A method for constructing a dynamic prediction model of flue-cured tobacco diseases
A dynamic tobacco disease prediction model using meteorological data and regression analysis addresses the limitations of existing models by accurately predicting disease progression based on short-term weather factors and disease severity, enabling effective real-time monitoring.
Patent Information
- Application Number
- CN202211334356.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-10-28
- Publication Date
- 2025-05-27
- Estimated Expiration
- 2042-10-28
AI Technical Summary
Existing tobacco disease prediction models for diseases like tobacco mosaic virus and tobacco streak virus fail to account for the dynamic influence of short-term meteorological factors during the growth season, leading to ineffective real-time monitoring and warning capabilities, and do not consider the varying impact of meteorological factors based on disease severity.
A method involving data processing and modeling that includes normalizing meteorological factors, calculating disease indices, and using Python for regression analysis to establish a dynamic prediction model that correlates current and past disease conditions with meteorological factors, specifically using equations to predict tobacco disease progression.
The model accurately predicts tobacco disease progression by considering short-term meteorological factors, enhancing real-time monitoring and warning capabilities, and improving predictive accuracy by accounting for varying disease severity.
Smart Images
Figure CN115754136B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of dynamic prediction models, and particularly to a method for constructing a dynamic prediction model for tobacco diseases. Background Art
[0002] Tobacco is one of the pillar industries in agriculture in some regions. In some areas, flue-cured tobacco is vulnerable to various diseases. Among them, common mosaic virus disease and brown spot are relatively common, being the first and second major infectious diseases, and the incidence trend has become increasingly serious in recent years.
[0003] Currently, research reports on tobacco diseases mainly focus on aspects such as infection and epidemic laws and conditions, prevention and control strategies and measures, and disaster loss estimation. After the conditions of the pathogen and susceptible plants are available, meteorological conditions are often the key factors determining whether a disease occurs. Regarding the influence of meteorological conditions, most use the methods of field investigation and field experiments to obtain the qualitative or semi-qualitative relationship between diseases and different meteorological conditions. In the relationship model between the development of tobacco diseases and meteorological factors, mainly through mathematical statistics methods such as stepwise regression and fuzzy mathematics methods such as membership function, a tobacco disease prediction model based on meteorological factors is constructed. In the prediction of diseases of other crops (such as wheat), methods such as grey system and neural network are also used for model construction. The existing prediction models for brown spot and mosaic virus disease of flue-cured tobacco mostly use the overall meteorological conditions in a growing season to predict the overall disease degree in that year, and cannot reflect the influence of short-term meteorological factors on the dynamics of diseases during the growing season of flue-cured tobacco. Therefore, it cannot be used for real-time monitoring and early warning of diseases, and the influence of meteorological factors on diseases in the existing models has nothing to do with the current disease degree. In actual production, under different disease degrees, there are great differences in the main pathogenic meteorological factors and their influence degrees, which are difficult to be reflected by the existing models. Therefore, a method for constructing a dynamic prediction model for tobacco diseases is proposed. Summary of the Invention
[0004] (1) Technical Problems to be Solved
[0005] In view of the deficiencies of the prior art, the present invention provides a method for constructing a dynamic prediction model for tobacco diseases, which solves the drawbacks of the existing prediction models for brown spot and mosaic virus disease of flue-cured tobacco, that is, it cannot reflect the influence of short-term meteorological factors on the dynamics of diseases during the growing season of flue-cured tobacco, so it cannot be used for real-time monitoring and early warning of diseases, and the influence of meteorological factors on diseases in the existing models has nothing to do with the current disease degree. In actual production, under different disease degrees, there are great differences in the main pathogenic meteorological factors and their influence degrees, which are difficult to be reflected by the existing models.
[0006] (2) Technical Solutions
[0007] To achieve the above object, the present invention provides the following technical solutions:
[0008] A method for constructing a dynamic prediction model of flue-cured tobacco diseases, comprising the following steps:
[0009] S1: Dimensionless processing of meteorological factors. First, collect meteorological factor data and flue-cured tobacco disease observation data in the pathological area, including survey records of brown spot disease and common mosaic virus disease. Meteorological observation station data, including the average value and extreme value of meteorological data, are calculated with a 5-day step. Therefore, the meteorological factors are dimensionless processed, and the calculation formula is: In the formula, Q is the meteorological factor, Q′ is the dimensionless value of this factor, Q max 、Q min are the maximum and minimum values within one end time of the factor;
[0010] S2: Calculation of disease factors. Use the disease index Y to represent the current disease degree, and the calculation method is: In the formula, a i represents the disease level value, n i represents the corresponding number of diseased plants (leaves). For brown spot disease, count the diseased leaves; for common mosaic virus disease, count the diseased plants. max(a) represents the maximum disease level value. The change in the disease degree Y within a period of time D is represented as: Y D = Y - Y′. The disease development speed V is represented as:
[0011] S3: Data processing. Analyze the correlation and perform stepwise regression analysis on the calculated data through Python 2.0 programming;
[0012] S4: Establishment of a disease dynamic prediction model. There is a significant positive or negative correlation between the changes in the disease degree and the disease development speed of brown spot disease and common mosaic virus disease and meteorological factors, that is:
[0013]
[0014] Through statistical transformation, we can get: Y ∝ ±Q ± QY′ + Y′
[0015] That is, the current disease index has a linear correlation with meteorological factors and the previous disease index, and at the same time, it also has a linear correlation with the composite factor of the previous disease index and meteorological factors. The relationship between the current disease index of brown spot disease and common mosaic virus disease and the previous disease index is represented by a quadratic polynomial equation, and the specific expression is as follows: Y = A(Q) + B(Y′) + C(Q)Y′ + D(Q)Y′ 2 + E
[0016] In the formula, Q represents meteorological factors such as temperature, humidity, and wind speed. A(Q), C(Q), and D(Q) are linear functions of meteorological factors, B(Y') is a quadratic function of the previous disease index, and E is a constant;
[0017] By the stepwise regression method, meteorological factors related to the development of brown spot and mosaic disease in the collected data are selected. With the disease index Y as the dependent variable, and Q, QY', QY' 2 as independent variables, variables causing multicollinearity and low contribution are screened and eliminated, and the dynamic models for the development of brown spot, common mosaic virus disease before topping, and common mosaic virus disease after topping based on the optimal explanatory variables are as follows:
[0018]
[0019]
[0020]
[0021] In the formula, Y 1 , Y 2 , Y 3 respectively represent the disease indices of brown spot, common mosaic virus disease before topping, and common mosaic virus disease after topping, and Y 1 ', Y 2 ', Y 3 ' respectively represent their previous disease indices (selecting the disease index 5 days ago as the previous disease index); T', TM', Tm', Vap', Rh', Wi' are the dimensionless values of the average temperature, maximum temperature, minimum temperature, average vapor pressure, average relative humidity, and maximum wind speed in the previous 5 days;
[0022] S5: Model testing. Using a total of 4 groups of independent observation data of brown spot and common mosaic virus disease in Longshan in the next year or two, the prediction effect of the dynamic prediction model for disease development is tested, and the time series of measured values and predicted values is obtained.
[0023] As a further solution of the present invention, in S2, Y is the current disease index, representing the current disease degree, and Y' is the disease index 5 days ago, representing the previous disease severity.
[0024] Furthermore, in S3, when specifically analyzing, the changes in the disease degrees Y D of brown spot in flue-cured tobacco (62 groups) and common mosaic virus disease (38 groups before topping and 40 groups after topping) are calculated, and the correlation coefficients with the meteorological factors such as temperature, humidity, precipitation, and wind speed in the previous period (5 days before the disease record) are calculated and significant tests are performed.
[0025] Based on the above solution, in S3, the common mosaic virus disease shows different characteristics, gradually worsening before topping and gradually lightening after topping in flue-cured tobacco, so separate analyses are carried out.
[0026] Further, in S4, the model is subjected to back substitution test, and multiple groups of models are used for experiments, and then the coefficient of determination is verified.
[0027] (III) Beneficial effects
[0028] Compared with the prior art, the present invention provides a method for constructing a dynamic prediction model for tobacco diseases, having the following beneficial effects:
[0029] 1. In the present invention, the dynamic prediction models for the development of Alternaria alternata and tobacco mosaic virus constructed by combining meteorological factors with the previous disease index have good prediction effects after being tested with independent data.
[0030] 2. In the present invention, the composite factor of meteorological factors and the previous disease condition is used, and the influence of wind is considered at the same time, which is more targeted. In terms of the prediction period, 5 days is used as the prediction period, which is consistent with the step length of short-term weather forecast, and the accuracy of short-term weather forecast is relatively high. Therefore, the reliability of the disease prediction model proposed by the present invention is significantly improved and the practicability is stronger. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] Figure 1 is a schematic flow chart of a method for constructing a dynamic prediction model for tobacco diseases proposed by the present invention;
[0032] Figure 2 is a schematic diagram of the fitting effect of the dynamic prediction model for the development of Alternaria alternata of a method for constructing a dynamic prediction model for tobacco diseases proposed by the present invention;
[0033] Figure 3 is a schematic diagram of the fitting effect of the dynamic prediction model for the development of tobacco mosaic virus before topping of a method for constructing a dynamic prediction model for tobacco diseases proposed by the present invention;
[0034] Figure 4 is a schematic diagram of the fitting effect of the dynamic prediction model for the development of tobacco mosaic virus after topping of a method for constructing a dynamic prediction model for tobacco diseases proposed by the present invention;
[0035] Figure 5 is a schematic diagram of the comparison of the disease index between the measured value and the predicted value of Alternaria alternata in 2019 of a method for constructing a dynamic prediction model for tobacco diseases proposed by the present invention;
[0036] Figure 6 is a schematic diagram of the comparison of the disease index between the measured value and the predicted value of Alternaria alternata in 2020 of a method for constructing a dynamic prediction model for tobacco diseases proposed by the present invention;
[0037] Figure 7 is a schematic diagram of the comparison of the disease index between the measured value and the predicted value of mosaic disease in 2019 of a method for constructing a dynamic prediction model for tobacco diseases proposed by the present invention;
[0038] Figure 8 Schematic diagram of the comparison between the measured values and predicted values of the disease index of mosaic disease in 2020 for a method of constructing a dynamic prediction model for flue-cured tobacco diseases proposed by the present invention. Detailed implementation manner
[0039] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.
[0040] The disease prediction data of flue-cured tobacco in Longshan County, Xiangxi Prefecture from 2012 to 2020 is sourced from the disease investigation records of Alternaria alternate and common mosaic virus of K326 flue-cured tobacco variety provided by the Longshan Branch of Xiangxi Tobacco Company. The investigation location of this record is Liangshui Village, Ciyan Town, Longshan County, with an altitude of 1100m. Among them, there are 83 groups of records of Alternaria alternate, 61 groups of records of common mosaic virus before topping, and 52 groups of records after topping.
[0041] The daily observation data of the Ciyantang Meteorological Observation Station in Longshan County, Xiangxi Prefecture from 2010 to 2021 is sourced from the Hunan Meteorological Information Center. This observation station is approximately 2km away from the investigation locations of Alternaria alternate and common mosaic disease (Liangshui Village, Ciyantang Town), with an altitude of 1146m. Since the Ciyantang Meteorological Observation Station is close to the investigation locations of Alternaria alternate and common mosaic virus disease, and the topographic and geomorphic environmental conditions are very similar, and there is no field microclimate station and meteorological observation data at the disease investigation location, the meteorological data of the Ciyantang Meteorological Observation Station is used in this study instead.
[0042] The average values, extreme values, etc. of the disease observation data and meteorological data are calculated with a step of 5 days.
[0043] Refer to Figure 1-8 , a method of constructing a dynamic prediction model for flue-cured tobacco diseases, includes the following steps:
[0044] S1: Nondimensionalization of meteorological factors. Since the units of different meteorological factors are different, the meteorological factors are nondimensionalized, and the calculation formula is: In the formula, Q is the meteorological factor, Q′ is the nondimensional value of this factor, Q max 、Q min are the maximum and minimum values of the factor from May to September in 2012 - 2020;
[0045] S2: Calculation of disease factors. Brown spot and common mosaic virus disease of flue-cured tobacco are caused by fungal or viral infections of flue-cured tobacco. The development of the disease not only refers to the change in the severity of the disease in flue-cured tobacco plants, but also refers to the change in the infection range of the disease. Therefore, the disease index Y is used to represent the current disease degree, and the calculation method is as follows: In the formula, a i represents the disease grade value, n i represents the corresponding number of diseased plants (leaves). For brown spot, diseased leaves are counted; for common mosaic virus disease, diseased plants are counted. max(a) represents the maximum disease grade value. The change in the disease degree Y within a period of time D is expressed as: Y D = Y - Y'. The disease development speed V is expressed as: where Y is the current disease index, representing the current disease degree, and Y' is the disease index 5 days ago, representing the severity of the previous disease;
[0046] S3: Data processing. The calculated data is subjected to correlation analysis and stepwise regression analysis through Python 2.0 programming. Specifically, when analyzing, calculate the change in the disease degree Y of brown spot (62 groups) and common mosaic virus disease (38 groups before topping and 40 groups after topping) of flue-cured tobacco from 2012 to 2018 D and the disease development speed V, and the correlation coefficients with meteorological factors such as temperature, humidity, precipitation, and wind speed in the previous period (5 days before the disease record) are calculated and a significance test is carried out (Table 1). Among them, the common mosaic virus disease gradually worsens before topping of flue-cured tobacco and gradually improves after topping, showing different characteristics, so separate analyses are carried out;
[0047] The results show that temperature is the meteorological factor that has the greatest impact on brown spot and common mosaic virus disease of flue-cured tobacco. When the temperature is relatively high from May to September, it is not conducive to the development of brown spot. For common mosaic virus disease before topping of flue-cured tobacco, within a certain range of high temperature, it is conducive to the aggravation of the disease, but it has an inhibitory effect on its infection speed. When the highest temperature is relatively low and the lowest temperature is relatively high after topping, it has an inhibitory effect on the development of common mosaic virus disease;
[0048] Humidity is also an important factor affecting the disease development. When the temperature is suitable, a high relative air humidity is conducive to the development of brown spot. When the temperature is relatively high, a high absolute humidity (vapor pressure) is not conducive to the development of brown spot and common mosaic virus disease before topping of flue-cured tobacco;
[0049] Wind speed also has a certain impact on the development of brown spot and common mosaic virus disease after topping. The extreme wind speed is significantly negatively correlated with both the change degree and development speed of the brown spot disease condition. A relatively large average wind speed is not conducive to the infection of common mosaic virus disease after topping of flue-cured tobacco;
[0050]
[0051]
[0052] Table 1 Correlation Coefficients and Significance Tests between Disease Factors and Meteorological Factors
[0053] S4: Establishment of the disease dynamic prediction model. As the disease condition of flue-cured tobacco plants deepens and the infection range of the disease expands, the sensitivity of the infection rates of Alternaria alternata and tobacco mosaic virus to meteorological factors will also change. Table 1 shows that there is a significant positive or negative correlation between the changes in the disease severity and the development speed of Alternaria alternata and tobacco mosaic virus and meteorological factors, that is:
[0054] Through statistical transformation, we can obtain: Y ∝ ±Q ± QY′ + Y′
[0055] That is, the current disease index has a linear correlation with meteorological factors and the previous disease index, and at the same time, it also has a linear correlation with the composite factor of the previous disease index and meteorological factors. Previous studies have pointed out that the dynamic changes of the disease index can be fitted using a polynomial equation. Therefore, the relationship between the current disease index of Alternaria alternata and tobacco mosaic virus and the previous disease index can be expressed by a quadratic polynomial equation, and the specific expression is as follows: Y = A(Q) + B(Y′) + C(Q)Y′ + D(Q)Y′ 2 +E
[0056] In the formula, Q represents meteorological factors such as temperature, humidity, and wind speed. A(Q), C(Q), and D(Q) are linear functions of meteorological factors, B(Y') is a quadratic function of the previous disease index, and E is a constant;
[0057] By the stepwise regression method, select the meteorological factors related to the disease development of Alternaria alternata and tobacco mosaic virus in Table 1. Taking the disease index Y as the dependent variable and Q, QY', QY' 2 as independent variables, screen and eliminate variables that cause multicollinearity and low contribution, and obtain the disease development dynamic models of Alternaria alternata, tobacco mosaic virus before topping, and tobacco mosaic virus after topping based on the optimal explanatory variables as follows:
[0058]
[0059]
[0060]
[0061] In the formula, Y 1 、Y 2 、Y 3 respectively represent the disease indexes of Alternaria alternata, tobacco mosaic virus before topping, and tobacco mosaic virus after topping, Y 1 '、Y 2 '、Y 3' respectively represent their early-stage disease indices (selecting the disease index 5 days ago as the early-stage disease index); T', TM', Tm', Vap', Rh', Wi' are the dimensionless values of the average temperature, maximum temperature, minimum temperature, average vapor pressure, average relative humidity, and maximum wind speed in the previous 5 days. The back-substitution test of the model shows that the determination coefficients of the three groups of models are all above 0.95. Therefore, the fitting effects of the dynamic models of the development of brown spot and common mosaic virus disease are good;
[0062] There is a close relationship between the development of brown spot and common mosaic virus disease and the early-stage disease index and early-stage meteorological factors. The more severe the early-stage disease, the faster the development of brown spot and common mosaic virus disease. At the same time, the development of the disease is also affected by the early-stage meteorological factors;
[0063] For brown spot, the development speed of brown spot is faster under low temperature and high humidity; when the disease index is low, a larger wind speed is conducive to the spread of brown spot, and when the disease index is high, strong winds are unfavorable to the development of brown spot;
[0064] For common mosaic virus disease, the main influencing factors before and after topping are different. Before topping, temperature has the greatest influence, followed by humidity, and high temperature and low humidity are conducive to its development; after topping, it is mainly affected by temperature, and high temperature is unfavorable to the development of the disease.
[0065] The coefficients of the humidity factors (Rh', Vap') are one to two orders of magnitude smaller than those of the temperature factors (T', TM', Tm'). Therefore, when the disease index reaches more than 10, the humidity has an obvious influence;
[0066] S5: Model test. Use a total of 4 groups of independent observation data of brown spot and common mosaic virus disease in Longshan from 2019 to 2020 to test the prediction effect of the dynamic prediction model of disease development. The time series of the measured values and predicted values are obtained. It can be seen that the disease indices predicted by the prediction models of brown spot and common mosaic virus disease proposed in this study are close to the measured values in size, and the change trends are basically the same. Since the transition time of common mosaic virus disease after topping of flue-cured tobacco is not considered in the disease index of common mosaic virus disease, the prediction deviation is large in the transition area from increasing to decreasing, but the predicted disease indices are basically the same as the measured disease indices in other periods. Therefore, the dynamic models of the development of brown spot and common mosaic virus disease proposed in this study have relatively accurate prediction ability for the degree and trend of disease development.
Claims
1. A method for constructing a dynamic prediction model of flue-cured tobacco diseases, characterized in that, it includes the following steps: S1: Nondimensionalization of meteorological factors. First, collect the meteorological factor data and flue-cured tobacco disease observation data in the pathological area, including the investigation records of brown spot and common mosaic virus disease. The data of the meteorological observation station include the average value and extreme value of meteorological data, which are calculated with a 5-day step. Therefore, the meteorological factors are nondimensionalized, and the calculation formula is: , where is the meteorological factor, is the nondimensional value of the factor, are the maximum and minimum values of the factor within a certain period of time; S2: Calculation of disease factors, using disease index Indicating the current disease level, the calculation method is as follows: , where Indicates the disease level value, Indicates the corresponding number of diseased plants or leaves. For brown spot, count diseased leaves; for common mosaic virus disease, count diseased plants, Indicates the maximum disease level value, the change in disease level over a period of time Is expressed as: , disease development rate Is expressed as: ; is the current disease index, indicating the current disease severity, is the disease index five days ago, indicating the severity of the previous disease; S3: Data processing, performing correlation analysis and stepwise regression analysis on the calculated data through Python 2.0 programming. The common mosaic virus disease gradually worsens before the topping of flue-cured tobacco and gradually improves after topping, showing different characteristics, so separate analyses are carried out; S4: Establishment of disease dynamic prediction model. There is a significant positive or negative correlation between the degree change and development speed of the disease severity of red star disease and common mosaic virus disease and meteorological factors, that is: It can be obtained through statistical transformation: That is, the current disease index has a linear correlation with meteorological factors and the previous disease index. At the same time, it also has a linear correlation with the composite factor of the previous disease index and meteorological factors. The relationship between the current disease index of brown spot and common mosaic virus disease and the previous disease index is expressed by a quadratic polynomial equation, and the specific expression is as follows: In the formula, Q represents meteorological factors such as temperature, humidity, and wind speed, A(Q), C(Q), D(Q) is a linear function of the meteorological factors, B(Y') is a quadratic function of the previous disease index, E is a constant; By the stepwise regression method, meteorological factors related to the development of brown spot and mosaic disease in the collected data are selected, with the disease index Y as the dependent variable, Q, QY', QY' 2 as the independent variables. Variables causing multicollinearity and low contribution are screened and excluded, and the dynamic models for the development of brown spot, common mosaic virus disease before topping, and common mosaic virus disease after topping based on the optimal explanatory variables are as follows: In the formula Y 1 、Y 2 and Y 3 respectively represent the disease index of cercospora leaf spot, the disease index of common mosaic virus before topping, and the disease index of common mosaic virus after topping Y 1 '、Y 2 ' and Y 3 ' respectively represent their disease indices in the early stage. The disease index 5 days ago is selected as the disease index in the early stage T' and TM' and Tm' and Vap', Rh' and Wi' are respectively the dimensionless values of the average temperature, the maximum temperature, the minimum temperature, the average water vapor pressure, the average relative humidity, and the maximum wind speed in the previous 5 days S5: Model testing, using 4 groups of independent observation data of Longshan brown spot disease and common mosaic virus disease in the next year or two to test the prediction effect of the disease development dynamic prediction model, and obtaining the time series of measured values and predicted values.
2. The method for constructing a dynamic prediction model of flue-cured tobacco diseases according to claim 1, characterized in that, in the specific analysis of S3, calculate the change YD of the disease severity and the disease development speed V of 62 groups of flue-cured tobacco brown spot disease and 38 groups before topping and 40 groups after topping of common mosaic virus disease, and the correlation coefficients with the meteorological factors of temperature, humidity, precipitation, and wind speed in the previous period, that is, 5 days before the disease record, and conduct significance tests.
3. The method for constructing a dynamic prediction model of flue-cured tobacco diseases according to claim 1, characterized in that, in S4, perform back substitution testing on the model, conduct experiments using multiple groups of models, and then verify their determination coefficients.