A prediction method for major silkworm diseases and insect pests based on meteorological big data

Abstract

The invention relates to the technical field of data processing, and proposes a method for predicting major diseases and insect pests of silkworms based on meteorological big data, including: obtaining a silkworm disease time series and different meteorological factor sequences; determining a time series lag redundancy coefficient according to two different meteorological factor sequences; screening a meteorological factor sequence that needs to be merged, and obtaining a de-redundant merged meteorological factor sequence; obtaining a silkworm disease covariance matrix according to the de-redundant merged meteorological factor sequence and the silkworm disease time series, performing eigenvalue decomposition, obtaining eigenvalues ​​and eigenvectors, screening the retained eigenvalues, and establishing a silkworm disease meteorological association matrix; training a neural network according to the silkworm disease meteorological association matrix, obtaining a silkworm disease incidence prediction network, and obtaining the incidence of silkworm blood-type pus disease according to the silkworm disease incidence prediction network and silkworm blood-type pus disease related data. The invention aims to solve the problem that the prediction accuracy of silkworm diseases is insufficient due to excessive redundant and noisy data in meteorological data.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method for predicting major diseases and insect pests of silkworms based on meteorological big data. Background Art

[0002] In the silkworm industry, due to the hot and humid climate and the restrictions and influence of the continuous silkworm rearing mode, it is very easy for pathogens to reproduce and infect, thus causing silkworm diseases. Silkworm blood-type pus disease is a viral disease caused by the nuclear polyhedrosis virus, also known as silkworm nuclear polyhedrosis. It is the most serious viral disease that harms silkworms and has a serious impact on the production and income of silkworm farmers. Therefore, strengthening the monitoring and prevention of silkworm blood-type pus disease is of vital importance to protecting the interests of silkworm farmers and promoting the sustainable development of the sericulture industry.

[0003] Meteorological factors such as temperature, precipitation and sunshine hours have an important impact on the incidence of blood-type pus disease of silkworms. Long short-term memory networks can be used to analyze meteorological factor data to predict the incidence of blood-type pus disease of silkworms. However, the amount of meteorological factor data in the past is large, containing a large amount of redundant and noisy data. Using too much data to train the long short-term memory network takes too long and consumes too much memory resources. At the same time, too much redundant and noisy data will also affect the training effect of the network model, thereby reducing the accuracy of silkworm disease prediction. Summary of the invention

[0004] The present invention provides a method for predicting major diseases and insect pests of silkworms based on meteorological big data, so as to solve the problem that the prediction accuracy of silkworm diseases is insufficient due to excessive redundancy and noise data in meteorological data. The technical scheme adopted is as follows:

[0005] An embodiment of the present invention provides a method for predicting major diseases and insect pests of silkworms based on meteorological big data, the method comprising the following steps:

[0006] Extract the relevant data of blood-type pus disease of silkworm and pre-process them to obtain the time series of silkworm disease and different meteorological factor series;

[0007] According to two different meteorological factor sequences, the time series lag redundancy coefficients of the two different meteorological factor sequences are determined;

[0008] The meteorological factor sequences to be merged are screened according to the time series lag redundancy coefficients, and the de-redundant merged meteorological factor sequences are obtained according to the time series lag redundancy coefficients of different meteorological factor sequences;

[0009] The covariance matrix of silkworm disease is obtained by removing redundant and merging meteorological factor sequences and silkworm disease time series, and the eigenvalue decomposition is performed on the covariance matrix of silkworm disease to obtain eigenvalues ​​and eigenvectors. The retained eigenvalues ​​are selected according to the values ​​of all eigenvalues, and the meteorological correlation matrix of silkworm disease is established according to the retained eigenvalues.

[0010] The neural network is trained according to the meteorological correlation matrix of silkworm disease to obtain the silkworm disease incidence prediction network, and the incidence of silkworm blood-type pus disease is obtained according to the silkworm disease incidence prediction network and the relevant data of silkworm blood-type pus disease.

[0011] Furthermore, the data related to the blood-type silkworm pus disease include records of the incidence of blood-type silkworm pus disease, temperature records, rainfall records and sunshine duration data records.

[0012] Further, the method of determining the time series lag redundancy coefficients of the two different meteorological factor sequences according to the two different meteorological factor sequences includes the following specific methods:

[0013] The time-lagged correlations of two different meteorological factor series at different time lags are calculated, and the time-lagged length corresponding to the maximum value of all time-lagged correlations is recorded as the peak length;

[0014] Obtain the first absolute values ​​of two different meteorological factor series based on the peak length and all time-lagged correlations;

[0015] The time series lag redundancy coefficients of two different meteorological factor sequences are obtained according to the first absolute value.

[0016] Furthermore, the first absolute value includes the following specific method:

[0017] The absolute value of the product of the maximum value of all time-lagged correlations and the peak length is recorded as the first absolute value of two different meteorological factor sequences.

[0018] Furthermore, the specific method of determining the time series lag redundancy coefficients of two different meteorological factor sequences includes:

[0019] The linear normalized value of the first absolute value is recorded as the time series lag redundancy coefficient of the first meteorological factor sequence and the second meteorological factor sequence.

[0020] Furthermore, the specific method of selecting the meteorological factor sequence to be merged according to the time series lag redundancy coefficient is as follows:

[0021] The meteorological factor sequences whose time series lag redundancy coefficient is greater than or equal to the preset meteorological factor redundancy threshold are taken as the meteorological factor sequences to be merged.

[0022] Furthermore, the method of obtaining a de-redundant combined meteorological factor sequence according to the time series lag redundancy coefficients of different meteorological factor sequences includes the following specific methods:

[0023] For the meteorological factor sequences to be merged, the meteorological data in each meteorological factor sequence are numbered with natural numbers starting from 1 according to the order of the meteorological data in the meteorological factor sequence, the absolute value of the difference between the meteorological data with the same number in the two meteorological factor sequences to be merged is recorded as the first difference with the same number, and the mean of all meteorological data contained in the two meteorological factor sequences to be merged is recorded as the first mean;

[0024] The product of the first difference value and the first mean value of the same number is recorded as the redundant removal and merging feature value of the same number;

[0025] Arrange all the de-redundant combined feature values ​​in the order of their numbers to obtain a de-redundant combined meteorological factor sequence.

[0026] Further, the method of screening the retained characteristic values ​​according to the values ​​of all characteristic values ​​and establishing the silkworm disease meteorological association matrix according to the retained characteristic values ​​includes the following specific methods:

[0027] According to all eigenvalues ​​of the covariance matrix of silkworm disease, the correlation threshold of silkworm disease is determined;

[0028] Among all the eigenvalues ​​of the silkworm disease covariance matrix, the eigenvalues ​​that are greater than or equal to the silkworm disease association threshold are recorded as retained eigenvalues;

[0029] The eigenvectors corresponding to the retained eigenvalues ​​are arranged from left to right to obtain the meteorological correlation matrix of silkworm disease.

[0030] Further, the silkworm disease association threshold is determined according to all eigenvalues ​​of the silkworm disease covariance matrix, including the specific method of:

[0031] The standard deviation and mean of all eigenvalues ​​of the silkworm disease covariance matrix were calculated, the coefficient of difference of all eigenvalues ​​was recorded as the first ratio, and the product of the first ratio and the median of all eigenvalues ​​of the silkworm disease covariance matrix was recorded as the silkworm disease association threshold.

[0032] Further, the neural network is trained according to the silkworm disease meteorological association matrix to obtain the silkworm disease incidence prediction network, and the incidence of silkworm blood-type pus disease is obtained according to the silkworm disease incidence prediction network and the silkworm blood-type pus disease related data, including the specific method of:

[0033] All the meteorological correlation matrix of silkworm disease and the incidence records of blood-type pus disease of silkworm are input into the long short-term memory network, and the long short-term memory network is trained using the dropout regularization technology, wherein the mean square error loss is used as the loss function and Adam is used as the optimizer to obtain the trained long short-term memory network, and the trained long short-term memory network is recorded as the silkworm disease incidence prediction network;

[0034] The silkworm blood-type pus disease related data of the silkworm blood-type pus disease to be predicted are arranged into a prediction matrix in one column, the prediction matrix is ​​input into a silkworm disease incidence prediction network, and the incidence of silkworm blood-type pus disease is obtained through the silkworm disease incidence prediction network.

[0035] The beneficial effects of the present invention are:

[0036] The invention starts from the problem that the prediction accuracy of mulberry silkworm diseases is insufficient due to too much redundant and noisy data in meteorological data. Firstly, according to the correlation between different meteorological data, the information redundancy brought by the data with significant correlation will reduce the efficiency and stability of the prediction of major mulberry silkworm diseases and insect pests. According to the significant degree of correlation between the data, the redundancy of information between different meteorological factor sequences is evaluated, and the time series lag redundancy coefficient of each two different meteorological factor sequences is obtained. Then, according to the evaluation of the redundancy degree, the different meteorological factor sequences are merged, and the redundant information in the two different meteorological factor sequences is weakened or eliminated as much as possible to obtain the de-redundant merged meteorological factor sequence; then, according to the characteristic that the correlation between the abnormal information contained in the de-redundant merged meteorological factor sequence and other information is small, the abnormal information therein is continuously screened out, and a mulberry silkworm disease meteorological association matrix is ​​established according to the information with greater correlation; finally, according to the mulberry silkworm disease meteorological association matrix, a neural network is trained to obtain a mulberry silkworm disease incidence prediction network, and according to the mulberry silkworm disease incidence prediction network and the silkworm blood type pus disease related data, the incidence of mulberry silkworm blood type pus disease is obtained, so as to realize the prediction of major mulberry silkworm diseases and insect pests based on big data, and solve the problem that the prediction accuracy of mulberry silkworm diseases is insufficient due to too much redundant and noisy data in meteorological data. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0038] Figure 1 A schematic flow chart of a method for predicting major diseases and insect pests of silkworms based on meteorological big data provided by one embodiment of the present invention;

[0039] Figure 2 A flow chart for obtaining a silkworm disease meteorological correlation matrix provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] See also Figure 1 , which shows a flow chart of a method for predicting major diseases and insect pests of silkworms based on meteorological big data provided by an embodiment of the present invention, the method comprising the following steps:

[0042] Step S001, extracting and preprocessing the relevant data of silkworm blood-type pus disease to obtain the silkworm disease time series and different meteorological factor sequences.

[0043] The daily data related to silkworm blood-type pus disease from 2013 to 2023 were extracted from the breeding records. The data related to silkworm blood-type pus disease included the incidence records of silkworm blood-type pus disease, temperature records, rainfall records and sunshine duration data records.

[0044] Arrange the incidence rate in chronological order to obtain the silkworm disease time series. Arrange the temperature, rainfall and sunshine duration data in chronological order and perform normalization processing to obtain the first meteorological factor sequence, the second meteorological factor sequence and the third meteorological factor sequence.

[0045] Among them, this embodiment uses the Z-Score normalization method to normalize the data. The implementer can also use other normalization methods to eliminate the dimension effect. This embodiment does not limit it. The Z-Score normalization method is a well-known technology and will not be repeated here. This embodiment selects 2013-2023 as the time corresponding to the extraction of data records in the breeding records. The implementer can also choose the collection time period according to actual needs, which is not limited here. This embodiment selects the incidence record of silkworm blood-type pus disease, temperature record, rainfall record and sunshine duration data record as the silkworm blood-type pus disease related data. The implementer can also choose to collect the data types included in the silkworm blood-type pus disease related data according to actual needs, which is not limited here.

[0046] The time series of silkworm disease, the first meteorological factor series, the second meteorological factor series, and the third meteorological factor series were processed using the cubic spline interpolation algorithm to fill in some missing values ​​and reduce the impact of missing data on subsequent analysis. Implementers can also choose the missing value filling algorithm according to actual needs, which is not limited here.

[0047] The first meteorological factor sequence, the second meteorological factor sequence and the third meteorological factor sequence are all recorded as meteorological factor sequences, and each data value contained in the meteorological factor sequence is recorded as meteorological data.

[0048] At this point, the meteorological factor series and silkworm disease time series are obtained.

[0049] Step S002: Determine the time series lag redundancy coefficients of the two different meteorological factor sequences according to the two different meteorological factor sequences.

[0050] There is a significant correlation between different meteorological data. For example, there is an indirect correlation between temperature and sunshine time. When the climate and weather influencing factors are the same, the longer the sunshine time, the higher the temperature. When the data related to blood-type purulent disease of silkworms contains a large amount of data with significant correlation, when these data are directly used to predict the main diseases and insect pests of silkworms, the information redundancy brought by these data with significant correlation will reduce the efficiency and stability of the prediction of the main diseases and insect pests of silkworms. Therefore, according to the correlation between the meteorological data contained in the meteorological factor sequence, redundant data can be identified and deleted, and the prediction ability of the prediction of the main diseases and insect pests of silkworms can be improved while maintaining the interpretability and generalization ability of the prediction of the main diseases and insect pests of silkworms.

[0051] Every two different meteorological factor sequences are analyzed respectively. This embodiment is described by taking the analysis of the first meteorological factor sequence and the second meteorological factor sequence as an example.

[0052] The time lag cross-correlation algorithm is used to analyze the time lag correlation of the first meteorological factor sequence and the second meteorological factor sequence, and the corresponding correlation of the first meteorological factor sequence and the second meteorological factor sequence at different time lags is obtained. The time lag length corresponding to the maximum value of all correlations is recorded as the peak length, and the absolute value of the product of the maximum value of all correlations and the peak length is recorded as the first absolute value of the first meteorological factor sequence and the second meteorological factor sequence.

[0053] According to the first absolute value of the first meteorological factor sequence and the second meteorological factor sequence, the temporal lag redundancy coefficient of the first meteorological factor sequence and the second meteorological factor sequence is obtained. The temporal lag redundancy coefficient is positively correlated with the first absolute value.

[0054] It can be understood that the positive correlation and negative correlation in this application refer to the relationship between the independent variable and the dependent variable. The positive correlation means that the independent variable increases (decreases) as the dependent variable increases (decreases), which can be an additive relationship, a multiplicative relationship, etc.

[0055] Preferably, as an embodiment of the present application, the linear normalization result of the first absolute value is recorded as the time-series lag redundancy coefficient of the first meteorological factor sequence and the second meteorological factor sequence.

[0056] Among them, using the time-lagged cross-correlation algorithm to calculate the correlation of two sequences is a well-known technology and will not be repeated here; this embodiment uses the Z-Score normalization method to normalize the first absolute value, and using the Z-Score normalization method for normalization is a well-known technology and will not be repeated here; the implementer can use other methods of the prior art to obtain the correlation corresponding to time lags of different lengths of two sequences, and can also use other normalization methods to normalize the first absolute value, and this application does not make any special restrictions.

[0057] At this point, the time series lag redundancy coefficients of every two different meteorological factor sequences can be obtained.

[0058] The time series lag redundancy coefficient can be used to evaluate the redundancy of the information contained in the two meteorological factor sequences. When the peak length is larger, the time series lag of the information contained in the two meteorological factor sequences is more obvious, and the degree of repetition and redundancy of the information contained in the two meteorological factor sequences is greater. That is, the more obvious the degree to which the information contained in one meteorological factor sequence has been reflected in the information contained in the other meteorological factor sequence, the greater the time series lag redundancy coefficient of the two meteorological factor sequences.

[0059] At this point, the time series lag redundancy coefficients of every two different meteorological factor sequences are obtained.

[0060] Step S003: Set a meteorological factor redundancy threshold, screen the meteorological factor sequences that need to be merged according to the meteorological factor redundancy threshold, and obtain a de-redundant merged meteorological factor sequence according to the meteorological factor sequence and the time lag redundancy coefficients of different meteorological factor sequences.

[0061] A meteorological factor redundancy threshold is set. In this embodiment, the value of the meteorological factor redundancy threshold is 0.9. The implementer can also set the meteorological factor redundancy threshold according to specific needs.

[0062] All meteorological factor sequences whose time series lag redundancy coefficients are greater than or equal to the meteorological factor redundancy threshold are selected, and the redundant information between these meteorological factor sequences is merged. The specific process of merging the redundant information is as follows. This embodiment is explained by taking the analysis of the first meteorological factor sequence and the second meteorological factor sequence as an example.

[0063] The meteorological data included in the first meteorological factor sequence and the second meteorological factor sequence are sorted according to the order of the meteorological data in the meteorological factor sequence, and each meteorological data is numbered with a natural number starting from 1. The absolute value of the difference between the meteorological data with the same number in the first meteorological factor sequence and the second meteorological factor sequence is recorded as the first difference with the same number. The mean of all meteorological data included in the first meteorological factor sequence and the second meteorological factor sequence is recorded as the first mean.

[0064] According to the first differences and first means of all the numbers, a de-redundant merged meteorological factor sequence is determined.

[0065] The product of the first difference and the first mean of the same number is recorded as the de-redundant combined feature value of the same number. All the de-redundant combined feature values ​​are arranged in the order of the numbers to obtain a de-redundant combined meteorological factor sequence.

[0066] It can be understood that the more redundant information there is between two different meteorological factor sequences, the larger the timing lag redundancy coefficient of the two different meteorological factor sequences. Therefore, the two different meteorological factor sequences are merged according to the repeated redundant information of the two different meteorological factor sequences to reduce or eliminate the redundant information in the two different meteorological factor sequences as much as possible.

[0067] At this point, the de-redundant merged meteorological factor sequences corresponding to all meteorological factor sequences whose time series lag redundancy coefficients are greater than or equal to the meteorological factor redundancy threshold are obtained.

[0068] Step S004, obtain the silkworm disease covariance matrix based on the de-redundant merged meteorological factor sequence and the silkworm disease time series, perform eigenvalue decomposition on the silkworm disease covariance matrix to obtain eigenvalues ​​and eigenvectors, select and retain eigenvalues ​​based on the values ​​of all eigenvalues, and establish a silkworm disease meteorological association matrix.

[0069] Each de-redundant merged meteorological factor sequence is analyzed separately, and the abnormal information therein is further screened out according to the de-redundant merged meteorological factor sequence.

[0070] The covariance matrix of the de-redundant combined meteorological factor sequence and the silkworm disease time series is obtained, and the obtained covariance matrix is ​​recorded as the silkworm disease covariance matrix. The silkworm disease covariance matrix is ​​used to reflect the correlation between the data contained in the de-redundant combined meteorological factor sequence and the silkworm disease time series. Among them, the process of calculating the covariance matrix of the two sequences is a well-known technology and will not be repeated here.

[0071] The covariance matrix of silkworm disease is decomposed using matrix eigenvalue decomposition algorithm to obtain eigenvalues ​​and eigenvectors.

[0072] When the meteorological factor sequence corresponding to the silkworm disease time series and the redundant merged meteorological factor sequence contains abnormal information, the correlation between the abnormal information and other information is small, so the corresponding value in the silkworm disease covariance matrix is ​​small.

[0073] Calculate the standard deviation and mean of all eigenvalues ​​of the silkworm disease covariance matrix, record the coefficient of difference of all eigenvalues ​​as the first ratio, record the product of the first ratio and the median of all eigenvalues ​​of the silkworm disease covariance matrix as the silkworm disease association threshold, and record the eigenvalues ​​of all eigenvalues ​​of the silkworm disease covariance matrix that are greater than or equal to the silkworm disease association threshold as the retained eigenvalues.

[0074] The eigenvectors corresponding to the retained eigenvalues ​​are arranged from left to right to obtain the silkworm disease meteorological association matrix. Among them, the eigenvectors of this embodiment are column vectors, and the implementer can also select other methods of constructing the silkworm disease meteorological association matrix using the eigenvectors corresponding to the retained eigenvalues ​​according to actual needs; the calculation of the difference coefficient is a well-known technology in the field of mathematics, and the calculation method of the difference coefficient of the eigenvalue is specifically the ratio of the standard deviation to the mean of all eigenvalues.

[0075] Furthermore, the silkworm disease association threshold is the eigenvalue screening threshold determined according to the data information contained in the silkworm disease covariance matrix. When the eigenvalue is determined to be a retained eigenvalue, it is considered that the retained eigenvalue is an eigenvalue with a greater correlation between the corresponding information. Therefore, the silkworm disease meteorological association matrix can eliminate the adverse effects of the eigenvalues ​​with less correlation corresponding to the silkworm disease covariance matrix on the subsequent silkworm disease and insect pest prediction process.

[0076] At this point, the meteorological correlation matrix of silkworm disease corresponding to each de-redundant combined meteorological factor sequence is obtained. The flowchart of obtaining the meteorological correlation matrix of silkworm disease is as follows: Figure 2 shown.

[0077] Step S005, training a neural network based on the meteorological association matrix of silkworm disease to obtain a silkworm disease incidence prediction network, obtaining the incidence of silkworm blood-type pus disease based on the silkworm disease incidence prediction network and data related to silkworm blood-type pus disease, and realizing the prediction of major silkworm diseases and pests based on big data.

[0078] The long short-term memory network (LSTM) was used to learn the long-term dependency between the meteorological association matrix of silkworm disease corresponding to the meteorological data and the incidence of blood-type pus disease of silkworm.

[0079] All the meteorological correlation matrices of silkworm disease and the daily incidence records of silkworm blood-type pus disease from 2013 to 2023 are input into the long short-term memory network, and the long short-term memory network is trained using the dropout regularization technique. The mean square error loss is used as the loss function and Adam is used as the optimizer to obtain the trained long short-term memory network, which is recorded as the silkworm disease incidence prediction network.

[0080] The silkworm blood-type pus disease related data on the same day that the incidence of silkworm blood-type pus disease needs to be predicted are arranged into a column matrix, recorded as a prediction matrix, and the prediction matrix is ​​input into the silkworm disease incidence prediction network, and the incidence of silkworm blood-type pus disease is obtained through the silkworm disease incidence prediction network. Among them, the column matrix is ​​to arrange all types of silkworm blood-type pus disease related data collected on the same day into one column.

[0081] Among them, the specific training process of the silkworm disease incidence prediction network is a well-known technology and will not be repeated here.

[0082] At this point, the prediction of major silkworm diseases and pests based on big data has been achieved.

[0083] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for predicting major diseases and insect pests of silkworms based on meteorological big data, characterized in that: The method comprises the following steps: Extract the relevant data of blood-type pus disease of silkworm and pre-process them to obtain the time series of silkworm disease and different meteorological factor series; According to two different meteorological factor sequences, the time series lag redundancy coefficients of the two different meteorological factor sequences are determined; The meteorological factor sequences to be merged are screened according to the time series lag redundancy coefficients, and the de-redundant merged meteorological factor sequences are obtained according to the time series lag redundancy coefficients of different meteorological factor sequences; The covariance matrix of silkworm disease is obtained by removing redundant and merging meteorological factor sequences and silkworm disease time series, and the eigenvalue decomposition of the silkworm disease covariance matrix is ​​performed to obtain eigenvalues ​​and eigenvectors, and the standard deviation and mean of all eigenvalues ​​of the silkworm disease covariance matrix are calculated, and the coefficient of difference of all eigenvalues ​​is recorded as the first ratio, and the product of the first ratio and the median of all eigenvalues ​​of the silkworm disease covariance matrix is ​​recorded as the silkworm disease association threshold; the eigenvalues ​​of all eigenvalues ​​of the silkworm disease covariance matrix that are greater than or equal to the silkworm disease association threshold are recorded as retained eigenvalues; the eigenvectors corresponding to the retained eigenvalues ​​are arranged in sequence from left to right to obtain the silkworm disease meteorological association matrix; The neural network is trained according to the meteorological correlation matrix of silkworm disease to obtain the silkworm disease incidence prediction network, and the incidence of silkworm blood-type pus disease is obtained according to the silkworm disease incidence prediction network and the relevant data of silkworm blood-type pus disease.

2. The method for predicting major diseases and insect pests of silkworms based on meteorological big data according to claim 1, characterized in that: The data related to the blood-type silkworm pus disease include records of the incidence of blood-type silkworm pus disease, temperature records, rainfall records and sunshine duration data records.

3. The method for predicting major diseases and insect pests of silkworms based on meteorological big data according to claim 1, characterized in that: The specific method of determining the time series lag redundancy coefficients of the two different meteorological factor sequences according to the two different meteorological factor sequences is as follows: The time-lagged correlations of two different meteorological factor series at different time lags are calculated, and the time-lagged length corresponding to the maximum value of all time-lagged correlations is recorded as the peak length; Obtain the first absolute values ​​of two different meteorological factor series based on the peak length and all time-lagged correlations; The time series lag redundancy coefficients of two different meteorological factor sequences are obtained according to the first absolute value.

4. The method for predicting major diseases and insect pests of silkworms based on meteorological big data according to claim 3, characterized in that: The first absolute value includes the following specific methods: The absolute value of the product of the maximum value of all time-lagged correlations and the peak length is recorded as the first absolute value of two different meteorological factor sequences.

5. The method for predicting major diseases and insect pests of silkworms based on meteorological big data according to claim 4, characterized in that: The specific method of determining the time series lag redundancy coefficients of two different meteorological factor sequences includes: The linear normalized value of the first absolute value is recorded as the time series lag redundancy coefficient of the first meteorological factor sequence and the second meteorological factor sequence.

6. The method for predicting major diseases and insect pests of silkworms based on meteorological big data according to claim 1, characterized in that: The specific method of screening the meteorological factor sequence to be merged according to the time series lag redundancy coefficient is as follows: The meteorological factor sequences whose time series lag redundancy coefficient is greater than or equal to the preset meteorological factor redundancy threshold are taken as the meteorological factor sequences to be merged.

7. The method for predicting major diseases and insect pests of silkworms based on meteorological big data according to claim 1, characterized in that: The specific method of obtaining the de-redundant combined meteorological factor sequence according to the time series lag redundancy coefficients of different meteorological factor sequences includes: For the meteorological factor sequences to be merged, the meteorological data in each meteorological factor sequence are numbered with natural numbers starting from 1 according to the order of the meteorological data in the meteorological factor sequence, the absolute value of the difference between the meteorological data with the same number in the two meteorological factor sequences to be merged is recorded as the first difference with the same number, and the mean of all meteorological data contained in the two meteorological factor sequences to be merged is recorded as the first mean; The product of the first difference value and the first mean value of the same number is recorded as the redundant removal and merging feature value of the same number; Arrange all the de-redundant combined feature values ​​in the order of their numbers to obtain a de-redundant combined meteorological factor sequence.

8. The method for predicting major diseases and insect pests of silkworms based on meteorological big data according to claim 1, characterized in that: The method of training a neural network according to the meteorological association matrix of silkworm disease to obtain a silkworm disease incidence prediction network, and obtaining the incidence of silkworm blood-type pus disease according to the silkworm disease incidence prediction network and the silkworm blood-type pus disease related data, includes the following specific methods: All the meteorological correlation matrix of silkworm disease and the incidence records of blood-type pus disease of silkworm are input into the long short-term memory network, and the long short-term memory network is trained using the dropout regularization technology, wherein the mean square error loss is used as the loss function and Adam is used as the optimizer to obtain the trained long short-term memory network, and the trained long short-term memory network is recorded as the silkworm disease incidence prediction network; The silkworm blood-type pus disease related data of the silkworm blood-type pus disease to be predicted are arranged into a prediction matrix in one column, the prediction matrix is ​​input into a silkworm disease incidence prediction network, and the incidence of silkworm blood-type pus disease is obtained through the silkworm disease incidence prediction network.

Images