Dynamic irrigation regulation and control method and system for drainage and drainage

Abstract

The invention discloses a dynamic irrigation regulation and control method and system for drainage and drainage, and the method comprises the steps: carrying out the interaction analysis of time series data among rainfall, river water level and farmland evapotranspiration through introducing a data processing and analysis algorithm based on big data analysis and deep learning, and taking the rainfall as a main variable; and the river water level and the farmland evapotranspiration are taken as covariables, so that time sequence fine-grained interactive semantic representation between irrigation main-covariables is captured, and dynamic analysis is carried out based on interactive semantic representation information so as to carry out adaptive adjustment on the irrigation frequency. Therefore, the drainage process can be optimized through time sequence dynamic interaction analysis among different irrigation parameter variables, more intelligent and adaptive irrigation strategy formulation is realized, and sustainable utilization of water resources is ensured.

Description

Technical Field

[0001] The present application relates to the field of intelligent analysis, and more specifically, to a dynamic irrigation regulation method and system for water diversion, drainage, and irrigation. Background Art

[0002] Irrigation is a crucial link in agricultural production, directly affecting the yield and quality of crops. Traditional irrigation methods often rely on experience or fixed irrigation schedules, which may lead to waste of water resources or insufficient water supply for crops. With the increasing demand for water resource management and the uncertainties brought about by climate change, more intelligent and efficient dynamic analysis and irrigation management systems have become particularly important.

[0003] Therefore, a dynamic irrigation regulation scheme for water diversion, drainage, and irrigation is desired. Summary of the Invention

[0004] To solve the above technical problems, the present application is proposed. Embodiments of the present application provide a dynamic irrigation regulation method and system for water diversion, drainage, and irrigation, which perform interactive analysis on the time-series data among rainfall, river water level, and farmland evapotranspiration by introducing data processing and analysis algorithms based on big data analysis and deep learning. By using rainfall as the main variable and river water level and farmland evapotranspiration as the covariables, the time-series fine-grained interactive semantic representation between irrigation main and covariables is captured, and based on the interactive semantic representation information, dynamic analysis is performed to adaptively adjust the irrigation frequency. In this way, the water diversion, drainage, and irrigation process can be optimized through the time-series dynamic interactive analysis between different irrigation parameter variables, and a more intelligent and adaptive irrigation strategy can be formulated to ensure the sustainable utilization of water resources.

[0005] According to one aspect of the present application, a dynamic irrigation regulation method for water diversion, drainage, and irrigation is provided, which includes:

[0006] Collect the time-series data sets of rainfall, river water level, and farmland evapotranspiration;

[0007] Use the rainfall as the main variable, and perform time-series encoding on the time-series data set of rainfall to obtain the time-series hidden encoding features of the irrigation main variable;

[0008] Use the river water level and the farmland evapotranspiration as the covariables, and perform time-series encoding on the time-series data sets of the river water level and the farmland evapotranspiration respectively to obtain the first time-series hidden encoding features of the irrigation covariable and the second time-series hidden encoding features of the irrigation covariable;

[0009] Perform feature alignment and fusion processing based on semantic information field modulation on the temporal implicit coding features of the main irrigation variables, the first irrigation covariate temporal implicit coding features, and the second irrigation covariate temporal implicit coding features to obtain fine-grained alignment and fusion features of the main-irrigation covariate time series;

[0010] Perform dynamic analysis based on the fine-grained alignment and fusion features of the main-irrigation covariate time series to determine whether the irrigation frequency should be maintained, reduced, or increased.

[0011] According to another aspect of the present application, a dynamic irrigation regulation system for water diversion, drainage, and irrigation is provided, which includes:

[0012] A data collection module for collecting time series datasets of rainfall, river water levels, and farmland evapotranspiration;

[0013] A main variable time series encoding module for using the rainfall as the main variable and performing time series encoding on the time series dataset of the rainfall to obtain temporal implicit coding features of the main irrigation variables;

[0014] A covariate time series encoding module for using the river water level and the farmland evapotranspiration as covariates and performing time series encoding on the time series datasets of the river water level and the farmland evapotranspiration respectively to obtain the first irrigation covariate temporal implicit coding features and the second irrigation covariate temporal implicit coding features;

[0015] A feature alignment and fusion module for performing feature alignment and fusion processing based on semantic information field modulation on the temporal implicit coding features of the main irrigation variables, the first irrigation covariate temporal implicit coding features, and the second irrigation covariate temporal implicit coding features to obtain fine-grained alignment and fusion features of the main-irrigation covariate time series;

[0016] An irrigation frequency determination module for performing dynamic analysis based on the fine-grained alignment and fusion features of the main-irrigation covariate time series to determine whether the irrigation frequency should be maintained, reduced, or increased.

[0017] Compared with the prior art, a dynamic irrigation regulation method and system for water diversion, drainage and irrigation provided by the present application perform interactive analysis on the time-series data among rainfall, river water level and farmland evapotranspiration by introducing data processing and analysis algorithms based on big data analysis and deep learning. In this way, rainfall is used as the main variable, and the river water level and farmland evapotranspiration are used as the covariates, so as to capture the time-series fine-grained interactive semantic representation between the irrigation main-covariate variables, and based on the interactive semantic representation information, dynamically analyze to adaptively adjust the irrigation frequency. In this way, the water diversion, drainage and irrigation process can be optimized through the time-series dynamic interactive analysis between different irrigation parameter variables, a more intelligent and adaptive irrigation strategy can be formulated, and the sustainable utilization of water resources can be ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other objects, features and advantages of the present application will become more obvious. The accompanying drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification, and are used to explain the present application together with the embodiments of the present application, and do not constitute a limitation to the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0019] Figure 1 It is a flowchart of a dynamic irrigation regulation method for water diversion, drainage and irrigation according to an embodiment of the present application;

[0020] Figure 2 It is a schematic diagram of data flow of a dynamic irrigation regulation method for water diversion, drainage and irrigation according to an embodiment of the present application;

[0021] Figure 3 It is a flowchart of sub-step S4 of a dynamic irrigation regulation method for water diversion, drainage and irrigation according to an embodiment of the present application;

[0022] Figure 4 It is a block diagram of a dynamic irrigation regulation system for water diversion, drainage and irrigation according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0024] As shown in this application and the claims, unless the context clearly indicates otherwise, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0025] Although this application makes various references to certain modules in the system according to the embodiments of this application, however, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0026] Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the operations before or below do not necessarily need to be executed precisely in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or several steps can be removed from these processes.

[0027] Next, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited by the exemplary embodiments described herein.

[0028] System Dynamics (SD) is an analytical tool for dealing with complex systems. It allows the establishment of a simulation model to dynamically simulate the irrigation process and can capture the dynamic interaction relationships between various variables within the system. By constructing an SD model, a deeper understanding of the operation mechanism of the irrigation system can be achieved, including the process of water introduction from the river to the farmland, the water migration in the soil, crop transpiration, etc., and how these processes respond to different environmental conditions and management decisions. For example, in a farmland irrigation project, water resources are introduced from a river. To optimize the irrigation process, a system dynamics model is used for dynamic analysis. First, key data such as rainfall, river water level, and farmland evapotranspiration are collected. Then, a system dynamics model is established and these data are input into the model for simulation. Through simulation, the interaction relationships and mutual influences between different irrigation parameter variables are analyzed, as well as their impacts on water resource utilization. For example, reduce the water diversion volume in the rainy season to save water resources; in the dry period, adjust the water diversion volume and irrigation frequency to ensure the water demand of crops, optimize the entire water diversion, transportation, and drainage process, and achieve the sustainable utilization of water resources.

[0029] In recent years, the development of deep learning and artificial intelligence technologies has provided new ideas for the optimization of irrigation systems. Therefore, in order to more deeply understand the irrigation process and capture the dynamic interaction relationships between various irrigation parameter variables to adaptively adjust the irrigation strategy, thereby optimizing the entire water diversion, conveyance, and drainage process and achieving the sustainable utilization of water resources.

[0030] Based on this, in the technical solution of this application, a dynamic irrigation regulation method for water diversion, conveyance, and drainage is proposed. Figure 1 It is a flowchart of the dynamic irrigation regulation method for water diversion, conveyance, and drainage according to an embodiment of this application. Figure 2 It is a schematic diagram of data flow of the dynamic irrigation regulation method for water diversion, conveyance, and drainage according to an embodiment of this application. As Figure 1 and Figure 2 shown, the dynamic irrigation regulation method for water diversion, conveyance, and drainage according to an embodiment of this application includes the steps of: S1, collecting the time series data sets of rainfall, river water level, and farmland evapotranspiration; S2, taking the rainfall as the main variable, performing time series encoding on the time series data set of rainfall to obtain the time series implicit encoding features of the irrigation main variable; S3, taking the river water level and the farmland evapotranspiration as the covariables, respectively performing time series encoding on the time series data set of the river water level and the time series data of the farmland evapotranspiration to obtain the first irrigation covariable time series implicit encoding features and the second irrigation covariable time series implicit encoding features; S4, performing feature alignment and fusion processing based on semantic information field modulation on the time series implicit encoding features of the irrigation main variable, the first irrigation covariable time series implicit encoding features, and the second irrigation covariable time series implicit encoding features to obtain the time series fine-grained alignment and fusion features of the irrigation main-covariable; S5, performing dynamic analysis based on the time series fine-grained alignment and fusion features of the irrigation main-covariable to determine whether the irrigation frequency should be maintained, reduced, or increased.

[0031] Specifically, the S1 collects the time - series datasets of rainfall, river water level, and farmland evapotranspiration. It should be understood that rainfall refers to the amount of water falling on the earth's surface within a certain period of time, usually measured in millimeters, and is one of the main ways for nature to provide water for farmland, which can significantly reduce the need for artificial irrigation. Understanding the temporal distribution and intensity of rainfall is crucial for reasonably planning irrigation strategies; in seasons or regions with sufficient rainfall, the irrigation frequency can be appropriately reduced, while in dry periods, additional irrigation is needed to supplement the water required by crops. Therefore, collecting the time - series dataset of rainfall is the basis for formulating an effective irrigation plan, which can reveal the changing trends of rainfall patterns and help farmers adjust irrigation operations according to actual situations. River water level refers to the height of the water surface in a river at a certain moment, which is affected by various factors such as seasonal changes, climatic conditions, and upstream precipitation. The time - series dataset of river water level describes the changing trend of river water level over time, which is particularly important for irrigation systems that rely on rivers as the main water source. Rivers are usually one of the important water sources for many irrigation systems, and the level of the river water directly affects the amount of water that can be pumped for irrigation. If the water level is too low, it may limit the quantity of available water resources, thus requiring an adjustment of the irrigation strategy. For example, during the dry season, alternative water sources may need to be found or the irrigation volume reduced to conserve limited water resources. In addition, the change in river water level also reflects the overall hydrological conditions within the basin, which helps predict future water resource availability and make corresponding preparations in advance. Farmland evapotranspiration includes two parts: soil evaporation and plant transpiration, and is an important indicator for measuring the water consumption of crops. It represents the total amount of water returned to the atmosphere from the ground and vegetation surfaces and is affected by various meteorological factors such as temperature, humidity, wind speed, and solar radiation. Accurately estimating evapotranspiration is crucial for determining the actual water requirements of crops, which can not only guide the irrigation volume but also improve water resource utilization efficiency. A high evapotranspiration rate means that crops need more water supply, and vice versa, the irrigation frequency can be reduced accordingly. By monitoring the changes in evapotranspiration, the water requirements of crops can be more precisely met, avoiding over - irrigation or under - irrigation. Therefore, collecting the time - series dataset of farmland evapotranspiration is also an indispensable part of building an intelligent irrigation system, providing first - hand information about the water requirements of crops and making irrigation decisions more scientific and reasonable.

[0032] Specifically, in step S2, taking the rainfall as the main variable, the time-series dataset of the rainfall is encoded in time series to obtain the time-series implicit encoding features of the main irrigation variable. It should be understood that since rainfall is one of the main ways for nature to provide water for farmland, it can significantly reduce the need for artificial irrigation. Therefore, understanding the temporal distribution and intensity of rainfall is crucial for reasonably planning irrigation strategies. Although traditional statistical methods can describe the trends of historical data to a certain extent, their ability to capture non-linear and long-term dependencies is limited. Based on this, by taking the rainfall as the main variable, the time-series dataset of the rainfall is input into a sequence encoder based on an LSTM-RNN hybrid model to obtain the time-series implicit encoding vector of the main irrigation variable. In this way, the time-series implicit correlation feature information of the main variable related to irrigation, i.e., rainfall, in the time dimension can be extracted, providing a basis for subsequent additional irrigation frequency decision-making tasks. Here, after the time-series dataset of the rainfall is input into the sequence encoder composed of the LSTM-RNN hybrid model, the encoder reads the rainfall values at each time point in sequence according to the time order, and gradually constructs an implicit representation form that can reflect the overall characteristics of the rainfall, namely, the time-series implicit encoding vector of the main irrigation variable.

[0033] Specifically, in step S3, taking the river water level and the farmland evapotranspiration as covariates, the time-series datasets of the river water level and the farmland evapotranspiration are encoded in time series respectively to obtain the first time-series implicit encoding features of the irrigation covariate and the second time-series implicit encoding features of the irrigation covariate. Considering irrigation requirements and decision-making, in addition to the main variable factor of rainfall, the river water level and the farmland evapotranspiration also need to be considered. This is because rivers are one of the main water sources for many irrigation systems. The level of the river water directly affects the amount of water that can be pumped for irrigation. If the water level is too low, it may limit the amount of available water resources, thus requiring adjustment of irrigation strategies. Evapotranspiration is an important indicator for measuring the water consumption of crops, which includes the process of plants absorbing water from the soil and evaporating it through leaves. Accurately estimating evapotranspiration helps determine the actual water requirements of crops, thus guiding the irrigation amount. Both the river water level and the farmland evapotranspiration are data sequences that change over time and have complex time-dependent relationships. Based on this, in the technical solution of this application, the river water level and the farmland evapotranspiration are further used as covariates, and the time-series datasets of the river water level and the farmland evapotranspiration are input into the sequence encoder based on the LSTM-RNN hybrid model to obtain the first time-series implicit encoding vector of the irrigation covariate and the second time-series implicit encoding vector of the irrigation covariate. In this way, the time-series dependent correlation feature information of the river water level and the farmland evapotranspiration in the time dimension can be captured respectively.

[0034] Specifically, in S4, a feature alignment and fusion process based on semantic information field modulation is performed on the irrigation main variable time series implicit coding feature, the first irrigation covariate time series implicit coding feature, and the second irrigation covariate time series implicit coding feature to obtain an irrigation main-covariate time series fine-grained alignment and fusion feature. It should be understood that the irrigation main variable time series implicit coding vector and the irrigation covariate time series concatenated implicit coding vector respectively contain the time series feature information of the irrigation main variable of rainfall amount and the time series fusion feature information of the two irrigation covariates of river water level and farmland evapotranspiration. There is a complex dynamic interaction between the time series semantics of the irrigation main variable and the covariates, and this dynamic interaction is of great significance for the formulation of irrigation strategies and the optimization of the water diversion, conveyance, and drainage process. However, traditional feature fusion methods may lead to information loss or inaccurate representation, especially when dealing with irrigation variable time series data from different sources or different modalities. Moreover, the dynamic analysis of the water diversion, conveyance, and drainage process and the formulation of irrigation strategies are not only affected by the internal interaction and correlation between the main variable and the covariates, but also by external key factors such as weather conditions, which means that there is a significant gap between the low-level time series perception features directly obtained from the irrigation main variable and the irrigation covariates and the high-level semantic concepts. Therefore, in the technical solution of this application, a feature alignment and fusion process based on semantic information field modulation is further performed on the irrigation main variable time series implicit coding feature and the irrigation covariate time series fusion feature to obtain an irrigation main-covariate time series fine-grained alignment and fusion feature.

[0035] In particular, through the feature alignment and fusion processing based on the semantic information field modulation, fine-grained alignment can be performed on the irrigation main variable time-series implicit encoding vector and the irrigation co-variable time-series concatenated implicit encoding vector, ensuring that the features extracted even in different time-series contexts can correspond precisely, improving the quality and expressive ability of the fused irrigation main-co-variable time-series fine-grained alignment and fusion features. This is very important for describing the time-series feature interaction and dynamic behavior semantics between irrigation variables, helping to better understand the actual irrigation needs to make corresponding decisions. Moreover, through the feature alignment and fusion processing, the semantic correlation and hidden connection between the irrigation main variable time-series implicit encoding features and the irrigation co-variable time-series fusion features can be captured at a higher level, enabling the model to more deeply understand the true meaning of the irrigation variable time-series. In addition, by constructing a space rich in semantic information to achieve the fine-grained alignment and deep interaction between the irrigation main variable time-series implicit encoding vector and the irrigation co-variable time-series concatenated implicit encoding vector, the provided semantic information field helps to alleviate the domain differences, promote the effective transfer of knowledge and the rapid adjustment of the model, so as to capture and strengthen the subtle semantic relationship between the irrigation main variable time-series implicit encoding features and the irrigation co-variable time-series fusion features, thereby generating a highly comprehensive and semantically rich irrigation main-co-variable time-series fine-grained alignment and fusion feature vector as the irrigation main-co-variable time-series fine-grained alignment and fusion feature. In this way, the dynamic analysis and irrigation management system can not only consider weather conditions (such as rainfall), but also combine the changing trend of river water levels and the actual water demand of farmland to dynamically adjust the irrigation strategy. Such a system can help make more scientific and reasonable irrigation decisions, such as reducing artificial irrigation when sufficient natural precipitation is expected, or reasonably allocating limited water resources during drought periods to ensure the healthy growth of crops while achieving the efficient utilization of water resources.

[0036] In a specific example of the present application, as Figure 3 shown, the S4 includes: S41, fusing the first irrigation co-variable time-series implicit encoding feature and the second irrigation co-variable time-series implicit encoding feature to obtain an irrigation co-variable time-series fusion feature; S42, performing semantic information field calculation on the irrigation main variable time-series implicit encoding feature and the irrigation co-variable time-series fusion feature to obtain a fine-grained semantic information field between the irrigation main-co-variable time-series encoding features; S43, based on the fine-grained semantic information field between the irrigation main-co-variable time-series encoding features, performing feature fusion with fine-grained alignment based on field mapping on the irrigation main variable time-series implicit encoding feature and the irrigation co-variable time-series fusion feature to obtain the irrigation main-co-variable time-series fine-grained alignment and fusion feature.

[0037] Specifically, in S41, the first irrigation covariate time-series implicit encoding features and the second irrigation covariate time-series implicit encoding features are fused to obtain irrigation covariate time-series fusion features. In the technical solution of this application, the first irrigation covariate time-series implicit encoding vector and the second irrigation covariate time-series implicit encoding vector are concatenated to obtain an irrigation covariate time-series concatenated implicit encoding vector as the irrigation covariate time-series fusion features. That is, by concatenating the first irrigation covariate time-series implicit encoding vector and the second irrigation covariate time-series implicit encoding vector to obtain an irrigation covariate time-series concatenated implicit encoding vector as the irrigation covariate time-series fusion features, the time-series dynamic information and the impact on irrigation of these two irrigation covariates, namely the river water level time-series features and the farmland evapotranspiration time-series features, are integrated in the time dimension, providing more comprehensive information support for the subsequent formulation of irrigation strategies.

[0038] Specifically, in S42, semantic information field calculation is performed on the irrigation main variable time-series implicit encoding features and the irrigation covariate time-series fusion features to obtain a fine-grained semantic information field between the irrigation main-covariate time-series encoding features. In the technical solution of this application, first, the irrigation main variable time-series implicit encoding vector and the irrigation covariate time-series concatenated implicit encoding vector are input into a dimension modulation module based on point convolution to obtain a modulated irrigation main variable time-series implicit encoding vector and a modulated irrigation covariate time-series concatenated implicit encoding vector, where the modulated irrigation main variable time-series implicit encoding vector and the modulated irrigation covariate time-series concatenated implicit encoding vector have the same feature dimension;

[0039] Furthermore, fine-grained association encoding and semantic information field encoding are performed on the modulated irrigation main variable time-series implicit encoding vector and the modulated irrigation covariate time-series concatenated implicit encoding vector to obtain a fine-grained semantic information field between the irrigation main-covariate time-series encoding features. Among them, the specific process of performing fine-grained association encoding and semantic information field encoding on the modulated irrigation main variable time-series implicit encoding vector and the modulated irrigation covariate time-series concatenated implicit encoding vector includes: performing fine-grained association encoding on the modulated irrigation main variable time-series implicit encoding vector and the modulated irrigation covariate time-series concatenated implicit encoding vector to obtain an irrigation main-covariate time-series fine-grained association feature matrix; inputting the irrigation main-covariate time-series fine-grained association feature matrix into a semantic information field encoding network based on a convolution kernel to obtain a fine-grained semantic information field between the irrigation main-covariate time-series encoding features. More specifically, the following semantic information field calculation formula is used to perform semantic information field calculation on the irrigation main variable time-series implicit encoding features and the irrigation covariate time-series fusion features to obtain a fine-grained semantic information field between the irrigation main-covariate time-series encoding features; where the semantic information field calculation formula is:

[0040] Among them, is the time-series implicit encoding vector of the irrigation main variable, is the time-series cascaded implicit encoding vector of the irrigation covariates obtained by cascading and fusing the first irrigation covariate time-series implicit encoding vector and the second irrigation covariate time-series implicit encoding vector. is point convolution processing, is the activation function, and are the modulated time-series implicit encoding vector of the irrigation main variable and the modulated time-series cascaded implicit encoding vector of the irrigation covariates respectively. is the length of the modulated time-series cascaded implicit encoding vector of the irrigation covariates. is the fine-grained association feature matrix of the irrigation main-covariate time series. is the convolutional layer with a convolution kernel of, and is the fine-grained semantic information field matrix between the irrigation main-covariate time series encoding features.

[0041] Specifically, in step S43, based on the fine-grained semantic information field between the irrigation main-covariate time series encoding features, feature fusion with fine-grained alignment based on field mapping is performed on the time-series implicit encoding features of the irrigation main variable and the time-series fusion features of the irrigation covariates to obtain the fine-grained alignment fusion features of the irrigation main-covariate time series. That is, in the technical solution of the present application, first, the modulated time-series implicit encoding vector of the irrigation main variable and the modulated time-series cascaded implicit encoding vector of the irrigation covariates are respectively multiplied by the fine-grained semantic information field matrix between the irrigation main-covariate time series encoding features to obtain the fine-grained aligned time-series implicit encoding vector of the irrigation main variable and the fine-grained aligned time-series cascaded implicit encoding vector of the irrigation covariates; then, the weighted sum by position between the fine-grained aligned time-series implicit encoding vector of the irrigation main variable and the fine-grained aligned time-series cascaded implicit encoding vector of the irrigation covariates is calculated to obtain the fine-grained alignment fusion feature vector of the irrigation main-covariate time series as the fine-grained alignment fusion features of the irrigation main-covariate time series. More specifically, based on the fine-grained semantic information field between the irrigation main-covariate time series encoding features, the following feature fusion formula is used to perform feature fusion with fine-grained alignment based on field mapping on the time-series implicit encoding features of the irrigation main variable and the time-series fusion features of the irrigation covariates to obtain the fine-grained alignment fusion features of the irrigation main-covariate time series; where the feature fusion formula is:

[0042] Among them, is matrix multiplication, and are the fine-grained aligned time-series implicit encoding vector of the irrigation main variable and the fine-grained aligned time-series cascaded implicit encoding vector of the irrigation covariates respectively, and are weight hyperparameters, and is the fine-grained alignment fusion feature vector of the irrigation main-covariate time series.

[0043] In particular, in step S5, dynamic analysis is performed based on the fine-grained alignment and fusion features of the irrigation main-covariate time series to determine whether the irrigation frequency should be maintained, decreased, or increased. That is, in a specific example of the present application, the fine-grained alignment and fusion feature vector of the irrigation main-covariate time series is input into a dynamic analysis module based on a classifier to obtain an analysis result, and the analysis result is used to indicate whether the irrigation frequency should be maintained, decreased, or increased. That is to say, the fine-grained interaction and fusion feature information between the main variable time series features and the covariate time series features of irrigation are used to dynamically analyze the irrigation demand, so as to adaptively adjust the irrigation frequency. In this way, the water diversion, conveyance, and drainage process can be optimized through the time series dynamic interaction analysis between different irrigation parameter variables, and a more intelligent and adaptive irrigation strategy can be formulated to ensure the sustainable utilization of water resources.

[0044] In particular, considering that the implicit coding vector of the irrigation main variable time series and the cascaded implicit coding vector of the irrigation covariate time series respectively represent the time series coding features of rainfall and the time series co-correlation features of river water level and the evapotranspiration of the farmland, when performing the alignment and interaction between features based on the semantic information field, the lack of correspondence in the semantic information positions of the time series features under different data source time series will lead to the loss of instance judgment of the interaction features of the fine-grained alignment and fusion feature vector of the irrigation main-covariate time series, thus affecting the accuracy of the analysis result obtained through the dynamic analysis module based on the classifier.

[0045] Preferably, inputting the fine-grained alignment and fusion feature vector of the irrigation main-covariate time series into a dynamic analysis module based on a classifier to obtain an analysis result includes:

[0046] Solving the similarity difference between the feature dimensions of the fine-grained alignment and fusion feature vector of the irrigation main-covariate time series, for example, using the Euclidean distance for measurement, and performing a square root operation on the difference value to obtain a fine-grained alignment and fusion difference measurement matrix of the irrigation main-covariate time series, that is,

[0047] where and respectively represent the th and th eigenvalue of the fine-grained alignment and fusion feature vector of the irrigation main-covariate time series, represents the fine-grained alignment and fusion feature vector of the irrigation main-covariate time series, represents solving the similarity difference between the feature dimensions of the fine-grained alignment and fusion feature vector of the irrigation main-covariate time series, and represents the eigenvalue at the position of the fine-grained alignment and fusion difference measurement matrix of the irrigation main-covariate time series;

[0048] Converting the fine-grained alignment and fusion feature vector of the irrigation main-covariate time series in the form of a row vector into a corresponding internal association matrix to construct a fine-grained alignment and fusion self-similarity matrix of the irrigation main-covariate time series, that is,, where represents matrix multiplication, represents transpose, and represents the fine-grained alignment and fusion self-similarity matrix of the irrigation main-covariate time series;

[0049] Multiply the fine-grained aligned fusion feature vector of the irrigation primary-covariate time series with the fine-grained aligned fusion difference metric matrix of the irrigation primary-covariate time series to obtain a fine-grained aligned fusion primary projection vector of the irrigation primary-covariate time series, that is, where represents the fine-grained aligned fusion primary projection vector of the irrigation primary-covariate time series;

[0050] Perform a hierarchical mapping operation on the joint product result of the fine-grained aligned fusion primary projection vector of the irrigation primary-covariate time series, the fine-grained aligned fusion difference metric matrix of the irrigation primary-covariate time series, and the fine-grained aligned fusion self-similarity matrix of the irrigation primary-covariate time series to obtain a fine-grained aligned fusion multi-layer mapping vector of the irrigation primary-covariate time series, where represents the fine-grained aligned fusion multi-layer mapping vector of the irrigation primary-covariate time series;

[0051] Dot the fine-grained aligned fusion multi-layer mapping vector of the irrigation primary-covariate time series with the fine-grained aligned fusion correlation basis vector composed of the eigenvalues of the fine-grained aligned fusion self-similarity matrix of the irrigation primary-covariate time series to obtain an optimized fine-grained aligned fusion feature vector of the irrigation primary-covariate time series, where interpolation or zero-padding is performed in the case of insufficient eigenvalues;

[0052] Input the optimized fine-grained aligned fusion feature vector of the irrigation primary-covariate time series into a dynamic analysis module based on a classifier to obtain an analysis result.

[0053] Therefore, by relying on the difference metric matrix of the fine-grained aligned fusion feature vector of the irrigation primary-covariate time series to construct a linear projection model, implementing a quadratic mapping reconstruction of the global instantiation expression of the self-similar features of this vector with a multi-level structure distribution, and integrating a fusion kernel correction mechanism to correct the feature correlation deviation, thereby strengthening the discriminant weight of the feature dimension of the fine-grained aligned fusion feature vector of the irrigation primary-covariate time series in iterative decision-making (i.e., enhancing the significant contribution of the feature instance to the regression decision), and improving the accuracy of the analysis result obtained by the fine-grained aligned fusion feature vector of the irrigation primary-covariate time series through a dynamic analysis module based on a classifier. In this way, it is possible to optimize the water diversion, transportation, and drainage process through the time-series dynamic interaction analysis between different irrigation parameter variables, realize a more intelligent and adaptive irrigation strategy formulation, and ensure the sustainable utilization of water resources.

[0054] In summary, the dynamic irrigation regulation method for water diversion, drainage, and irrigation according to the embodiments of the present application is elucidated. It performs interactive analysis on the time-series data among rainfall, river water level, and farmland evapotranspiration by introducing data processing and analysis algorithms based on big data analysis and deep learning. By using rainfall as the main variable and river water level and farmland evapotranspiration as the covariates, the time-series fine-grained interactive semantic representation between the irrigation main and covariate variables is captured, and based on this interactive semantic representation information, dynamic analysis is performed to adaptively adjust the irrigation frequency. In this way, the water diversion, drainage, and irrigation process can be optimized through the time-series dynamic interactive analysis of different irrigation parameter variables, enabling the formulation of a more intelligent and adaptive irrigation strategy and ensuring the sustainable utilization of water resources.

[0055] Furthermore, a dynamic irrigation regulation system for water diversion, drainage, and irrigation is also provided.

[0056] Figure 4 The block diagram of the dynamic irrigation regulation system for water diversion, drainage, and irrigation according to the embodiments of the present application is as follows. Figure 4 As shown, the dynamic irrigation regulation system 300 according to the embodiments of the present application includes: a data collection module 310 for collecting the time-series data sets of rainfall, the time-series data sets of river water level, and the time-series data sets of farmland evapotranspiration; a main variable time-series encoding module 320 for using the rainfall as the main variable and performing time-series encoding on the time-series data set of rainfall to obtain the irrigation main variable time-series implicit encoding feature; a covariate time-series encoding module 330 for using the river water level and the farmland evapotranspiration as the covariates and respectively performing time-series encoding on the time-series data set of the river water level and the time-series data of the farmland evapotranspiration to obtain the first irrigation covariate time-series implicit encoding feature and the second irrigation covariate time-series implicit encoding feature; a feature alignment and fusion module 340 for performing feature alignment and fusion processing based on semantic information field modulation on the irrigation main variable time-series implicit encoding feature, the first irrigation covariate time-series implicit encoding feature, and the second irrigation covariate time-series implicit encoding feature to obtain the irrigation main-covariate time-series fine-grained alignment and fusion feature; an irrigation frequency determination module 350 for performing dynamic analysis based on the irrigation main-covariate time-series fine-grained alignment and fusion feature to determine whether the irrigation frequency should be maintained, decreased, or increased.

[0057] As described above, the dynamic irrigation regulation system 300 for water diversion and drainage according to the embodiments of the present application can be implemented in various wireless terminals, such as a server with a dynamic irrigation regulation algorithm for water diversion and drainage. In a possible implementation manner, the dynamic irrigation regulation system 300 for water diversion and drainage according to the embodiments of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the dynamic irrigation regulation system 300 for water diversion and drainage can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the dynamic irrigation regulation system 300 for water diversion and drainage can also be one of the many hardware modules of the wireless terminal.

[0058] Alternatively, in another example, the dynamic irrigation regulation system 300 for water diversion and drainage and the wireless terminal can also be separate devices, and the dynamic irrigation regulation system 300 for water diversion and drainage can be connected to the wireless terminal through a wired and / or wireless network and transmit interaction information according to a predefined data format.

[0059] The embodiments of the present disclosure have been described above. The above description is exemplary and not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art in the technical field without departing from the scope and spirit of the described embodiments. The selection of the terms used herein is intended to best explain the principles of the embodiments, the practical application, or the improvement of the technology in the market, or to enable other ordinary skill in the art in the technical field to understand the embodiments disclosed herein.

Claims

1. A dynamic irrigation regulation method and system for water diversion, drainage and irrigation, characterized in that, Including: Collecting a time-series dataset of rainfall, a time-series dataset of river water level, and a time-series dataset of farmland evapotranspiration; Taking the rainfall as the main variable, performing time-series encoding on the time-series dataset of the rainfall to obtain an irrigation main variable time-series implicit encoding feature; Taking the river water level and the farmland evapotranspiration as covariables, performing time-series encoding on the time-series dataset of the river water level and the time-series dataset of the farmland evapotranspiration respectively to obtain a first irrigation covariable time-series implicit encoding feature and a second irrigation covariable time-series implicit encoding feature; Performing feature alignment and fusion processing based on semantic information field modulation on the irrigation main variable time-series implicit encoding feature, the first irrigation covariable time-series implicit encoding feature, and the second irrigation covariable time-series implicit encoding feature to obtain an irrigation main-covariable time-series fine-grained alignment and fusion feature, including: fusing the first irrigation covariable time-series implicit encoding feature and the second irrigation covariable time-series implicit encoding feature to obtain an irrigation covariable time-series fusion feature; performing semantic information field calculation on the irrigation main variable time-series implicit encoding feature and the irrigation covariable time-series fusion feature to obtain a fine-grained semantic information field between the irrigation main-covariable time-series encoding features; based on the fine-grained semantic information field between the irrigation main-covariable time-series encoding features, performing feature fusion with fine-grained alignment based on field mapping on the irrigation main variable time-series implicit encoding feature and the irrigation covariable time-series fusion feature to obtain the irrigation main-covariable time-series fine-grained alignment and fusion feature; Performing dynamic analysis based on the irrigation main-covariable time-series fine-grained alignment and fusion feature to determine whether the irrigation frequency should be maintained, reduced, or increased.

2. The dynamic irrigation regulation method for water diversion, drainage and conveyance according to claim 1, wherein, Taking the rainfall as the main variable, performing time-series encoding on the time-series dataset of the rainfall to obtain an irrigation main variable time-series implicit encoding feature, including: taking the rainfall as the main variable, inputting the time-series dataset of the rainfall into a sequence encoder based on an LSTM-RNN hybrid model to obtain an irrigation main variable time-series implicit encoding vector as the irrigation main variable time-series implicit encoding feature.

3. The dynamic irrigation regulation method for water diversion and drainage according to claim 2, characterized in that, Taking the river water level and the farmland evapotranspiration as covariables, performing time-series encoding on the time-series dataset of the river water level and the time-series data of the farmland evapotranspiration respectively to obtain a first irrigation covariable time-series implicit encoding feature and a second irrigation covariable time-series implicit encoding feature, including: taking the river water level and the farmland evapotranspiration as covariables, inputting the time-series dataset of the river water level and the time-series dataset of the farmland evapotranspiration into the sequence encoder based on the LSTM-RNN hybrid model to obtain a first irrigation covariable time-series implicit encoding vector as the first irrigation covariable time-series implicit encoding feature and a second irrigation covariable time-series implicit encoding vector as the second irrigation covariable time-series implicit encoding feature.

4. The dynamic irrigation regulation method for water diversion, drainage and irrigation according to claim 3, characterized in that Fusing the first irrigation covariate time series implicit coding feature and the second irrigation covariate time series implicit coding feature to obtain an irrigation covariate time series fusion feature, including: concatenating the first irrigation covariate time series implicit coding vector and the second irrigation covariate time series implicit coding vector to obtain an irrigation covariate time series concatenated implicit coding vector as the irrigation covariate time series fusion feature.

5. The dynamic irrigation regulation method for water diversion and drainage according to claim 4, characterized in that, Calculating a fine-grained semantic information field between the irrigation main variable time series implicit coding feature and the irrigation covariate time series fusion feature to obtain a fine-grained semantic information field between the irrigation main-covariate time series coding features, including: Inputting the irrigation main variable time series implicit coding vector and the irrigation covariate time series concatenated implicit coding vector into a dimension modulation module based on point convolution to obtain a modulated irrigation main variable time series implicit coding vector and a modulated irrigation covariate time series concatenated implicit coding vector, wherein the modulated irrigation main variable time series implicit coding vector and the modulated irrigation covariate time series concatenated implicit coding vector have the same feature dimension; Performing fine-grained association coding and semantic information field coding on the modulated irrigation main variable time series implicit coding vector and the modulated irrigation covariate time series concatenated implicit coding vector to obtain the fine-grained semantic information field between the irrigation main-covariate time series coding features.

6. The dynamic irrigation regulation method for water diversion, drainage and irrigation according to claim 5, characterized in that Performing fine-grained association coding and semantic information field coding on the modulated irrigation main variable time series implicit coding vector and the modulated irrigation covariate time series concatenated implicit coding vector to obtain the fine-grained semantic information field between the irrigation main-covariate time series coding features, including: Performing fine-grained association coding on the modulated irrigation main variable time series implicit coding vector and the modulated irrigation covariate time series concatenated implicit coding vector to obtain an irrigation main-covariate time series fine-grained association feature matrix; Inputting the irrigation main-covariate time series fine-grained association feature matrix into a semantic information field coding network based on a convolution kernel to obtain the fine-grained semantic information field between the irrigation main-covariate time series coding features.

7. The dynamic irrigation regulation method for water diversion and drainage according to claim 6, characterized in that Based on the fine-grained semantic information field between the irrigation main-covariate time series coding features, performing feature fusion with fine-grained alignment based on field mapping on the irrigation main variable time series implicit coding feature and the irrigation covariate time series fusion feature to obtain the irrigation main-covariate time series fine-grained alignment fusion feature, including: Mapping the modulated irrigation main variable time series implicit coding vector and the modulated irrigation covariate time series concatenated implicit coding vector to the fine-grained semantic information field between the irrigation main-covariate time series coding features to obtain a fine-grained aligned irrigation main variable time series implicit coding vector and a fine-grained aligned irrigation covariate time series concatenated implicit coding vector; Calculating the position-wise weighted sum between the fine-grained aligned irrigation main variable time series implicit coding vector and the fine-grained aligned irrigation covariate time series concatenated implicit coding vector to obtain an irrigation main-covariate time series fine-grained alignment fusion feature vector as the irrigation main-covariate time series fine-grained alignment fusion feature.

8. The dynamic irrigation regulation method for water diversion, drainage and conveyance according to claim 7, characterized in that, Mapping the modulated irrigation main variable time - series implicit encoding vector and the modulated irrigation co - variable time - series concatenated implicit encoding vector to the fine - grained semantic information field between the irrigation main - co - variable time - series encoding features respectively to obtain a fine - grained aligned irrigation main variable time - series implicit encoding vector and a fine - grained aligned irrigation co - variable time - series concatenated implicit encoding vector, including: multiplying the modulated irrigation main variable time - series implicit encoding vector and the modulated irrigation co - variable time - series concatenated implicit encoding vector by the fine - grained semantic information field matrix between the irrigation main - co - variable time - series encoding features respectively to obtain the fine - grained aligned irrigation main variable time - series implicit encoding vector and the fine - grained aligned irrigation co - variable time - series concatenated implicit encoding vector.

9. The dynamic irrigation regulation method for water diversion and drainage according to claim 8, characterized in that Performing dynamic analysis based on the irrigation main - co - variable time - series fine - grained aligned fusion feature to determine whether the irrigation frequency should be maintained, decreased or increased, including: inputting the irrigation main - co - variable time - series fine - grained aligned fusion feature vector into a dynamic analysis module based on a classifier to obtain an analysis result, where the analysis result is used to indicate whether the irrigation frequency should be maintained, decreased or increased.

10. A dynamic irrigation regulation system for water diversion, drainage and irrigation, characterized in that, Including: A data collection module, configured to collect a time - series data set of rainfall, a time - series data set of river water level, and a time - series data set of farmland evapotranspiration; A main variable time - series encoding module, configured to use the rainfall as the main variable and perform time - series encoding on the time - series data set of rainfall to obtain an irrigation main variable time - series implicit encoding feature; A co - variable time - series encoding module, configured to use the river water level and the farmland evapotranspiration as co - variables, and perform time - series encoding on the time - series data set of the river water level and the time - series data of the farmland evapotranspiration respectively to obtain a first irrigation co - variable time - series implicit encoding feature and a second irrigation co - variable time - series implicit encoding feature; A feature alignment and fusion module, configured to perform feature alignment and fusion processing based on semantic information field modulation on the irrigation main variable time - series implicit encoding feature, the first irrigation co - variable time - series implicit encoding feature, and the second irrigation co - variable time - series implicit encoding feature to obtain an irrigation main - co - variable time - series fine - grained aligned fusion feature; An irrigation frequency determination module, configured to perform dynamic analysis based on the irrigation main - co - variable time - series fine - grained aligned fusion feature to determine whether the irrigation frequency should be maintained, decreased or increased.

Images