CN115860285B - Prediction method and device for optimal transplanting period of tobacco - Google Patents
Prediction method and device for optimal transplanting period of tobacco Download PDFInfo
- Publication number
- CN115860285B CN115860285B CN202310183279.0A CN202310183279A CN115860285B CN 115860285 B CN115860285 B CN 115860285B CN 202310183279 A CN202310183279 A CN 202310183279A CN 115860285 B CN115860285 B CN 115860285B
- Authority
- CN
- China
- Prior art keywords
- time
- tobacco
- data set
- factor
- prediction
- 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
Links
- 235000002637 Nicotiana tabacum Nutrition 0.000 title claims abstract description 109
- 241000208125 Nicotiana Species 0.000 title claims abstract description 101
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000009331 sowing Methods 0.000 claims abstract description 16
- 238000011156 evaluation Methods 0.000 claims description 41
- 239000002352 surface water Substances 0.000 claims description 29
- 238000001556 precipitation Methods 0.000 claims description 18
- 238000004364 calculation method Methods 0.000 claims description 17
- 238000012545 processing Methods 0.000 claims description 15
- 230000001186 cumulative effect Effects 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 11
- 238000007781 pre-processing Methods 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 8
- 238000010899 nucleation Methods 0.000 claims description 6
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 4
- 238000002310 reflectometry Methods 0.000 claims description 4
- 238000012216 screening Methods 0.000 claims description 4
- 239000002344 surface layer Substances 0.000 claims description 4
- 238000004590 computer program Methods 0.000 claims description 3
- 238000003860 storage Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 claims description 2
- 238000005520 cutting process Methods 0.000 claims description 2
- 238000011160 research Methods 0.000 abstract description 6
- 238000009825 accumulation Methods 0.000 abstract description 2
- 239000010410 layer Substances 0.000 description 11
- 244000061176 Nicotiana tabacum Species 0.000 description 8
- 238000009826 distribution Methods 0.000 description 7
- 238000011161 development Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 238000004519 manufacturing process Methods 0.000 description 4
- 239000002689 soil Substances 0.000 description 4
- 238000010276 construction Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000000701 chemical imaging Methods 0.000 description 1
- 235000019504 cigarettes Nutrition 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application provides a method and a device for predicting the optimal transplanting period of tobacco, belonging to the field of tobacco planting management, and comprising the following steps: the method comprises the steps of combining a sensor-2 remote sensing image and various meteorological data, constructing a multi-factor dataset for tobacco transplanting period prediction, combining data such as a history record and the like, determining an optimum transplanting period judgment standard in a mode of superposition of a plurality of factor accumulation values, and predicting each factor through an LSTM prediction model, so that dynamic prediction of the optimum transplanting period of tobacco is realized. The scheme uses a mode of superposition of accumulated values of multiple factors from the sowing time to the most suitable transplanting time to determine the final most suitable transplanting time judging standard, and compared with the existing research, the method adopts a mode of determining by adopting single or multiple factor value thresholds, and effectively improves judging accuracy and anti-interference capability.
Description
Technical Field
The application belongs to the field of tobacco planting management, and particularly relates to a method and a device for predicting the optimal transplanting period of tobacco.
Background
The quality of tobacco leaves has decisive influence on the quality of cigarette products, and the production of high-quality tobacco leaves not only needs proper ecological conditions, but also needs to fully utilize limited natural resources according to the growth and development characteristics of tobacco leaves. The distribution of climate factors in different growth periods in the tobacco field plays a vital role in the production of high-quality tobacco, and the climate conditions mainly comprise temperature, precipitation and illumination, and the proper climate conditions are necessary for the tobacco to complete normal growth and development. In addition, the relative humidity can influence the tobacco morbidity, and excessive relative humidity in the early stage of tobacco field growth period can obviously increase the tobacco field morbidity.
The transplanting period is an important link in tobacco cultivation measures, and has obvious influence on the field characters, yield and quality of tobacco. The tobacco plants in different transplanting periods have different climatic conditions such as temperature, illumination, rainfall and the like in each growth period, so that the growth and development of the tobacco plants and the production quality are greatly influenced. Transplanting is a key link of tobacco production, if the transplanting is too early, the tobacco plants after the transplanting are in a low-temperature illumination condition for a long time, the normal growth and development of the tobacco plants can be seriously influenced, and even early flowers appear; if the tobacco plants are transplanted too late, the tobacco plants are in a relatively high-temperature and high-humidity condition in the early stage after the tobacco plants are transplanted, the growth and development speed is too high, the dry matters are not accumulated enough, the leaves are thinner, the temperature of the tobacco plants in the late stage of the growth is reduced, the normal maturity of the leaves is affected, and the yield and quality of tobacco leaves are reduced. Therefore, a proper transplanting period is one of key factors of high quality and high yield of tobacco.
A method and system for determining the date of tobacco transplants in the fujian tobacco sector (patent application number: CN 201910445302.2) is disclosed, comprising: fitting a function of the change of the average temperature of the day of the previous year along with the date according to the average temperature data of the day of the previous natural year; establishing a prediction function of the annual daily average temperature variation along with the date according to the fitted daily average temperature variation along with the date; and determining the tobacco transplanting date based on a specific rule according to the prediction function of the annual daily average temperature variation along with the date of the agriculture. The scheme adopts a single factor (temperature) mode to determine the tobacco transplanting period, is easy to be interfered by other factors, has poor anti-interference capability and lower accuracy.
The method comprises the steps of constructing a simulation method of a spatial-temporal change pattern of weather and soil moisture in a tobacco region by ten days through mathematical modeling by utilizing MODIS remote sensing image data and DEM elevation data and combining tobacco region ground observation data, extracting a spatial-temporal distribution pattern of main weather elements and soil moisture, combining optimum weather factors and soil moisture ranges required by tobacco leaf transplanting, and determining the spatial distribution pattern of the optimum transplanting time of the tobacco region. The scheme uses MODIS remote sensing image data for data processing, but the data is usually used for large-scale research at the global or national level, the data product is not suitable for research in a small scale range at the regional level, the result is lack of effectiveness, and a transplanting period prediction result with high spatial resolution and continuous spatial distribution cannot be generated well.
Disclosure of Invention
The application provides a method and a device for predicting the optimal transplanting period of tobacco, and aims to solve the problems that the method for determining the transplanting period of tobacco by adopting a single factor is easy to be interfered by other factors, has poor anti-interference capability and low accuracy, and cannot well generate a transplanting period prediction result with high spatial resolution and continuous spatial distribution.
In order to achieve the above object, the present application adopts the following technical scheme, including:
acquiring meteorological data and remote sensing image data of a tobacco field to be detected, and performing first preprocessing to obtain a first historical data set and a second predicted data set which are divided according to time attributes;
acquiring the sowing time and the optimal transplanting period of tobacco in a tobacco field to be detected in the past year, obtaining a first time interval, intercepting a first historical data set based on the first time interval, and performing second pretreatment to obtain an evaluation standard of the optimal transplanting period of tobacco;
respectively constructing a prediction model for each factor in the first historical data set based on the first historical data set and adopting a mode of inputting at multiple time points and outputting at multiple time points, and establishing an LSTM prediction model, wherein each factor is respectively air temperature, precipitation, sunshine hours and surface layer moisture content index;
inputting the second prediction data set into an LSTM prediction model for calculation to obtain predicted values of all factors of continuous R days after the current time node, calculating a first accumulated value of all factors between the annual seeding time and the last day of prediction time node according to the predicted values of all factors, comparing and analyzing the first accumulated value with an evaluation standard to obtain first prediction time and second prediction time, and obtaining a predicted result of the optimal tobacco transplanting period in the tobacco field to be detected by taking the first prediction time as a starting point and the second prediction time as an ending point, wherein R is a natural number.
Preferably, the first pretreatment is specifically:
sequentially carrying out time screening, space screening and cutting, cloud removal processing and abnormal value removal of images on the remote sensing image data to obtain processed remote sensing image data;
calculating a surface water content index based on the processed remote sensing image data to obtain SCWI time sequence data, and performing time sequence reconstruction on the SCWI time sequence data by using an S-G filtering algorithm to obtain space-time continuous SCWI data, wherein a calculation formula of the surface water content index is as followsWherein SCWI is the surface water content index, and b11 and b12 respectively correspond to the reflectivity of the 11 th wave band and the 12 th wave band of Sentinel-2;
dividing meteorological DATA and space-time continuous SCWI DATA according to N-1 years and N years to respectively obtain N-1 year meteorological DATA, surface water content index, N year meteorological DATA and surface water content index, combining N-1 year DATA to obtain a DATA set DATA1, performing Kriging interpolation processing on N year meteorological DATA in space and combining the N year surface water content index to obtain a DATA set DATA2, wherein N is the current year, N-1 is the past year, the DATA set DATA1 is a first historical DATA set, the DATA set DATA2 is a second predicted DATA set, and the meteorological DATA comprises air temperature, precipitation and sunshine hours.
Preferably, the second pretreatment is specifically:
sequentially calculating the accumulated values of all factors in the first time interval of each year by using the intercepted first historical data set, and summarizing to obtain a second accumulated value;
respectively summarizing and counting the second accumulated values according to the types of the factors to obtain confidence intervals of the accumulated values of the factors;
performing superposition analysis on the confidence intervals of the accumulated values of the factors to obtain a time intersection;
and obtaining the accumulated value range of each factor based on the intersection in time so as to establish the evaluation standard of the optimal transplanting period of the tobacco.
Preferably, the calculation formula in the second pretreatment process is as follows
Wherein Ax represents the x factor, i.e. the set of accumulated values of stations in each year from ts to te in the past year, N is the current year, N-1 is the past year, ts is the sowing time, te is the most suitable transplanting period>The value of factor x on day t, y represents the y-th year of the previous N-1 years, i represents the i-th weather site, μ Ax Sum sigma Ax The mean and standard deviation of the x-factor cumulative values are shown, bAx and pAx are the lower and upper limits of the confidence interval of the x-factor cumulative value, T bAx And T pAx Respectively representing the time corresponding to the lower limit and the upper limit of the confidence interval of the accumulated value of each factor, at is the air temperature, pr is the precipitation, su is the sunshine hours, wa is the surface water content index, T b And Tp is the intersection of the upper and lower time limits, and the factor x is one of the factors.
Preferably, the first historical data set is divided into a training data set and a test data set, which are used for construction and verification of the LSTM prediction model, respectively.
Preferably, the first accumulated value and the evaluation standard are compared and analyzed to obtain a first predicted time and a second predicted time, specifically:
comparing the first accumulated value with an evaluation standard, recording a first time point when the first accumulated values are in the range of the evaluation standard, and recording a second time point when the first accumulated value of any factor is greater than the upper limit of the evaluation standard;
the first predicted time and the second predicted time are obtained based on the first time point and the second time point.
Preferably, the first cumulative value of each factor is obtained by overlapping each factor cumulative value of the current time node with the predicted value of each factor in the continuous R days day by day.
A device for predicting an optimal transplanting period of tobacco, comprising:
a data set dividing module: the method comprises the steps of acquiring meteorological data and remote sensing image data of a tobacco field to be detected, and performing first preprocessing to obtain a first historical data set and a second predicted data set which are divided according to time attributes;
the evaluation standard establishment module: the method comprises the steps of acquiring the last annual sowing time and the optimal transplanting period of tobacco in a tobacco field to be detected, obtaining a first time interval, intercepting a first historical data set based on the first time interval, and performing second pretreatment to obtain an evaluation standard of the optimal transplanting period of tobacco;
the prediction model building module: the method comprises the steps of respectively constructing a prediction model for each factor in a first historical data set based on the first historical data set and adopting a mode of inputting at multiple time points and outputting at multiple time points, and establishing an LSTM prediction model for each factor which is respectively air temperature, precipitation, sunshine hours and surface water content index;
and calculating an optimal transplanting period: the method comprises the steps of inputting a second prediction data set into an LSTM prediction model for calculation to obtain predicted values of all factors of continuous R days after a current time node, calculating a first accumulated value of all factors between a annual seeding time and a final day prediction time node according to the predicted values of all factors, comparing and analyzing the first accumulated value with an evaluation standard to obtain a first prediction time and a second prediction time, and obtaining a prediction result of the optimal tobacco transplanting period in a tobacco field to be detected by taking the first prediction time as a starting point and the second prediction time as an end point, wherein R is a natural number.
An electronic device comprising a memory and a processor, the memory configured to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement a method of predicting an optimal transplanting period of tobacco as defined in any one of the above.
A computer readable storage medium storing a computer program which when executed by a computer implements a method of predicting an optimal transplanting period of tobacco as defined in any one of the above.
The application has the following beneficial effects:
(1) The 11 and 12 wave bands of the Sentinel-2 data used in the scheme have the spatial resolution of 20m, and the block scale of the regional small scale range can be achieved through interpolation of the meteorological site data, so that more refined prediction is realized, and the transplanting period prediction result of the spatial continuous distribution with high spatial resolution can be generated;
(2) According to the scheme, a large number of related researches are referred, a plurality of factors (air temperature, precipitation, sunshine hours and surface water content indexes) related to the tobacco transplanting period are screened out, and the transplanting period of the tobacco is predicted comprehensively by adopting a plurality of factors, so that the result consideration is more comprehensive and reliable;
(3) In the aspect of the evaluation standard of the suitable transplanting period, a mode of superposing the accumulated values of a plurality of factors from the sowing period to the most suitable transplanting period is used for determining the final most suitable transplanting period evaluation standard, and compared with the existing research, the mode of determining by adopting a single or multi-factor value threshold value, the evaluation accuracy and the anti-interference capability are effectively improved;
(4) The stable information of a plurality of factors in the process of year-to-year change of tobacco during sowing and transplanting is captured by using the time series data of many years and adopting an LSTM model, so that the prediction of each factor is more accurately realized, the uncontrollable accidental factor interference is reduced, and the prediction accuracy of the optimum transplanting period is enhanced.
Drawings
FIG. 1 is a flow chart of a method for predicting the optimal transplanting period of tobacco in the application;
FIG. 2 is a schematic diagram of the embodiment of the application in example 1;
FIG. 3 is a schematic diagram of the reconstruction of time series in embodiment 1 of the present application;
FIG. 4 is a schematic diagram showing the determination of the optimum transplanting period of tobacco in example 1 of the present application;
fig. 5 is a schematic structural diagram of a device for predicting the optimal transplanting period of tobacco in the application.
Detailed Description
Example 1
As shown in fig. 1, a method for predicting the optimal transplanting period of tobacco comprises the following steps:
s11, acquiring meteorological data and remote sensing image data of a tobacco field to be detected, and performing first preprocessing to obtain a first historical data set and a second predicted data set which are divided according to time attributes;
s12, acquiring the annual sowing time and the optimal transplanting period of tobacco in a tobacco field to be detected, obtaining a first time interval, intercepting a first historical data set based on the first time interval, and performing second pretreatment to obtain an evaluation standard of the optimal transplanting period of tobacco;
s13, respectively constructing a prediction model for each factor in the first historical data set based on the first historical data set and adopting a mode of inputting at multiple time points and outputting at multiple time points, and establishing an LSTM prediction model for each factor which is respectively the air temperature, precipitation, sunshine hours and surface water content index;
s14, inputting the second prediction data set into an LSTM prediction model for calculation to obtain predicted values of all factors of continuous R days after the current time node, calculating a first accumulated value of all factors between the annual seeding time and the predicted time node of the last day according to the predicted values of all factors, comparing and analyzing the first accumulated value with an evaluation standard to obtain first prediction time and second prediction time, and obtaining a predicted result of the optimal tobacco transplanting period in the tobacco field to be detected by taking the first prediction time as a starting point and the second prediction time as an ending point, wherein R is a natural number.
In the scheme, a multi-factor dataset for tobacco transplanting period prediction is constructed by combining a sensor-2 remote sensing image and various meteorological data, a history record and other data are combined, an optimal transplanting period judgment standard is determined by adopting a mode of superposition of a plurality of factor accumulation values, and each factor is predicted by an LSTM model, so that the dynamic prediction of the optimal transplanting period of tobacco is realized. The thinking guide chart is shown in fig. 2, and the specific flow is as follows: 1. data acquisition and processing
1.1 Meteorological data
The meteorological data is derived from the official meteorological site data (the number of the meteorological sites is a plurality of) of the China meteorological bureau, and the meteorological data comprises daily air temperature, precipitation and sunshine hours data.
1.2sentinel-2
Sentinel-2 is a multispectral imaging satellite and is divided into two satellites, namely 2A and 2B, wherein 2A is launched and lifted off by a 'Zhenjingshi' carrier rocket in 2015, 6 months, 23 days, 01:52UTC, and 2B is launched and lifted off by a 'Zhenjingshi' carrier rocket in 2017, 3 months, 07 days, beijing time, 9 minutes, 49 minutes UTC. The revisitation period of one satellite is 10 days, the two complementary satellites are 5 days. The Sentinel-2 satellite carries a multispectral instrument (MSI) which can cover 13 spectral bands from visible light to short wave infrared, and the ground resolutions are respectively 10m, 20m and 60m. The scheme selects the Level-2A product as one of basic data of subsequent work, and the product is the atmospheric bottom layer reflectivity data subjected to atmospheric correction.
In order to reduce the time of data processing to the greatest extent, the working efficiency is improved. The scheme utilizes a GEE (Google earth engine) remote sensing cloud computing platform to complete data acquisition and processing. In terms of data acquisition, first, annual Sentinel-2L2A data is temporally screened in a specific time range (described later), then the data is spatially screened and cut by using vector data in a research area range, then a cloud mask is constructed by using a QA60 band of an image to perform cloud removal processing on the image, and on the basis, the identification of crops is influenced by taking account of an abnormal value of the image, so that the abnormal value of the image is removed by further using an SLC band. Related studies have demonstrated that the surface moisture index (SCWI) is sensitive to changes in soil moisture, and this scheme uses SCWI to describe tobacco field moisture content, which is calculated as shown in equation 1 below
Wherein SCWI is the surface water content index, and b11 and b12 respectively correspond to the reflectivity of the 11 th band and the 12 th band of Sentinel-2.
In order to avoid interference of the preprocessed null region on the integrity expression of the SCWI time sequence data, the scheme carries out time sequence reconstruction processing on the calculated SCWI time sequence data through S-G filtering. The time series reconstruction diagram is shown in fig. 3, in which square data points represent the original SCWI values (i.e., calculated SCWI time series data), and circular data points represent the SCWI values after reconstruction processing, which are continuous after reconstruction. The S-G filtering is an algorithm which is firstly proposed by Savizky and Golay and then widely applied to reconstructing time series data, and the value in a window is weighted and averaged in a sliding window mode by setting the size of the time window with a certain length, so that the value at the moment to be reconstructed is obtained. The basic formula is shown in formula 2:
wherein SCWI i Represents the fitting value of SCWI at the i-th time, SCWI i+j Representing the original value of SCWI at the i+j time, N being the window size, satisfying n=2m+1; w (w) j The coefficients representing the S-G polynomial fit have values that depend on the least squares fit of the given higher order polynomial within a filter window.
In a specific reconstruction process, only the data missing position is obtained by a mask mode, then time sequence reconstruction is carried out, so that the interference to the original data values of other positions is reduced as much as possible, and SCWI data which are continuous in time-space every year is obtained after processing.
1.3 data integration
All data is divided into two parts (the first N-1 year and the current year N) according to the time attribute. For the DATA of the previous N-1 year, firstly, the spatial position information of a site (site number M) is utilized, the surface water content index DATA is combined, the surface water content index value of the site is obtained, then the air temperature, precipitation and sunshine hours DATA of the site are combined to form a DATA set DATA1 (namely a first historical DATA set), the DATA expression form of the DATA is shown in a table 1, and the DATA are mainly used for training of a prediction model of each factor (air temperature: at; precipitation: pr; sunshine hours: su; surface water content index: wa) of the surface water content index of the surface water index of the site, and the surface water content index of the site is combined to form an optimal transplanting period evaluation standard. For the DATA of the nth year, the temperature, precipitation and sunshine hours DATA of the meteorological site are interpolated in a Kriging way in space, so that the interpolated grid captures the grid layer of the surface water content index, the grid layer of the surface water content index is sampled to be consistent with the grid of the surface water content index in space resolution, the four DATA together form a DATA set DATA2 (namely a second prediction DATA set), the DATA set is identical to the DATA set DATA1 in appearance, but is continuous in space due to interpolation, and the DATA is used for predicting various factors in the unknown time period of the nth year and determining the optimum transplanting period by combining the prediction result and the evaluation standard of the optimum transplanting period.
TABLE 1 Integrated schematic form of various factor data
The scheme in the above 1.2 and 1.3 is a specific process of the first pretreatment.
2. Evaluation criterion of optimum transplanting period (for different varieties of tobacco and tobacco field vector range)
And taking the accumulated value of each factor between the sowing period and the optimal transplanting period as an evaluation standard of the optimal transplanting period.
According to the historical record DATA and the related literature DATA, in the previous N-1 year, acquiring the sowing time ts and the optimal transplanting period te of a certain variety of tobacco C in each year, intercepting a DATA set DATA1 in the range of the year by using the time range ts to te (namely a first time interval), and then calculating the accumulated value of each factor in the time range ts to te of each year by using the intercepted DATA1 DATA; the same method calculates the accumulated value of all factors in the previous N-1 years, namely a second accumulated value, and the calculation process is shown in a formula 3; then, summarizing and counting all factor accumulated values of all meteorological sites in all years respectively to obtain a confidence interval of each factor accumulated value, wherein the calculation process is shown in a formula 4; further acquiring the intersection of confidence intervals of the accumulated values of the factors in time, wherein the calculation process is shown in a formula 5; finally, the accumulated value range of each factor corresponding to the time intersection is obtained and used as an evaluation standard of the optimum transplanting period of the variety of tobacco, and the calculation process is shown in a formula 6. Fig. 4 shows the determination process of the optimum transplanting period intuitively, in which the graph is an exemplary graph of any factor of the factors, that is, any factor is obtained by determining the range from the maximum accumulated value to the minimum accumulated value (the range is the confidence interval of the factor), so as to determine the optimum transplanting time of a single factor, repeating all the factors to obtain the optimum transplanting time of each factor, and taking the intersection of all the optimum transplanting times to obtain the evaluation standard of the optimum transplanting period.
Wherein Ax represents the x factor, N-1 years ago, i.e. the set of accumulated values of stations in each year in the past year in the time from ts to te, N is the current year, N-1 is the past year, ts is the sowing time, te is the most suitable transplanting period,the value of factor x on day t, y represents the y-th year of the previous N-1 years, i represents the i-th weather site, μ Ax Sum sigma Ax The mean and standard deviation of the x-factor cumulative values are shown, bAx and pAx are the lower and upper limits of the confidence interval of the x-factor cumulative value, T bAx And T pAx Respectively representing the time corresponding to the lower limit and the upper limit of the confidence interval of the accumulated value of each factor, at is the air temperature, pr is the precipitation, su is the sunshine hours, wa is the surface water content index, T b And Tp is the intersection of the upper and lower time limits, and the factor x is one of the factors.
The above-mentioned process of processing the DATA set DATA1 (first historical DATA set) to obtain the evaluation criteria of the most suitable transplanting period of tobacco is a specific process of the second pretreatment, and the above-mentioned formulas 3, 4, 5 and 6 are calculation formulas used in the calculation of the second pretreatment process.
3. LSTM prediction model
The LSTM is an excellent time sequence prediction model derived on the basis of a traditional cyclic neural network (RNN), solves the problems of gradient elimination, long-term dependence and the like existing in the process of processing data by the RNN, can well grasp the change rule of the data in the process of processing data such as weather factors of long-time sequences and gives a more accurate prediction result.
The construction of the LSTM prediction model of four factors involved in the scheme mainly comprises four steps: (1) The DATA1 DATA set in 1.3 is further sorted into an input DATA set and divided into a training DATA set DATA1_Train and a Test DATA set DATA1_test; (2) Constructing an LSTM network model, and confirming parameters and targets required to be used in model training and prediction; (3) Training the model by using training data, and establishing a prediction model of each factor; (4) inspecting the model with the test dataset.
In order to predict daily DATA of each factor, the scheme adopts a mode of inputting at multiple time points and outputting at multiple time points to respectively construct models for different factors, respectively arranges DATA1_Train time sequence numbers containing each factor into m groups of input sequence DATA, standardizes the input DATA, takes DATA with fixed length in front of each group of DATA as an input part of a network, and takes DATA at the time points needing to be predicted as theoretical output values of the models. The hidden layer is formed by combining the double-layer LSTM and the Dense full-connection layer together so as to weaken the influence of the data generated by the large-range dynamic property as far as possible. Because the scheme trains the models of the factors respectively, the feature dimension of the input and the output of the prediction model of each factor is 1. The learning rate is adjusted finely after confirming the magnitude of the change in loss value (loss) from 0.1 to 0.000001. The training times are determined according to the convergence of the loss values with the increase of the training times. The number of nodes of the hidden layer is then determined according to empirical formula 7 from prior studies.
Wherein l is the number of hidden layer nodes, alpha and beta are the number of input layer nodes and output layer nodes respectively, mu is an integer between 0 and 9, and a traversal optimizing method is used for finding the l value corresponding to the minimum RMSE.
And continuously updating the weight of the network model by using the determined parameters and the gradient descent algorithm in the process of training the data to obtain the hidden layer network. Finally, the Test DATA set DATA1_Test is predicted by using a trained network, and is denormalized, then a predicted value of a predicted time point is output, and then the predicted value is compared with an actual value, so that the effect of the model is checked.
4. Spatial distribution of optimal transplanting period
The DATA set DATA2 (second prediction DATA set) is used as input DATA of the LSTM prediction model trained in the section 3, the values of all factors of the continuous R days after the current time node are predicted, then the accumulated value Ax (namely the first accumulated value) of all factors between the sowing time of the year (the N year) and the prediction time node of the last day is calculated, the range (namely the evaluation standard) of the accumulated values of all factors of the optimal transplanting time obtained in the section 2 is compared with the first accumulated value Ax, and under the two conditions that all values of Ax are respectively in the range shown in the formula 6 and one Ax value is larger than the upper limit of the formula 6, two prediction days R1 and R2 are obtained, and R2> R1 is calculated, so that the optimal transplanting date of the tobacco variety is obtained in the days R1 to R2 from the current date, and the optimal transplanting time is closer to the current date, and the prediction result is more accurate.
The process comprises the following steps: comparing the first accumulated value with an evaluation standard, recording a first time point when the first accumulated values are in the range of the evaluation standard, and recording a second time point when the first accumulated value of any factor is greater than the upper limit of the evaluation standard; the first time point is R1, and the second time point is R2;
the first predicted time and the second predicted time are obtained based on the first time point and the second time point. The first predicted time is the current date plus the first time point R1, and the second predicted time is the current date plus the second time point R2.
The R day is an uncertain time, and the node of the predicted time of the last day is a term, which means when the cumulative value Ax satisfies the two conditions, and when the cumulative value Ax is the last day, that is, the time of reaching the transplanting date standard is determined according to the values of various factors in the predicted future.
The first accumulated values of the factors are sequentially overlapped from the predicted time node to obtain the accumulated values, so that when the two conditions are met in the process of overlapping the accumulated values, two time points appear, and the optimal transplanting date is calculated.
Example 2
As shown in fig. 5, a device for predicting the optimal transplanting period of tobacco comprises:
the data set dividing module 10: the method comprises the steps of acquiring meteorological data and remote sensing image data of a tobacco field to be detected, and performing first preprocessing to obtain a first historical data set and a second predicted data set which are divided according to time attributes;
evaluation criterion establishment module 20: the method comprises the steps of acquiring the last annual sowing time and the optimal transplanting period of tobacco in a tobacco field to be detected, obtaining a first time interval, intercepting a first historical data set based on the first time interval, and performing second pretreatment to obtain an evaluation standard of the optimal transplanting period of tobacco;
prediction model construction module 30: the method comprises the steps of respectively constructing a prediction model for each factor in a first historical data set based on the first historical data set and adopting a mode of inputting at multiple time points and outputting at multiple time points, and establishing an LSTM prediction model for each factor which is respectively air temperature, precipitation, sunshine hours and surface water content index;
the optimal transplanting period calculation module 40: the method comprises the steps of inputting a second prediction data set into an LSTM prediction model for calculation to obtain predicted values of all factors of continuous R days after a current time node, calculating a first accumulated value of all factors between a annual seeding time and a final day prediction time node according to the predicted values of all factors, comparing and analyzing the first accumulated value with an evaluation standard to obtain a first prediction time and a second prediction time, and obtaining a prediction result of the optimal tobacco transplanting period in a tobacco field to be detected by taking the first prediction time as a starting point and the second prediction time as an end point, wherein R is a natural number.
In one embodiment of the above device, in the data set dividing module 10, weather data and remote sensing image data of a tobacco field to be measured are obtained and subjected to first preprocessing to obtain a first historical data set and a second predictive data set which are divided according to time attributes, in the evaluation criterion establishing module 20, sowing time and optimum transplanting period of tobacco in the tobacco field to be measured in the past year are obtained, a first time interval is obtained, the first historical data set is intercepted and subjected to second preprocessing based on the first time interval to obtain an evaluation criterion of the optimum transplanting period of the tobacco, in the predictive model constructing module 30, the first historical data set is based on the first historical data set and the mode of outputting multiple time points is adopted to respectively construct a predictive model, each factor is respectively air temperature, precipitation, sunshine hours and surface layer moisture content index, in the optimum transplanting period calculating module 40, the second predictive data set is input into the predictive model to calculate predicted values of each factor of continuous R days after the current time node, according to the predicted values of each factor, the first time node is calculated to be measured, and the first time is calculated, the predicted values of the first time node is calculated and the predicted values are accumulated, the predicted values of each time of tobacco is obtained are calculated, and the first time is calculated to be the predicted, and the predicted values of the first time node is calculated to be the predicted, and the predicted values are calculated, and the first time is calculated, and the predicted and the estimated time is calculated.
Example 3
On the basis of the above embodiments, the present embodiment provides an electronic device.
Example 4
On the basis of the above embodiments, the present embodiment provides a storage medium.
The above embodiments are merely illustrative embodiments of the present application, but the technical features of the present application are not limited thereto, and any changes or modifications made by those skilled in the art within the scope of the present application are included in the scope of the present application.
Claims (8)
1. The method for predicting the optimal transplanting period of the tobacco is characterized by comprising the following steps of:
acquiring meteorological data and remote sensing image data of a tobacco field to be detected, and performing first preprocessing to obtain a first historical data set and a second predicted data set which are divided according to time attributes;
acquiring the sowing time and the optimal transplanting period of tobacco in a tobacco field to be detected in the past year, obtaining a first time interval, intercepting a first historical data set based on the first time interval, and performing second pretreatment to obtain an evaluation standard of the optimal transplanting period of tobacco;
respectively constructing a prediction model for each factor in the first historical data set based on the first historical data set and adopting a mode of inputting at multiple time points and outputting at multiple time points, and establishing an LSTM prediction model, wherein each factor is respectively air temperature, precipitation, sunshine hours and surface layer moisture content index;
inputting the second prediction data set into an LSTM prediction model for calculation to obtain predicted values of all factors of continuous R days after the current time node, calculating a first accumulated value of all factors between the annual seeding time and the last day of prediction time node according to the predicted values of all factors, comparing and analyzing the first accumulated value with an evaluation standard to obtain a first prediction time and a second prediction time, and obtaining a prediction result of the optimal tobacco transplanting period in the tobacco field to be detected by taking the first prediction time as a starting point and the second prediction time as an end point, wherein R is a natural number;
the first pretreatment is specifically as follows:
sequentially carrying out time screening, space screening and cutting, cloud removal processing and abnormal value removal of images on the remote sensing image data to obtain processed remote sensing image data;
calculating surface layer moisture content based on processed remote sensing image dataObtaining SCWI time sequence data, and performing time sequence reconstruction on the SCWI time sequence data by using an S-G filtering algorithm to obtain time-space continuous SCWI data, wherein the calculation formula of the surface water content index is as followsWherein SCWI is the surface water content index, and b11 and b12 respectively correspond to the reflectivity of the 11 th wave band and the 12 th wave band of Sentinel-2;
dividing meteorological DATA and space-time continuous SCWI DATA according to N-1 years and N years to respectively obtain N-1 year meteorological DATA, surface water content index, N year meteorological DATA and surface water content index, combining N-1 year DATA to obtain a DATA set DATA1, performing Kriging interpolation processing on N year meteorological DATA in space and combining the N year surface water content index to obtain a DATA set DATA2, wherein N is the current year, N-1 is the past year, the DATA set DATA1 is a first historical DATA set, the DATA set DATA2 is a second predicted DATA set, and the meteorological DATA comprises air temperature, precipitation and sunshine hours;
the second pretreatment is specifically as follows:
sequentially calculating the accumulated values of all factors in the first time interval of each year by using the intercepted first historical data set, and summarizing to obtain a second accumulated value;
respectively summarizing and counting the second accumulated values according to the types of the factors to obtain confidence intervals of the accumulated values of the factors;
performing superposition analysis on the confidence intervals of the accumulated values of the factors to obtain a time intersection;
and obtaining the accumulated value range of each factor based on the intersection in time so as to establish the evaluation standard of the optimal transplanting period of the tobacco.
2. The method for predicting the optimal transplanting period of tobacco as claimed in claim 1, wherein the calculation formula in the second pretreatment process is as followsWherein Ax represents the factor xN-1 years before, i.e. the collection of accumulated values of stations in each year in the last year in the time from ts to te, N is the current year, N-1 is the last year, ts is the sowing time, te is the optimum transplanting period,/-, and>the value of factor x on day t, y represents the y-th year of the previous N-1 years, i represents the i-th weather site, μ Ax Sum sigma Ax The mean and standard deviation of the x-factor cumulative values are shown, bAx and pAx are the lower and upper limits of the confidence interval of the x-factor cumulative value, T bAx And T pAx Respectively representing the time corresponding to the lower limit and the upper limit of the confidence interval of the accumulated value of each factor, at is the air temperature, pr is the precipitation, su is the sunshine hours, wa is the surface water content index, T b And Tp is the intersection of the upper and lower time limits, and the factor x is one of the factors.
3. The method of claim 1, wherein the first historical data set is divided into a training data set and a test data set, and the training data set and the test data set are used for constructing and checking an LSTM prediction model, respectively.
4. The method for predicting the optimal transplanting period of tobacco according to claim 1, wherein the first accumulated value is compared with an evaluation criterion to obtain a first predicted time and a second predicted time, specifically:
comparing the first accumulated value with an evaluation standard, recording a first time point when the first accumulated values are in the range of the evaluation standard, and recording a second time point when the first accumulated value of any factor is greater than the upper limit of the evaluation standard;
the first predicted time and the second predicted time are obtained based on the first time point and the second time point.
5. The method according to claim 4, wherein the first cumulative value of each factor is obtained by stacking the cumulative value of each factor of the current time node and the predicted value of each factor of R consecutive days day by day.
6. A device for predicting an optimal transplanting period of tobacco, comprising:
a data set dividing module: the method comprises the steps of acquiring meteorological data and remote sensing image data of a tobacco field to be detected, and performing first preprocessing to obtain a first historical data set and a second predicted data set which are divided according to time attributes;
the evaluation standard establishment module: the method comprises the steps of acquiring the last annual sowing time and the optimal transplanting period of tobacco in a tobacco field to be detected, obtaining a first time interval, intercepting a first historical data set based on the first time interval, and performing second pretreatment to obtain an evaluation standard of the optimal transplanting period of tobacco;
the prediction model building module: the method comprises the steps of respectively constructing a prediction model for each factor in a first historical data set based on the first historical data set and adopting a mode of inputting at multiple time points and outputting at multiple time points, and establishing an LSTM prediction model for each factor which is respectively air temperature, precipitation, sunshine hours and surface water content index;
and calculating an optimal transplanting period: the method comprises the steps of inputting a second prediction data set into an LSTM prediction model for calculation to obtain predicted values of all factors of continuous R days after a current time node, calculating a first accumulated value of all factors between a annual seeding time and a final day prediction time node according to the predicted values of all factors, comparing and analyzing the first accumulated value with an evaluation standard to obtain a first prediction time and a second prediction time, and obtaining a prediction result of the optimal tobacco transplanting period in a tobacco field to be detected by taking the first prediction time as a starting point and the second prediction time as an end point, wherein R is a natural number.
7. An electronic device comprising a memory and a processor, the memory configured to store one or more computer instructions, wherein the one or more computer instructions are executable by the processor to implement a method of predicting an optimal transplanting period of tobacco according to any one of claims 1-5.
8. A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a computer, implements a method of predicting an optimal transplanting period of tobacco according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310183279.0A CN115860285B (en) | 2023-03-01 | 2023-03-01 | Prediction method and device for optimal transplanting period of tobacco |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310183279.0A CN115860285B (en) | 2023-03-01 | 2023-03-01 | Prediction method and device for optimal transplanting period of tobacco |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115860285A CN115860285A (en) | 2023-03-28 |
| CN115860285B true CN115860285B (en) | 2023-10-31 |
Family
ID=85659488
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310183279.0A Active CN115860285B (en) | 2023-03-01 | 2023-03-01 | Prediction method and device for optimal transplanting period of tobacco |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115860285B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117313993B (en) * | 2023-09-15 | 2024-07-12 | 湖南省烟草公司湘西自治州公司 | Flue-cured tobacco cultivation management method based on temperature and altitude |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104851048A (en) * | 2015-05-28 | 2015-08-19 | 四川农业大学 | Method for determining spatial distribution of suitable transplanting period of flue-cured tobacco in complicated hilly area |
| CN108564309A (en) * | 2018-05-07 | 2018-09-21 | 江西省农业科学院农业经济与信息研究所（江西省农业工程咨询中心、江西省农业科技图书馆） | A kind of Weather Risk appraisal procedure of tobacco leaf planting |
| CN110150078A (en) * | 2019-05-27 | 2019-08-23 | 福建中烟工业有限责任公司 | A kind of method and system on determining northwestern Fujian tobacco transplant date |
| CN111275569A (en) * | 2020-03-13 | 2020-06-12 | 中国农业科学院烟草研究所 | Method and system for determining ecological characteristics of flue-cured tobacco producing area, storage medium and terminal |
| CN111837863A (en) * | 2020-07-28 | 2020-10-30 | 福建中烟工业有限责任公司 | Transplanting period-based cultivation method and system for Fujian fresh, sweet and fragrant high-quality tobacco leaves |
| CN112749627A (en) * | 2020-12-09 | 2021-05-04 | 北京星衡科技有限公司 | Method and device for dynamically monitoring tobacco based on multi-source remote sensing image |
| WO2021180925A1 (en) * | 2020-03-13 | 2021-09-16 | Basf Agro Trademarks Gmbh | Method and system for determining a plant protection treatment plan of an agricultural plant |
| CN114186423A (en) * | 2021-12-14 | 2022-03-15 | 湖北省烟草科学研究院 | Method and system for predicting and evaluating suitable planting area of cigar smoking product |
| CN115376006A (en) * | 2022-08-10 | 2022-11-22 | 中联智慧农业股份有限公司 | Method, storage medium, and processor for predicting crop harvest date |
| CN115545305A (en) * | 2022-10-08 | 2022-12-30 | 中化现代农业有限公司 | Crop transplanting period time prediction method and system |
| CN115688997A (en) * | 2022-10-21 | 2023-02-03 | 浙江领见数智科技有限公司 | Accumulated temperature-based tea leaf picking period prediction method and system |
-
2023
- 2023-03-01 CN CN202310183279.0A patent/CN115860285B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104851048A (en) * | 2015-05-28 | 2015-08-19 | 四川农业大学 | Method for determining spatial distribution of suitable transplanting period of flue-cured tobacco in complicated hilly area |
| CN108564309A (en) * | 2018-05-07 | 2018-09-21 | 江西省农业科学院农业经济与信息研究所（江西省农业工程咨询中心、江西省农业科技图书馆） | A kind of Weather Risk appraisal procedure of tobacco leaf planting |
| CN110150078A (en) * | 2019-05-27 | 2019-08-23 | 福建中烟工业有限责任公司 | A kind of method and system on determining northwestern Fujian tobacco transplant date |
| CN111275569A (en) * | 2020-03-13 | 2020-06-12 | 中国农业科学院烟草研究所 | Method and system for determining ecological characteristics of flue-cured tobacco producing area, storage medium and terminal |
| WO2021180925A1 (en) * | 2020-03-13 | 2021-09-16 | Basf Agro Trademarks Gmbh | Method and system for determining a plant protection treatment plan of an agricultural plant |
| CN111837863A (en) * | 2020-07-28 | 2020-10-30 | 福建中烟工业有限责任公司 | Transplanting period-based cultivation method and system for Fujian fresh, sweet and fragrant high-quality tobacco leaves |
| CN112749627A (en) * | 2020-12-09 | 2021-05-04 | 北京星衡科技有限公司 | Method and device for dynamically monitoring tobacco based on multi-source remote sensing image |
| CN114186423A (en) * | 2021-12-14 | 2022-03-15 | 湖北省烟草科学研究院 | Method and system for predicting and evaluating suitable planting area of cigar smoking product |
| CN115376006A (en) * | 2022-08-10 | 2022-11-22 | 中联智慧农业股份有限公司 | Method, storage medium, and processor for predicting crop harvest date |
| CN115545305A (en) * | 2022-10-08 | 2022-12-30 | 中化现代农业有限公司 | Crop transplanting period time prediction method and system |
| CN115688997A (en) * | 2022-10-21 | 2023-02-03 | 浙江领见数智科技有限公司 | Accumulated temperature-based tea leaf picking period prediction method and system |
Non-Patent Citations (2)
| Title |
|---|
| 皖南烟区不同移栽期气象因子与农艺性状的关系分析;冉法芬;裴洲洋;朱启法;刘国侠;;安徽农学通报(第02期);全文 * |
| 移栽期对烤烟生长发育及品质的影响;杨亚;朱列书;朱静娴;杨威;冯连军;刘伟;邓亚飞;;作物研究(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115860285A (en) | 2023-03-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110751094B (en) | Crop yield estimation method based on GEE comprehensive remote sensing image and deep learning method | |
| CN108921885B (en) | Method for jointly inverting forest aboveground biomass by integrating three types of data sources | |
| CN111727443A (en) | Soil available nutrient inversion method based on deep neural network | |
| CN109711102B (en) | Method for rapidly evaluating crop disaster loss | |
| CN112348812B (en) | Forest stand age information measurement method and device | |
| CN114062439B (en) | Method for jointly estimating salinity of soil profile by using time series remote sensing images | |
| CN111898922B (en) | Multi-scale crop yield assessment method and system | |
| CN109800921B (en) | Regional winter wheat yield estimation method based on remote sensing phenological assimilation and particle swarm optimization | |
| CN114387516B (en) | Single-season rice SAR (synthetic aperture radar) identification method for small and medium-sized fields in complex terrain environment | |
| CN115860285B (en) | Prediction method and device for optimal transplanting period of tobacco | |
| CN112861435B (en) | Mangrove quality remote sensing inversion method and intelligent terminal | |
| CN114120137A (en) | Wetland element space-time evolution monitoring method based on time sequence main remote sensing image | |
| CN118072178B (en) | Corn yield estimation method and system based on classified percentage data assimilation | |
| CN117455062A (en) | Crop yield prediction algorithm based on multi-source heterogeneous agricultural data | |
| CN114882361A (en) | Deep learning forest overground biological estimation method based on multi-source remote sensing fusion | |
| CN118364975A (en) | Wheat yield prediction method of multi-source data-driven hybrid mechanism learning model | |
| CN117556695B (en) | Crop root soil water content simulation method based on deep learning | |
| CN115661601A (en) | Agricultural soil water stress discrimination and monitoring method based on CEEMDAN | |
| CN117953959B (en) | Carbon reserve calculation method of hybrid forest ecosystem | |
| Zhang et al. | Enhanced Feature Extraction From Assimilated VTCI and LAI With a Particle Filter for Wheat Yield Estimation Using Cross-Wavelet Transform | |
| Wu et al. | An online deep convolutional model of gross primary productivity and net ecosystem exchange estimation for global forests | |
| CN117571626A (en) | Method and system for evaluating remote sensing estimation saturation of grassland biomass by MODIS NDVI | |
| CN111737876A (en) | Tea leaf mining period and picking time prediction method with space-time distribution characteristics | |
| Sharma et al. | NGFS rainfall forecast verification over India using the contiguous rain area (CRA) method | |
| CN114707412B (en) | SWAT model optimization method based on vegetation canopy time-varying characteristics |
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 |