A method and system for farmland irrigation after extreme weather

Abstract

The present application discloses a method and system for irrigation of farmland after extreme weather, which relates to the field of agricultural irrigation management, including: obtaining weather monitoring data and farmland monitoring data; the farmland monitoring data includes multiple frames of farmland monitoring images, crop growth monitoring data, farmland soil monitoring data and farmland air monitoring data; determining the type of farmland irrigation according to the weather monitoring data; extracting features from multiple frames of farmland monitoring images to obtain a first damaged feature, and extracting features from farmland soil monitoring data and farmland air monitoring data to obtain a second damaged feature; constructing a multivariate output regression model for crop growth according to the crop growth monitoring data, and obtaining a basic water demand and an irrigation time table according to the weather monitoring data, the farmland irrigation type and the multivariate output regression model for crop growth; obtaining a corrected water demand based on the first damaged feature, the second damaged feature and the basic water demand, and irrigating the farmland to be irrigated in combination with the irrigation time table.

Description

Technical Field

[0001] The present application relates to the field of agricultural irrigation management, and in particular to a method and system for irrigating farmland after extreme weather. Background Art

[0002] In agricultural production, irrigation is the most important link and is closely related to the yield and quality of crops. Appropriate and reasonable irrigation can ensure the stability of crop yield and quality and reduce the waste of agricultural water resources. Traditional agricultural farmland irrigation methods mainly rely on the long-term practical experience of agricultural production workers, and there are problems such as decision-making difficulties and irrigation inaccuracy. Therefore, a farmland irrigation method that does not rely on subjective judgment is needed.

[0003] In addition to traditional agricultural farmland irrigation methods, there are irrigation methods based on digital twins or knowledge graphs in the existing technology, but they are mainly focused on irrigation under normal climate conditions, and the main optimization goal is water resource utilization. There is less involvement in irrigation under the influence of extreme weather, and the impact of damage to farmland after extreme weather on irrigation is not considered, resulting in poor irrigation effect when irrigating farmland after extreme weather, and poor irrigation and maintenance effect on farmland. Therefore, how to reasonably irrigate farmland after extreme weather is still a technical problem that needs to be solved urgently. Summary of the invention

[0004] The present application provides a method and system for irrigating farmland after extreme weather, so as to solve the technical problem that farmland cannot be reasonably irrigated after extreme weather.

[0005] According to a first aspect of the embodiments of the present application, a method for irrigating farmland after extreme weather is provided, comprising:

[0006] Acquire weather monitoring data and farmland monitoring data of the farmland to be irrigated; wherein the farmland monitoring data includes multiple frames of farmland monitoring images, crop growth monitoring data, farmland soil monitoring data, and farmland air monitoring data;

[0007] Determining the type of farmland irrigation according to the weather monitoring data;

[0008] Performing feature extraction on the multiple frames of farmland monitoring images to obtain a first damaged feature, and performing feature extraction on the farmland soil monitoring data and the farmland air monitoring data to obtain a second damaged feature;

[0009] Constructing a multivariate output regression model for crop growth according to the crop growth monitoring data, and obtaining a basic water demand and an irrigation schedule for the farmland to be irrigated according to the weather monitoring data, the farmland irrigation type and the multivariate output regression model for crop growth;

[0010] Based on the first damage characteristic and the second damage characteristic, the basic water demand is corrected to obtain a corrected water demand, and the farmland to be irrigated is irrigated according to the corrected water demand and the irrigation schedule.

[0011] This application first obtains weather monitoring data and farmland monitoring data of the farmland to be irrigated, and determines the farmland irrigation type according to the weather monitoring data, so as to preliminarily determine the water demand according to the farmland irrigation type, and then extracts features of multiple frames of farmland monitoring images, farmland soil monitoring data and farmland air monitoring data in the farmland monitoring data to obtain damaged features, taking into account the impact of farmland damage on farmland irrigation, and then constructs a multivariate output regression model for crop growth according to the crop growth monitoring data, and then obtains the basic water demand and irrigation schedule of the farmland to be irrigated according to the weather monitoring data and the farmland irrigation type. By constructing a multivariate output regression model, it is possible to focus on the growth of crops and the water demand of farmland, and then solve the basic water demand and irrigation schedule to better meet the actual irrigation needs of the farmland to be irrigated. Then, the basic water demand is corrected according to the damaged features, and the impact of farmland damage on irrigation is expanded, so that the corrected water demand is more in line with the actual irrigation needs under the current weather conditions, thereby improving the irrigation and maintenance effect of farmland crops and improving the rationality of farmland irrigation.

[0012] In certain embodiments of the present application, determining the farmland irrigation type according to the weather monitoring data specifically includes:

[0013] Performing cluster analysis on the weather monitoring data to obtain multiple clusters, and obtaining a first cluster corresponding to the one with the largest timestamp;

[0014] Characterizing all weather monitoring data in the first cluster respectively to obtain a plurality of weather monitoring features;

[0015] Calculate the cosine similarities of the multiple weather monitoring features and other weather monitoring features respectively to obtain multiple cosine similarities, and obtain first data and second data; wherein the first data and the second data are the weather monitoring features corresponding to the smallest one among the multiple cosine similarities;

[0016] According to the mean values ​​of the multiple cosine similarities, the first data and the second data, the farmland irrigation type is matched in a preset farmland irrigation database.

[0017] The present application first performs cluster analysis on the weather monitoring data to obtain the first cluster corresponding to the one with the largest timestamp, which can aggregate data with similar characteristics, and then characterizes all weather monitoring data in the first cluster to obtain multiple weather monitoring features, and calculates the cosine similarity of each of them to obtain the two weather monitoring features corresponding to the one with the smallest cosine similarity, which can find the two feature data with the largest difference, and then perform irrigation type matching based on the mean of multiple cosine similarities and the two feature data with the largest difference, which can more accurately obtain the farmland irrigation type and reduce the probability of false matching, which is conducive to preliminarily determining the water demand based on the farmland irrigation type.

[0018] In certain embodiments of the present application, the extracting features of the multiple frames of farmland monitoring images to obtain the first damaged features specifically includes:

[0019] Characterizing the multiple frames of farmland monitoring images respectively to obtain multiple image monitoring features;

[0020] Based on a convolutional neural network, feature extraction is performed on the multiple image monitoring features to obtain multiple image damaged sub-features; wherein the multiple image damaged sub-features are arranged in ascending order according to the timestamps of the multiple frames of farmland monitoring images;

[0021] respectively calculating the Pearson correlation coefficients between the plurality of damaged image sub-features and adjacent damaged image sub-features, and dividing the plurality of damaged image sub-features into a plurality of damaged feature domains according to a preset number of classifications based on the Pearson correlation coefficients;

[0022] The multiple damaged image sub-features in the multiple damaged feature domains are respectively merged to obtain multiple damaged image features, and the multiple damaged image features are concatenated in ascending order of timestamps to obtain a first damaged feature.

[0023] The present application first characterizes multiple frames of farmland monitoring images to obtain multiple image monitoring features, then performs feature extraction based on a convolutional neural network, and then calculates the Pearson correlation coefficient for each image and divides the multiple image damaged sub-features into multiple damaged feature domains according to the Pearson correlation coefficient and a preset number of classifications. The damaged sub-features of images with similar damaged features can be aggregated, and the various types of damage corresponding to each damaged feature domain can be better characterized. The first damaged feature can then be merged and then spliced ​​in ascending order of timestamps to obtain the first damaged feature, which can make the first damaged feature better reflect the changes in the damage situation in time sequence.

[0024] In certain embodiments of the present application, the feature extraction of the farmland soil monitoring data and the farmland air monitoring data to obtain the second damaged feature specifically includes:

[0025] Aligning and merging the farmland soil monitoring data and the farmland air monitoring data according to timestamps to obtain farmland environment monitoring data;

[0026] The farmland environment monitoring data is divided multiple times based on multiple time scales to obtain multiple farmland environment monitoring data subsets; wherein the multiple time scales include 3 days, 5 days and 10 days;

[0027] Extracting features from the plurality of farmland environment monitoring data subsets respectively to obtain a plurality of environmental damage features;

[0028] The plurality of environmental impairment features are fused to obtain a second impairment feature.

[0029] This application first aligns and merges the farmland soil monitoring data and the farmland air monitoring data to obtain the farmland environment monitoring data, and then divides it multiple times based on multiple time scales to obtain environmental monitoring data of multiple granularities. Then, when performing feature extraction, multiple environmental damage features are obtained through multiple granularities, so as to better characterize the actual farmland environmental monitoring and make the fused second damaged features more in line with the actual needs of farmland environmental monitoring.

[0030] In certain embodiments of the present application, constructing a multivariate output regression model for crop growth based on the crop growth monitoring data specifically includes:

[0031] According to the crop growth monitoring data, a multivariate output regression model of crop growth is constructed based on the gradient boosting regression tree model, with crop plant height, irrigation time and irrigation water volume as independent variables and crop water requirement at each growth stage as dependent variable.

[0032] This application constructs a multivariate output regression model for crop growth based on crop growth monitoring data, with crop plant height, irrigation time and irrigation water amount as independent variables, and water requirement of crops at each growth stage as dependent variable, based on the gradient boosting regression tree model. It takes into account the impact of changes in crop plant height, irrigation time and irrigation water amount on water requirement of crops at each growth stage, and constructs a model based on the gradient boosting regression tree model, which can make the constructed multivariate output regression model for crop growth more comprehensive and accurate.

[0033] In certain embodiments of the present application, the basic water requirement and irrigation schedule of the farmland to be irrigated are obtained according to the weather monitoring data, the farmland irrigation type and the crop growth multivariate output regression model, specifically including:

[0034] Based on a preset time series prediction model, the weather monitoring data is subjected to data prediction expansion to obtain weather prediction data;

[0035] According to the farmland irrigation type, an initial irrigation water volume is matched in a preset farmland irrigation database;

[0036] The weather monitoring data, the weather forecast data and the initial irrigation water volume are input into the crop growth multivariate output regression model to obtain the basic water demand and irrigation schedule of the farmland to be irrigated.

[0037] This application first performs data forecasting and expansion on the weather monitoring data based on a preset time series forecasting model to obtain weather forecast data, which can prevent solution distortion caused by too small a sample data volume, and then obtains the initial irrigation water volume by matching the farmland irrigation type. Furthermore, the basic water demand and irrigation time series table are obtained from the multivariate output regression model of crop growth, which is more in line with the actual irrigation needs of farmland.

[0038] In certain embodiments of the present application, the modifying the basic water demand to obtain the modified water demand based on the first damage characteristic and the second damage characteristic specifically includes:

[0039] According to the weather monitoring data and the first damage characteristic, a first correction value is obtained by matching in a preset farmland irrigation database;

[0040] Inputting the second damage feature into a preset crop irrigation time series prediction model to predict the irrigation water volume to obtain a second correction amount;

[0041] A corrected water demand is obtained according to the first correction amount, the second correction amount, the weather monitoring data and the basic water demand.

[0042] This application first obtains each correction amount according to the matching of each damage feature, and then corrects the basic water demand according to each correction amount and weather monitoring data, which can expand the impact of farmland damage on irrigation, so that the corrected water demand is more in line with the actual irrigation needs after extreme weather, thereby improving the irrigation and maintenance effect of farmland crops and improving the rationality of farmland irrigation.

[0043] In certain embodiments of the present application, the corrected water demand is obtained according to the first correction amount, the second correction amount, the weather monitoring data and the basic water demand, specifically:

[0044] ;

[0045] in, are the modified water demand and the basic water demand, is the correction factor, is the daily average temperature, are the first correction and the second correction respectively, is the characteristic coefficient.

[0046] According to a second aspect of the implementation of the present application, there is provided a farmland irrigation system after extreme weather, comprising a monitoring data acquisition module, an irrigation type determination module, a damaged feature extraction module, a data regression solution module and an irrigation result correction module;

[0047] The monitoring data acquisition module is used to acquire weather monitoring data and farmland monitoring data of the farmland to be irrigated; wherein the farmland monitoring data includes multiple frames of farmland monitoring images, crop growth monitoring data, farmland soil monitoring data and farmland air monitoring data;

[0048] The irrigation type determination module is used to determine the farmland irrigation type according to the weather monitoring data;

[0049] The damaged feature extraction module is used to extract features from the multiple frames of farmland monitoring images to obtain a first damaged feature, and to extract features from the farmland soil monitoring data and the farmland air monitoring data to obtain a second damaged feature;

[0050] The data regression solution module is used to construct a multivariate output regression model for crop growth according to the crop growth monitoring data, and obtain the basic water demand and irrigation time table of the farmland to be irrigated according to the weather monitoring data, the farmland irrigation type and the multivariate output regression model for crop growth;

[0051] The irrigation result correction module is used to correct the basic water demand to obtain a corrected water demand based on the first damaged characteristic and the second damaged characteristic, and irrigate the farmland to be irrigated according to the corrected water demand and the irrigation schedule.

[0052] In some embodiments of the present application, the damaged feature extraction module includes a first damaged feature extraction submodule; the first damaged feature extraction submodule includes an image characterization unit, an image feature extraction unit, a feature domain division unit and a damaged feature merging unit;

[0053] The image characterization unit is used to characterize the multiple frames of farmland monitoring images respectively to obtain multiple image monitoring features;

[0054] The image feature extraction unit is used to extract features from the multiple image monitoring features based on a convolutional neural network to obtain multiple image damaged sub-features; wherein the multiple image damaged sub-features are arranged in ascending order according to the timestamps of the multiple frames of farmland monitoring images;

[0055] The feature domain division unit is used to calculate the Pearson correlation coefficients between the multiple damaged image sub-features and adjacent damaged image sub-features respectively, and divide the multiple damaged image sub-features into multiple damaged feature domains according to the Pearson correlation coefficients and a preset number of classifications;

[0056] The damaged feature merging unit is used to merge the multiple damaged image sub-features in the multiple damaged feature domains respectively to obtain multiple damaged image features, and to splice the multiple damaged image features in ascending order of timestamps to obtain a first damaged feature. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 : A schematic diagram of a method for irrigation of farmland after extreme weather according to certain embodiments of the present application;

[0058] Figure 2 : A modular structure diagram of a farm irrigation system after extreme weather as shown in certain embodiments of the present application. DETAILED DESCRIPTION

[0059] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals represent the same or similar elements or elements having the same or similar functions from beginning to end. The embodiments described below in conjunction with the accompanying drawings are exemplary and are only used to explain some embodiments of the present application, and should not be construed as limitations on the embodiments of the present application. Based on the embodiments shown in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0060] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, unless otherwise clearly and specifically defined, "multiple" and "several" mean two or more.

[0061] Existing irrigation methods mainly focus on irrigation under normal climate conditions, and the main optimization goal is water resource utilization. There is little involvement in irrigation under the influence of extreme weather. At the same time, the impact of damage to farmland after extreme weather on irrigation is not considered, resulting in poor irrigation effect when irrigating farmland after extreme weather, and poor irrigation and maintenance effect on farmland. Therefore, how to reasonably irrigate farmland after extreme weather is still a technical problem that needs to be solved urgently.

[0062] Based on the above technical background, please refer to Figure 1 The present application embodiment provides a method for irrigation of farmland after extreme weather, including steps S101 to S105, each of which is as follows:

[0063] Step S101: Acquire weather monitoring data and farmland monitoring data of the farmland to be irrigated; wherein the farmland monitoring data includes multiple frames of farmland monitoring images, crop growth monitoring data, farmland soil monitoring data and farmland air monitoring data.

[0064] Specifically, the preferred method for collecting the weather monitoring data, the crop growth monitoring data, the farmland soil monitoring data and the farmland air monitoring data is to collect them through sensors of corresponding types; the preferred method for collecting the multiple frames of farmland monitoring images is to collect them through a fixed-height camera set up at the edge of the farmland.

[0065] Step S102: Determine the farmland irrigation type according to the weather monitoring data.

[0066] In certain embodiments of the present application, determining the farmland irrigation type according to the weather monitoring data specifically includes:

[0067] Performing cluster analysis on the weather monitoring data to obtain multiple clusters, and obtaining a first cluster corresponding to the one with the largest timestamp;

[0068] Characterizing all weather monitoring data in the first cluster respectively to obtain a plurality of weather monitoring features;

[0069] Calculate the cosine similarities of the multiple weather monitoring features and other weather monitoring features respectively to obtain multiple cosine similarities, and obtain first data and second data; wherein the first data and the second data are the weather monitoring features corresponding to the smallest one among the multiple cosine similarities;

[0070] According to the mean values ​​of the multiple cosine similarities, the first data and the second data, the farmland irrigation type is matched in a preset farmland irrigation database.

[0071] In certain embodiments of the present application, when calculating the cosine similarity between the multiple weather monitoring features and other weather monitoring features, the cosine similarity may also be replaced by Euclidean distance, Manhattan distance or log-likelihood similarity, which is not limited in the present application.

[0072] The present application first performs cluster analysis on the weather monitoring data to obtain the first cluster corresponding to the one with the largest timestamp, which can aggregate data with similar characteristics, and then characterizes all weather monitoring data in the first cluster to obtain multiple weather monitoring features, and calculates the cosine similarity of each of them to obtain the two weather monitoring features corresponding to the one with the smallest cosine similarity, which can find the two feature data with the largest difference, and then perform irrigation type matching based on the mean of multiple cosine similarities and the two feature data with the largest difference, which can more accurately obtain the farmland irrigation type and reduce the probability of false matching, which is conducive to preliminarily determining the water demand based on the farmland irrigation type.

[0073] Step S103: extracting features from the multiple frames of farmland monitoring images to obtain a first damaged feature, and extracting features from the farmland soil monitoring data and the farmland air monitoring data to obtain a second damaged feature.

[0074] In certain embodiments of the present application, the extracting features of the multiple frames of farmland monitoring images to obtain the first damaged features specifically includes:

[0075] Characterizing the multiple frames of farmland monitoring images respectively to obtain multiple image monitoring features;

[0076] Based on a convolutional neural network, feature extraction is performed on the multiple image monitoring features to obtain multiple image damaged sub-features; wherein the multiple image damaged sub-features are arranged in ascending order according to the timestamps of the multiple frames of farmland monitoring images;

[0077] respectively calculating the Pearson correlation coefficients between the plurality of damaged image sub-features and adjacent damaged image sub-features, and dividing the plurality of damaged image sub-features into a plurality of damaged feature domains according to a preset number of classifications based on the Pearson correlation coefficients;

[0078] The multiple damaged image sub-features in the multiple damaged feature domains are respectively merged to obtain multiple damaged image features, and the multiple damaged image features are concatenated in ascending order of timestamps to obtain a first damaged feature.

[0079] In certain embodiments of the present application, the convolutional neural network includes but is not limited to a VGG16 model or a ResNet model, and a preferred embodiment is the VGG16 model.

[0080] In certain embodiments of the present application, when calculating the Pearson correlation coefficient between the multiple image-impaired sub-features and adjacent image-impaired sub-features, the Pearson correlation coefficient may also be replaced by a Spearman rank correlation coefficient, a mean square error, a peak signal-to-noise ratio, or a structural similarity algorithm, and the present application does not impose any limitation thereto.

[0081] The present application first characterizes multiple frames of farmland monitoring images to obtain multiple image monitoring features, then performs feature extraction based on a convolutional neural network, and then calculates the Pearson correlation coefficient for each image and divides the multiple image damaged sub-features into multiple damaged feature domains according to the Pearson correlation coefficient and a preset number of classifications. The damaged sub-features of images with similar damaged features can be aggregated, and the various types of damage corresponding to each damaged feature domain can be better characterized. The first damaged feature can then be merged and then spliced ​​in ascending order of timestamps to obtain the first damaged feature, which can make the first damaged feature better reflect the changes in the damage situation in time sequence.

[0082] In certain embodiments of the present application, the feature extraction of the farmland soil monitoring data and the farmland air monitoring data to obtain the second damaged feature specifically includes:

[0083] Aligning and merging the farmland soil monitoring data and the farmland air monitoring data according to timestamps to obtain farmland environment monitoring data;

[0084] The farmland environment monitoring data is divided multiple times based on multiple time scales to obtain multiple farmland environment monitoring data subsets; wherein the multiple time scales include 3 days, 5 days and 10 days;

[0085] Extracting features from the plurality of farmland environment monitoring data subsets respectively to obtain a plurality of environmental damage features;

[0086] The plurality of environmental impairment features are fused to obtain a second impairment feature.

[0087] This application first aligns and merges the farmland soil monitoring data and the farmland air monitoring data to obtain the farmland environment monitoring data, and then divides it multiple times based on multiple time scales to obtain environmental monitoring data of multiple granularities. Then, when performing feature extraction, multiple environmental damage features are obtained through multiple granularities, so as to better characterize the actual farmland environmental monitoring and make the fused second damaged features more in line with the actual needs of farmland environmental monitoring.

[0088] Step S104: construct a multivariate output regression model for crop growth based on the crop growth monitoring data, and obtain the basic water demand and irrigation schedule for the farmland to be irrigated based on the weather monitoring data, the farmland irrigation type and the multivariate output regression model for crop growth.

[0089] In certain embodiments of the present application, constructing a multivariate output regression model for crop growth based on the crop growth monitoring data specifically includes:

[0090] According to the crop growth monitoring data, a multivariate output regression model of crop growth is constructed based on the gradient boosting regression tree model, with crop plant height, irrigation time and irrigation water volume as independent variables and crop water requirement at each growth stage as dependent variable.

[0091] This application constructs a multivariate output regression model for crop growth based on crop growth monitoring data, with crop plant height, irrigation time and irrigation water amount as independent variables, and water requirement of crops at each growth stage as dependent variable, based on the gradient boosting regression tree model. It takes into account the impact of changes in crop plant height, irrigation time and irrigation water amount on water requirement of crops at each growth stage, and constructs a model based on the gradient boosting regression tree model, which can make the constructed multivariate output regression model for crop growth more comprehensive and accurate.

[0092] In certain embodiments of the present application, the basic water requirement and irrigation schedule of the farmland to be irrigated are obtained according to the weather monitoring data, the farmland irrigation type and the crop growth multivariate output regression model, specifically including:

[0093] Based on a preset time series prediction model, the weather monitoring data is subjected to data prediction expansion to obtain weather prediction data;

[0094] According to the farmland irrigation type, an initial irrigation water volume is matched in a preset farmland irrigation database;

[0095] The weather monitoring data, the weather forecast data and the initial irrigation water volume are input into the crop growth multivariate output regression model to obtain the basic water demand and irrigation schedule of the farmland to be irrigated.

[0096] In certain embodiments of the present application, the preset time series prediction model includes but is not limited to a VAR vector autoregression model, an LSTM long short-term memory model or a Transformer model, and a preferred embodiment is the LSTM long short-term memory model.

[0097] This application first performs data forecasting and expansion on the weather monitoring data based on a preset time series forecasting model to obtain weather forecast data, which can prevent solution distortion caused by too small a sample data volume, and then obtains the initial irrigation water volume by matching the farmland irrigation type. Furthermore, the basic water demand and irrigation time series table are obtained from the multivariate output regression model of crop growth, which is more in line with the actual irrigation needs of farmland.

[0098] Step S105: Based on the first damage characteristic and the second damage characteristic, the basic water demand is corrected to obtain a corrected water demand, and the farmland to be irrigated is irrigated according to the corrected water demand and the irrigation schedule.

[0099] In certain embodiments of the present application, the modifying the basic water demand to obtain the modified water demand based on the first damage characteristic and the second damage characteristic specifically includes:

[0100] According to the weather monitoring data and the first damage characteristic, a first correction value is obtained by matching in a preset farmland irrigation database;

[0101] Inputting the second damage feature into a preset crop irrigation time series prediction model to predict the irrigation water volume to obtain a second correction amount;

[0102] A corrected water demand is obtained according to the first correction amount, the second correction amount, the weather monitoring data and the basic water demand.

[0103] In certain embodiments of the present application, the corrected water demand is obtained according to the first correction amount, the second correction amount, the weather monitoring data and the basic water demand, specifically:

[0104] ;

[0105] in, are the modified water demand and the basic water demand, is the correction factor, is the daily average temperature, are the first correction and the second correction respectively, is the characteristic coefficient.

[0106] In certain embodiments of the present application, the correction factor The preferred value of is 0.95.

[0107] This application first obtains each correction amount according to the matching of each damage feature, and then corrects the basic water demand according to each correction amount and weather monitoring data, which can expand the impact of farmland damage on irrigation, so that the corrected water demand is more in line with the actual irrigation needs after extreme weather, thereby improving the irrigation and maintenance effect of farmland crops and improving the rationality of farmland irrigation.

[0108] Compared with the prior art, the present application first obtains weather monitoring data and farmland monitoring data of the farmland to be irrigated, and determines the farmland irrigation type according to the weather monitoring data, so as to preliminarily determine the water demand according to the farmland irrigation type, and then extracts features of multiple frames of farmland monitoring images, farmland soil monitoring data and farmland air monitoring data in the farmland monitoring data to obtain damaged features, taking into account the impact of farmland damage on farmland irrigation, and then constructs a multivariate output regression model for crop growth according to the crop growth monitoring data, and then obtains the basic water demand and irrigation schedule of the farmland to be irrigated according to the weather monitoring data and the farmland irrigation type. By constructing a multivariate output regression model, it is possible to focus on the growth of crops and the water demand of farmland, and then solve the basic water demand and irrigation schedule to better meet the actual irrigation needs of the farmland to be irrigated. Then, the basic water demand is corrected according to the damaged features to expand the impact of farmland damage on irrigation, so that the corrected water demand is more in line with the actual irrigation needs under the current weather conditions, thereby improving the irrigation and maintenance effect of farmland crops and improving the rationality of farmland irrigation.

[0109] Corresponding to the previous method, see Figure 2 , the embodiment of the present application provides a farmland irrigation system after extreme weather, including a monitoring data acquisition module 210, an irrigation type determination module 220, a damaged feature extraction module 230, a data regression solution module 240 and an irrigation result correction module 250;

[0110] The monitoring data acquisition module 210 is used to acquire weather monitoring data and farmland monitoring data of the farmland to be irrigated; wherein the farmland monitoring data includes multiple frames of farmland monitoring images, crop growth monitoring data, farmland soil monitoring data and farmland air monitoring data;

[0111] The irrigation type determination module 220 is used to determine the farmland irrigation type according to the weather monitoring data;

[0112] The damaged feature extraction module 230 is used to extract features from the multiple frames of farmland monitoring images to obtain a first damaged feature, and to extract features from the farmland soil monitoring data and the farmland air monitoring data to obtain a second damaged feature;

[0113] The data regression solution module 240 is used to construct a multivariate output regression model for crop growth according to the crop growth monitoring data, and obtain the basic water demand and irrigation time table of the farmland to be irrigated according to the weather monitoring data, the farmland irrigation type and the multivariate output regression model for crop growth;

[0114] The irrigation result correction module 250 is used to correct the basic water demand to obtain a corrected water demand based on the first damaged feature and the second damaged feature, and irrigate the farmland to be irrigated according to the corrected water demand and the irrigation schedule.

[0115] In certain embodiments of the present application, the irrigation type determination module 220 includes a cluster analysis submodule, a weather characterization submodule, a similarity calculation submodule, and a type matching submodule;

[0116] The cluster analysis submodule is used to perform cluster analysis on the weather monitoring data to obtain multiple clusters, and obtain a first cluster corresponding to the one with the largest timestamp;

[0117] The weather characterization submodule is used to characterize all weather monitoring data in the first cluster respectively to obtain multiple weather monitoring features;

[0118] The similarity calculation submodule is used to calculate the cosine similarities between the multiple weather monitoring features and other weather monitoring features respectively, obtain multiple cosine similarities, and obtain first data and second data; wherein the first data and the second data are the weather monitoring features corresponding to the smallest one among the multiple cosine similarities;

[0119] The type matching submodule is used to match the farmland irrigation type in a preset farmland irrigation database according to the mean of the multiple cosine similarities, the first data and the second data.

[0120] In some embodiments of the present application, the damaged feature extraction module 230 includes a first damaged feature extraction submodule; the first damaged feature extraction submodule includes an image characterization unit, an image feature extraction unit, a feature domain division unit and a damaged feature merging unit;

[0121] The image characterization unit is used to characterize the multiple frames of farmland monitoring images respectively to obtain multiple image monitoring features;

[0122] The image feature extraction unit is used to extract features from the multiple image monitoring features based on a convolutional neural network to obtain multiple image damaged sub-features; wherein the multiple image damaged sub-features are arranged in ascending order according to the timestamps of the multiple frames of farmland monitoring images;

[0123] The feature domain division unit is used to calculate the Pearson correlation coefficients between the multiple damaged image sub-features and adjacent damaged image sub-features respectively, and divide the multiple damaged image sub-features into multiple damaged feature domains according to the Pearson correlation coefficients and a preset number of classifications;

[0124] The damaged feature merging unit is used to merge the multiple damaged image sub-features in the multiple damaged feature domains respectively to obtain multiple damaged image features, and to splice the multiple damaged image features in ascending order of timestamps to obtain a first damaged feature.

[0125] In some embodiments of the present application, the damaged feature extraction module 230 includes a second damaged feature extraction submodule; the second damaged feature extraction submodule includes a timestamp alignment unit, a multi-scale division unit, a feature extraction unit and a feature fusion unit;

[0126] The timestamp alignment unit is used to align and merge the farmland soil monitoring data and the farmland air monitoring data according to the timestamps to obtain farmland environment monitoring data;

[0127] The multi-scale division unit is used to divide the farmland environment monitoring data multiple times based on multiple time scales to obtain multiple farmland environment monitoring data subsets; wherein the multiple time scales include 3 days, 5 days and 10 days;

[0128] The feature extraction unit is used to extract features from the plurality of farmland environment monitoring data subsets respectively to obtain a plurality of environmental damage features;

[0129] The feature fusion unit is used to fuse the multiple environmental damage features to obtain a second damage feature.

[0130] In certain embodiments of the present application, the data regression solution module 240 includes a multivariate output regression model construction submodule; the multivariate output regression model construction submodule includes a model construction unit;

[0131] The model building unit is used to build a multivariate output regression model for crop growth based on the crop growth monitoring data, with crop plant height, irrigation time and irrigation water volume as independent variables, and water demand of crops at each growth stage as dependent variable, based on a gradient boosting regression tree model.

[0132] In certain embodiments of the present application, the data regression solution module 240 includes a regression model solution submodule; the regression model solution submodule includes a time series expansion unit, an initial water volume matching unit and a model solution unit;

[0133] The time series expansion unit is used to perform data prediction expansion on the weather monitoring data based on a preset time series prediction model to obtain weather prediction data;

[0134] The initial water volume matching unit is used to match the initial irrigation water volume in a preset farmland irrigation database according to the farmland irrigation type;

[0135] The model solving unit is used to input the weather monitoring data, the weather forecast data and the initial irrigation water volume into the crop growth multivariate output regression model to solve and obtain the basic water demand and irrigation schedule of the farmland to be irrigated.

[0136] In certain embodiments of the present application, the irrigation result correction module 250 includes a first correction submodule, a second correction submodule, and a water demand correction submodule;

[0137] The first correction submodule is used to match the weather monitoring data and the first damage feature in a preset farmland irrigation database to obtain a first correction value;

[0138] The second correction submodule is used to input the second damage feature into a preset crop irrigation time series prediction model to predict the irrigation water volume and obtain a second correction amount;

[0139] The water demand correction submodule is used to obtain a corrected water demand according to the first correction amount, the second correction amount, the weather monitoring data and the basic water demand.

[0140] In certain embodiments of the present application, the water demand correction submodule is specifically:

[0141] ;

[0142] in, are the modified water demand and the basic water demand, is the correction factor, is the daily average temperature, are the first correction and the second correction respectively, is the characteristic coefficient

[0143] This application first obtains weather monitoring data and farmland monitoring data of the farmland to be irrigated, and determines the farmland irrigation type according to the weather monitoring data, so as to preliminarily determine the water demand according to the farmland irrigation type, and then extracts features of multiple frames of farmland monitoring images, farmland soil monitoring data and farmland air monitoring data in the farmland monitoring data to obtain damaged features, taking into account the impact of farmland damage on farmland irrigation, and then constructs a multivariate output regression model for crop growth according to the crop growth monitoring data, and then obtains the basic water demand and irrigation schedule of the farmland to be irrigated according to the weather monitoring data and the farmland irrigation type. By constructing a multivariate output regression model, it is possible to focus on the growth of crops and the water demand of farmland, and then solve the basic water demand and irrigation schedule to better meet the actual irrigation needs of the farmland to be irrigated. Then, the basic water demand is corrected according to the damaged features, and the impact of farmland damage on irrigation is expanded, so that the corrected water demand is more in line with the actual irrigation needs under the current weather conditions, thereby improving the irrigation and maintenance effect of farmland crops and improving the rationality of farmland irrigation.

[0144] It should be understood that the device provided in the embodiments of the present application corresponds to the aforementioned method, and the farmland irrigation system after extreme weather provided in the embodiments of the present application can implement the farmland irrigation method after extreme weather provided in any embodiment of the present application.

[0145] Adaptively, the embodiments of the present application also provide a computer device and a computer-readable storage medium.

[0146] The computer device comprises: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor;

[0147] Wherein, when the processor executes the computer program, a method for farmland irrigation after extreme weather of the present application is implemented.

[0148] The computer-readable storage medium stores a plurality of instructions, and the instructions are suitable for a processor to load to execute a method of farmland irrigation after extreme weather of the present application.

[0149] The above are some implementation methods of the present application, and the purpose, technical scheme and beneficial effects of the present application are further described in detail. It should be clear that the above-mentioned partial implementation methods of the present application cannot be understood as limiting the present application. It is particularly pointed out that for those skilled in the art, any changes, modifications, equivalent substitutions and variations made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for irrigation of farmland after extreme weather, characterized in that: include: Acquire weather monitoring data and farmland monitoring data of the farmland to be irrigated; wherein the farmland monitoring data includes multiple frames of farmland monitoring images, crop growth monitoring data, farmland soil monitoring data, and farmland air monitoring data; Determining the type of farmland irrigation according to the weather monitoring data; Performing feature extraction on the multiple frames of farmland monitoring images to obtain a first damaged feature, and performing feature extraction on the farmland soil monitoring data and the farmland air monitoring data to obtain a second damaged feature; Constructing a multivariate output regression model for crop growth according to the crop growth monitoring data, and obtaining a basic water demand and an irrigation schedule for the farmland to be irrigated according to the weather monitoring data, the farmland irrigation type and the multivariate output regression model for crop growth; Based on the first damage characteristic and the second damage characteristic, the basic water demand is modified to obtain a modified water demand, and the farmland to be irrigated is irrigated according to the modified water demand and the irrigation schedule; The extracting features of the multiple frames of farmland monitoring images to obtain the first damaged features specifically includes: Characterizing the multiple frames of farmland monitoring images respectively to obtain multiple image monitoring features; Based on a convolutional neural network, feature extraction is performed on the multiple image monitoring features to obtain multiple image damaged sub-features; wherein the multiple image damaged sub-features are arranged in ascending order according to the timestamps of the multiple frames of farmland monitoring images; respectively calculating the Pearson correlation coefficients between the plurality of damaged image sub-features and adjacent damaged image sub-features, and dividing the plurality of damaged image sub-features into a plurality of damaged feature domains according to a preset number of classifications based on the Pearson correlation coefficients; Merging the multiple damaged image sub-features in the multiple damaged feature domains respectively to obtain multiple damaged image features, and concatenating the multiple damaged image features in ascending order of timestamps to obtain a first damaged feature; The step of constructing a multivariate output regression model for crop growth based on the crop growth monitoring data specifically includes: According to the crop growth monitoring data, a multivariate output regression model of crop growth is constructed based on a gradient boosting regression tree model, taking crop plant height, irrigation time and irrigation water volume as independent variables and crop water requirement at each growth stage as dependent variable; According to the weather monitoring data, the farmland irrigation type and the crop growth multivariate output regression model, the basic water demand and irrigation time table of the farmland to be irrigated are obtained, which specifically include: Based on a preset time series prediction model, the weather monitoring data is subjected to data prediction expansion to obtain weather prediction data; According to the farmland irrigation type, an initial irrigation water volume is matched in a preset farmland irrigation database; The weather monitoring data, the weather forecast data and the initial irrigation water volume are input into the crop growth multivariate output regression model to obtain the basic water demand and irrigation schedule of the farmland to be irrigated.

2. The method for irrigation of farmland after extreme weather according to claim 1, characterized in that: Determining the farmland irrigation type according to the weather monitoring data specifically includes: Performing cluster analysis on the weather monitoring data to obtain multiple clusters, and obtaining a first cluster corresponding to the one with the largest timestamp; Characterizing all weather monitoring data in the first cluster respectively to obtain a plurality of weather monitoring features; Calculate the cosine similarities of the multiple weather monitoring features and other weather monitoring features respectively to obtain multiple cosine similarities, and obtain first data and second data; wherein the first data and the second data are the weather monitoring features corresponding to the smallest one among the multiple cosine similarities; According to the mean values ​​of the multiple cosine similarities, the first data and the second data, the farmland irrigation type is matched in a preset farmland irrigation database.

3. The method for irrigation of farmland after extreme weather according to claim 1, characterized in that: The feature extraction of the farmland soil monitoring data and the farmland air monitoring data to obtain the second damaged feature specifically includes: Aligning and merging the farmland soil monitoring data and the farmland air monitoring data according to timestamps to obtain farmland environment monitoring data; The farmland environment monitoring data is divided multiple times based on multiple time scales to obtain multiple farmland environment monitoring data subsets; wherein the multiple time scales include 3 days, 5 days and 10 days; Extracting features from the plurality of farmland environment monitoring data subsets respectively to obtain a plurality of environmental damage features; The plurality of environmental impairment features are fused to obtain a second impairment feature.

4. The method for irrigation of farmland after extreme weather according to claim 1, characterized in that: The step of correcting the basic water demand based on the first damage characteristic and the second damage characteristic to obtain a corrected water demand specifically includes: According to the weather monitoring data and the first damage characteristic, a first correction value is obtained by matching in a preset farmland irrigation database; Inputting the second damage feature into a preset crop irrigation time series prediction model to predict the irrigation water volume to obtain a second correction amount; A corrected water demand is obtained according to the first correction amount, the second correction amount, the weather monitoring data and the basic water demand.

5. The method for irrigation of farmland after extreme weather according to claim 4, characterized in that: The corrected water demand is obtained according to the first correction amount, the second correction amount, the weather monitoring data and the basic water demand, specifically: ； in, are the modified water demand and the basic water demand, is the correction factor, is the daily average temperature, are the first correction and the second correction respectively, is the characteristic coefficient.

6. A farmland irrigation system after extreme weather, the farmland irrigation system is used to implement the farmland irrigation method after extreme weather as claimed in claim 1, characterized in that: It includes a monitoring data acquisition module, an irrigation type determination module, a damaged feature extraction module, a data regression solution module and an irrigation result correction module; The monitoring data acquisition module is used to acquire weather monitoring data and farmland monitoring data of the farmland to be irrigated; wherein the farmland monitoring data includes multiple frames of farmland monitoring images, crop growth monitoring data, farmland soil monitoring data and farmland air monitoring data; The irrigation type determination module is used to determine the farmland irrigation type according to the weather monitoring data; The damaged feature extraction module is used to extract features from the multiple frames of farmland monitoring images to obtain a first damaged feature, and to extract features from the farmland soil monitoring data and the farmland air monitoring data to obtain a second damaged feature; The data regression solution module is used to construct a multivariate output regression model for crop growth according to the crop growth monitoring data, and obtain the basic water demand and irrigation time table of the farmland to be irrigated according to the weather monitoring data, the farmland irrigation type and the multivariate output regression model for crop growth; The irrigation result correction module is used to correct the basic water demand to obtain a corrected water demand based on the first damage characteristic and the second damage characteristic, and irrigate the farmland to be irrigated according to the corrected water demand and the irrigation schedule; The damaged feature extraction module includes a first damaged feature extraction submodule; the first damaged feature extraction submodule includes an image characterization unit, an image feature extraction unit, a feature domain division unit and a damaged feature merging unit; The image characterization unit is used to characterize the multiple frames of farmland monitoring images respectively to obtain multiple image monitoring features; The image feature extraction unit is used to extract features from the multiple image monitoring features based on a convolutional neural network to obtain multiple image damaged sub-features; wherein the multiple image damaged sub-features are arranged in ascending order according to the timestamps of the multiple frames of farmland monitoring images; The feature domain division unit is used to calculate the Pearson correlation coefficients between the multiple damaged image sub-features and adjacent damaged image sub-features respectively, and divide the multiple damaged image sub-features into multiple damaged feature domains according to the Pearson correlation coefficients and a preset number of classifications; The damaged feature merging unit is used to merge the multiple damaged image sub-features in the multiple damaged feature domains respectively to obtain multiple damaged image features, and to splice the multiple damaged image features in ascending order of timestamps to obtain a first damaged feature.

Images