Search

CN-121998193-A - Artificial intelligence-based Jin Xiutian tea whole industry chain optimization method

CN121998193ACN 121998193 ACN121998193 ACN 121998193ACN-121998193-A

Abstract

The invention relates to a Jin Xiutian tea full-industry chain optimization method based on artificial intelligence, which belongs to the field of intersection of agricultural biotechnology and artificial intelligence, and sequentially comprises the steps of constructing five AI subsystems and a distributed data center, achieving AI directional breeding, shortening a breeding period to 18 months based on improved random forest algorithm association genes and characters and combining reinforcement learning, AI embryo breeding, improving regression optimization environment through the Internet of things and gradient, improving survival rate to more than 90%, AI planting management and protection, fusing AI-GIS site selection, LSTM water fertilizer decision, YOLOv8 pest identification and near infrared spectrum harvesting decision, achieving accurate planting, AI deep processing, utilizing reinforcement learning optimization extraction technology, accelerating product research and development based on online spectrum and GAN, and achieving supply and demand accurate matching through LSTM demand prediction, intelligent recommendation and block chain source tracing. The method realizes the cooperative optimization and the comprehensive efficiency improvement of the whole industrial chain.

Inventors

  • HUANG YUYUAN
  • WEI YULIN

Assignees

  • 黑马当先(金秀)绿色农业科技有限公司

Dates

Publication Date
20260508
Application Date
20260127

Claims (10)

  1. 1. The full industrial chain optimization method of Jin Xiutian tea based on artificial intelligence is characterized by comprising the following steps of: S1, constructing an AI oriented seed breeding system, namely collecting Jin Xiu sweet tea gene sequences and character data, constructing a gene-character database, adopting an improved random forest algorithm to train a gene-character association model, positioning high-value gene segments, predicting the characters of filial generation through an AI breeding simulation system, dynamically adjusting breeding parameters by combining reinforcement learning, and directionally breeding special varieties with the purity of rubusoside more than or equal to 98 percent and the grade of felt disease resistance less than or equal to 2; S2, constructing an AI group embryo seedling intelligent regulation subsystem, namely deploying an Internet of things sensor in a tissue culture room, collecting environmental parameters in real time, adopting a gradient lifting regression model to predict the survival rate of test tube seedlings, and optimizing the environmental parameter combination through grid search; S3, constructing an AI planting management and protection standardized subsystem, namely, whole alloy elegant area environment data, screening planting plots with suitability scores more than or equal to 80 minutes through AIGIS space analysis models, deploying unmanned aerial vehicles and soil sensors, generating a precise water and fertilizer scheme based on an LSTM time sequence prediction model, adopting an improved YOLOv algorithm to construct a disease and pest control system by combining meteorological data, detecting the rubusoside content of leaves through near infrared spectrum, and triggering intelligent picking instructions; S4, constructing an AI rubusoside deep processing optimization subsystem, namely constructing a reinforcement learning environment based on rubusoside physical and chemical properties, training a process optimization model through a PPO algorithm, deploying on-line spectrum detection equipment in an extraction and purification link, adjusting process parameters in real time by AI, constructing a rubusoside application database, and adopting generation of an antagonistic network to simulate the mouthfeel of a product; s5, constructing an AI supply chain collaboration subsystem, namely analyzing downstream enterprise order data by adopting an LSTM time sequence model to generate a raw material purchasing and inventory plan, pushing a customization scheme by an AI recommendation algorithm based on enterprise product types and target groups, recording full-flow data by adopting an AI and blockchain technology, and generating a unique traceability two-dimensional code; S6, constructing a distributed data center, realizing real-time synchronization of five subsystem data by adopting a MySQL+Redis architecture, and establishing a data cleaning rule and a right management module to ensure data quality and safety.
  2. 2. The method of claim 1, wherein the improved random forest algorithm in step S1 uses the feature bag outside error to screen the gene segment features, and the number and the maximum depth of the decision tree are adjusted by grid search, so that the coefficient of the model R 2 is more than or equal to 0.92 and the confidence coefficient is more than or equal to 90%.
  3. 3. The method according to claim 1, wherein the input features of the LSTM timing prediction model in step S3 include plant NDVI value, soil moisture content, nitrogen, phosphorus and potassium content, near 7 balance average temperature and precipitation, and output as a watering time and fertilization scheme for 7 days in the future.
  4. 4. The method according to claim 1, wherein the bonus function calculation formula of the reinforcement learning environment in step S4 is: reward = 0.4 x extraction +0.5 x purity-0.1 x (energy consumption/2) Wherein the extraction rate and purity are calculated in percentage, and the energy consumption is calculated in kW.
  5. 5. The method according to claim 1, wherein the blockchain technique in step S5 employs a federated chain architecture, and the nodes include planting bases, deep processing workshops, downstream enterprises and regulatory authorities, and the data modification is validated by at least 3 nodes.
  6. 6. The method according to claim 1, wherein the data cleansing rules in step S6 include sensor outlier rejection using 3σ principle, normalization of gene sequence format to FASTQ format, and filling of trait data missing values using K-nearest neighbor algorithm.
  7. 7. A system for implementing the method of any one of claims 16, comprising: the gene sequencing equipment is used for collecting sweet tea gene sequences; the sensor of the Internet of things is used for detecting illumination, temperature and soil parameters; The multispectral unmanned aerial vehicle is used for monitoring plant growth; the online spectrum detector is used for detecting the purity of rubusoside; The server is used for deploying five AI subsystems and the distributed data center; And the client is used for the user to check the data, receive the early warning and give the instruction.
  8. 8. The system of claim 7, wherein the hardware configuration of the server comprises CPU not less than Intel Xeon Gold 6330, memory not less than 64GB, hard disk not less than 2TB SSD, GPU not less than NVIDIA A100, client support Windows 10/11, macOS 12+ operating system, browser support Chrome 90+, edge 90+.
  9. 9. A computer readable storage medium having stored thereon a computer program, which when executed by a processor performs the steps of the method of any of claims 16.
  10. 10. The computer-readable storage medium of claim 9, wherein the storage medium comprises at least one of a U-disk, a removable hard disk, ROM, RAM, SSD, and the computer program comprises an AI model training module, a data processing module, and a user interaction module.

Description

Artificial intelligence-based Jin Xiutian tea whole industry chain optimization method Technical Field The invention relates to the technical field of intersection of agricultural biotechnology and artificial intelligence, in particular to a Jin Xiutian tea full-industry chain optimization method based on artificial intelligence. Background The Jin Xiutian tea of Rubus of Rosaceae is a special medicinal and edible plant of Guangxi Jin Xiu county of Yao nationality, and the rubusoside rich in leaves has the characteristics of high sweetness and low calorie, is a natural non-sugar sweetener with great potential, and can be used as a new food raw material in the fields of sugarless beverage, health food and the like. However, in the whole chain from variety breeding to market supply, the current Jin Xiu sweet tea industry has a plurality of technical bottlenecks to be broken through, and the standardization, scale and industrialization development of the sweet tea industry are restricted. Firstly, the variety directional breeding efficiency is low. The traditional breeding method is severely dependent on phenotype selection and artificial hybridization, the breeding period is as long as 3 to 5 years, and complex characters such as high rubusoside content (target is more than or equal to 98%), high stress resistance (such as resistance to felting disease) and the like are difficult to accurately polymerize, so that the rigidity requirements of the downstream food industry on uniform raw material components and stable quality cannot be met. Secondly, the intelligent level of the production and processing links is insufficient. The survival rate of embryo seedling is generally only 60% -70%, the water and fertilizer management in field planting depends on experience, so that the resource waste is more than 30%, the optimization of the extraction process of rubusoside depends on trial and error, the extraction rate is less than 80%, the product purity fluctuation is large, and the defective rate is often higher than 5%. And finally, the data of each link of the industrial chain is split, and the synergy is poor. The lack of an effective data linkage mechanism between breeding, planting, processing and terminal market demands can not dynamically adjust production parameters according to specific formula demands (such as specific taste suitability and purity standards) of downstream enterprises, so that problems such as delayed supply chain response, stock backlog or raw material shortage are frequent. In the prior art, the application of artificial intelligence in the agricultural field is concentrated on local optimization of a single link, such as pest and disease diagnosis based on image recognition, auxiliary breeding by using a simulation model, and the like. Although the technical schemes can improve the efficiency in a specific link, the technical schemes lack systematic design aiming at Jin Xiutian tea whole industry chains, and especially do not combine the special genetic characters, tissue culture physiological requirements and physicochemical properties of rubusoside to construct a special AI model, so that the industrial pain point cannot be systematically solved. In summary, there is a need for an integrated artificial intelligence optimization method that can traverse the whole process of biological breeding, intelligent seedling raising, precise planting, efficient extraction, supply chain collaboration. Disclosure of Invention The invention provides a Jin Xiutian tea whole industry chain optimization method based on artificial intelligence for solving the problems in the prior art. In order to solve the technical problems, the invention is realized through the following technical scheme that in the first aspect, the Jin Xiutian tea full-industry chain optimization method based on artificial intelligence comprises the following steps: S1, constructing an AI oriented seed breeding system, namely collecting Jin Xiu sweet tea gene sequences and character data, constructing a gene-character database, adopting an improved random forest algorithm to train a gene-character association model, positioning high-value gene segments, predicting the characters of filial generation through an AI breeding simulation system, dynamically adjusting breeding parameters by combining reinforcement learning, and directionally breeding special varieties with the purity of rubusoside more than or equal to 98 percent and the grade of felt disease resistance less than or equal to 2; S2, constructing an AI group embryo seedling intelligent regulation subsystem, namely deploying an Internet of things sensor in a tissue culture room, collecting environmental parameters in real time, adopting a gradient lifting regression model to predict the survival rate of test tube seedlings, and optimizing the environmental parameter combination through grid search; S3, constructing an AI planting management and protection standardized subs