A monitoring and management system for tea garden pests and growth conditions

By deploying cameras and environmental acquisition modules in tea gardens, combined with deep neural network models, accurate monitoring and prediction of tea tree growth status and pests and diseases are achieved, the problems of low irrigation and fertilizer application and deviation of pests and diseases in traditional tea garden management systems are solved, and the problems of low accuracy of irrigation and fertilizer application and pest prediction deviations are improved. The stability of tea yield and quality.

CN116486253BActive Publication Date: 2024-11-19CHONGQING ACAD OF AGRI SCI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310215004.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-03-08
Publication Date
2024-11-19
Estimated Expiration
2043-03-08

AI Technical Summary

Technical Problem

Due to differences in regional and growth stages, traditional tea garden management systems have low accuracy of irrigation or fertilizer application, which can easily lead to reduced yield of tea trees, over-eutrophication of soil and pest prediction deviations, making it difficult to meet the requirements of intelligent and ecological management.

Method used

A monitoring and management system including camera module, environmental acquisition module, growth state quantization module, pre-regulation module, pest and disease warning module and self-update module was designed. The deep neural network model is self-supervised and adaptively adjusted to accurately monitor tea tree growth. Status and pests and diseases to achieve precise control of irrigation and fertilization.

Benefits of technology

It improves the stability of tea yield and quality, avoids errors in irrigation and fertilizer application, timely prevents pests and diseases, ensures high accuracy and adaptability of the system, and reduces labor costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116486253B_ABST
    Figure CN116486253B_ABST
Patent Text Reader

Abstract

The present invention provides a monitoring and management system for tea garden pests and growth conditions, including a camera module, an environment acquisition module, a growth state quantification module, a pre-regulation module, a pest warning module and a self-update module; the camera module is set up above the tea shed, the environment acquisition module is evenly distributed inside the tea garden, the growth state quantification module is connected to the camera module, the pre-regulation module is respectively connected to the environment acquisition module and the growth state quantification module, the pest warning module is respectively connected to the camera module and the growth state quantification module, and the self-update module is respectively connected to the growth state quantification module and the pre-regulation module. The monitoring and management system can perform self-supervision according to the different growth conditions of tea trees, complete adaptive adjustment and update, and the monitoring and management system has high accuracy and small error.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of tea planting, and in particular to a monitoring and management system for tea garden pests and diseases and growth conditions. Background Art

[0002] Tea has been discovered and used for thousands of years and has become one of the important beverages in my country. Scientific management of tea is the fundamental guarantee of tea quality and safety, and will directly affect the yield and taste of tea. As consumers have higher requirements for tea quality, the management of tea gardens has become more important.

[0003] Traditional tea gardens rely on a large number of manual labor for regular monitoring and management, which greatly wastes manpower and material resources. Intelligent tea gardens rely on image recognition technology to monitor tea gardens in real time, such as: identification of pests and diseases in tea gardens, weed identification, monitoring of tea tree growth status, etc., and cooperate with mechanized systems to complete the management of tea gardens, which helps to reduce labor costs, increase tea production, and improve tea quality. However, the growth of tea trees varies according to the planting environment, that is, tea trees in different regions have certain differences. At the same time, tea trees in the same region have different planting conditions due to different tea varieties and growth stages. Therefore, if a unified and fixed management system is used to apply irrigation or fertilizer to tea trees, or to predict pests and diseases, the differences in regions or growth stages may lead to too much or too little input into the tea garden. The irrigation or fertilizer amounts have low accuracy and large errors, which may easily lead to reduced production of tea trees, over-nutrition of the soil, and destruction of the ecological balance. This does not meet the requirements of intelligent and ecological tea garden management. At the same time, there will be deviations in the prediction of pests and diseases, which may easily lead to large-scale occurrence of pests and diseases and reduce tea production. Summary of the invention

[0004] In view of the problems existing in the above-mentioned prior art, the purpose of the present invention is to provide a monitoring and management system for tea garden pests and diseases and growth conditions, which can perform self-supervision according to the different growth conditions of tea trees, and complete adaptive adjustment and updating, so as to avoid the prediction deviation (i.e. irrigation, fertilization amount and pest and disease prediction) caused by different growth stages of tea trees or different planting environments, ensure the high accuracy and small error of the monitoring and management system, and thus guarantee the yield and quality of tea.

[0005] The purpose of the present invention is achieved through the following technical solutions:

[0006] A monitoring and management system for tea garden pests and growth conditions, characterized by comprising a camera module, an environment acquisition module, a growth status quantification module, a pre-regulation module, a pest warning module and a self-update module;

[0007] The camera module is mounted above the tea tent to obtain images of the tea trees;

[0008] The environmental collection modules are distributed inside the tea garden and are used to monitor environmental factors in the tea garden and obtain environmental monitoring parameters;

[0009] The growth state quantification module is connected to the camera module and is used to convert the image acquired by the camera module into a growth state quantification value;

[0010] The pre-regulation module is connected to the environment acquisition module and the growth state quantification module respectively, and is used to obtain the tea tree regulation value through the growth state quantification value and the environmental monitoring parameters;

[0011] The pest warning module is connected to the camera module and the growth status quantification module respectively, and is used to issue pest warnings through the growth status quantification value and the image of the camera module;

[0012] The self-update module is connected to the growth state quantization module and the pre-regulation module respectively, and is used to perform self-supervision through the growth state quantization value, the growth state pre-regulation effect and the growth state stability, complete the self-update of the pre-regulation module, and realize the adaptive adjustment of the entire system.

[0013] For further optimization, the camera module adopts a wide-angle camera, which is mounted above the tea tent, and the camera module can obtain clear images of tea trees and tea leaves in the tea garden.

[0014] For further optimization, the growth state quantification module adopts a pre-trained deep neural network model, whose input is two frames of images of the camera module at a fixed time interval, and the output is the growth state quantification value; wherein, the two frames of images of the camera module at a fixed time interval are the images taken by the camera module at the evaluation time, and the images captured by the camera before the fixed time interval T.

[0015] Preferably, the fixed time interval T is obtained through a large amount of experimental data.

[0016] For further optimization, the environmental acquisition module includes a soil pH sensor, a soil organic matter detector, a soil trace element detector, a soil temperature and humidity sensor, a tea garden light intensity sensor, and a tea garden temperature and humidity sensor, which are used to monitor the pH, organic matter, trace elements, temperature and humidity of the tea garden soil as well as the light intensity and air environment temperature and humidity of the tea garden.

[0017] For further optimization, the pre-regulation model adopts a pre-trained deep neural network model, whose input is the tea garden environmental monitoring parameters and growth status quantified values, and whose output is the (current) required amount of watering and fertilization for the tea trees.

[0018] Preferably, the monitoring and management system also includes spray pipes and fertilizer pipes evenly laid in the tea garden, and electromagnetic control valves are evenly distributed on the spray pipes and fertilizer pipes. The electromagnetic control valves are connected to the pre-control module, and the opening and closing and opening degree of each electromagnetic control valve are controlled by the pre-control module, thereby achieving precise control of the amount of fertilizer and irrigation.

[0019] For further optimization, the pest and disease warning module adopts a pre-trained deep neural network model, whose input is the fusion feature and the output is the pest and disease warning value; the fusion feature adopts a preprocessing method to fuse the growth state quantization value with the image obtained by the camera module (at the current moment).

[0020] Preferably, the preprocessing method is specifically as follows: resizing the image taken by the wide-angle camera to fixed pixels, adding a fourth channel of pixels of unfixed size, and all of its values ​​are quantitative evaluation values ​​of the growth status of the tea trees.

[0021] Preferably, the image acquired by the camera module (at the current moment) is a three-channel RGB image.

[0022] For further optimization, the pest warning module is connected to an alarm, and an alarm threshold value is preset in the alarm. When the pest warning value output by the pest warning module is greater than the alarm threshold value, the alarm issues a pest alarm.

[0023] For further optimization, the growth state pre-regulation effect e(t) is obtained by the quantitative value of the growth state of the tea tree, specifically:

[0024]

[0025]

[0026] Where: ρ(t) represents the quantitative value of the growth state of the tea tree at time t, ρ′(t) is its first-order derivative, which represents the change of the quantitative value of the growth state of the tea tree and is used to reflect the control effect of the pre-control model; Δt represents the statistical cycle time of the tea garden model error, t0 represents the delay time when the pre-control model has the greatest impact on the growth of the tea tree, which is obtained by experimental statistics; λ is the time influence decay factor, which means that the farther the time interval from time t0, the smaller the impact on the error calculation, which is obtained through experiments.

[0027] For further optimization, the growth state stability is obtained by calculating the variance of the quantitative values ​​of the growth state of the tea trees at all times within the statistical stage.

[0028] For further optimization, the specific steps of the self-update module for self-supervision are:

[0029] First, the comprehensive score threshold S is preset in the self-update module. d ;

[0030] Then, the self-update module obtains a comprehensive score S by randomly extracting the quantitative value of the growth state, the pre-regulation effect of the growth state, and the stability of the growth state at M moments in the system, specifically:

[0031]

[0032] Where: ω1, ω2, ω3 represent the weight of pre-regulation parameter effect evaluation, the minimum tolerance weight of tea tree growth and the weight of tea tree growth stability, respectively, where ω1+ω2+ω3=1; ρ th Indicates the minimum tolerance threshold of tea tree growth, obtained through a large amount of experimental data; var th It indicates the maximum variance of tea tree growth fluctuation, obtained through a large amount of experimental data;

[0033] in:

[0034] Finally, the self-update module performs S and S d Comparison:

[0035] When S<S d When S ≥ S d , no changes are made.

[0036] For further optimization, the self-update module performs self-update of the pre-regulation module specifically by: including automatic update data set generation of the pre-regulation model, and online tuning method of pre-regulation model parameters;

[0037] The automatic update data set generation of the pre-control model mainly generates a parameter set for model tuning; the parameter set is in the form of a data pair, including two sets of pre-control model inputs and the corresponding intelligent tea garden parameter adjustment effects;

[0038] The online tuning of the pre-control model parameters is to fine-tune the model through a newly added tuning parameter set based on the original pre-control model to obtain model parameters that meet the requirements.

[0039] In the above monitoring and management system, the self-update of the pre-regulation model is completed only by relying on the self-supervision of the system. In the whole process, the growth state quantification module is relied on to quantify the growth state of the tea tree. It is obvious that the accuracy of the growth state quantification module plays an extremely important role in the whole system. For further optimization, the growth state quantification module introduces artificially assisted reinforcement learning into the self-update mechanism of the growth state quantification module, thereby further ensuring the stable operation of the system.

[0040] The artificially assisted reinforcement learning includes active confirmation method, growth state quantization module error evaluation and self-update of growth state quantization module;

[0041] The active confirmation method is as follows: the system pushes the data to be manually confirmed according to the active confirmation data push time period, and waits for the manual confirmation result; the data to be manually confirmed is the growth state quantitative model, that is, two frames of images taken by the camera at a fixed time interval; the manual confirmation result is the quantitative scoring of the growth state of the tea tree based on the above two frames of images;

[0042] The specific error assessment of the growth state quantification model is:

[0043] err=ρ-ρ l ;

[0044] Where: ρ represents the quantitative value of the growth state of tea tree obtained by the growth state quantitative model, ρ l Indicates the corresponding manual confirmation result;

[0045] The data push time period is dynamically adjusted according to the error evaluation value of the growth state quantization model. The larger the error, the shorter the push period. Specifically:

[0046]

[0047] Where: T i+1 represents the (i+1)th active recommendation cycle, T min Indicates the minimum push cycle time, α is the attenuation factor, err th Indicates the maximum allowable error (err th >0), all obtained in advance from experimental data;

[0048] The self-update of the growth state quantification module is as follows: based on the parameters of the original growth state quantification module, the model is fine-tuned using the latest evaluation results confirmed with human assistance to obtain model parameters that meet the error tolerance requirements and complete the update of the system model parameters.

[0049] The present invention has the following technical effects:

[0050] The present application quantifies the growth status of tea trees in a tea garden through the cooperation of a camera module and a growth status quantification module, thereby reflecting the growth status of tea trees more clearly and intuitively; through the cooperation of an environment acquisition module, a growth status quantification module and a pre-control module, the irrigation amount and the fertilization amount are regulated according to the growth status of tea trees, thereby avoiding the problem of excessive or insufficient irrigation and fertilization, and ensuring the full growth of tea trees while avoiding waste of inputs, thereby ensuring the yield and quality of tea leaves; through the cooperation of a camera module, a growth status quantification module and a pest and disease prediction module, pests and diseases in the tea garden can be predicted according to the growth status of tea trees. The system can predict the pests and diseases and take timely prevention measures in the tea garden to avoid the damage of tea trees and the reduction of tea production caused by pests and diseases. Through the cooperation of the self-renewal module, the growth state quantification module and the pre-control module, the growth state quantification value is used to obtain the growth state pre-control effect and growth state stability, so as to realize the self-supervision and self-renewal of the pre-control module, and effectively avoid the problem of incompatibility of irrigation and fertilization amount caused by different environmental factors in different growth stages of tea trees or different regions, so as to ensure the accuracy and adaptability of the pre-control module and avoid the problems of low tea quality and low tea tree yield caused by error accumulation.

[0051] At the same time, this application uses human-assisted reinforcement learning and human intervention to further improve the accuracy of the entire system and avoid errors, thereby ensuring the smooth and accurate operation of the monitoring and management system. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 Schematic diagram of a monitoring and management system in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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.

[0054] Embodiment 1:

[0055] like Figure 1 As shown: A monitoring and management system for tea garden pests and growth conditions, characterized by: including a camera module, an environment acquisition module, a growth status quantification module, a pre-regulation module, a pest warning module and a self-update module;

[0056] The camera module is mounted above the tea tent to obtain images of the tea garden. The camera module specifically adopts a wide-angle camera, which is mounted above the tea tent and can obtain clear images of tea trees and tea leaves in the tea garden. At the same time, the image obtained by the wide-angle camera is a three-channel RGB image.

[0057] The environmental collection modules are distributed inside the tea garden and are used to monitor the environmental factors in the tea garden and obtain environmental monitoring parameters; the environmental collection modules include soil pH sensors (or pH sensors), soil organic matter detectors, soil trace element detectors, soil temperature and humidity sensors, tea garden light intensity sensors, and tea garden temperature and humidity sensors (all of the above are common models of existing technologies, or commonly available equipment on the market), which are used to monitor the pH (i.e. pH value), organic matter, trace elements, temperature and humidity of the tea garden soil, as well as the light intensity, and air environment temperature and humidity of the tea garden.

[0058] The growth state quantification module is connected to the camera module, and is used for converting the image collected by the camera module into a growth state quantification value; the growth state quantification module adopts a pre-trained deep neural network model, whose input is two frames of images of the camera module at a fixed time interval, and the output is a growth state quantification value; wherein, the two frames of images of the camera module at a fixed time interval are images taken by the camera module at the evaluation time, and images captured by the camera before the fixed time interval T; the fixed time interval T is obtained through a large amount of experimental data.

[0059] The pre-control module is connected to the environment acquisition module and the growth state quantification module respectively, and is used to obtain the tea tree control value through the growth state quantification value and the environmental monitoring parameters; the pre-control model adopts a pre-trained deep neural network model, whose input is the tea garden environmental monitoring parameters and the growth state quantification value, and the output is the (current) required watering and fertilization amount of the tea tree. The monitoring and management system also includes spray pipes and fertilizer pipes evenly laid in the tea garden (the discharge ports of the spray pipes and fertilizer pipes are also evenly distributed on the pipes, and the spacing between two adjacent discharge ports is determined according to the actual tea tree planting spacing and area), and electromagnetic control valves are evenly distributed on the spray pipes and fertilizer pipes. The electromagnetic control valves are connected to the pre-control module, and the opening and closing and opening degree of each electromagnetic control valve are controlled by the pre-control module, so as to achieve precise control of the fertilizer and irrigation amount.

[0060] The pest and disease warning module is connected to the camera module and the growth status quantification module respectively, and is used to issue pest and disease warnings through the growth status quantification value and the image of the camera module; the pest and disease warning module adopts a pre-trained deep neural network model, whose input is the fusion feature and the output is the pest and disease warning value; the fusion feature adopts a preprocessing method to fuse the growth status quantification value with the image obtained by the camera module (at the current moment).

[0061] The preprocessing method is specifically as follows: resizing the image taken by the wide-angle camera to fixed pixels, adding a fourth channel with pixels of unfixed size, and all of its values ​​are quantitative evaluation values ​​of the growth status of the tea trees.

[0062] The pest and disease warning module is connected to an alarm (the alarm can be set inside the tea garden or outside the tea garden where it is easily observed by operators). The alarm has a preset alarm threshold value (in the embodiment of the present application, the alarm threshold value is 0.5). When the pest and disease warning value output by the pest and disease warning module is greater than the alarm threshold value, the alarm will sound a pest and disease alarm; the alarm can adopt a structure combining a flashing light and a buzzer.

[0063] The self-update module is connected to the growth state quantization module and the pre-control module respectively, and is used to perform self-supervision through the growth state quantization value, the growth state pre-control effect and the growth state stability, complete the self-update of the pre-control module, and realize the adaptive adjustment of the whole system.

[0064] The growth state pre-regulation effect e(t) is obtained through the quantitative value of the growth state of the tea tree, specifically:

[0065]

[0066]

[0067] Where: ρ(t) represents the quantitative value of the growth state of the tea tree at time t, ρ′(t) is its first-order derivative, which represents the change of the quantitative value of the growth state of the tea tree and is used to reflect the control effect of the pre-control model; Δt represents the statistical cycle time of the tea garden model error, t0 represents the delay time when the pre-control model has the greatest impact on the growth of the tea tree, which is obtained by experimental statistics; λ is the time influence decay factor, which means that the farther the time interval from time t0, the smaller the impact on the error calculation, which is obtained through experiments.

[0068] The stability of growth state is obtained by statistically analyzing the variance of the quantitative value of tea tree growth state at all moments in the stage, that is, var{ρ(t1),ρ(t2),...,ρ(t M )}(quantized value of growth state at M moments).

[0069] The specific steps of self-supervision by the self-update module are:

[0070] First, the comprehensive score threshold S is preset in the self-update module. d (obtained based on actual experience data and experimental data);

[0071] Then, the self-update module obtains a comprehensive score S by randomly extracting the quantitative value of the growth state, the pre-regulation effect of the growth state, and the stability of the growth state at M moments in the system, specifically:

[0072]

[0073] Where: ω1, ω2, ω3 represent the weight of pre-regulation parameter effect evaluation, the minimum tolerance weight of tea tree growth and the weight of tea tree growth stability, respectively, where ω1+ω2+ω3=1 (each weight value is assigned according to the actual situation and experiment); ρ th Indicates the minimum tolerance threshold of tea tree growth, obtained through a large amount of experimental data; var th It represents the maximum variance of tea tree growth fluctuation, obtained through a large amount of experimental data;

[0074] in:

[0075] Finally, the self-update module performs S and S d Comparison:

[0076] When S<S d When S ≥ S d , no changes are made.

[0077] The self-update module performs self-update of the pre-control module specifically including: automatically updating the data set generation of the pre-control model, and online tuning method of the pre-control model parameters;

[0078] The automatic update data set generation of the pre-control model mainly generates the parameter set for model tuning; the parameter set is in the form of data pairs, including the input of two sets of pre-control models and the corresponding intelligent tea garden parameter adjustment effects;

[0079] The online tuning of the pre-control model parameters is to fine-tune the model based on the original pre-control model through the newly added tuning parameter set to obtain the model parameters that meet the requirements.

[0080] Embodiment 2:

[0081] In the above monitoring and management system, the self-update of the pre-regulation model is completed only by relying on the self-supervision of the system. In the whole process, the growth state quantification module is relied on to quantify the growth state of the tea tree. It is obvious that the accuracy of the growth state quantification module plays an extremely important role in the whole system. For further optimization, the growth state quantification module introduces artificially assisted reinforcement learning into the self-update mechanism of the growth state quantification module, thereby further ensuring the stable operation of the system.

[0082] Human-assisted reinforcement learning includes active confirmation methods, growth state quantification model error assessment methods, and growth state quantification model automatic update methods;

[0083] The active confirmation method is as follows: the system pushes the data to be manually confirmed according to the active confirmation data push time period, and waits for the manual confirmation result; the data to be manually confirmed is the growth state quantitative model, that is, two frames of images taken by the camera at a fixed time interval; the manual confirmation result is the quantitative scoring of the growth state of the tea tree based on the above two frames of images;

[0084] The specific error evaluation method of the growth state quantification model is as follows:

[0085] err=ρ-ρ l ;

[0086] Where: ρ represents the quantitative value of the growth state of tea tree obtained by the growth state quantitative model, ρ l Indicates the corresponding manual confirmation result;

[0087] The data push time period is dynamically adjusted according to the error evaluation value of the growth state quantitative model. The larger the error, the shorter the push period. Specifically:

[0088]

[0089] Where: T i+1 represents the (i+1)th active recommendation cycle, T min Indicates the minimum push cycle time, α is the attenuation factor, err th Indicates the maximum allowable error (err th >0), all obtained in advance from experimental data;

[0090] The automatic updating method of the growth state quantification model is as follows: based on the parameters of the original growth state quantification model, the model is fine-tuned using the latest evaluation results confirmed with human assistance to obtain model parameters that meet the error tolerance requirements and complete the updating of the system model parameters.

[0091] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A monitoring and management system for tea garden pests and growth conditions, characterized by: It includes camera module, environment acquisition module, growth status quantification module, pre-regulation module, pest and disease early warning module and self-renewal module; The camera module is mounted above the tea tent to obtain images of the tea garden; The environmental collection modules are distributed inside the tea garden and are used to monitor environmental factors in the tea garden and obtain environmental monitoring parameters; The growth state quantification module is connected to the camera module and is used to convert the image acquired by the camera module into a growth state quantification value; The pre-regulation module is connected to the environment acquisition module and the growth state quantification module respectively, and is used to obtain the tea tree regulation value through the growth state quantification value and the environmental monitoring parameters; The pest warning module is connected to the camera module and the growth status quantification module respectively, and is used to issue pest warnings through the growth status quantification value and the image of the camera module; The self-update module is connected to the growth state quantification module and the pre-regulation module respectively, and is used to perform self-supervision through the growth state quantification value, the growth state pre-regulation effect and the growth state stability, so as to complete the self-update of the pre-regulation module; The growth state pre-regulation effect e(t) It is obtained through the quantitative value of the growth status of the tea tree, specifically: Where: express t The quantitative value of the growth status of the tea tree at the moment, is its first-order derivative, which indicates the change of the quantitative value of the growth state of the tea tree and is used to reflect the control effect of the pre-control module; represents the statistical cycle time of the tea garden model error, t 0 indicates the delay time when the pre-regulation module has the greatest impact on the growth of tea trees, obtained by experimental statistics; is the time-effect decay factor, which represents the time t The farther the time interval is from 0, the smaller the impact on the error calculation is, which is obtained through experiments.

2. A monitoring and management system for tea garden pests and growth conditions according to claim 1, characterized in that: The camera module adopts a wide-angle camera and is mounted above the tea tent.

3. A monitoring and management system for tea garden pests and growth conditions according to claim 1 or 2, characterized in that: The growth state quantification module adopts a pre-trained deep neural network model, whose input is two frames of images taken by the camera module at a fixed time interval, and output is a growth state quantification value; wherein, the two frames of images taken by the camera module at a fixed time interval are the images taken by the camera module at the evaluation time, and the fixed time interval T Images captured by the front camera.

4. A monitoring and management system for tea garden pests and growth conditions according to claim 3, characterized in that: The environmental acquisition module includes a soil pH sensor, a soil organic matter detector, a soil trace element detector, a soil temperature and humidity sensor, a tea garden light intensity sensor, and a tea garden temperature and humidity sensor, which are used to monitor the pH, organic matter, trace elements, temperature and humidity of the tea garden soil as well as the light intensity and air environment temperature and humidity of the tea garden.

5. A monitoring and management system for tea garden pests and growth conditions according to claim 4, characterized in that: The pre-regulation module adopts a pre-trained deep neural network model, whose input is the tea garden environmental monitoring parameters and growth status quantified values, and whose output is the amount of watering and fertilization required for the tea trees.

6. A monitoring and management system for tea garden pests and growth conditions according to claim 5, characterized in that: The pest and disease early warning module adopts a pre-trained deep neural network model, whose input is fusion features and output is pest and disease warning values; the fusion features use a preprocessing method to fuse the growth state quantization value with the image obtained by the camera module.

7. A monitoring and management system for tea garden pests and growth conditions according to claim 1, characterized in that: The growth state stability is obtained by statistically analyzing the variance of the quantitative values ​​of the growth state of the tea trees at all times during the stage.

8. A monitoring and management system for tea garden pests and growth conditions according to claim 7, characterized in that: The specific steps of the self-update module for self-supervision are: First, the comprehensive score threshold is preset in the self-update module S d ; Then, the self-update module randomly extracts M The quantitative value of the growth state at each moment, the pre-regulation effect of the growth state and the stability of the growth state are used to obtain a comprehensive score S , specifically: Where: They represent the evaluation weight of the effect of pre-regulation parameters, the minimum tolerance weight of tea tree growth, and the stability weight of tea tree growth, respectively. ; Indicates the minimum tolerance threshold for tea tree growth, obtained through a large amount of experimental data; var th It represents the maximum variance of tea tree growth fluctuation, obtained through a large amount of experimental data; in: Finally, the self-update module performs S and S d Comparison: when When , the self-update of the pre-regulation module is triggered; when , no changes are made.

9. A monitoring and management system for tea garden pests and growth conditions according to claim 8, characterized in that: The self-update module performs self-update of the pre-regulation module specifically by: including automatic update data set generation of the pre-regulation module, and an online tuning method of parameters of the pre-regulation module; The pre-control module automatically updates the data set to generate a parameter set mainly for model tuning; the parameter set is in the form of a data pair, including two sets of inputs of the pre-control module and the corresponding intelligent tea garden parameter tuning effects; The online tuning of the parameters of the pre-control module is to fine-tune the model through a newly added tuning parameter set on the basis of the original pre-control module to obtain model parameters that meet the requirements.