CN-121189829-B - Tobacco planting risk prediction system based on meteorological big data analysis

CN121189829BCN 121189829 BCN121189829 BCN 121189829BCN-121189829-B

Abstract

The invention relates to the technical field of tobacco planting, in particular to a tobacco planting risk prediction system based on meteorological big data analysis, which comprises a meteorological module for collecting temperature and humidity light data, eliminating repeated deletion, generating a standardized sequence according to the hour calculation amplitude recording direction, the crop module extracts leaf color texture boundary judgment change rate and jump degree marking fluctuation labels, the association module analyzes the fluctuation direction and the healthy trend extraction consistent section mapping dry edge screen intersection to generate an association interval, and the risk module extracts a suspicious section generation trend set corresponding to the change track and the abnormality judgment consistent trend label. According to the invention, the dynamic expression of the growth state is realized by carrying out standardized coding on the fluctuation amplitude and direction of the temperature and humidity light change data and combining with the blade tone texture feature sequence, the risk-causing section is constructed based on the consistency comparison of the environmental factors and the plant state direction, the consistency trend section is extracted to generate a risk-causing factor set, and the early warning area is identified by combining with the trend path overlap ratio comparison, so that the weather risk identification degree and pertinence are improved.

Inventors

ZHANG LUMIN
Xiao changdong
DONG ZERONG
OuYang Chengren
SUN WEIDONG
JI CHUNTAO
Shan Shuanglv

Assignees

云南省烟草公司红河州公司
云南农业大学

Dates

Publication Date: 20260512
Application Date: 20250928

Claims (8)

1. A tobacco planting risk prediction system based on meteorological big data analysis, the system comprising: The meteorological data acquisition module acquires temperature, humidity and illumination data of a tobacco planting area, eliminates repeated sampling and missing time periods, calculates variation amplitude according to the time period, records fluctuation direction and combines the fluctuation direction into a time segment, and generates a standardized meteorological factor sequence; The crop state monitoring module is used for extracting leaf tone, texture gray level and boundary shape in a tobacco plant image, judging tone change rate and texture jump density, marking trend offset time intervals and generating plant growth fluctuation time interval labels; The data association recognition module is used for constructing a fixed period sliding window to analyze the relation between the fluctuation direction of each item of data and the healthy change trend of the plant based on the standardized meteorological factor sequence and the plant growth fluctuation period label, extracting a direction consistent section and marking the direction consistent section as a fluctuation section, combining a top blade dry edge time node mapping section, screening time intersections, and generating a staged meteorological risk association section; And the risk factor extraction module is used for calling the change time period of each item of data in the staged risk factor association interval, extracting a change track according to the maximum fluctuation range, corresponding to the abnormal plant time period, judging the consistent trend, marking the same as the suspicious risk factor concentrated fluctuation period if the same exists, and generating a suspicious risk factor trend set.
2. The tobacco planting risk prediction system based on weather big data analysis according to claim 1, wherein the standardized weather factor sequence comprises a temperature variation amplitude sequence, a humidity variation amplitude sequence, an illumination intensity variation amplitude sequence, a fluctuation direction coding sequence and a time slice normalization tag, the plant growth fluctuation period tag comprises a tone variation rate characteristic, a texture jump density mark and a trend offset duration period, the staged weather risk correlation interval comprises a temperature and humidity light fluctuation matching section, a health trend synchronization section and a top blade dry edge mapping node, and the suspicious risk factor trend set comprises a temperature concentrated fluctuation trend, a humidity concentrated fluctuation trend, an illumination concentrated fluctuation trend and a trend direction consistency mark.
3. The tobacco planting risk prediction system based on meteorological big data analysis according to claim 1, wherein the meteorological data acquisition module comprises: The data acquisition sub-module is used for monitoring a tobacco planting area, acquiring time sequences of temperature data, humidity data and illumination intensity data, comparing sampling result values for adjacent time periods, eliminating current points as repetition if the sampling result values are consistent, scanning the whole sequence, identifying a segment formed by continuously missing data points, and screening out the corresponding time period when the number of the missing points exceeds one to generate an effective meteorological data sequence; The change calculation sub-module is used for extracting a temperature value set according to the time intervals of natural hours based on the effective meteorological data sequence, calculating a maximum value and minimum value to obtain a temperature change amplitude, and processing the humidity value set and the illumination value set to obtain the humidity change amplitude and the illumination change amplitude to generate a meteorological change amplitude sequence; And the sequence generation sub-module is used for judging the fluctuation direction according to whether the temperature change amplitude value is larger than zero or smaller than zero for each small period based on the weather change amplitude sequence, and also judging the humidity fluctuation direction and the illumination fluctuation direction, and combining the direction values into a time segment unit to generate a standardized weather factor sequence.
4. The tobacco planting risk prediction system based on meteorological big data analysis of claim 1, wherein the crop condition monitoring module comprises: The image feature extraction submodule is used for acquiring image data of tobacco plants in a planting area, extracting tone channel values of leaf areas of the tobacco plants in the image, extracting edge texture gray level distribution values, extracting area boundary shape values and generating a leaf feature set; The change rate calculation submodule is used for calculating the adjacent point difference ratio of the tone value of the blade as the tone change rate according to the continuous time segment grouping data based on the blade characteristic set, counting the jump point number in the texture gray level distribution value and dividing the time segment length to obtain the texture jump point concentration, and generating a change rate concentration index; And the trend deviation labeling sub-module is used for comparing numerical symbols of tone change rate and texture jump point density for each time period based on the change rate density index, labeling the time period if the two numerical symbols are consistent, scanning continuous labeling time periods, labeling the continuous labeling time periods as trend deviation time periods when the continuous time period length exceeds a set continuous time period threshold value, and generating a plant growth fluctuation time period label.
5. The tobacco planting risk prediction system based on meteorological big data analysis according to claim 1, wherein the data association identification module comprises: The matching analysis sub-module is used for constructing a sliding analysis window according to a fixed period length based on the temperature change, the humidity and the illumination change amplitude in the standardized meteorological factor sequence and the leaf health change rate and the growth stage change mode in the plant growth fluctuation period label, comparing whether the meteorological fluctuation direction is consistent with the direction sign of the plant health change trend or not according to each window, and marking the window if so, so as to generate a matching marking sequence; a continuous segment extraction sub-module for identifying segments formed by continuous marking windows based on the matching marking sequence, scanning the time sequence, marking the segments when the segment length exceeds a set continuous length threshold value, and generating a synchronous fluctuation segment sequence; And the interval generation sub-module is used for calling a time node set of the dry edge phenomenon of the top blade based on the synchronous fluctuation section sequence, mapping the time range of each node into the synchronous fluctuation section sequence, screening time periods of overlapping nodes and sections, and generating a staged weather risk association interval.
6. The tobacco planting risk prediction system based on meteorological big data analysis according to claim 1, wherein the risk factor extraction module comprises: Invoking temperature, humidity and illumination change time sections in the staged weather risk association interval, scanning all fluctuation amplitude values in a window for each time window, extracting the maximum absolute value of the temperature as a temperature change track point, extracting the maximum absolute value of the humidity as a humidity change track point, extracting the maximum absolute value of illumination as an illumination change track point, connecting all window track points to form a complete track, and generating a weather change track set; positioning a time point corresponding to the maximum fluctuation amplitude of the temperature in each time window based on the meteorological change track set, mapping the time point to the starting point of the abnormal period of the tobacco plant, and also processing the maximum fluctuation time of the humidity and the maximum fluctuation time of the illumination to generate an alignment period set containing the abnormal period of the temperature, the abnormal period of the humidity and the abnormal period of the illumination; based on the alignment period set, reading a temperature trend direction symbol, a humidity trend direction symbol and an illumination trend direction symbol in a time overlapping interval, comparing whether the three symbols are identical, and if so, marking the time overlapping interval as a suspicious risk factor concentrated fluctuation section to generate a suspicious risk factor trend set.
7. The tobacco planting risk prediction system based on meteorological big data analysis of claim 1, wherein the system further comprises: The risk trend prediction module is used for calculating the coincidence degree of each temperature, humidity and illumination actual change path and the trend set path in the current period based on the risk factor types in the suspicious risk factor trend set and the trend direction sequences corresponding to the risk factor types, judging the number and span of continuous coincidence periods, marking the continuous coincidence periods as a risk concentration section if the continuous coincidence periods exceed a threshold value, and generating a short-term tobacco planting weather risk early warning result; The short-term tobacco planting meteorological risk early warning result comprises a risk concentration section identification result, a trend path overlap ratio index, a risk factor type combination and an early warning period range identification.
8. The tobacco planting risk prediction system based on meteorological big data analysis of claim 7, wherein the risk trend prediction module comprises: The coincidence quantifying sub-module is used for comparing the consistency of the sign with the sign of the corresponding time marking in the trend set point by point according to the sign of the actual change of the temperature per hour in the current period based on the temperature, humidity and illumination trend direction sequence in the suspicious risk factor trend set, processing the sign of the actual change of the humidity and the sign of the actual change of the illumination, counting the sign points of which all three direction signs are consistent at each time, and generating a direction coincidence sequence; Based on the direction coincidence sequence, scanning a time axis, identifying a time period formed by continuous coincidence marks, calculating a length value of each continuous time period, namely the number of contained hours, comparing the length value with a preset continuous time period threshold value, and marking the time period when the length value exceeds the threshold value to generate a risk candidate section set; And extracting the time span starting points and the time span ending points of all the marking time periods based on the risk candidate section sets, merging the overlapping time periods according to time sequence, establishing a current risk set section set, and generating a short-term tobacco planting meteorological risk early warning result.

Description

Tobacco planting risk prediction system based on meteorological big data analysis Technical Field The invention relates to the technical field of tobacco planting, in particular to a tobacco planting risk prediction system based on meteorological big data analysis. Background The technical field of tobacco planting comprises core contents such as soil management, climate adaptation, pest control, growth monitoring and the like, and the field systematically relates to the whole processes of seed selection and seedling raising, field management, fertigation, environment monitoring and harvesting treatment of tobacco crops, and the influence of meteorological conditions, soil quality, biological factors and the like on the growth of tobacco is focused to ensure the stable yield and quality. Overall, tobacco planting technology enables optimal management from planting preparation to final harvesting by integrating agricultural science principles and practical experience, where risk control such as weather disasters and disease outbreaks are key challenges, relying on data-driven decision support. The tobacco planting risk prediction system based on weather big data analysis is characterized in that historical weather records and real-time weather observation data are collected, weather parameters such as temperature, precipitation and humidity change trend are analyzed by combining tobacco growth stage characteristics, so that abnormal modes or threshold exceeding events are identified, potential risk prediction of tobacco planting by weather disasters such as drought, flood and frost is covered by aiming at technical matters, and the specific solution is to utilize multi-source weather data sources such as ground weather stations and satellite remote sensing information, and a data integration and statistical comparison method is adopted, so that risk probability assessment is completed based on the historical data modes to match current conditions. In the prior art, the weather risk early warning is often dependent on parameter threshold judgment and history mode comparison, a time matching mechanism between the change direction of a weather factor and the response trend of a plant is not involved, under the condition of coexistence of continuous multivariable fluctuation, the association path of the abnormal change of an environmental factor and the plant is difficult to distinguish, the actual evolution process of a potential risk factor cannot be positioned in the face of continuous deviation or trend reversal, especially in the compound action period of common fluctuation of temperature, humidity and light, the situations of early warning trigger delay and inaccurate intervention range can occur, and the accurate intervention capability of rapid identification and local abnormal change in the early stage of risk formation is limited. Disclosure of Invention The invention aims to solve the defects in the prior art, and provides a tobacco planting risk prediction system based on meteorological big data analysis. In order to achieve the aim, the invention adopts the following technical scheme that the tobacco planting risk prediction system based on meteorological big data analysis comprises: The meteorological data acquisition module acquires temperature, humidity and illumination data of a tobacco planting area, eliminates repeated sampling and missing time periods, calculates variation amplitude according to the time period, records fluctuation direction and combines the fluctuation direction into a time segment, and generates a standardized meteorological factor sequence; The crop state monitoring module is used for extracting leaf tone, texture gray level and boundary shape in a tobacco plant image, judging tone change rate and texture jump density, marking trend offset time intervals and generating plant growth fluctuation time interval labels; The data association recognition module is used for constructing a fixed period sliding window to analyze the relation between the fluctuation direction of each item of data and the healthy change trend of the plant based on the standardized meteorological factor sequence and the plant growth fluctuation period label, extracting a direction consistent section and marking the direction consistent section as a fluctuation section, combining a top blade dry edge time node mapping section, screening time intersections, and generating a staged meteorological risk association section; And the risk factor extraction module is used for calling the change time period of each item of data in the staged risk factor association interval, extracting a change track according to the maximum fluctuation range, corresponding to the abnormal plant time period, judging the consistent trend, marking the same as the suspicious risk factor concentrated fluctuation period if the same exists, and generating a suspicious risk factor trend set. As a further aspect of the present inventio