CN110276482B - Method for forecasting starting and ending time of growth period of plant flowers and fruits - Google Patents

Method for forecasting starting and ending time of growth period of plant flowers and fruits Download PDF

Info

Publication number
CN110276482B
CN110276482B CN201910474904.0A CN201910474904A CN110276482B CN 110276482 B CN110276482 B CN 110276482B CN 201910474904 A CN201910474904 A CN 201910474904A CN 110276482 B CN110276482 B CN 110276482B
Authority
CN
China
Prior art keywords
forecasting
factors
year
linear
sample set
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201910474904.0A
Other languages
Chinese (zh)
Other versions
CN110276482A (en
Inventor
姚日升
黄鹤楼
杨栋
顾思南
涂小萍
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ning Boshiqixiangtai
Original Assignee
Ning Boshiqixiangtai
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Ning Boshiqixiangtai filed Critical Ning Boshiqixiangtai
Priority to CN201910474904.0A priority Critical patent/CN110276482B/en
Publication of CN110276482A publication Critical patent/CN110276482A/en
Application granted granted Critical
Publication of CN110276482B publication Critical patent/CN110276482B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G06F16/2462—Approximate or statistical queries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/02—Agriculture; Fishing; Forestry; Mining

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Strategic Management (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • General Physics & Mathematics (AREA)
  • Probability & Statistics with Applications (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • Marketing (AREA)
  • Databases & Information Systems (AREA)
  • Fuzzy Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Game Theory and Decision Science (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Software Systems (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Mathematical Physics (AREA)
  • Development Economics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Agronomy & Crop Science (AREA)
  • Animal Husbandry (AREA)
  • Marine Sciences & Fisheries (AREA)
  • Mining & Mineral Resources (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a forecasting method for the starting and ending time of the growth period of flowers and fruits of plants, which collects historical data of the starting or ending time of the growth period of flowers and fruits of plants in the past year, wherein the historical data comprises the sequence of the starting or ending time of the growth period of flowers and fruits of plants in the past year, and a plurality of meteorological factors and physiological factors related to the starting or ending time of the growth period of flowers and fruits of plants in the past year; determining candidate forecasting factors according to the meteorological factors and the phenological factors; selecting one model type from three model types of a linear model, a nonlinear model and a mixed model as a forecasting model construction type according to the candidate forecasting factors; constructing an optimal forecasting model according to the forecasting model construction type; forecasting the beginning or ending time of the growth period of the flower and fruit of the plant to be forecasted according to the optimal forecasting model; the method has the advantages that the method can forecast the starting time or the ending time of the growth period of the plant flowers and fruits by fully utilizing long-time meteorological element forecast, and the forecasting accuracy is high.

Description

Method for forecasting starting and ending time of growth period of plant flowers and fruits
Technical Field
The invention relates to a technology for forecasting the growth period of plant flowers and fruits, in particular to a method for forecasting the starting and ending time of the growth period of the plant flowers and fruits.
Background
With the rapid development of social economy and the continuous improvement of the living standard of people, various local festivals and tourism projects with the themes of ornamental flowers and picking activities are vigorously developed all over the country, and the development of plant flower and fruit growth period (initial flowering period, full flowering period, fruit mature period and the like) forecast also becomes one of the works of the weather service society. The flowering season of cherry blossom in spring every year is a very much concerned topic of Japan. The growth of the plants depends on factors such as temperature, humidity, rainfall, illumination and the like, and agricultural and meteorological workers study the cold and heat requirements of deciduous fruit trees in a specific growth period to obtain the effective accumulated temperature required by different plant varieties in the growth period. In order to improve the capacity of the weather service society, meteorological workers in various regions also carry out a great deal of research on the flowering phase and fruit maturity forecast of various flowers and trees, and establish various forecast models, wherein the forecast models basically use linear regression models and mainly comprise two types of forecast factors: one is based on the factors of the phenology, such as a certain phenological period of other plants or relatively early-maturing varieties, Yao Dong et al (journal of ecology, 2006) also studies the relationship between the leaf-picking period of the peach tree in the previous year and the full-bloom period of the peach tree in the current year; the other is a meteorological factor which mainly comprises the average temperature of early months and ten days, the date of stable passing of a certain temperature, the accumulated temperature and the ground temperature of a certain period of time and the like, and the meteorological factor is also extracted from the high-altitude circulation of the early period to make the medium-long term forecast of the peony flowering phase (Wenxilaulan (meteorology, 2001) and Koufacu (Chinese agricultural meteorology, 2011). The meteorological factors mainly aim at the accumulation of a certain meteorological element in the early stage of the growth period of the plant flowers and fruits, wherein the most important is the accumulated temperature, the accumulated temperature is the positive accumulated temperature, namely the accumulated temperature obtained from a certain day, such as 1 month and 1 day, in fact, the correlation between the growth period of the plant flowers and fruits and the reverse accumulated temperature is higher, and the reverse accumulated temperature refers to the accumulated temperature of a specific time period from the growth period of the plant flowers and fruits and back to the time. For the plant growth period forecast, if only the meteorological elements of a certain period of time in the earlier stage are used, the influence of the weather factors close to the growth period of flowers and fruits cannot be reflected, and if a more accurate forecast is to be made, the meteorological elements of a period of time in the future need to be used for forecasting, the current weather forecast time can reach more than 10 days, and with the development of a numerical forecasting technology and the improvement of a weather forecasting technology, a more precise and longer-time forecast can be obtained in the future, so that a more reasonable forecasting model needs to be established to fully utilize the long-time meteorological elements for forecasting the growth period of the flowers and fruits.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a method for forecasting the starting and ending time of the growth period of the plant flowers and fruits, which can forecast the starting or ending time of the growth period of the plant flowers and fruits by fully utilizing long-time meteorological elements and has high forecasting accuracy.
The technical scheme adopted by the invention for solving the technical problems is as follows: a forecasting method for the starting and ending time of the growth period of plant flowers and fruits is characterized by comprising the following steps:
step 1: collecting historical data of the beginning or ending time of the growth period of the flowers and fruits of the plants, wherein the historical data comprises the sequence of the beginning or ending time of the growth period of the flowers and fruits of the plants in the past year, a plurality of meteorological factors related to the beginning or ending time of the growth period of the flowers and fruits of the plants in the past year, and a phenological factor related to the beginning or ending time of the growth period of the flowers and fruits of the plants in the past year, wherein the meteorological factor is a linear forecasting factor or a nonlinear forecasting factor, and the phenological factor is a linear forecasting factor or a nonlinear forecasting factor; if the linear forecasting factor and the nonlinear forecasting factor do not exist, the forecasting is not carried out, otherwise, the step 2 is executed;
the historical range is 10-30 years, the meteorological factors which are linear forecasting factors comprise forward accumulated temperature, reverse accumulated temperature, forward accumulated precipitation, reverse accumulated precipitation, forward accumulated humidity, reverse accumulated humidity, forward accumulated illumination hours and reverse accumulated illumination hours, and the meteorological factors which are non-linear forecasting factors are obtained according to historical experience;
step 2: estimating the earliest possible sequence and the latest possible sequence of the growth period start or end time of the flowers and fruits of the plants according to the sequence of the growth period start or end time of the flowers and fruits of the plants in the past year, and correspondingly marking the sequence as D1And D2(ii) a Wherein D is1＜D2；
And step 3: under the condition that only linear forecasting factors exist, calculating a linear correlation coefficient of each linear forecasting factor of the sequence of the growth start or end time of the flower and fruit of each year plant in the calendar year and the growth start or end time of the flower and fruit of the corresponding year plant; then, carrying out correlation significance check on all linear correlation coefficients; then taking the linear forecasting factors corresponding to all linear correlation coefficients passing through the correlation significance verification as candidate forecasting factors;
under the condition that only the nonlinear forecasting factors exist, all the nonlinear forecasting factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year are used as candidate forecasting factors;
under the condition that both the linear forecasting factors and the non-linear forecasting factors exist, calculating a linear correlation coefficient of each linear forecasting factor of the starting or ending time sequence of the growth period of the flower and fruit of each year plant in the calendar year and the starting or ending time of the growth period of the flower and fruit of the corresponding year plant; then, carrying out correlation significance check on all linear correlation coefficients; then, taking linear forecast factors corresponding to all linear correlation coefficients passing correlation significance verification and all nonlinear forecast factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year as candidate forecast factors;
and 4, step 4: selecting one model type from three model types of a linear model, a nonlinear model and a mixed model as a forecasting model construction type according to the candidate forecasting factors, wherein the mixed model consists of a linear module and a nonlinear module; if the candidate forecasting factors are all nonlinear forecasting factors, selecting a nonlinear model as a forecasting model construction type; if the candidate forecasting factors have both linear forecasting factors and nonlinear forecasting factors, selecting any model type of a nonlinear model and a mixed model as a forecasting model construction type, and when the mixed model is selected as the forecasting model construction type, only the linear forecasting factors can be selected by a linear module in the mixed model for construction; then, constructing an optimal forecasting model according to the forecasting model construction type;
if the construction type of the forecasting model is a linear model, the construction process of the optimal forecasting model is as follows:
a1, forming a history sample set by the sequence of the starting time or the ending time of the growth period of the flowers and fruits of the plants in the past year and all candidate forecasting factors of the starting time or the ending time of the growth period of the flowers and fruits of the plants in the past year;
a2, dividing a history sample set into two subsets according to the year, and respectively using the two subsets as a modeling sample set and a test sample set;
a3, establishing an initial forecasting model by adopting a regression equation method and utilizing a modeling sample set;
a4, inputting all candidate forecasting factors of each year in the test sample set into an initial forecasting model according to the year to obtain a sequential forecasting value of the corresponding year;
a5, calculating the difference value of the sequence forecast value of each year and the sequence of the corresponding year in the test sample set according to the year; then calculating the average value of the difference values corresponding to all the years; then judging whether the average value is less than or equal to a set value Ds1If so, determining the initial forecasting model as an optimal forecasting model, determining all candidate forecasting factors in the historical sample set as final forecasting factors, and executing the step 5; otherwise, go to step A6; wherein D iss1∈[1,3]Day;
a6, deleting the candidate forecasting factors corresponding to the minimum linear correlation coefficient in the historical sample set, then returning to the step A2 to continue execution, and if all the candidate forecasting factors in the historical sample set are processed and the optimal forecasting model is not determined, not forecasting;
if the construction type of the forecasting model is a nonlinear model, the construction process of the optimal forecasting model is as follows:
b1, defining the current year to be processed as the current year according to the year in the calendar year;
b2, D in the current year1～D2Defining the current sequence to be processed in the sequence as the current sequence;
b3, marking the current sequence as Dcur(ii) a Current sequence DcurAs an additional forecasting factor and the current order DcurWhether or not it is greater than or equal to D0The return value of (2) is used as a forecast object; wherein D iscur∈[D1,D2]，D0The sequence of the beginning or end of the growth period of the flower and fruit of the plant in the current year, D1＜D0＜D2If D iscur≥D0The forecast object value is 1, if Dcur＜D0The forecast object value is 0;
b4, forming a historical sample by the forecast object, the additional forecast factors and all candidate forecast factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year;
b5, D in the current year1～D2The next to-be-processed sequence in the sequence is taken as the current sequence, and then the step B3 is returned to continue execution until D in the current year1～D2After each sequence in the sequence is processed, executing the step B6;
b6, taking the next year to be processed in the past year as the current year, then returning to the step B2 to continue execution until each year of the past year is processed, and D in all the years in the past year1～D2All history samples corresponding to the sequence form a history sample set;
b7, dividing the historical sample set into two subsets according to the year, and respectively using the two subsets as a modeling sample set and a testing sample set;
b8, establishing an initial forecasting model by adopting an artificial neural network or a support vector machine method and utilizing a modeling sample set;
b9, D in each year in the set of test samples1～D2Inputting the additional forecasting factors and all candidate forecasting factors corresponding to the additional forecasting factors in the daily historical samples of the sequence into an initial forecasting model to obtain forecasting values corresponding to the additional forecasting factors, and setting the forecasting values to be 1 if the forecasting values are more than or equal to 0.5; if the predicted value is reported<Setting the forecast value to 0.5; judging whether the forecast value corresponding to the additional forecast factor is the same as the forecast object corresponding to the additional forecast factor in the test sample set, if so, indicating that the forecast is correct, and if not, indicating that the forecast is wrong; if the total prediction accuracy corresponding to all historical samples in the test sample set reaches 80%, determining the initial prediction modelDetermining the prediction model as an optimal prediction model, determining all candidate prediction factors in the historical sample set as final selection prediction factors, and executing the step 5; otherwise, go to step B10;
b10, if the candidate forecasting factors in the historical sample set are all linear forecasting factors, deleting the linear forecasting factor corresponding to the minimum linear correlation coefficient in the historical sample set; if all the candidate forecasting factors in the historical sample set are nonlinear forecasting factors, deleting any nonlinear forecasting factor in the historical sample set; if the candidate forecasting factors in the historical sample set have linear forecasting factors and nonlinear forecasting factors, deleting the linear forecasting factor or any nonlinear forecasting factor corresponding to the minimum linear correlation coefficient in the historical sample set;
b11, returning to the step B7 to continue execution, and if all candidate forecasting factors in the historical sample set are processed and the optimal forecasting model is not determined, determining not to forecast;
if the construction type of the forecasting model is a mixed model, the construction process of the forecasting model is as follows:
c1, if the candidate forecasting factors are linear forecasting factors, forming a linear module history sample set by the sequence of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year and part of the candidate forecasting factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year; if the candidate forecasting factors have linear forecasting factors and nonlinear forecasting factors, all or part of linear forecasting factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year and the starting or ending time of the growth period of the flowers and fruits of the plants in the past year form a linear module historical sample set;
c2, dividing the linear module historical sample set into two subsets according to the year, and respectively using the two subsets as a linear module modeling sample set and a linear module testing sample set;
c3, modeling a sample set by using a regression equation method and a linear module, and establishing an initial linear module;
c4, inputting all candidate forecasting factors of each year in the linear module test sample set into the initial linear module according to the year to obtain the sequential forecasting value of the corresponding year;
c5, calculating the difference value of the sequence forecast value of each year and the sequence of the corresponding year in the linear module test sample set according to the year; then calculating the average value of the difference values corresponding to all the years; then judging whether the average value is less than or equal to a set value Ds1If so, determining the initial linear module as an optimal linear module, determining all candidate forecasting factors in the linear module historical sample set as final forecasting factors of the optimal linear module, inputting all final forecasting factors of each year in the linear module historical sample set into the optimal linear module to obtain a sequential forecasting value of the corresponding year, and executing the step C7; otherwise, go to step C6; wherein D iss1∈[1,3]Day;
c6, deleting the candidate forecasting factors corresponding to the minimum linear correlation coefficient in the linear module historical sample set, then returning to the step C2 to continue execution, if all the candidate forecasting factors in the linear module historical sample set are processed and the optimal linear module is not determined, the linear module is invalid, the mixed model only contains the nonlinear module, and the forecasting model construction type is changed into the nonlinear model;
c7, defining the current year to be processed as the current year according to the year in the calendar year;
c8, D in the current year1～D2Defining the current sequence to be processed in the sequence as the current sequence;
c9, marking the current sequence as Dcur(ii) a Current sequence DcurAs a first additional forecasting factor and the current order DcurGreater than or equal to D0The return value of (2) is used as a forecast object; taking the sequential forecast value of the current year obtained by the optimal linear module in the step C5 as a second additional forecast factor; wherein D iscur∈[D1,D2]，D0The sequence of the beginning or end of the growth period of the flower and fruit of the plant in the current year, D1＜D0＜D2If D iscur≥D0The forecast object value is 1, if Dcur＜D0Then forecast pairThe image value is 0;
c10, forming a nonlinear module history sample by the forecast object, the first additional forecast factor, the second additional forecast factor, the sequence of the growth start or end time of the flower and fruit of the plant in the past year and the remaining candidate forecast factors of the growth start or end time of the flower and fruit of the plant in the past year except for forming the linear module history sample set;
c11, D in the current year1～D2The next to-be-processed sequence in the sequence is taken as the current sequence, and then the step C9 is returned to continue execution until D in the current year1～D2After each of the sequences is processed, executing step C12;
c12, taking the next year to be processed in the past year as the current year, then returning to the step C8 to continue executing until each year of the past year is processed, and D in all the years in the past year1～D2All historical samples corresponding to the sequence form a non-linear module historical sample set;
c13, dividing the nonlinear module historical sample set into two subsets according to the year, and respectively using the two subsets as a nonlinear module modeling sample set and a nonlinear module testing sample set;
c14, establishing an initial nonlinear module by adopting an artificial neural network or a support vector machine method and utilizing a nonlinear module modeling sample set;
c15, checking the non-linear module for D in each year in the sample set1～D2Inputting a first additional forecasting factor and all candidate forecasting factors corresponding to the first additional forecasting factor in the historical sample of the day-by-day nonlinear module of the sequence into an initial nonlinear module to obtain a forecasting value corresponding to the first additional forecasting factor, and setting the forecasting value to be 1 if the forecasting value is more than or equal to 0.5; if the predicted value is reported<Setting the forecast value to 0.5; judging whether the forecast value corresponding to the first additional forecast factor is the same as the forecast object corresponding to the first additional forecast factor in the nonlinear module test sample set, if so, indicating that the forecast is correct, and if not, indicating that the forecast is wrong; if the non-linear module tests the sum corresponding to all the non-linear module historical samples in the sample setIf the body prediction accuracy reaches 80%, determining the initial nonlinear module as an optimal nonlinear module, determining all candidate prediction factors in the nonlinear module historical sample set as final selection prediction factors of the optimal nonlinear module, taking a mixed model formed by the optimal linear module and the optimal nonlinear module as an optimal prediction model, and executing the step 5; otherwise, go to step C16;
c16, if the candidate forecasting factors in the non-linear module historical sample set are all linear forecasting factors, deleting the linear forecasting factor corresponding to the minimum linear correlation coefficient in the non-linear module historical sample set; if all the candidate forecasting factors in the non-linear module historical sample set are non-linear forecasting factors, deleting any non-linear forecasting factor in the non-linear module historical sample set; if the candidate forecasting factors in the non-linear module historical sample set have both linear forecasting factors and non-linear forecasting factors, deleting the linear forecasting factor or any one of the non-linear forecasting factors corresponding to the minimum linear correlation coefficient in the non-linear module historical sample set;
c17, returning to the step C13 to continue execution, and if all candidate forecasting factors in the non-linear module historical sample set are processed and the best non-linear module is not determined, determining not to forecast;
and 5: defining the year of the beginning or ending time of the growth period of the flower and fruit of the plant to be forecasted as a forecast year; defining the date sequence of the date of starting to make forecast as the starting date sequence and recording as Ds(ii) a Recording the day of the longest effective time of weather forecast provided by weather as Dmax；
When the type of the optimal forecasting model is a linear model, calculating the factor value of each final forecasting factor corresponding to the optimal forecasting model by using the data of the forecasting year, which are used for calculating the final forecasting factors, inputting the factor values of all the final forecasting factors corresponding to the optimal forecasting model into the optimal forecasting model, and calculating to obtain a sequence value; if the calculated sequence value is in the time range D of the growth period of the flowers and fruits of the plants to be forecasted1～D2And if so, the calculated sequence value is considered to be valid,the forecasting result of the sequence of the starting time or the ending time of the growth period of the flowers and fruits of the plants is obtained, and then step 6 is executed; otherwise, the calculated sequence value is considered to be invalid, and the prediction is not carried out;
when the type of the best prediction model is a non-linear model, if Ds＜D1-DmaxOr Ds＞D2If yes, the forecast is not carried out; otherwise, the following steps are executed:
step 5_1a, the order D of the dates currently calculatedjIs defined as the calculation order, wherein DjHas an initial value of Ds；
Step 5_1b, according to DjCalculating the factor value of each final selection forecasting factor corresponding to the optimal forecasting model, inputting the factor values of all final selection forecasting factors corresponding to the optimal forecasting model into the optimal forecasting model, and calculating to obtain a forecasting value;
step 5_1c, if the calculated forecast value is more than or equal to 0.5, when D is equal tojTime range D of flower and fruit growth period of plant to be forecasted1～D2When it is internal, consider DjIf the prediction result is valid, the prediction result is the prediction result of the sequence of the starting or ending time of the growth period of the flower and fruit of the plant to be predicted, and then step 6 is executed; when D is presentjTime range D of not in the growth period of flowers and fruits of the plants to be forecasted1～D2When it is internal, consider DjInvalid, not forecasting; if the calculated predicted value is obtained<0.5, executing the step 5_1 d;
step 5_1D, if Dj＞D2Or Dj＞Ds+DmaxIf yes, the forecast is not carried out; otherwise, let Dj＝Dj+1, taking the sequence of the next calculation date as the calculation sequence, and returning to the step 5_1b for continuous execution; wherein D isj＝DjThe "═ in + 1" is an assignment symbol;
when the type of the optimal forecasting model is a mixed model, calculating the factor value of each final forecasting factor corresponding to the linear module of the optimal forecasting model by using the information of the forecasting year required for calculating the final forecasting factors, and calculating the factor value of each final forecasting factor corresponding to the linear module of the optimal forecasting modelInputting the factor values of all final selection forecasting factors into a linear module of the optimal forecasting model, and calculating to obtain the sequence values forecasted by the linear module; if the calculated sequence value predicted by the linear module is not in the time range D of the growth period of the flower and fruit of the plant to be predicted1～D2In the method, the calculated sequence value predicted by the linear module is considered invalid, and the prediction is not carried out; if the calculated sequence value predicted by the linear module is in the time range D of the growth period of the flower and fruit of the plant to be predicted1～D2And considering that the calculated sequence value predicted by the linear module is valid, and executing the following steps:
step 5_2a, if Ds＜D1-DmaxOr Ds＞D2If yes, the forecast is not carried out; otherwise, executing step 5_2 b;
step 5_2b, the order D of the current calculation datesjIs defined as the calculation order, wherein DjHas an initial value of Ds；
Step 5_2c, according to DjCalculating the factor value of each final selection forecasting factor corresponding to the optimal forecasting model, inputting the factor values of all final selection forecasting factors corresponding to the optimal forecasting model into the optimal forecasting model, and calculating to obtain a forecasting value;
step 5_2D, if the calculated forecast value is more than or equal to 0.5, when D is equal tojTime range D of flower and fruit growth period of plant to be forecasted1～D2When it is internal, consider DjIf the prediction result is valid, the prediction result is the prediction result of the sequence of the starting or ending time of the growth period of the flower and fruit of the plant to be predicted, and then step 6 is executed; when D is presentjTime range D of not in the growth period of flowers and fruits of the plants to be forecasted1～D2When it is internal, consider DjInvalid, not forecasting; if the calculated predicted value is obtained<0.5, executing the step 5_2 e;
step 5_2e, if Dj＞D2Or Dj＞Ds+DmaxIf yes, the forecast is not carried out; otherwise, let Dj＝Dj+1, using the next calculation date as the calculation date, returning to step 5_2c to continue executionA row; wherein D isj＝DjThe "═ in + 1" is an assignment symbol;
step 6: and converting the obtained forecast result into a date, and taking the date as the beginning or ending date of the growth period of the flower and fruit of the plant of the forecast year to finish the forecast.
In the step 1, the phenological factor is based on the phenological period of other plants or relatively early-maturing varieties.
In the step a2 and the step B7, the number of the samples in the modeling sample set is 70% of the number of the samples in the history sample set, and the number of the samples in the testing sample set is 30% of the number of the samples in the history sample set.
In the step C2, the number of samples in the linear module modeling sample set is 70% of the number of samples in the linear module history sample set, and the number of samples in the linear module testing sample set is 30% of the number of samples in the linear module history sample set; in the step C13, the number of samples in the non-linear module modeling sample set is 70% of the number of samples in the non-linear module history sample set, and the number of samples in the non-linear module testing sample set is 30% of the number of samples in the non-linear module history sample set.
Compared with the prior art, the invention has the advantages that:
1) the method can select a linear model, a nonlinear model or a mixed model, finally determines a proper forecasting model according to the type of the candidate forecasting factor and the test result, and improves the forecasting accuracy.
2) The forecasting equation constructed by linear regression can only select the linearly related forecasting factors, namely the linear forecasting factors, and the nonlinear forecasting factors can be selected by the nonlinear model in the method, so that the selection range of the forecasting factors is enlarged.
3) When the linear regression method is used for constructing the plant flower and fruit growth period forecasting model, the sequence is taken as a forecasting object, and the plant flower and fruit growth period data at one time can only form one historical sample.
4) The candidate forecasting factors used in the method can use weather forecasting products, and can also use the reverse accumulated value of a certain physical quantity as the candidate forecasting factors, so that the influence of weather factors near the growth period of the plant flowers and fruits can be fully reflected, and in fact, the influence is larger than that of the weather factors in the early period, and the forecasting accuracy is improved.
5) The typical plant flower and fruit growing period comprises a flowering period, a fruit mature period and the like, the method can independently forecast the starting time or the ending time of the flower and fruit growing period, and can also forecast the process time of the plant flower and fruit growing period simultaneously and respectively, thereby providing technical support for nursing the plant flower and fruit growing period and various social activities taking ornamental flowers and picking activities as themes.
Drawings
FIG. 1 is a block diagram of the overall implementation of the method of the present invention.
Detailed Description
The invention is described in further detail below with reference to the accompanying examples.
The invention provides a method for forecasting the starting and ending time of the growth period of plant flowers and fruits, which comprises the following steps:
step 1: collecting historical data of the beginning or ending time of the growth period of the flowers and fruits of the plants, wherein the historical data comprises the sequence of the beginning or ending time of the growth period of the flowers and fruits of the plants in the past year, a plurality of meteorological factors related to the beginning or ending time of the growth period of the flowers and fruits of the plants in the past year, and a phenological factor related to the beginning or ending time of the growth period of the flowers and fruits of the plants in the past year, wherein the meteorological factor is a linear forecasting factor or a nonlinear forecasting factor, and the phenological factor is a linear forecasting factor or a nonlinear forecasting factor; if the linear forecasting factor and the non-linear forecasting factor do not exist, the forecasting is not carried out, otherwise, the step 2 is executed.
The range of the previous year is 10-30 years, the meteorological factors which are linear forecasting factors comprise forward accumulated temperature, reverse accumulated temperature, forward accumulated precipitation, reverse accumulated precipitation, forward accumulated humidity, reverse accumulated humidity, forward accumulated illumination hours and reverse accumulated illumination hours, the forward or reverse accumulated starting and ending time of the meteorological factors is determined by sea selection according to a principle of priority of a correlation relation with the date sequence of the starting or ending time of the growth period of the flowers and fruits of the plants, one kind of the accumulated starting and ending time is fixed, and the accumulated starting and ending time of each year in the previous year is the same and is called as forward accumulation; the other is that the date of the beginning or the ending of the growth period of the flowers and fruits of the plants is taken as the accumulation starting date and is accumulated back to the previous DcDay, if taking Dc30, referred to as reverse accumulation, the starting time or the ending time of the growth period of the flowers and fruits of each year in the past year are different, so that the starting time or the ending time of the reverse accumulation is different due to the difference of the starting time or the ending time of the growth period of the flowers and fruits of the plants, the meteorological factors which are the nonlinear forecasting factors are obtained according to historical experience, and the composition and the expression mode of each meteorological factor which is the nonlinear forecasting factor are determined according to specific situations; the sequence refers to the sequence number of the date relative to 1 month and 1 day, for example, the sequence of 1 month and 1 day is 1, the sequence of 1 month and 2 days is 2, and so on; the phenological factor is based on the phenological period of other plants or relatively early-maturing species.
Step 2: estimating the earliest possible sequence and the latest possible sequence of the growth period start or end time of the flowers and fruits of the plants according to the sequence of the growth period start or end time of the flowers and fruits of the plants in the past year, and correspondingly marking the sequence as D1And D2(ii) a Wherein D is1＜D2。
And step 3: under the condition that only linear forecasting factors exist, calculating a linear correlation coefficient of each linear forecasting factor of the sequence of the growth start or end time of the flower and fruit of each year plant in the calendar year and the growth start or end time of the flower and fruit of the corresponding year plant; then, carrying out correlation significance check on all linear correlation coefficients; and then taking the linear forecasting factors corresponding to all linear correlation coefficients passing through the correlation significance verification as candidate forecasting factors.
And in the case that only the nonlinear forecasting factors exist, all the nonlinear forecasting factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year are used as candidate forecasting factors.
Under the condition that both the linear forecasting factors and the non-linear forecasting factors exist, calculating a linear correlation coefficient of each linear forecasting factor of the starting or ending time sequence of the growth period of the flower and fruit of each year plant in the calendar year and the starting or ending time of the growth period of the flower and fruit of the corresponding year plant; then, carrying out correlation significance check on all linear correlation coefficients; and then taking the linear forecasting factors corresponding to all linear correlation coefficients passing the correlation significance check and all nonlinear forecasting factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year as candidate forecasting factors.
And 4, step 4: selecting one model type from three model types of a linear model, a nonlinear model and a mixed model as a forecasting model construction type according to the candidate forecasting factors, wherein the mixed model consists of a linear module and a nonlinear module; if the candidate forecasting factors are all nonlinear forecasting factors, selecting a nonlinear model as a forecasting model construction type; if the candidate forecasting factors have both linear forecasting factors and nonlinear forecasting factors, selecting any model type of a nonlinear model and a mixed model as a forecasting model construction type, and when the mixed model is selected as the forecasting model construction type, only the linear forecasting factors can be selected by a linear module in the mixed model for construction; and then constructing an optimal forecasting model according to the forecasting model construction type.
If the construction type of the forecasting model is a linear model, the construction process of the optimal forecasting model is as follows:
a1, forming a historical sample set by the sequence of the starting time or the ending time of the flower growth period of the plants in the past years and all candidate forecasting factors of the starting time or the ending time of the flower growth period of the plants in the past years.
And A2, dividing the historical sample set into two subsets by year, and respectively serving as a modeling sample set and a testing sample set.
And A3, establishing an initial forecasting model by adopting a regression equation method and utilizing a modeling sample set.
And A4, inputting all candidate forecasting factors of each year in the test sample set into the initial forecasting model according to the year to obtain the sequential forecasting value of the corresponding year.
A5, calculating the difference value of the sequence forecast value of each year and the sequence of the corresponding year in the test sample set according to the year; then calculating the average value of the difference values corresponding to all the years; then judging whether the average value is less than or equal to a set value Ds1If so, determining the initial forecasting model as an optimal forecasting model, determining all candidate forecasting factors in the historical sample set as final forecasting factors, and executing the step 5; otherwise, go to step A6; wherein D iss1∈[1,3]Day, if taking Ds1Day 2.
And A6, deleting the candidate forecasting factors corresponding to the minimum linear correlation coefficient in the historical sample set, then returning to the step A2 to continue execution, and if all the candidate forecasting factors in the historical sample set are processed and the optimal forecasting model is not determined, not forecasting.
If the construction type of the forecasting model is a nonlinear model, the construction process of the optimal forecasting model is as follows:
b1, defining the current year to be processed as the current year according to the year in the calendar year.
B2, D in the current year1～D2The current sequence to be processed in the sequence is defined as the current sequence.
B3, marking the current sequence as Dcur(ii) a Current sequence DcurAs an additional forecasting factor and the current order DcurWhether or not it is greater than or equal to D0The return value of (2) is used as a forecast object; wherein D iscur∈[D1,D2]，D0The sequence of the beginning or end of the growth period of the flower and fruit of the plant in the current year, D1＜D0＜D2If D iscur≥D0The forecast object value is 1, if Dcur＜D0The forecast object takes a value of 0.
B4, forming a historical sample by the forecast object, the additional forecast factors and all candidate forecast factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year; the method judges whether the processed forecast sequence is greater than or equal to the sequence of the starting or ending date of the growth period of the flower and fruit when the actual forecast is carried out, and the forecast object is the current sequence D when the historical samples are formedcurWhether or not it is greater than or equal to D0Return value of, current order of day DcurIs first preset as the date sequence of the growth period start or end date of the flower and fruit, and then is accumulated back to the front D by taking the growth period start or end date of the flower and fruit as the start datecThe forecast factor of the reverse accumulation of days is needed for the current day sequence DcurAnd calculating the corresponding factor value.
B5, D in the current year1～D2The next to-be-processed sequence in the sequence is taken as the current sequence, and then the step B3 is returned to continue execution until D in the current year1～D2After each of the sequences is processed, step B6 is performed.
B6, taking the next year to be processed in the past year as the current year, then returning to the step B2 to continue execution until each year of the past year is processed, and D in all the years in the past year1～D2All history samples corresponding to the sequence form a history sample set.
And B7, dividing the historical sample set into two subsets by year, and respectively serving as a modeling sample set and a testing sample set.
And B8, establishing an initial forecasting model by adopting an artificial neural network or a support vector machine method and utilizing a modeling sample set.
B9, D in each year in the set of test samples1～D2Inputting the additional forecasting factors and all candidate forecasting factors corresponding to the additional forecasting factors in the daily historical samples of the sequence into an initial forecasting model to obtain forecasting values corresponding to the additional forecasting factors, and setting the forecasting values to be 1 if the forecasting values are more than or equal to 0.5; if it is notForecast value<Setting the forecast value to 0.5; judging whether the forecast value corresponding to the additional forecast factor is the same as the forecast object corresponding to the additional forecast factor in the test sample set, if so, indicating that the forecast is correct, and if not, indicating that the forecast is wrong; if the total prediction accuracy corresponding to all the historical samples in the test sample set reaches 80%, determining the initial prediction model as the optimal prediction model, determining all the candidate prediction factors in the historical sample set as final prediction factors, and then executing the step 5; otherwise, step B10 is performed.
B10, if the candidate forecasting factors in the historical sample set are all linear forecasting factors, deleting the linear forecasting factor corresponding to the minimum linear correlation coefficient in the historical sample set; if all the candidate forecasting factors in the historical sample set are nonlinear forecasting factors, deleting any nonlinear forecasting factor in the historical sample set; and if the candidate forecasting factors in the historical sample set have both linear forecasting factors and nonlinear forecasting factors, deleting the linear forecasting factor or any nonlinear forecasting factor corresponding to the minimum linear correlation coefficient in the historical sample set.
B11, returning to the step B7 to continue execution, and if all candidate forecasting factors in the historical sample set are processed and the optimal forecasting model is not determined, determining not to forecast.
If the construction type of the forecasting model is a mixed model, the construction process of the forecasting model is as follows:
c1, if the candidate forecasting factors are linear forecasting factors, forming a linear module history sample set by the sequence of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year and part of the candidate forecasting factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year; if the candidate forecasting factors have linear forecasting factors and nonlinear forecasting factors, all or part of linear forecasting factors of the starting time or the ending time of the growth period of the flowers and fruits of the plants in the past year and the starting time or the ending time of the growth period of the flowers and fruits of the plants in the past year form a linear module historical sample set.
And C2, dividing the linear module historical sample set into two subsets according to the year, and respectively serving as a linear module modeling sample set and a linear module testing sample set.
And C3, modeling the sample set by adopting a regression equation method and utilizing a linear module to establish an initial linear module.
And C4, inputting all candidate forecasting factors of each year in the linear module test sample set into the initial linear module according to the year to obtain the sequential forecasting value of the corresponding year.
C5, calculating the difference value of the sequence forecast value of each year and the sequence of the corresponding year in the linear module test sample set according to the year; then calculating the average value of the difference values corresponding to all the years; then judging whether the average value is less than or equal to a set value Ds1If so, determining the initial linear module as an optimal linear module, determining all candidate forecasting factors in the linear module historical sample set as final forecasting factors of the optimal linear module, inputting all final forecasting factors of each year in the linear module historical sample set into the optimal linear module to obtain a sequential forecasting value of the corresponding year, and executing the step C7; otherwise, go to step C6; wherein D iss1∈[1,3]Day, if taking Ds1Day 2.
And C6, deleting the candidate forecasting factors corresponding to the minimum linear correlation coefficient in the linear module historical sample set, then returning to the step C2 to continue execution, if all the candidate forecasting factors in the linear module historical sample set are processed and the optimal linear module is not determined, the linear module is invalid, the mixed model only contains the nonlinear module, the forecasting model construction type is changed into the nonlinear model, and the steps B1 to B11 are executed.
And C7, defining the current year to be processed as the current year according to the year in the calendar year.
C8, D in the current year1～D2The current sequence to be processed in the sequence is defined as the current sequence.
C9, marking the current sequence as Dcur(ii) a Current sequence DcurAs a first additional forecasting factor and the current order DcurGreater than or equal to D0As a forecast pairLike; taking the sequential forecast value of the current year obtained by the optimal linear module in the step C5 as a second additional forecast factor; wherein D iscur∈[D1,D2]，D0The sequence of the beginning or end of the growth period of the flower and fruit of the plant in the current year, D1＜D0＜D2If D iscur≥D0The forecast object value is 1, if Dcur＜D0The forecast object takes a value of 0.
C10, forming a nonlinear module history sample by the forecast object, the first additional forecast factor, the second additional forecast factor, the sequence of the growth start or end time of the flower and fruit of the plant in the past year and the remaining candidate forecast factors of the growth start or end time of the flower and fruit of the plant in the past year except for forming the linear module history sample set; the method judges whether the processed forecast sequence is greater than or equal to the sequence of the starting or ending date of the growth period of the flower and fruit when the actual forecast is carried out, and the forecast object is the current sequence D when the nonlinear module historical sample is formedcurWhether or not it is greater than or equal to D0Return value of, current order of day DcurIs first preset as the date sequence of the growth period start or end date of the flower and fruit, and then is accumulated back to the front D by taking the growth period start or end date of the flower and fruit as the start datecThe forecast factor of the reverse accumulation of days is needed for the current day sequence DcurAnd calculating the corresponding factor value.
C11, D in the current year1～D2The next to-be-processed sequence in the sequence is taken as the current sequence, and then the step C9 is returned to continue execution until D in the current year1～D2After each of the sequences is processed, step C12 is performed.
C12, taking the next year to be processed in the past year as the current year, then returning to the step C8 to continue executing until each year of the past year is processed, and D in all the years in the past year1～D2And all the historical samples corresponding to the sequence form a non-linear module historical sample set.
And C13, dividing the nonlinear module historical sample set into two subsets according to the year, and respectively using the two subsets as a nonlinear module modeling sample set and a nonlinear module testing sample set.
And C14, establishing an initial nonlinear module by adopting an artificial neural network or a support vector machine method and utilizing a nonlinear module modeling sample set.
C15, checking the non-linear module for D in each year in the sample set1～D2Inputting a first additional forecasting factor and all candidate forecasting factors corresponding to the first additional forecasting factor in the historical sample of the day-by-day nonlinear module of the sequence into an initial nonlinear module to obtain a forecasting value corresponding to the first additional forecasting factor, and setting the forecasting value to be 1 if the forecasting value is more than or equal to 0.5; if the predicted value is reported<Setting the forecast value to 0.5; judging whether the forecast value corresponding to the first additional forecast factor is the same as the forecast object corresponding to the first additional forecast factor in the nonlinear module test sample set, if so, indicating that the forecast is correct, and if not, indicating that the forecast is wrong; if the overall prediction accuracy corresponding to all the non-linear module historical samples in the non-linear module test sample set reaches 80%, determining the initial non-linear module as the optimal non-linear module, determining all candidate prediction factors in the non-linear module historical sample set as final selection prediction factors of the optimal non-linear module, taking a mixed model formed by the optimal linear module and the optimal non-linear module as the optimal prediction model, and executing the step 5; otherwise, step C16 is performed.
C16, if the candidate forecasting factors in the non-linear module historical sample set are all linear forecasting factors, deleting the linear forecasting factor corresponding to the minimum linear correlation coefficient in the non-linear module historical sample set; if all the candidate forecasting factors in the non-linear module historical sample set are non-linear forecasting factors, deleting any non-linear forecasting factor in the non-linear module historical sample set; and if the candidate forecasting factors in the non-linear module historical sample set have both linear forecasting factors and non-linear forecasting factors, deleting the linear forecasting factor or any one of the non-linear forecasting factors corresponding to the minimum linear correlation coefficient in the non-linear module historical sample set.
And C17, returning to the step C13 to continue execution, and if all candidate forecasting factors in the non-linear module historical sample set are processed and the best non-linear module is not determined, determining not to forecast.
And 5: defining the year of the beginning or ending time of the growth period of the flower and fruit of the plant to be forecasted as a forecast year; defining the date sequence of the date of starting to make forecast as the starting date sequence and recording as Ds(ii) a Recording the day of the longest effective time of weather forecast provided by weather as Dmax。
When the type of the optimal forecasting model is a linear model, calculating the factor value of each final forecasting factor corresponding to the optimal forecasting model by using the data of the forecasting year, which are used for calculating the final forecasting factors, inputting the factor values of all the final forecasting factors corresponding to the optimal forecasting model into the optimal forecasting model, and calculating to obtain a sequence value; if the calculated sequence value is in the time range D of the growth period of the flowers and fruits of the plants to be forecasted1～D2If the calculated sequence value is valid, the result is a forecast result of the sequence of the starting time or the ending time of the growth period of the flowers and fruits of the plants, and then step 6 is executed; otherwise, the calculated sequence value is considered invalid, and the prediction is not carried out.
When the type of the best prediction model is a non-linear model, if Ds＜D1-DmaxOr Ds＞D2If yes, the forecast is not carried out; otherwise, the following steps are executed:
step 5_1a, the order D of the dates currently calculatedjIs defined as the calculation order, wherein DjHas an initial value of Ds。
Step 5_1b, according to DjAnd calculating the factor value of each final selection forecasting factor corresponding to the optimal forecasting model, inputting the factor values of all final selection forecasting factors corresponding to the optimal forecasting model into the optimal forecasting model, and calculating to obtain a forecasting value.
Step 5_1c, if the calculated forecast value is more than or equal to 0.5, when D is equal tojTime range D of flower and fruit growth period of plant to be forecasted1～D2When it is internal, consider DjIs effective inForecasting the start or end time sequence of the growth period of the flowers and fruits of the plants to be forecasted, and then executing the step 6; when D is presentjTime range D of not in the growth period of flowers and fruits of the plants to be forecasted1～D2When it is internal, consider DjInvalid, not forecasting; if the calculated predicted value is obtained<0.5, step 5_1d is performed.
Step 5_1D, if Dj＞D2Or Dj＞Ds+DmaxIf yes, the forecast is not carried out; otherwise, let Dj＝Dj+1, taking the sequence of the next calculation date as the calculation sequence, and returning to the step 5_1b for continuous execution; wherein D isj＝DjIn +1, "═ is an assigned symbol.
When the type of the optimal forecasting model is a mixed model, calculating the factor value of each final forecasting factor corresponding to the linear module of the optimal forecasting model by using the data of the forecasting year, which are used for calculating the final forecasting factors, inputting the factor values of all the final forecasting factors corresponding to the linear module of the optimal forecasting model into the linear module of the optimal forecasting model, and calculating to obtain the sequence value forecasted by the linear module; if the calculated sequence value predicted by the linear module is not in the time range D of the growth period of the flower and fruit of the plant to be predicted1～D2In the method, the calculated sequence value predicted by the linear module is considered invalid, and the prediction is not carried out; if the calculated sequence value predicted by the linear module is in the time range D of the growth period of the flower and fruit of the plant to be predicted1～D2And considering that the calculated sequence value predicted by the linear module is valid, and executing the following steps:
step 5_2a, if Ds＜D1-DmaxOr Ds＞D2If yes, the forecast is not carried out; otherwise, step 5_2b is performed.
Step 5_2b, the order D of the current calculation datesjIs defined as the calculation order, wherein DjHas an initial value of Ds。
Step 5_2c, according to DjCalculating the factor of each final selected forecasting factor corresponding to the optimal forecasting modelAnd the sub-value is used for inputting the factor values of all final selection forecasting factors corresponding to the optimal forecasting model into the optimal forecasting model and calculating to obtain a forecasting value.
Step 5_2D, if the calculated forecast value is more than or equal to 0.5, when D is equal tojTime range D of flower and fruit growth period of plant to be forecasted1～D2When it is internal, consider DjIf the prediction result is valid, the prediction result is the prediction result of the sequence of the starting or ending time of the growth period of the flower and fruit of the plant to be predicted, and then step 6 is executed; when D is presentjTime range D of not in the growth period of flowers and fruits of the plants to be forecasted1～D2When it is internal, consider DjInvalid, not forecasting; if the calculated predicted value is obtained<0.5, step 5_2e is performed.
Step 5_2e, if Dj＞D2Or Dj＞Ds+DmaxIf yes, the forecast is not carried out; otherwise, let Dj＝Dj+1, taking the sequence of the next calculation date as the calculation sequence, and returning to the step 5_2c to continue the execution; wherein D isj＝DjIn +1, "═ is an assigned symbol.
Step 6: and converting the obtained forecast result into a date, and taking the date as the beginning or ending date of the growth period of the flower and fruit of the plant of the forecast year to finish the forecast.
In the present embodiment, in step a2 and step B7, the number of samples in the modeling sample set is 70% of the number of samples in the history sample set, and the number of samples in the testing sample set is 30% of the number of samples in the history sample set.
In the present embodiment, in step C2, the number of samples in the linear model modeling sample set is 70% of the number of samples in the linear model history sample set, and the number of samples in the linear model testing sample set is 30% of the number of samples in the linear model history sample set; in the step C13, the number of samples in the non-linear module modeling sample set is 70% of the number of samples in the non-linear module history sample set, and the number of samples in the non-linear module testing sample set is 30% of the number of samples in the non-linear module history sample set.

Claims (3)

1. A forecasting method for the starting and ending time of the growth period of plant flowers and fruits is characterized by comprising the following steps:
step 1: collecting historical data of the beginning or ending time of the growth period of the flowers and fruits of the plants, wherein the historical data comprises the sequence of the beginning or ending time of the growth period of the flowers and fruits of the plants in the past year, a plurality of meteorological factors related to the beginning or ending time of the growth period of the flowers and fruits of the plants in the past year, and a phenological factor related to the beginning or ending time of the growth period of the flowers and fruits of the plants in the past year, wherein the meteorological factor is a linear forecasting factor or a nonlinear forecasting factor, and the phenological factor is a linear forecasting factor or a nonlinear forecasting factor; if the linear forecasting factor and the nonlinear forecasting factor do not exist, the forecasting is not carried out, otherwise, the step 2 is executed;
the range of the past years is 10-30 years, the meteorological factors which are linear forecasting factors comprise forward accumulated temperature, reverse accumulated temperature, forward accumulated precipitation, reverse accumulated precipitation, forward accumulated humidity, reverse accumulated humidity, forward accumulated illumination hours and reverse accumulated illumination hours, the meteorological factors which are non-linear forecasting factors are obtained according to historical experience, and the phenological factors are phenological factors which are phenological periods of other plants or relatively early-maturing varieties;
step 2: estimating the earliest possible sequence and the latest possible sequence of the growth period start or end time of the flowers and fruits of the plants according to the sequence of the growth period start or end time of the flowers and fruits of the plants in the past year, and correspondingly marking the sequence as D1And D2(ii) a Wherein D is1＜D2；
And step 3: under the condition that only linear forecasting factors exist, calculating a linear correlation coefficient of each linear forecasting factor of the sequence of the growth start or end time of the flower and fruit of each year plant in the calendar year and the growth start or end time of the flower and fruit of the corresponding year plant; then, carrying out correlation significance check on all linear correlation coefficients; then taking the linear forecasting factors corresponding to all linear correlation coefficients passing through the correlation significance verification as candidate forecasting factors;
under the condition that only the nonlinear forecasting factors exist, all the nonlinear forecasting factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year are used as candidate forecasting factors;
under the condition that both the linear forecasting factors and the non-linear forecasting factors exist, calculating a linear correlation coefficient of each linear forecasting factor of the starting or ending time sequence of the growth period of the flower and fruit of each year plant in the calendar year and the starting or ending time of the growth period of the flower and fruit of the corresponding year plant; then, carrying out correlation significance check on all linear correlation coefficients; then, taking linear forecast factors corresponding to all linear correlation coefficients passing correlation significance verification and all nonlinear forecast factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year as candidate forecast factors;
and 4, step 4: selecting one model type from three model types of a linear model, a nonlinear model and a mixed model as a forecasting model construction type according to the candidate forecasting factors, wherein the mixed model consists of a linear module and a nonlinear module; if the candidate forecasting factors are all nonlinear forecasting factors, selecting a nonlinear model as a forecasting model construction type; if the candidate forecasting factors have both linear forecasting factors and nonlinear forecasting factors, selecting any model type of a nonlinear model and a mixed model as a forecasting model construction type, and when the mixed model is selected as the forecasting model construction type, only the linear forecasting factors can be selected by a linear module in the mixed model for construction; then, constructing an optimal forecasting model according to the forecasting model construction type;
if the construction type of the forecasting model is a linear model, the construction process of the optimal forecasting model is as follows:
a1, forming a history sample set by the sequence of the starting time or the ending time of the growth period of the flowers and fruits of the plants in the past year and all candidate forecasting factors of the starting time or the ending time of the growth period of the flowers and fruits of the plants in the past year;
a2, dividing a history sample set into two subsets according to the year, and respectively using the two subsets as a modeling sample set and a test sample set;
a3, establishing an initial forecasting model by adopting a regression equation method and utilizing a modeling sample set;
a4, inputting all candidate forecasting factors of each year in the test sample set into an initial forecasting model according to the year to obtain a sequential forecasting value of the corresponding year;
a5, calculating the difference value of the sequence forecast value of each year and the sequence of the corresponding year in the test sample set according to the year; then calculating the average value of the difference values corresponding to all the years; then judging whether the average value is less than or equal to a set value Ds1If so, determining the initial forecasting model as an optimal forecasting model, determining all candidate forecasting factors in the historical sample set as final forecasting factors, and executing the step 5; otherwise, go to step A6; wherein D iss1∈[1,3]Day;
a6, deleting the candidate forecasting factors corresponding to the minimum linear correlation coefficient in the historical sample set, then returning to the step A2 to continue execution, and if all the candidate forecasting factors in the historical sample set are processed and the optimal forecasting model is not determined, not forecasting;
if the construction type of the forecasting model is a nonlinear model, the construction process of the optimal forecasting model is as follows:
b1, defining the current year to be processed as the current year according to the year in the calendar year;
b2, D in the current year1～D2Defining the current sequence to be processed in the sequence as the current sequence;
b3, marking the current sequence as Dcur(ii) a Current sequence DcurAs an additional forecasting factor and the current order DcurWhether or not it is greater than or equal to D0The return value of (2) is used as a forecast object; wherein D iscur∈[D1,D2]，D0The sequence of the beginning or end of the growth period of the flower and fruit of the plant in the current year, D1＜D0＜D2If D iscur≥D0The forecast object value is 1, if Dcur＜D0The forecast object value is 0;
b4, forming a historical sample by the forecast object, the additional forecast factors and all candidate forecast factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year;
b5, D in the current year1～D2The next to-be-processed sequence in the sequence is taken as the current sequence, and then the step B3 is returned to continue execution until D in the current year1～D2After each sequence in the sequence is processed, executing the step B6;
b6, taking the next year to be processed in the past year as the current year, then returning to the step B2 to continue execution until each year of the past year is processed, and D in all the years in the past year1～D2All history samples corresponding to the sequence form a history sample set;
b7, dividing the historical sample set into two subsets according to the year, and respectively using the two subsets as a modeling sample set and a testing sample set;
b8, establishing an initial forecasting model by adopting an artificial neural network or a support vector machine method and utilizing a modeling sample set;
b9, D in each year in the set of test samples1～D2Inputting the additional forecasting factors and all candidate forecasting factors corresponding to the additional forecasting factors in the daily historical samples of the sequence into an initial forecasting model to obtain forecasting values corresponding to the additional forecasting factors, and setting the forecasting values to be 1 if the forecasting values are more than or equal to 0.5; if the predicted value is reported<Setting the forecast value to 0.5; judging whether the forecast value corresponding to the additional forecast factor is the same as the forecast object corresponding to the additional forecast factor in the test sample set, if so, indicating that the forecast is correct, and if not, indicating that the forecast is wrong; if the total prediction accuracy corresponding to all the historical samples in the test sample set reaches 80%, determining the initial prediction model as the optimal prediction model, determining all the candidate prediction factors in the historical sample set as final prediction factors, and then executing the step 5; otherwise, go to step B10;
b10, if the candidate forecasting factors in the historical sample set are all linear forecasting factors, deleting the linear forecasting factor corresponding to the minimum linear correlation coefficient in the historical sample set; if all the candidate forecasting factors in the historical sample set are nonlinear forecasting factors, deleting any nonlinear forecasting factor in the historical sample set; if the candidate forecasting factors in the historical sample set have linear forecasting factors and nonlinear forecasting factors, deleting the linear forecasting factor or any nonlinear forecasting factor corresponding to the minimum linear correlation coefficient in the historical sample set;
b11, returning to the step B7 to continue execution, and if all candidate forecasting factors in the historical sample set are processed and the optimal forecasting model is not determined, determining not to forecast;
if the construction type of the forecasting model is a mixed model, the construction process of the forecasting model is as follows:
c1, if the candidate forecasting factors are linear forecasting factors, forming a linear module history sample set by the sequence of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year and part of the candidate forecasting factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year; if the candidate forecasting factors have linear forecasting factors and nonlinear forecasting factors, all or part of linear forecasting factors of the starting or ending time of the growth period of the flowers and fruits of the plants in the past year and the starting or ending time of the growth period of the flowers and fruits of the plants in the past year form a linear module historical sample set;
c2, dividing the linear module historical sample set into two subsets according to the year, and respectively using the two subsets as a linear module modeling sample set and a linear module testing sample set;
c3, modeling a sample set by using a regression equation method and a linear module, and establishing an initial linear module;
c4, inputting all candidate forecasting factors of each year in the linear module test sample set into the initial linear module according to the year to obtain the sequential forecasting value of the corresponding year;
c5, calculating the difference value of the sequence forecast value of each year and the sequence of the corresponding year in the linear module test sample set according to the year; then calculating the average value of the difference values corresponding to all the years; then judging whether the average value is less than or equal to a set value Ds1If so, determining the initial linear module as an optimal linear module, determining all candidate forecasting factors in the linear module historical sample set as final forecasting factors of the optimal linear module, inputting all final forecasting factors of each year in the linear module historical sample set into the optimal linear module to obtain a sequential forecasting value of the corresponding year, and executing the step C7; otherwise, go to step C6; wherein D iss1∈[1,3]Day;
c6, deleting the candidate forecasting factors corresponding to the minimum linear correlation coefficient in the linear module historical sample set, then returning to the step C2 to continue execution, if all the candidate forecasting factors in the linear module historical sample set are processed and the optimal linear module is not determined, the linear module is invalid, the mixed model only contains the nonlinear module, and the forecasting model construction type is changed into the nonlinear model;
c7, defining the current year to be processed as the current year according to the year in the calendar year;
c8, D in the current year1～D2Defining the current sequence to be processed in the sequence as the current sequence;
c9, marking the current sequence as Dcur(ii) a Current sequence DcurAs a first additional forecasting factor and the current order DcurGreater than or equal to D0The return value of (2) is used as a forecast object; taking the sequential forecast value of the current year obtained by the optimal linear module in the step C5 as a second additional forecast factor; wherein D iscur∈[D1,D2]，D0The sequence of the beginning or end of the growth period of the flower and fruit of the plant in the current year, D1＜D0＜D2If D iscur≥D0The forecast object value is 1, if Dcur＜D0The forecast object value is 0;
c10, forming a nonlinear module history sample by the forecast object, the first additional forecast factor, the second additional forecast factor, the sequence of the growth start or end time of the flower and fruit of the plant in the past year and the remaining candidate forecast factors of the growth start or end time of the flower and fruit of the plant in the past year except for forming the linear module history sample set;
c11, D in the current year1～D2The next to-be-processed sequence in the sequence is taken as the current sequence, and then the step C9 is returned to continue execution until D in the current year1～D2After each of the sequences is processed, executing step C12;
c12, taking the next year to be processed in the past year as the current year, then returning to the step C8 to continue executing until each year of the past year is processed, and D in all the years in the past year1～D2All historical samples corresponding to the sequence form a non-linear module historical sample set;
c13, dividing the nonlinear module historical sample set into two subsets according to the year, and respectively using the two subsets as a nonlinear module modeling sample set and a nonlinear module testing sample set;
c14, establishing an initial nonlinear module by adopting an artificial neural network or a support vector machine method and utilizing a nonlinear module modeling sample set;
c15, checking the non-linear module for D in each year in the sample set1～D2Inputting a first additional forecasting factor and all candidate forecasting factors corresponding to the first additional forecasting factor in the historical sample of the day-by-day nonlinear module of the sequence into an initial nonlinear module to obtain a forecasting value corresponding to the first additional forecasting factor, and setting the forecasting value to be 1 if the forecasting value is more than or equal to 0.5; if the predicted value is reported<Setting the forecast value to 0.5; judging whether the forecast value corresponding to the first additional forecast factor is the same as the forecast object corresponding to the first additional forecast factor in the nonlinear module test sample set, if so, indicating that the forecast is correct, and if not, indicating that the forecast is wrong; if the overall prediction accuracy corresponding to all the non-linear module historical samples in the non-linear module test sample set reaches 80%, determining the initial non-linear module as the optimal non-linear module, determining all candidate prediction factors in the non-linear module historical sample set as final selection prediction factors of the optimal non-linear module, taking a mixed model formed by the optimal linear module and the optimal non-linear module as the optimal prediction model, and executing the step 5; otherwise, it is heldLine step C16;
c16, if the candidate forecasting factors in the non-linear module historical sample set are all linear forecasting factors, deleting the linear forecasting factor corresponding to the minimum linear correlation coefficient in the non-linear module historical sample set; if all the candidate forecasting factors in the non-linear module historical sample set are non-linear forecasting factors, deleting any non-linear forecasting factor in the non-linear module historical sample set; if the candidate forecasting factors in the non-linear module historical sample set have both linear forecasting factors and non-linear forecasting factors, deleting the linear forecasting factor or any one of the non-linear forecasting factors corresponding to the minimum linear correlation coefficient in the non-linear module historical sample set;
c17, returning to the step C13 to continue execution, and if all candidate forecasting factors in the non-linear module historical sample set are processed and the best non-linear module is not determined, determining not to forecast;
and 5: defining the year of the beginning or ending time of the growth period of the flower and fruit of the plant to be forecasted as a forecast year; defining the date sequence of the date of starting to make forecast as the starting date sequence and recording as Ds(ii) a Recording the day of the longest effective time of weather forecast provided by weather as Dmax；
When the type of the optimal forecasting model is a linear model, calculating the factor value of each final forecasting factor corresponding to the optimal forecasting model by using the data of the forecasting year, which are used for calculating the final forecasting factors, inputting the factor values of all the final forecasting factors corresponding to the optimal forecasting model into the optimal forecasting model, and calculating to obtain a sequence value; if the calculated sequence value is in the time range D of the growth period of the flowers and fruits of the plants to be forecasted1～D2If the calculated sequence value is valid, the result is a forecast result of the sequence of the starting time or the ending time of the growth period of the flowers and fruits of the plants, and then step 6 is executed; otherwise, the calculated sequence value is considered to be invalid, and the prediction is not carried out;
when the type of the best prediction model is a non-linear model, if Ds＜D1-DmaxOr Ds＞D2Then do notForecasting; otherwise, the following steps are executed:
step 5_1a, the order D of the dates currently calculatedjIs defined as the calculation order, wherein DjHas an initial value of Ds；
Step 5_1b, according to DjCalculating the factor value of each final selection forecasting factor corresponding to the optimal forecasting model, inputting the factor values of all final selection forecasting factors corresponding to the optimal forecasting model into the optimal forecasting model, and calculating to obtain a forecasting value;
step 5_1c, if the calculated forecast value is more than or equal to 0.5, when D is equal tojTime range D of flower and fruit growth period of plant to be forecasted1～D2When it is internal, consider DjIf the prediction result is valid, the prediction result is the prediction result of the sequence of the starting or ending time of the growth period of the flower and fruit of the plant to be predicted, and then step 6 is executed; when D is presentjTime range D of not in the growth period of flowers and fruits of the plants to be forecasted1～D2When it is internal, consider DjInvalid, not forecasting; if the calculated predicted value is obtained<0.5, executing the step 5_1 d;
step 5_1D, if Dj＞D2Or Dj＞Ds+DmaxIf yes, the forecast is not carried out; otherwise, let Dj＝Dj+1, taking the sequence of the next calculation date as the calculation sequence, and returning to the step 5_1b for continuous execution; wherein D isj＝DjThe "═ in + 1" is an assignment symbol;
when the type of the optimal forecasting model is a mixed model, calculating the factor value of each final forecasting factor corresponding to the linear module of the optimal forecasting model by using the data of the forecasting year, which are used for calculating the final forecasting factors, inputting the factor values of all the final forecasting factors corresponding to the linear module of the optimal forecasting model into the linear module of the optimal forecasting model, and calculating to obtain the sequence value forecasted by the linear module; if the calculated sequence value predicted by the linear module is not in the time range D of the growth period of the flower and fruit of the plant to be predicted1～D2In the method, the calculated sequence value predicted by the linear module is considered invalid, and the prediction is not carried out; if calculated to obtainThe time range D of the sequence value predicted by the linear module in the growth period of the flower and fruit of the plant to be predicted1～D2And considering that the calculated sequence value predicted by the linear module is valid, and executing the following steps:
step 5_2a, if Ds＜D1-DmaxOr Ds＞D2If yes, the forecast is not carried out; otherwise, executing step 5_2 b;
step 5_2b, the order D of the current calculation datesjIs defined as the calculation order, wherein DjHas an initial value of Ds；
Step 5_2c, according to DjCalculating the factor value of each final selection forecasting factor corresponding to the optimal forecasting model, inputting the factor values of all final selection forecasting factors corresponding to the optimal forecasting model into the optimal forecasting model, and calculating to obtain a forecasting value;
step 5_2D, if the calculated forecast value is more than or equal to 0.5, when D is equal tojTime range D of flower and fruit growth period of plant to be forecasted1～D2When it is internal, consider DjIf the prediction result is valid, the prediction result is the prediction result of the sequence of the starting or ending time of the growth period of the flower and fruit of the plant to be predicted, and then step 6 is executed; when D is presentjTime range D of not in the growth period of flowers and fruits of the plants to be forecasted1～D2When it is internal, consider DjInvalid, not forecasting; if the calculated predicted value is obtained<0.5, executing the step 5_2 e;
step 5_2e, if Dj＞D2Or Dj＞Ds+DmaxIf yes, the forecast is not carried out; otherwise, let Dj＝Dj+1, taking the sequence of the next calculation date as the calculation sequence, and returning to the step 5_2c to continue the execution; wherein D isj＝DjThe "═ in + 1" is an assignment symbol;
step 6: and converting the obtained forecast result into a date, and taking the date as the beginning or ending date of the growth period of the flower and fruit of the plant of the forecast year to finish the forecast.
2. A method as claimed in claim 1, wherein in step a2 and step B7, the number of samples in the modeling sample set is 70% of the number of samples in the history sample set, and the number of samples in the testing sample set is 30% of the number of samples in the history sample set.
3. A method as claimed in claim 2, wherein in step C2, the number of samples in the linear model modeling sample set is 70% of the number of samples in the linear model history sample set, and the number of samples in the linear model testing sample set is 30% of the number of samples in the linear model history sample set; in the step C13, the number of samples in the non-linear module modeling sample set is 70% of the number of samples in the non-linear module history sample set, and the number of samples in the non-linear module testing sample set is 30% of the number of samples in the non-linear module history sample set.
CN201910474904.0A 2019-06-03 2019-06-03 Method for forecasting starting and ending time of growth period of plant flowers and fruits Active CN110276482B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910474904.0A CN110276482B (en) 2019-06-03 2019-06-03 Method for forecasting starting and ending time of growth period of plant flowers and fruits

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910474904.0A CN110276482B (en) 2019-06-03 2019-06-03 Method for forecasting starting and ending time of growth period of plant flowers and fruits

Publications (2)

Publication Number Publication Date
CN110276482A CN110276482A (en) 2019-09-24
CN110276482B true CN110276482B (en) 2020-02-21

Family

ID=67960385

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910474904.0A Active CN110276482B (en) 2019-06-03 2019-06-03 Method for forecasting starting and ending time of growth period of plant flowers and fruits

Country Status (1)

Country Link
CN (1) CN110276482B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113348943B (en) * 2021-06-17 2022-12-09 宁波市奉化区气象局 Method for determining cold demand and heat demand of peach trees based on meteorological data
CN113761708A (en) * 2021-07-21 2021-12-07 南京林业大学 Flowering phase forecasting method based on rolling weather forecast

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106960265A (en) * 2017-04-06 2017-07-18 侯鲁文 Utilize the method that March, average air temperature value carried out Tokyo oriental cherry Florescence forecast
CN107356569A (en) * 2017-06-06 2017-11-17 河南农业大学 The construction method of method and its model based on chlorophyll fluorescence prediction wheat grain yield
CN109345016A (en) * 2018-09-30 2019-02-15 成都信息工程大学 Wisdom serve for agriculture system and method based on Fructus Forsythiae prediction of flowering period

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080077487A1 (en) * 2006-09-21 2008-03-27 Mark Davis Targeted Incentives Based Upon Predicted Behavior
CN109325630B (en) * 2018-10-12 2022-04-29 南京农业大学 Morphological parameter-based rice yield prediction method under high-temperature stress

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106960265A (en) * 2017-04-06 2017-07-18 侯鲁文 Utilize the method that March, average air temperature value carried out Tokyo oriental cherry Florescence forecast
CN107356569A (en) * 2017-06-06 2017-11-17 河南农业大学 The construction method of method and its model based on chlorophyll fluorescence prediction wheat grain yield
CN109345016A (en) * 2018-09-30 2019-02-15 成都信息工程大学 Wisdom serve for agriculture system and method based on Fructus Forsythiae prediction of flowering period

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
Design of an image analysis website for phenological and meteorological monitoring;E. Bradley 等;《Environmental Modelling & Software》;20100131;第25卷(第1期);第107-116页 *
浙江水蜜桃成熟期集合预报模型;杨栋 等;《中国农业气象》;20180630;第39卷(第6期);第421-430页 *
白玉兰始花期与气象因子的关系分析;舒素芳 等;《浙江农业学报》;20130208;第25卷(第2期);第248-251页 *
莎车巴旦姆物候期对气象条件的响应及花期预测模型;徐相明 等;《应用生态学报》;20160227;第27卷(第2期);第421-428页 *

Also Published As

Publication number Publication date
CN110276482A (en) 2019-09-24

Similar Documents

Publication Publication Date Title
Rafferty et al. Changing climate drives divergent and nonlinear shifts in flowering phenology across elevations
Matthews Modeling the impact of climate change on rice production in Asia
CN111932402B (en) Short-term power load bidirectional combination prediction method based on similar day and LSTM
CN108877905B (en) Hospital outpatient quantity prediction method based on Xgboost framework
Gao et al. Rice clock model—a computer model to simulate rice development
CN110276482B (en) Method for forecasting starting and ending time of growth period of plant flowers and fruits
CN109800921A (en) A kind of Regional Fall Wheat yield estimation method based on remote sensing phenology assimilation and particle swarm optimization algorithm
CN112215716A (en) Crop growth intervention method, device, equipment and storage medium
CN110246051A (en) The method that crop cycle model and variety classes agricultural product establish growth cycle
Qiu et al. Latitudinal pattern of flowering synchrony in an invasive wind-pollinated plant
Pearson et al. Phenological trends in the California poppy (Eschscholzia californica): digitized herbarium specimens reveal intraspecific variation in the sensitivity of flowering date to climate change
Richardson et al. Remotely-sensed spectral indicators of sorghum development and their use in growth modeling
CN115688997A (en) Accumulated temperature-based tea leaf picking period prediction method and system
CN109615150B (en) Method and system for determining rice meteorological output
CN109615149B (en) Method and system for determining beet meteorological yield
US20100286973A1 (en) Real-Time Process For Targeting Trait Phenotyping of Plant Breeding Experiments
CN109858678B (en) Method and system for determining meteorological yield of sunflowers
CN114219183A (en) Construction method of northern runoff litchi yield major-minor year type grade region prediction model based on meteorological conditions
CN113935542A (en) Method for predicting cotton yield per unit based on climate suitability
CN109840623B (en) Method and system for determining meteorological yield of sesame
JP2021174062A (en) Machine learning program, machine learning method, and machine learning apparatus
CN117436710B (en) Drought assessment method based on crop model
CN113762768B (en) Agricultural drought dynamic risk assessment method based on natural gas generator and crop model
Kim Development of examination model of weather factors on garlic yield using big data analysis
CN114091331A (en) Method and system for simulating forest leaf expanding period by using meteorological factors

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant