0088] Furthermore, in the orchard picking adaptive management system, if p=2, and Indicates the linear impact of the orchard yield in the past two time periods on the current yield. =0.5, which means that for every 1 unit increase in output in the previous cycle, output in the current cycle is expected to increase by 0.5 units. This effect is based on the autocorrelation in the data.

[0089] As new fruit trees are planted in an orchard, the yield increases year by year. This trend can be eliminated by performing a differential operation. If d=1, then $(1-B)y_t=y_t-y_{t-1}$, which is the difference in yield between adjacent time points.

[0090] If q=1, it indicates the degree of influence of the random error of the previous period on the prediction error of the current period.

[0091] Embodiment 4:

[0092] An orchard picking adaptive management system, the automation control module includes a control instruction generation unit, an equipment control unit and a status monitoring unit;

[0093] The control instruction generating unit is used to generate control instructions for the picking equipment according to the analysis results, and the control instructions include picking equipment start instructions, picking equipment stop instructions, picking path planning instructions, equipment parameter adjustment instructions, pesticide spraying instructions and fertilizer spraying instructions;

[0094] The equipment control unit is used to send control instructions to the picking equipment;

[0095] The status monitoring unit is used to monitor the working status of orchard picking equipment in real time.

[0096] Furthermore, the control instruction generation unit generates a start instruction according to the picking demand of the orchard and the state of the picking equipment to ensure that the equipment starts working in the best state. When the picking task is completed or the equipment fails, a stop instruction is generated to protect the equipment and the orchard. The control instruction generation unit can also generate an optimal picking path based on the map information of the orchard, the distribution of fruit trees and the maturity of the fruit to improve the picking efficiency. The control instruction generation unit can also dynamically adjust the parameters of the picking equipment according to the orchard environment and picking demand to ensure the picking quality. The control instruction generation unit can also generate pesticide spraying instructions according to the pest and disease monitoring results and the pesticide use plan to ensure the health of the orchard. The control instruction generation unit can also generate fertilizer spraying instructions according to the soil nutrient analysis results and crop growth requirements to promote crop growth;

[0097] The equipment control unit sends control instructions to the picking equipment wirelessly or wired to ensure accurate transmission of the instructions. It can also receive response signals from the picking equipment to confirm that the picking equipment has correctly executed the instructions. When a fault occurs in the picking equipment, it can automatically or manually trigger the fault handling process, including alarm, shutdown and restart.

[0098] The condition monitoring unit uses sensors and data analysis technology to monitor the working status of the picking equipment in real time, including operating speed, picking efficiency and energy consumption, and then uses machine learning algorithms to detect abnormal conditions of the equipment, including overheating, overload and failure, and generate alarm information. At the same time, it can conduct in-depth analysis of the monitoring data and extract key indicators to provide data support for the generation and optimization of control instructions. It also provides an intuitive visual interface to display the real-time status and historical data of the orchard picking equipment, which facilitates remote monitoring and decision-making by orchard managers.

[0099] Embodiment five:

[0100] An orchard picking adaptive management system, the intelligent picking module includes an image recognition unit, a mechanical arm control unit and a fruit sorting unit;

[0101] The image recognition unit includes an image acquisition subunit, an image preprocessing subunit, a fruit feature extraction subunit and a fruit recognition and classification subunit;

[0102] The image acquisition subunit includes a high-resolution color camera and a light compensation device. The high-resolution color camera is used to obtain images of fruit trees and fruits in the orchard. The light compensation device is used to automatically adjust the brightness and color temperature according to the data of the ambient light sensor.

[0103] The image preprocessing subunit is used to perform image filtering and noise reduction, image enhancement processing and image geometric transformation processing on the image acquired by the image acquisition subunit;

[0104] The fruit feature extraction subunit is used to extract the geometric features and color features of the fruit using computer vision technology, and to extract the high-level abstract features of the fruit based on the CNN deep learning algorithm;

[0105] The fruit recognition and classification subunit is used to collect images of fruits of different varieties and different degrees of maturity as training data sets to train the classification model, establish a fruit feature database, and input the images after preprocessing and feature extraction into the trained classification model for real-time recognition and classification.

[0106] The robot arm control unit is used to control the robot arm to pick according to the recognition result;

[0107] The fruit sorting unit is used to sort and pack the picked fruits.

[0108] Furthermore, the high-resolution color camera of the image acquisition subunit has a high dynamic range and wide angle, which can capture clear images of fruit trees and fruits in the orchard. The illumination compensation device integrates an ambient light sensor to automatically adjust the brightness and color temperature of LED lights or other light sources according to real-time illumination data to ensure the quality of image acquisition;

[0109] When the image preprocessing subunit performs image filtering and noise reduction, it applies Gaussian filtering and mean filtering algorithms to remove noise and interference in the image, then uses histogram equalization and contrast stretching technology to enhance the details and contrast of the image, and finally performs operations such as image rotation, scaling and cropping to correct geometric distortion in the image;

[0110] The fruit feature extraction subunit first uses traditional computer vision technology to extract features such as shape, size, edge and color of the fruit. Then, based on the CNN deep learning algorithm, it extracts high-level features of the fruit, including texture and surface defects, to improve the accuracy and robustness of recognition.

[0111] The fruit recognition and classification subunit collects fruit images of different varieties, different maturity and different lighting conditions, builds a rich training data set, uses the classification model to train the training data set, establishes a fruit feature database and classification model, inputs the preprocessed and feature-extracted images into the trained classification model, performs real-time recognition and classification, and can output the variety, maturity and location information of the fruit;

[0112] The robot control unit plans the movement path of the robot according to the output results of the fruit recognition and classification subunit to ensure that the robot can accurately reach the fruit position. The robot uses the end effector of the robot to accurately pick the fruit according to its shape and size. At the same time, the robot can also integrate sensors to monitor the environment around the robot in real time to avoid collisions with fruit trees, other robots or obstacles.

[0113] The fruit sorting unit uses machine vision or sensor technology to inspect the quality of picked fruits, including size, shape, color, surface defects, etc. According to the quality inspection results, the fruits are classified into different grades, and corresponding sorting strategies are formulated. The sorted fruits are packaged and identification information is added to the packaging for subsequent management and sales.

[0114] Embodiment six:

[0115] An orchard picking adaptive management system, wherein the pest and disease monitoring module includes a pest and disease identification unit and an early warning unit;

[0116] The pest identification unit includes an image acquisition device, an image preprocessing subunit, a pest feature extraction subunit and a pest classification model.

[0117] The image acquisition device is used to capture the characteristics of pests and diseases on fruit trees. The image preprocessing subunit is used to crop, scale and enhance the acquired images. The pest and disease feature extraction subunit is used to extract the morphological characteristics of pests and diseases using computer vision technology, and automatically extract abstract features in the image based on the CNN deep learning algorithm. The pest and disease classification model is used to accurately classify pests and diseases into different categories based on the extracted features.

[0118] The early warning unit includes an early warning rule setting subunit, an early warning information generating subunit and an early warning information sending subunit;

[0119] The early warning rule setting subunit is used to set the classification standards according to the types of pests and diseases and their potential harm to fruit tree growth, fruit yield and quality, set the early warning threshold of the pest and disease infection range in combination with the area of ​​the orchard and the planting density of fruit trees, and determine the occurrence time of pests and diseases according to the growth cycle of fruit trees;

[0120] The warning information generation subunit is used to generate warning messages according to the warning rules and the pest identification results. The formats of the warning messages include SMS warning, mobile application push and email;

[0121] The warning information sending subunit is used to determine the recipients of the warning information. Different recipients are set with different receiving permissions and notification methods. Orchard field workers receive SMS warnings, and managers and technicians receive mobile application push and emails through mobile applications and emails.

[0122] Furthermore, the image acquisition equipment in the pest and disease identification unit uses a high-resolution and low-noise camera to ensure that the captured pest and disease images are clear and rich in details. The camera has the characteristics of waterproof, dustproof and sun-proof to adapt to the complex and changeable environment of the orchard. In addition to basic processing such as cropping, scaling and enhancement, the image preprocessing subunit can also perform advanced processing such as image denoising and color correction to improve image quality. At the same time, image registration technology is introduced to align images of the same pest and disease taken at different times or angles to facilitate subsequent feature extraction and classification.

[0123] The pest and disease feature extraction subunit first uses edge detection and contour extraction technology to extract features such as shape, size, and texture of pests and diseases. Then, based on the CNN deep learning algorithm, it automatically learns and extracts high-level features in the image, including local details and color distribution of pests and diseases, and integrates morphological features and abstract features to form a more comprehensive and discriminative feature vector for subsequent pest and disease classification.

[0124] The pest and disease classification model adopts transfer learning and few-sample learning techniques to improve the generalization ability and accuracy of the model in orchard pest and disease classification tasks. It introduces an online learning mechanism to continuously update and optimize the model based on newly collected pest and disease images and labels to adapt to changes in orchard pest and disease types and new identification needs.

[0125] The early warning rule setting sub-unit in the early warning unit sets different early warning levels, including mild, moderate and severe, according to factors such as the type of pests and diseases, the degree of damage and the speed of spread. At the same time, it sets early warning thresholds based on the area of ​​the orchard, the planting density of fruit trees and the historical occurrence of pests and diseases to trigger corresponding early warning measures. It also predicts the occurrence time of pests and diseases based on factors such as the growth cycle of fruit trees and climatic conditions, and makes early warning preparations in advance.

[0126] In addition to SMS warnings, mobile application push notifications and emails, the warning message format of the warning information generation subunit also includes voice broadcasts and on-site display screens to meet the warning needs in different scenarios. The warning content includes the type of pests and diseases, the degree of damage, the warning level and the recommended prevention and control measures, so that orchard managers and workers can quickly understand and take action.

[0127] The recipients of the early warning information sending subunit determine the receiving rights and notification methods of different recipients according to the organizational structure and division of responsibilities of the orchard. In addition to text messages, mobile applications and e-mails, an internal communication platform or group chat can be established in the orchard to transmit early warning information more quickly and directly. A feedback mechanism for early warning information can also be established to collect information such as the orchard managers and workers' reception of early warning information, the prevention and control measures taken, and the prevention and control effects, so as to continuously optimize and improve the early warning system.

[0128] Embodiment seven:

[0129] An orchard picking adaptive management system, the pest control module includes a pest knowledge base, a control suggestion generation unit and a control suggestion output unit;

[0130] The pest knowledge base is used to store the biological information of various pests and diseases, including the life habits of pests and the pathogenesis of diseases, as well as the prevention and control methods of different pests and diseases, including chemical control, biological control and physical control;

[0131] The control suggestion generation unit is used to retrieve the most suitable combination of control methods from the pest knowledge base according to the pest types determined by the pest identification unit;

[0132] The prevention and control suggestion output unit is used to output the prevention and control suggestions in the format of text reports, operation guides and charts, and send them to orchard on-site workers, managers and technicians through the early warning information sending unit.

[0133] Furthermore, the pest knowledge base stores a large amount of biological information on pests and diseases, including pest life habits and disease pathogenesis, records the pest reproduction cycle, activity patterns, preferred plants and migration paths, etc., explains the pathogens of the disease, infection routes, disease conditions and symptoms, etc., and prevention and control methods include chemical control, biological control and physical control. A knowledge review mechanism can also be established to ensure that the stored pest control methods are scientific, effective, safe and accurate;

[0134] The prevention and control suggestion generation unit identifies the types of pests and diseases, receives the output results of the pest and disease identification unit, and confirms the types and degree of damage of the pests and diseases. If there is uncertainty in the identification results, an expert consultation or manual review mechanism can be initiated;

[0135] According to the types and severity of pests and diseases, appropriate prevention and control methods are retrieved from the pest and disease knowledge base. At the same time, factors such as the ecological environment of the orchard, crop types and growth cycle need to be considered to select the best combination of prevention and control methods. After the prevention and control is completed, the cost-effectiveness, environmental impact and sustainability of the prevention and control methods are evaluated to ensure the rationality and feasibility of the prevention and control recommendations;

[0136] According to the orchard's historical prevention records, weather conditions, soil conditions and other factors, the prevention and control plan is personalized and optimized, and machine learning algorithms are introduced to predict the prevention and control effects and adjust the prevention and control strategy based on historical data and real-time data;

[0137] The prevention and control suggestion output unit provides detailed prevention and control suggestions, including the types of pests and diseases, degree of damage, prevention and control methods and precautions, etc. Through the early warning information sending unit, the prevention and control suggestions are sent to the orchard site workers, managers and technicians in the form of text messages, emails, mobile application push, etc., to ensure the accuracy and timeliness of the information, provide clear guidance and operational suggestions, and collect feedback and practical effects from the orchard site workers, managers and technicians on the prevention and control suggestions. Based on the feedback and practical effects, the prevention and control suggestions are continuously optimized and adjusted to improve the prevention and control effects and management level.

[0138] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, the embodiments should be considered exemplary and non-restrictive in all respects, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims be included in the present invention.

Claims

1. An orchard picking adaptive management system, characterized in that: It includes environmental monitoring module, data analysis and prediction module, automation control module, intelligent picking module, pest and disease monitoring module and pest and disease control module; The environmental monitoring module is used to collect orchard environmental data and transmit it to the data analysis and prediction module. The data analysis and prediction module is used to prepare and preprocess data, determine the model order, establish and test the ARIMA model for yield prediction. The automation control module is used to generate and send picking equipment control instructions and monitor the working status of the equipment according to the analysis results. The intelligent picking module is used for image recognition, robotic arm control picking, and fruit sorting and packaging. The pest and disease monitoring module is used to shoot, preprocess, extract features and classify pests and diseases on fruit trees. The pest and disease control module is used to set early warning rules, generate and send early warning information, and generate and output prevention and control suggestions; The environmental monitoring module includes a sensor unit, a data acquisition unit and a communication unit; The data analysis and prediction module includes a data preparation and preprocessing unit, a model order determination unit, an ARIMA model establishment unit, a model testing and tuning unit, and a prediction and result analysis unit.

2. The orchard picking adaptive management system according to claim 1, characterized in that: The sensor unit includes an ambient temperature and humidity sensor, an ambient light sensor, a soil moisture sensor, and a soil pH sensor. The temperature and humidity sensor and the ambient light sensor are used to monitor the temperature, humidity, and light intensity of the orchard, respectively. The soil moisture sensor and the soil pH sensor are used to monitor the moisture content and pH value in the soil, respectively. The data acquisition unit is used to receive the data collected by the sensor unit and convert it into digital signals for processing and storage; The communication unit is used to transmit the processed data to the data analysis and prediction module.

3. The orchard picking adaptive management system according to claim 1, characterized in that: The data preparation and preprocessing unit includes a data preparation subunit, a data visualization subunit, and a stationarity test subunit; The data preparation subunit is used to collect historical time series data of the orchard; The data visualization subunit is used to display data using charts and graphs to preliminarily determine the trend and periodicity of the data; The stationarity test subunit is used to determine whether the data is stationary using ADF and other test methods. If it is not stationary, differential processing is performed; The model order determination unit is used to preliminarily determine the values ​​of p and q by calculating ACF and PACF, and to determine the value of d by observing the difference results; The ARIMA model building unit is used to build an ARIMA model for orchard yield prediction using the ARIMA function according to the determined model order; The model testing and tuning unit is used to evaluate the model using residual test and goodness of fit test. If the residual test and goodness of fit test results show that the model is not suitable, the model order is re-determined to establish the model and the model test is re-performed to select the optimal model. The forecasting and result analysis unit is used to use the verified ARIMA model to perform time series forecasting, calculate the confidence interval and forecast error of the forecast value, and formulate an orchard picking plan based on the forecast results.

4. The orchard picking adaptive management system according to claim 3, characterized in that: The expression of the ARIMA model is: ; Among them, B is the backshift operator, , is the autoregressive coefficient, and p is the autoregressive order, which means that the model considers the impact of the observations at the past p time points on the current observations; is the difference operator, d is the difference order, which is used to transform the non-stationary time series into a stationary time series; , is the moving average coefficient, q is the moving average order, and the moving average part is used to deal with random fluctuations in the time series, indicating that the current prediction error is the white noise of the past q time points Linear combinations of is a white noise sequence with a mean of 0 , the variance is constant , and the white noise at different time points are independent of each other , in the orchard yield prediction model, represents the random factors that the model cannot explain through the autoregression and moving average parts.

5. The orchard picking adaptive management system according to claim 1, characterized in that: The automation control module includes a control instruction generation unit, an equipment control unit and a status monitoring unit; The control instruction generating unit is used to generate control instructions for the picking equipment according to the analysis results, and the control instructions include picking equipment start instructions, picking equipment stop instructions, picking path planning instructions, equipment parameter adjustment instructions, pesticide spraying instructions and fertilizer spraying instructions; The equipment control unit is used to send control instructions to the picking equipment; The status monitoring unit is used to monitor the working status of orchard picking equipment in real time.

6. The orchard picking adaptive management system according to claim 1, characterized in that: The intelligent picking module includes an image recognition unit, a robotic arm control unit, and a fruit sorting unit; The image recognition unit includes an image acquisition subunit, an image preprocessing subunit, a fruit feature extraction subunit and a fruit recognition and classification subunit; The robot arm control unit is used to control the robot arm to pick according to the recognition result; The fruit sorting unit is used to sort and pack the picked fruits.

7. The orchard picking adaptive management system according to claim 6, characterized in that: The image acquisition subunit includes a high-resolution color camera and a light compensation device. The high-resolution color camera is used to obtain images of fruit trees and fruits in the orchard. The light compensation device is used to automatically adjust the brightness and color temperature according to the data of the ambient light sensor. The image preprocessing subunit is used to perform image filtering and noise reduction, image enhancement processing and image geometric transformation processing on the image acquired by the image acquisition subunit; The fruit feature extraction subunit is used to extract the geometric features and color features of the fruit using computer vision technology, and to extract the high-level abstract features of the fruit based on the CNN deep learning algorithm; The fruit recognition and classification subunit is used to collect images of fruits of different varieties and different degrees of maturity as training data sets to train the classification model, establish a fruit feature database, and input the images after preprocessing and feature extraction into the trained classification model for real-time recognition and classification.

8. The orchard picking adaptive management system according to claim 1, characterized in that: The pest and disease monitoring module includes a pest and disease identification unit and an early warning unit; The pest identification unit includes an image acquisition device, an image preprocessing subunit, a pest feature extraction subunit and a pest classification model. The image acquisition device is used to capture the characteristics of pests and diseases on fruit trees. The image preprocessing subunit is used to crop, scale and enhance the acquired images. The pest and disease feature extraction subunit is used to extract the morphological characteristics of pests and diseases using computer vision technology, and automatically extract abstract features in the image based on the CNN deep learning algorithm. The pest and disease classification model is used to accurately classify pests and diseases into different categories based on the extracted features.

9. The orchard picking adaptive management system according to claim 8, characterized in that: The early warning unit includes an early warning rule setting subunit, an early warning information generating subunit and an early warning information sending subunit; The early warning rule setting subunit is used to set the classification standards according to the types of pests and diseases and their potential harm to fruit tree growth, fruit yield and quality, set the early warning threshold of the pest and disease infection range in combination with the area of ​​the orchard and the planting density of fruit trees, and determine the occurrence time of pests and diseases according to the growth cycle of fruit trees; The warning information generation subunit is used to generate warning messages according to the warning rules and the pest identification results. The formats of the warning messages include SMS warning, mobile application push and email; The warning information sending subunit is used to determine the recipients of the warning information. Different recipients are set with different receiving permissions and notification methods. Orchard field workers receive SMS warnings, and managers and technicians receive mobile application push and emails through mobile applications and emails.

10. The orchard picking adaptive management system according to claim 1, characterized in that: The pest control module includes a pest knowledge base, a control suggestion generation unit, and a control suggestion output unit; The pest knowledge base is used to store the biological information of various pests and diseases, including the life habits of pests and the pathogenesis of diseases, as well as the prevention and control methods of different pests and diseases, including chemical control, biological control and physical control; The control suggestion generation unit is used to retrieve the most suitable combination of control methods from the pest knowledge base according to the pest types determined by the pest identification unit; The prevention and control suggestion output unit is used to output the prevention and control suggestions in the format of text reports, operation guides and charts, and send them to orchard on-site workers, managers and technicians through the early warning information sending unit.

Images