Intelligent prediction system and method for groundwater reserve change based on space-time sequence analysis

Abstract

The invention provides an intelligent groundwater reserve change prediction system and method based on space-time sequence analysis, and relates to the technical field of groundwater reserve prediction, and the system comprises a data collection module which is used for carrying out the data collection and analysis of groundwater reserves of a target region; performing time sequence analysis and space analysis on the underground water reserve change data set; the space-time mapping module is used for performing space-time mapping on the underground water reserve change trend characteristics and the underground water reserve change sensitive factors; and the change prediction module is used for modeling according to the space-time law map, synchronizing the groundwater reserve change data set to the depth prediction model for prediction, and generating groundwater reserve change prediction data. According to the method and the device, the technical problem that the prediction accuracy of the groundwater reserve change is relatively low due to the fact that the time dynamic and spatial distribution characteristics of the groundwater reserve change are difficult to capture at the same time in the prior art can be solved, and the prediction accuracy of the groundwater reserve change is improved by integrating the time trend and the spatial sensitive factors.

Description

Technical Field

[0001] This application relates to the technical field of groundwater storage prediction, and particularly to an intelligent prediction system and method for groundwater storage change based on spatio-temporal sequence analysis. Background Art

[0002] Groundwater storage refers to the total amount of water existing in underground aquifers, which is usually determined by various factors such as precipitation, evaporation, groundwater flow, and pumping. The change of groundwater storage can be divided into short-term fluctuations and long-term trends. Short-term fluctuations are usually caused by precipitation, evaporation, and human activities (such as pumping, irrigation, etc.). Long-term trends may be affected by factors such as climate change, regional hydrogeological characteristics, and land use change. The change of groundwater storage is a typical spatio-temporal coupling problem, which is affected by seasonal changes, groundwater flow laws, and regional differences. Most of the existing prediction methods focus on either the time dimension or the space dimension, and it is difficult to capture the interaction and complex relationship between the two at the same time. On the spatial scale, the change of groundwater storage usually has significant spatial heterogeneity, and the hydrogeological conditions in different regions vary greatly, resulting in difficulty for the model to generalize in multiple regions. On the time scale, the change of groundwater storage is often a long-term process, while many prediction methods tend to short-term prediction and are difficult to capture long-term trends. Modeling solely relying on time series or spatial characteristics cannot fully understand and predict the comprehensive dynamic behavior of the groundwater system, resulting in a decrease in prediction accuracy.

[0003] In summary, there is a technical problem in the prior art that the prediction accuracy of groundwater storage change is relatively low because it is difficult to capture the time dynamics and spatial distribution characteristics of groundwater storage change simultaneously. Summary of the Invention

[0004] The purpose of this application is to provide an intelligent prediction system and method for groundwater storage change based on spatio-temporal sequence analysis, so as to solve the technical problem in the prior art that the prediction accuracy of groundwater storage change is relatively low because it is difficult to capture the time dynamics and spatial distribution characteristics of groundwater storage change simultaneously.

[0005] In view of the above problems, this application provides an intelligent prediction system and method for groundwater storage change based on spatio-temporal sequence analysis.

[0006] In a first aspect, the present application provides an intelligent prediction system for groundwater storage change based on spatio-temporal sequence analysis. Among them, the intelligent prediction system for groundwater storage change based on spatio-temporal sequence analysis includes: a data acquisition module, which is used to collect and analyze data on the groundwater storage in the target area through multiple data sources to obtain a groundwater storage change data set; a time series analysis module, which is used to perform time series analysis based on the groundwater storage change data set to generate groundwater storage change trend characteristics; a spatial analysis module, which is used to perform spatial analysis based on the groundwater storage change data set to generate groundwater storage change sensitive factors; a spatio-temporal mapping module, which is used to perform spatio-temporal mapping on the groundwater storage change trend characteristics and the groundwater storage change sensitive factors to construct a spatio-temporal regularity map; a change prediction module, which is used to build a depth prediction model according to the spatio-temporal regularity map, synchronize the groundwater storage change data set to the depth prediction model for prediction, and generate groundwater storage change prediction data.

[0007] In a second aspect, the present application also provides an intelligent prediction method for groundwater storage change based on spatio-temporal sequence analysis. Among them, the intelligent prediction method for groundwater storage change based on spatio-temporal sequence analysis includes: collecting and analyzing data on the groundwater storage in the target area through multiple data sources to obtain a groundwater storage change data set; performing time series analysis based on the groundwater storage change data set to generate groundwater storage change trend characteristics; performing spatial analysis based on the groundwater storage change data set to generate groundwater storage change sensitive factors; performing spatio-temporal mapping on the groundwater storage change trend characteristics and the groundwater storage change sensitive factors to construct a spatio-temporal regularity map; building a depth prediction model according to the spatio-temporal regularity map, synchronizing the groundwater storage change data set to the depth prediction model for prediction, and generating groundwater storage change prediction data.

[0008] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0009] Through a data acquisition module, the data acquisition module is used to collect and analyze data on the groundwater reserves in the target area through multiple data sources to obtain a dataset of groundwater reserve changes; a time series analysis module, the time series analysis module is used to perform time series analysis based on the dataset of groundwater reserve changes to generate characteristics of the groundwater reserve change trend; a spatial analysis module, the spatial analysis module is used to perform spatial analysis based on the dataset of groundwater reserve changes to generate sensitive factors for groundwater reserve changes; a spatio-temporal mapping module, the spatio-temporal mapping module is used to perform spatio-temporal mapping on the characteristics of the groundwater reserve change trend and the sensitive factors for groundwater reserve changes to construct a spatio-temporal law map; a change prediction module, the change prediction module is used to build a depth prediction model according to the spatio-temporal law map, synchronize the dataset of groundwater reserve changes to the depth prediction model for prediction, and generate prediction data on groundwater reserve changes. That is to say, by collecting multiple data, performing time series analysis and spatial analysis on the groundwater reserve change data, performing spatio-temporal mapping on the time trend characteristics and spatial sensitive factors of groundwater reserve changes, constructing a spatio-temporal law map reflecting the groundwater reserve change law in the target area, and accordingly building a depth prediction model to predict the groundwater reserve changes, the accuracy of predicting groundwater reserve changes is improved.

[0010] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, the following specifically gives the specific implementation manners of the present application. It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become easily understandable through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings described below are only exemplary, and for those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0012] Figure 1 It is a schematic structural diagram of an intelligent prediction system for groundwater reserve changes based on spatio-temporal sequence analysis of the present application;

[0013] Figure 2 It is a schematic flow diagram of an intelligent prediction method for groundwater reserve changes based on spatio-temporal sequence analysis of the present application.

[0014] Explanation of the accompanying drawings: data collection module 11, time series analysis module 12, space analysis module 13, space-time mapping module 14, change prediction module 15. DETAILED DESCRIPTION

[0015] This application solves the technical problem in the prior art that the prediction accuracy of groundwater reserve changes is low due to the difficulty in simultaneously capturing the temporal dynamics and spatial distribution characteristics of groundwater reserve changes by providing an intelligent prediction system and method for groundwater reserve changes based on spatiotemporal sequence analysis. By collecting data from multiple sources, performing temporal and spatial analysis on groundwater reserve change data, and performing spatiotemporal mapping of the temporal trend characteristics of groundwater reserve changes and spatial sensitive factors, a spatiotemporal law map reflecting the law of groundwater reserve changes in the target area is constructed, and a deep prediction model is constructed based on this to predict groundwater reserve changes, thereby improving the accuracy of groundwater reserve change predictions.

[0016] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0017] For example, please refer to the attached Figure 1 The present application provides a groundwater reserve change intelligent prediction system based on spatiotemporal sequence analysis, wherein the groundwater reserve change intelligent prediction system based on spatiotemporal sequence analysis is used to implement the steps of the groundwater reserve change intelligent prediction method based on spatiotemporal sequence analysis, and the groundwater reserve change intelligent prediction system based on spatiotemporal sequence analysis includes:

[0018] The data collection module 11 is used to collect and analyze the groundwater reserves in the target area through multiple data sources to obtain a groundwater reserve change data set.

[0019] Specifically, the groundwater storage in the target area is obtained through a variety of different data collection methods and technical means. Multiple data sources usually include multiple information sources such as remote sensing equipment groups and ground sensing equipment groups, which respectively provide data on different dimensions of groundwater storage changes. The groundwater storage in the target area is collected separately by the remote sensing equipment group and the sensing equipment group to obtain a multi-source remote sensing data set and a ground observation data set. The multi-source remote sensing data set and the ground observation data set from the remote sensing equipment group and the sensing equipment group are integrated and learned. Multiple decision trees are trained through the integrated learning. Each decision tree will learn the relationship between the change in groundwater storage and other factors (such as meteorological data, vegetation cover, soil moisture, etc.). By combining the results of multiple decision trees, the integrated learning can greatly improve the prediction accuracy of the model.

[0020] Set the first resolution of the target data set according to the actual needs. Combine the low-resolution remote sensing data with the ground observation data through the downscaling inversion model to generate the prediction data of the groundwater storage change with a higher spatial resolution, and obtain the groundwater storage change data set of the target area according to the first resolution. The specific process is introduced in detail in the subsequent corresponding refinement steps and will not be elaborated here. By combining the remote sensing data with the ground observation data, the model can more comprehensively capture the spatial and temporal dynamics of the groundwater storage change, making the prediction results more reliable and comprehensive. The integrated learning can integrate the information of multiple data sources, avoid the bias brought by a single data source, and the model can adapt to the data in different regions and different time periods, enhancing the generalization ability of the prediction results.

[0021] A time series analysis module 12, where the time series analysis module 12 is used to perform time series analysis based on the groundwater storage change data set to generate the characteristics of the groundwater storage change trend.

[0022] Specifically, perform a time series analysis on the groundwater storage change dataset to identify features such as patterns, trends, seasonality, and periodicity in the data. According to the temporal changes in groundwater storage, divide the groundwater storage change dataset in chronological order to construct a time series dataset. Use a long short-term memory network to train the time series dataset to capture the long-term and short-term trends in groundwater storage changes, and observe the overall change pattern and short-term fluctuation state of groundwater storage over time. Conduct a periodic analysis on the long-term trend data to identify whether there are periodic changes in the data, such as fixed periodic fluctuations in groundwater storage over multiple years. Conduct a fluctuation analysis on the short-term trend data to calculate the rate of change in groundwater storage, such as the growth rate or reduction rate, and the rate of change of these rates. Add the obtained periodic characteristics and rate characteristics of groundwater storage changes to the groundwater storage change trend characteristics to form a more comprehensive feature set, obtaining the groundwater storage change trend characteristics. Through time series analysis, reveal the change trend of groundwater storage over time, help identify the long-term water level change trend, and provide a more accurate basis for the long-term and short-term prediction of groundwater storage by identifying long-term trends and periodic fluctuations, thereby improving the accuracy of prediction.

[0023] A spatial analysis module 13, where the spatial analysis module 13 is used to perform spatial analysis based on the groundwater storage change dataset to generate groundwater storage change sensitive factors.

[0024] Specifically, for the spatial analysis of the groundwater storage change dataset, first divide the groundwater storage change dataset into multiple equally sized grid cells according to the geographical information of the target area. Each grid cell represents a basic unit for spatial analysis. Traverse each grid cell, and in combination with the groundwater storage change dataset, analyze the relationship between the groundwater storage change in each grid cell and other variables (such as geological structure, land use type, hydrological characteristics, etc.), that is, calculate the correlation coefficients between the groundwater storage change in each grid cell and these variables to obtain multiple correlation coefficients. Use the geographically weighted regression method and combine the above-generated correlation coefficients to perform weighted regression analysis on each grid cell.

[0025] Through geographically weighted regression analysis, the spatial influence weight values of each grid cell are obtained, and these weight values reflect the influence degree of different variables on the change of groundwater storage. The obtained multiple spatial influence weight values are sorted in descending order to generate a spatial influence weight sequence. Based on the spatial influence weight sequence, the target area is identified to generate multiple identified areas. According to the identified areas, spatial weight distribution data is generated, and a sensitivity analysis is performed on the groundwater storage change data set to evaluate the influence degree of different geographical information on the change of groundwater storage, and finally a groundwater storage change sensitive factor is generated. That is to say, through sensitivity analysis, the variables with the greatest influence on the change of groundwater storage are identified, and these variables are the groundwater storage change sensitive factors. Through geographically weighted regression, the accuracy of spatial analysis is improved, making the analysis results more in line with the actual situation.

[0026] A spatio-temporal mapping module 14, and the spatio-temporal mapping module 14 is used to perform spatio-temporal mapping on the groundwater storage change trend characteristics and the groundwater storage change sensitive factors to construct a spatio-temporal regularity map.

[0027] Specifically, a spatio-temporal mapping relationship between the groundwater storage change trend characteristics and the groundwater storage change sensitive factors is constructed, and the groundwater change laws at different positions and time periods are presented in the form of a visualized map. First, the groundwater storage change trend characteristics and the groundwater storage change sensitive factors are aligned in time to ensure that the two match under the same time scale. Then, the groundwater storage change trend characteristics and the groundwater storage change sensitive factors are aligned in space. According to the matching between the spatial area grid corresponding to the groundwater storage change sensitive factor and the geographical location of the groundwater storage change trend characteristics, it is ensured that the groundwater storage data at the same geographical location can match its related sensitive factors (such as meteorological data, land use, etc.). According to the first alignment parameter obtained by time alignment and the second alignment parameter obtained by spatial alignment, the groundwater storage change trend characteristics and the sensitive factors are aligned in space and time to construct a spatio-temporal mapping relationship.

[0028] According to the spatio-temporal mapping relationship, the target area is divided into spatio-temporal grids, and each grid cell contains data of a specific time period and spatial position. Based on the spatio-temporal grid data, dynamic mapping calculations are performed to obtain the dynamic mapping values of each grid cell at different time points, which reflect the spatio-temporal laws of the change of groundwater storage. According to the results of the spatio-temporal mapping model, groundwater storage change maps for different time periods and spatial positions are generated, clearly showing the change trends of groundwater storage in different regions at different time periods. The dynamic mapping values are synchronized to the spatio-temporal regularity map to visually display the groundwater change laws at different positions and time periods. Through the spatio-temporal regularity map, hot spots of groundwater storage change can be accurately identified, especially areas with groundwater resource shortages, providing more accurate groundwater storage predictions, reducing uncertainty, and improving the reliability of predictions.

[0029] A change prediction module 15, which is used to model according to the spatio-temporal law map, construct a depth prediction model, synchronize the groundwater storage change data set to the depth prediction model for prediction, and generate groundwater storage change prediction data.

[0030] Specifically, model according to the spatio-temporal law map to construct a depth prediction model. Perform time analysis on the spatio-temporal law map according to the dynamic mapping value to extract the characteristics of the time series. Traverse each spatial grid unit of the spatio-temporal law map, analyze the dynamic mapping value of each grid unit, and extract spatial characteristics. Normalize the time feature set and the spatial feature set to eliminate the dimensionality influence between different features and ensure that the data is on the same scale. Determine the data ratio information according to the normalization result, construct multiple training data and multiple validation data sets to ensure the representativeness of the data set. Usually, the proportion of training data is higher than that of validation data, which can be 7:3 or 6:4, depending on the influence degree of time and space. Use the gradient descent algorithm to learn multiple training data to generate initial training parameters. Evaluate according to the initial training parameters of multiple validation data to generate a training score. If the training score is less than the expected score, adjust the training data, which may include feature selection, data augmentation, parameter tuning, etc. Repeat the iterative training process until the training score is greater than or equal to the expected score. When the training score meets the expectation, stop the iteration, and the model at this time is the depth prediction model.

[0031] Input the collected groundwater storage change data set into the depth prediction model for training so that the model can learn the change law in the data. Combine the historical groundwater storage data, meteorological factors, etc. with the spatio-temporal law map, and through steps such as preprocessing and standardization, convert the data format into an input format acceptable to the deep learning model. Input the data into the depth prediction model for training, and optimize the model parameters during the training process to ensure that the model can accurately capture the spatio-temporal change law in the data. Through the trained depth prediction model, predict the change of groundwater storage in the future for a period of time. The prediction results include information such as the value and change trend of groundwater storage in the future for a period of time. The prediction results are usually presented in the form of a time series. The groundwater storage change prediction data is the future groundwater storage change data generated by the prediction model and is used to predict the dynamic change of groundwater in the future for a period of time. Through the combination of the spatio-temporal law map and the deep learning model, the change trend of future groundwater storage can be accurately predicted. It not only considers the time series data but also comprehensively considers the spatial characteristics, and can more comprehensively reflect the change of groundwater storage.

[0032] Furthermore, the data acquisition module 11 in the intelligent prediction system for groundwater storage change based on spatio-temporal sequence analysis is further used for:

[0033] A remote sensing data acquisition unit for acquiring data on the groundwater storage in a target area based on a remote sensing device group to obtain a multi-source remote sensing data set; a sensing data acquisition unit for sensing the groundwater storage in the target area based on a sensing device group to obtain a ground observation data set; an integrated learning unit for performing integrated learning on the multi-source remote sensing data set and the ground observation data set to construct a downscaling inversion model; a change analysis unit for setting a first resolution and performing change analysis according to the first resolution through the downscaling inversion model to obtain a groundwater storage change data set, and there is a corresponding relationship between the groundwater storage change data set and the first resolution.

[0034] Specifically, data on the groundwater storage in a target area is acquired through a remote sensing device group (such as satellites, aerial cameras, radars, etc.) to obtain a multi-source remote sensing data set. The multi-source remote sensing data set refers to data obtained from multiple different types of remote sensing devices or sensors, including data with different resolutions, different bands, or different platforms (such as satellite remote sensing, aerial remote sensing, etc.), such as satellite images, thermal infrared images, radar data, etc. A remote sensing device is a device that senses ground objects through electromagnetic waves (such as optics, infrared, microwave, etc.) and acquires data. The remote sensing device group refers to multiple remote sensing devices working together to collect groundwater-related data in a large area, including satellite remote sensing and unmanned aerial vehicle remote sensing. For example, the GRACE (Gravity Recovery and Climate Experiment) satellite indirectly infers changes in groundwater storage by measuring changes in the Earth's gravity field.

[0035] The groundwater storage in the target area is monitored in real time through a sensing device group to obtain a ground observation data set, including air temperature, precipitation, humidity, etc. The sensing device group can be a visual sensor or other sensors combined with meteorological data or hydrogeological data, etc. for acquisition, and is usually installed at different locations within the target area to supplement the deficiencies of remote sensing data.

[0036] Integrate the multi-source data collected by the remote sensing device group with the ground observation data collected by the sensing device group, and train multiple weak learners (such as decision trees) through gradient boosting decision trees to gradually improve the prediction accuracy of the groundwater storage change. Ensemble learning is a machine learning method that improves the prediction performance by combining multiple learners (i.e., multiple models). Essentially, it combines multiple weak models to form a stronger model to improve the accuracy and robustness of the model. Build a downscaling inversion model to convert low-resolution remote sensing data into high-resolution data, and infer finer data through the existing data. It is usually used to improve the coarse-resolution remote sensing data to a higher spatial resolution for more detailed analysis. The downscaling inversion model captures the spatial characteristics of groundwater storage changes through the integrated analysis of remote sensing data and ground observation data.

[0037] Set the first resolution. For example, the first resolution is usually selected as 1 km, which means that the groundwater storage change data output by the model will be carried out on a 1 km spatial scale. This is because in water resources management and groundwater monitoring, a 1 km spatial scale can provide sufficient fineness and ensure the efficiency of data processing. Analyze the changes in groundwater storage in the target area according to the first resolution through the downscaling inversion model. That is, for the entire target area, analyze the changes in groundwater storage according to a 1 km resolution, and output the groundwater storage change data set corresponding to the first resolution, indicating the groundwater storage changes in different geographical locations. During the entire downscaling inversion process, there is a strict correspondence between the generated groundwater storage change data set and the set first resolution, which means that each grid has a specific groundwater storage change value, and these values are closely related to the spatial characteristics of the grid.

[0038] By combining multi-source remote sensing data and ground observation data through ensemble learning, the model can make full use of the information of various data sources, reduce the errors caused by a single data source, and significantly improve the prediction accuracy of groundwater storage changes. By setting the first resolution and conducting change analysis, the model can adapt to the data requirements of different resolutions, provide multi-level groundwater storage predictions from macro to micro, and enhance the adaptability and accuracy of the model at different spatial scales.

[0039] Furthermore, the data acquisition module 11 in the intelligent prediction system for groundwater storage change based on spatio-temporal sequence analysis is also used for:

[0040] A meteorological factor extraction subunit, configured to extract spatial features of a target area based on the multi-source remote sensing dataset and extract meteorological factors of the target area based on the ground observation dataset; a decision tree construction subunit, configured to map and match the spatial features and the meteorological factors to generate a mapping connection network, and construct a gradient boosting decision tree according to the mapping connection network; a cross-validation subunit, configured to perform cross-validation on the gradient boosting decision tree at a coarse resolution based on the validation result, prune the gradient boosting decision tree, and construct the downscaling inversion model.

[0041] Specifically, spatial features of the target area are extracted from the multi-source remote sensing dataset, and topographic features (such as altitude, slope, etc.), vegetation coverage, land use types, etc. are extracted from the remote sensing images. Meteorological factors, such as precipitation, temperature, humidity, evaporation, etc., are extracted from the ground observation dataset. Meteorological factors refer to meteorological elements that affect the change of groundwater storage, such as temperature, precipitation, humidity, evaporation, etc., and these factors have a direct impact on the recharge and consumption of groundwater. The spatial features and the meteorological factors are mapped and matched to generate a mapping connection network.

[0042] Mapping and matching means establishing a mathematical or statistical relationship between spatial features and meteorological factors, such as correlation analysis or regression analysis, associating data from different sources, and representing the relationship between spatial features and meteorological factors in a graph or network structure. Each node represents a feature or factor, and the edges between the nodes represent the correlation between them. For example, the correlation coefficient between spatial features and meteorological factors is calculated through the Pearson correlation coefficient to find the linear or non-linear relationship between them. The Pearson correlation coefficient measures the linear relationship between two variables. Its value ranges from -1 to 1, where 1 indicates a perfect positive correlation, -1 indicates a perfect negative correlation, and 0 indicates no linear relationship. If the correlation coefficient is close to 1 or -1, it indicates a strong linear relationship between the two variables. If the correlation coefficient is close to 0, it indicates no obvious linear relationship between the two variables.

[0043] Based on the mapping connection network, the mapping relationship between the input space features and meteorological factors is used to build a model with gradient boosting decision trees. During the construction process, each decision tree is trained based on the previous one, gradually optimizing the prediction performance of the model. Gradient boosting decision tree is an ensemble learning method that improves prediction accuracy by constructing a series of weak learners (usually decision trees). Each new tree is built on the basis of the previous tree to optimize and reduce the model error. The predictive ability of each tree is weak, but by integrating multiple decision trees, a powerful predictive model is finally formed. By combining multiple models into a strong model, the advantages of each model are fully utilized, thus improving the prediction accuracy. Gradient boosting decision tree is one of the commonly used ensemble learning methods. By constructing gradient boosting decision trees through multiple iterations, the final model can more accurately capture the complex relationship between features and targets.

[0044] Use coarse-resolution data for cross-validation of gradient boosting decision trees. Divide the dataset into several subsets, and each time use a different subset for training and testing to evaluate the generalization ability of the model. During the cross-validation process, adjust the relevant hyperparameters of the decision tree, such as tree depth (determining the maximum depth of the tree), number of leaf nodes (the minimum number of samples in each leaf node of each tree), etc. Tree depth controls the maximum depth of the decision tree. When the tree depth is large, the model may overfit; when the tree depth is small, it may underfit. The number of leaf nodes determines the minimum number of samples required for each leaf node, which is used to control the minimum number of samples required for each leaf node to split, avoiding too many small-sample nodes. Coarse resolution means that the spatial resolution of the model or data is low, indicating that less information is represented by each pixel or data unit within a given spatial range. Cross-validation is a method to verify the performance of the model. By dividing the dataset into multiple subsets and using different subsets for training and testing one by one, the generalization ability of the model can be evaluated.

[0045] According to the results of cross-validation, determine the pruning strategy of the model to remove nodes that contribute less to the model prediction and reduce the model complexity. Pruning includes pre-pruning and post-pruning. Pre-pruning means that during the tree growth process, the expansion of the tree is stopped in advance (such as restricting the maximum depth or the minimum number of samples in each node). Post-pruning means that after the tree growth is completed, unnecessary branches are removed based on the importance of the tree nodes and their contribution to the error. The decision tree can be pruned by setting the number of leaf nodes and the minimum number of node samples to remove redundant nodes. Pruning is an optimization measure during the decision tree training process, used to remove those nodes that contribute less to the model prediction. Pruning can prevent the decision tree from overfitting and improve the generalization ability of the model. For example, assume that after cross-validation of the gradient boosting decision tree, the maximum depth is set to 6 and the number of leaf nodes is set to 15. During the post-pruning stage, it is found that the influence of some nodes is very small (such as the error reduction is very small), and these nodes will be removed, finally obtaining a simplified tree.

[0046] Based on the optimized gradient boosting decision tree model, a downscaling technique is used to convert low-resolution data into higher-resolution data. Through the inversion model, detailed prediction data of the groundwater storage in the target area is obtained. The downscaling inversion model is a method for converting coarse-resolution data into high-resolution data, which can improve the spatial resolution of the data while maintaining the overall trend of the data, facilitating more refined analysis and prediction. Through the optimization and pruning of the gradient boosting decision tree, the model can effectively improve the accuracy of predicting the change of groundwater storage. Cross-validation ensures that the model will not overfit and can better adapt to the data in different regions.

[0047] Furthermore, the time series analysis module 12 in the intelligent prediction system for groundwater storage change based on spatio-temporal sequence analysis is also used for:

[0048] The time series division unit is used to divide the groundwater storage change data set according to the change time series to construct a time series data set; the data capture unit is used to capture the data in the time series data set by using a long short-term memory network to obtain multiple trend data, and the multiple trend data includes long-term trend data and short-term trend data; the period analysis unit is used to perform period analysis according to the long-term trend data to obtain the period characteristics of the groundwater storage change; the fluctuation analysis unit is used to perform fluctuation analysis according to the short-term trend data to obtain the change rate characteristics of the groundwater storage; the feature integration unit is used to add the period characteristics of the groundwater storage change and the change rate characteristics of the groundwater storage to the trend characteristics of the groundwater storage change.

[0049] Specifically, the groundwater storage change data set is divided according to the time sequence to construct a time series data set. That is to say, the groundwater storage change data set is sliced into continuous time periods according to time, and the data within each time period represents the state at a time point. Ensure that the data format of each time window is consistent. Usually, each time point corresponds to a data record, including the groundwater storage value and other relevant features.

[0050] Convert the time series data set into a format suitable for the input of the long short-term memory network, usually a three-dimensional array (number of samples, time steps, number of features). The long short-term memory network (LSTM) is a special recurrent neural network that can capture long-term dependencies in time series data and effectively avoid the problem of gradient disappearance, so it has significant advantages in time series prediction. Use the long short-term memory network to capture the data in the time series data set and capture the long-term dependencies in the time series data.

[0051] During the training process, the LSTM gradually adjusts the weights and biases to minimize the error between the model's predicted values and the actual values. Through multiple iterations of training, the LSTM can accurately capture the long-term and short-term trends in the changes of groundwater storage. After being trained, the LSTM model can output long-term trend data, which describes the long-term change trend of groundwater storage. Usually, the changes in long-term trend data are smooth, reflecting the gradual changes in groundwater storage. Short-term trend data captures the fluctuations within a shorter time span in the data, usually seasonal or periodic. Long-term trend data refers to the change trend within a long time period in a time series, usually used to describe the overall change direction, such as the long-term upward or downward trend of groundwater storage. Short-term trend data refers to the change trend within a short time range in a time series, usually used to describe rapid fluctuations or seasonal changes. For example, the fluctuations in groundwater storage within a quarter or a year.

[0052] Based on the long-term trend data generated by the LSTM, perform periodic analysis to identify the periodic characteristics of the changes in groundwater storage, such as seasonal changes (such as water level fluctuations in spring, summer, autumn, and winter) or change patterns of multi-year cycles (such as long-term water level changes caused by climate change or geological changes). Periodic analysis is the process of analyzing the periodic change characteristics in time series data, aiming to identify the periodic laws in the data, such as seasonal changes or long-term cycle changes. The goal of periodic analysis is to identify the periodic change patterns in groundwater storage in the long-term trend, such as the influence of factors such as seasonal changes, climate cycle changes, and land use changes. Use the long-term trend data obtained by training with the LSTM model as the input for periodic analysis, which reflects the overall change trend of groundwater storage and can extract the change patterns of longer cycles from it. Periodic analysis methods can be analyzed through Fourier transform and wavelet transform. Fourier transform is a commonly used periodic analysis tool. By converting the time series into the frequency domain, it can identify the periodic characteristics of different frequency components; wavelet transform can analyze the signal in both the time-frequency domain, which helps to identify the periodic characteristics in non-linear and non-stationary data.

[0053] Based on the short-term trend data generated by LSTM, perform volatility analysis to identify rapid changes or drastic fluctuations in groundwater storage. For example, the groundwater storage may change rapidly after certain seasons or meteorological events (such as extreme precipitation). Volatility analysis refers to the analysis of the rapidly changing part of the data, aiming to identify drastic changes in the short term, mainly focusing on the short-term change rate of groundwater storage, such as seasonal fluctuations or rapid changes caused by climate events. Volatility analysis can be carried out through methods such as the difference method, standard deviation analysis, and change rate analysis. Among them, the difference method analyzes the volatility by calculating the change amount of time series data, that is, calculating the difference between the data at each moment and the previous moment to obtain the volatility amplitude of the data; the standard deviation can measure the degree of data volatility, and a larger standard deviation indicates a larger volatility, evaluating the severity of water level changes; the change rate analysis evaluates the speed of water level changes by calculating the change rates at different time points. For example, the instantaneous change rate (i.e., the water level change per unit time) is used to quantify the change rate of groundwater storage.

[0054] Integrate the characteristics of the groundwater storage change cycle extracted from the long-term trend data and the characteristics of the groundwater storage change rate extracted from the short-term trend data to form a complete set of groundwater storage change trend characteristics, including the change periodicity of the long-term trend, the rate of short-term fluctuations, and the change pattern of the overall trend. Through the training of the LSTM model, accurately capture the long-term trend and short-term fluctuations of groundwater storage changes. Cycle analysis can effectively identify the periodic laws in groundwater storage changes, and volatility analysis can identify sudden fluctuations or anomalies in groundwater storage changes, providing more comprehensive information for the prediction of groundwater storage changes, and helping to improve the accuracy and reliability of groundwater storage change prediction.

[0055] Furthermore, the spatial analysis module 13 in the intelligent prediction system for groundwater storage change based on spatio-temporal sequence analysis is also used for:

[0056] A grid processing unit for performing grid processing on the groundwater storage change data set according to the geographical information of the target area to generate a plurality of grid cells; a correlation analysis unit for traversing the plurality of grid cells and performing correlation analysis in combination with the groundwater storage change data set to generate a plurality of correlation coefficients; a weighted regression unit for performing geographically weighted regression on the plurality of grid cells according to the plurality of correlation coefficients to generate a plurality of spatial influence weight values; a regional identification unit for sorting the plurality of spatial influence weight values in descending order to generate a spatial influence weight sequence, and identifying the target area based on the spatial influence weight sequence to generate a plurality of identified areas; a sensitivity analysis unit for determining spatial weight distribution data according to the plurality of identified areas, and performing data sensitivity analysis on the groundwater storage change data set based on the spatial weight distribution data to generate the groundwater storage change sensitivity factor.

[0057] Specifically, the groundwater storage change data set is grid-processed according to the geographical information of the target area, such as geological structure (specific location), land use type (such as cultivated land, forest land, city, wetland), hydrological characteristics (such as distribution of rivers, lakes, groundwater reservoirs), etc., to divide the area into a plurality of grid cells, and each grid cell represents a small area of the area. Grid processing refers to dividing a geographical area into a plurality of small grid cells. Each grid cell represents a part of the area, and the groundwater storage change is analyzed grid by grid according to the geographical information data.

[0058] Traverse a plurality of grid cells, and for each grid cell, perform correlation analysis in combination with its corresponding groundwater storage change data set (including groundwater level, precipitation, land use type, etc.). Calculate the correlation coefficients between different variables within each grid cell. For example, the correlation coefficient between groundwater storage change and land use type. Correlation analysis is a statistical analysis method used to study the relationship between two or more variables. The correlation degree between different variables can be quantified through the correlation coefficient. For example, the relationship between groundwater storage change and geological structure, land use type, hydrological characteristics, etc. The closer the correlation coefficient is to 1 or -1, the stronger the relationship between the two variables; approaching 0 indicates that there is almost no linear relationship between the variables.

[0059] Based on the calculated correlation coefficients, perform geographically weighted regression on the grid cells, considering the spatial location of each grid cell, and generating different regression models according to different locations, so as to obtain the spatial influence weight values of each region. Through geographically weighted regression, the spatial regression coefficients corresponding to each unit grid are obtained, which represent the influence of the local characteristics of each region on the change of groundwater storage, and a spatial influence weight value is assigned to each grid cell. The larger this value is, the greater the influence of the region on the change of groundwater storage. Through the spatial influence weight values, evaluate the influence intensity of each grid cell in the target area on the change of groundwater storage.

[0060] Arrange the generated multiple spatial influence weight values in descending order to generate a spatial influence weight sequence. The higher the weight of the grid cell, the greater the influence on the change of groundwater storage. According to the spatial influence weight sequence, identify the target area to generate multiple identified areas. According to the level of the spatial weight value, some areas can be identified as high-influence areas or low-influence areas, the grid cells with greater influence are identified as high-influence areas, and the grid cells with smaller influence are identified as low-influence areas.

[0061] According to the multiple identified areas, determine the spatial weight distribution data of the entire target area, which reflects the spatial influence intensity of each grid cell. According to the spatial weight distribution data, conduct data sensitivity analysis. By perturbing the input data (such as meteorological data, land use data, etc. in the groundwater storage change dataset), observe how the changes in these data affect the prediction results of groundwater storage changes. Usually, local sensitivity analysis or global sensitivity analysis is used to analyze the sensitivity of the data. Local sensitivity analysis observes the change of the prediction result by changing a single input variable. Global sensitivity analysis usually changes multiple input variables simultaneously and simulates and calculates the comprehensive influence of multiple input variables on the change of groundwater storage. According to the results of the data sensitivity analysis, generate a groundwater storage change sensitivity factor, which reflects the influence degree of a certain factor (such as temperature, precipitation, land use type, etc.) on the change of groundwater storage. The larger the sensitivity factor is, the stronger the influence of the factor on the change of groundwater storage.

[0062] Through grid processing and geographically weighted regression, extract the spatial characteristics of different geographical units in the target area, reveal the relationship between geographical factors and the change of groundwater storage, understand the influence degree of different geographical factors on the change of groundwater storage, and identify the factors with greater influence on the change of groundwater storage according to the sensitivity analysis, which helps to understand and capture the spatial characteristics of the change of groundwater storage.

[0063] Furthermore, the spatio-temporal mapping module 14 in the intelligent prediction system for groundwater storage change based on spatio-temporal sequence analysis is also used for:

[0064] A time alignment unit for time-aligning the groundwater storage change trend feature and the groundwater storage change sensitive factor to obtain a first alignment parameter; a spatial alignment unit for spatially aligning the groundwater storage change trend feature and the groundwater storage change sensitive factor to obtain a second alignment parameter; a spatio-temporal alignment unit for spatio-temporally aligning the groundwater storage change trend feature and the groundwater storage change sensitive factor according to the first alignment parameter and the second alignment parameter to construct a spatio-temporal mapping relationship; a grid division unit for dividing a target area according to the spatio-temporal mapping relationship to construct spatio-temporal grid data; a mapping calculation unit for performing dynamic mapping calculation based on the spatio-temporal grid data to obtain a dynamic mapping value and synchronizing the dynamic mapping value to the spatio-temporal law map.

[0065] Specifically, the groundwater storage change trend feature obtained through time series analysis and the groundwater storage change sensitive factor obtained through spatial analysis are aligned in time to obtain a first alignment parameter. Through the matching of the collected timestamps, the groundwater storage change trend feature and the sensitive factor are aligned in time, which involves the time synchronization of data and may require interpolation or resampling to ensure data consistency. For example, assuming that the time span of the groundwater storage change is from 2010 to 2020, and the meteorological data is provided daily, the daily data can be converted into annual data for alignment by taking the average or other interpolation methods. If the time spans of the two are different, weights are assigned to the data in certain time periods to ensure that they have an appropriate proportion in the spatio-temporal model. Time alignment refers to the matching of data with different time scales or time points. For the groundwater storage change trend feature and the sensitive factor, time alignment means aligning these two data sets in chronological order to ensure that the change features at the same time point can correspond.

[0066] The groundwater storage change trend feature obtained through time series analysis and the groundwater storage change sensitive factor obtained through spatial analysis are aligned in space to obtain a second alignment parameter. That is to say, the spatial area grid corresponding to the groundwater storage change sensitive factor is matched with the geographical location of the groundwater storage change trend feature to ensure that the data at each spatial position can correspond. Spatial alignment refers to the matching of data at different spatial positions. For the groundwater storage change trend feature and the sensitive factor, spatial alignment means matching the spatial area (such as a grid) represented by the sensitive factor with the geographical location of the trend feature to ensure that these features and factors correspond at the same geographical location.

[0067] Comprehensively align the characteristics of groundwater storage change trends with sensitive factors in terms of time and space through the first alignment parameter (time alignment) and the second alignment parameter (space alignment), and construct a spatio-temporal mapping relationship. That is, combine the groundwater storage change trends at each time point and spatial unit with their corresponding sensitive factors to construct a comprehensive spatio-temporal dataset. The spatio-temporal mapping relationship is the relationship between the datasets after time alignment and space alignment, which describes the mutual influence between the groundwater storage change trends and their sensitive factors at different times and spaces. Spatio-temporal alignment combines time and space information to simultaneously align the characteristics of groundwater storage change trends and sensitive factors. Through time alignment and space alignment, it is ensured that at a specific time point and specific location, the groundwater storage change trends and their influencing factors are matched, thus obtaining a more accurate spatio-temporal mapping relationship.

[0068] Divide the target area according to the spatio-temporal mapping relationship, divide the target area into multiple spatio-temporal grid cells, and generate spatio-temporal characteristics of groundwater storage change for each cell. Each grid cell contains data for a specific time period and spatial location. Through the spatio-temporal mapping relationship, map the groundwater storage change data and the corresponding sensitive factors (such as meteorological data, land use, etc.) into each grid cell. Each grid cell not only contains location information in space but also combines data in the time dimension (such as the groundwater storage change and related sensitive factors in a certain year or season).

[0069] Conduct dynamic mapping calculations based on the spatio-temporal grid data. Based on the spatio-temporal grid data, perform dynamic prediction and calculation through the spatio-temporal mapping relationship, so as to generate dynamic mapping values of groundwater storage change, indicating the change trends of groundwater storage at different times and spaces and the degree to which they are affected by sensitive factors. The dynamic mapping calculation is deduced through the spatio-temporal grid data and the spatio-temporal mapping relationship. Use the historical data of these grid cells, combined with relevant sensitive factors, to predict and calculate the groundwater storage change. Based on the historical data and sensitive factors, select appropriate models for dynamic mapping calculation, including regression models (such as linear regression, ridge regression), machine learning models (such as random forest, support vector machine), and deep learning models (such as long short-term memory network LSTM). Use the historical spatio-temporal grid data for model training. The training process is to learn the relationship between the input data and the groundwater storage change by minimizing the prediction error. Use the trained model to predict the groundwater storage change at a future moment or a certain time period, and obtain the dynamic mapping values of each grid cell. Assume that a linear regression model is selected, calculate the regression equation through the historical data, and use the sensitive factors as inputs to predict the future groundwater storage change.

[0070] Combine the dynamic mapping values with the spatio-temporal regular map, and finally present the variation laws of groundwater in different time and space segments on the map. Through the spatio-temporal regular map, the distribution laws of groundwater storage changes in time and space can be visually seen, helping to predict the future trends of groundwater storage changes. Through spatio-temporal alignment and spatio-temporal mapping calculations, the variation laws of groundwater storage can be captured more accurately, improving the accuracy of the prediction model.

[0071] Further, the change prediction module 15 in the intelligent prediction system for groundwater storage change based on spatio-temporal sequence analysis is further configured to:

[0072] A time analysis unit is configured to traverse the spatio-temporal regular map to perform time analysis according to the dynamic mapping value to obtain a time feature set; a space analysis unit is configured to traverse the spatio-temporal regular map to perform space analysis according to the dynamic mapping value to obtain a space feature set; a normalization processing unit is configured to perform normalization processing on the time feature set and the space feature set, and determine data ratio information according to the processing result; a gradient descent unit is configured to construct a plurality of training data and a plurality of verification data according to the data ratio information, perform gradient descent based on the plurality of training data, and generate initial training parameters; a model training unit is configured to evaluate the initial training parameters based on the plurality of verification data to generate a training score, and when the training score is less than the expected score, adjust the plurality of training data, and thus perform iteration until the training score is greater than or equal to the expected score to obtain the depth prediction model.

[0073] Specifically, traverse the spatio-temporal regular map, perform time analysis according to the dynamic mapping value, analyze the change trends of groundwater storage in different time periods, and extract a time feature set, such as trends, cycles, seasonality, etc. The time feature set is feature data related to the time dimension extracted from the spatio-temporal regular map. For example, it may contain information such as the trends, periodic fluctuations, and seasonal changes of groundwater storage in different time periods. According to the dynamic mapping value in the spatio-temporal regular map, analyze the spatial dimension, and extract spatial distribution features, such as spatial autocorrelation, spatial heterogeneity, etc. The space feature set is feature data related to the spatial dimension extracted from the spatio-temporal regular map. For example, it may include information such as the change patterns, spatial distributions, and hot spots of groundwater storage in different geographical locations.

[0074] Normalize the time feature set and the spatial feature set to eliminate the scale differences between features, map the values of all features to the range [0, 1], so that the influence of each feature is the same, thus avoiding some features from dominating the model training due to their large numerical values. Determine the data ratio information according to the processing results. The proportion of training data should be higher than that of validation data, which can be 7:3 or 6:4, depending on the influence degrees of time and space. Construct multiple sets of training data and validation data according to the data ratio information. Training data refers to selecting a part (such as 80%) from the normalized dataset as training data for model training. The input data for training the model can enable the model to learn the relationship between features and results through the training data. Validation data refers to selecting the remaining part (such as 20%) as validation data for evaluating the training effect and generalization ability of the model, which is a test dataset for evaluating the model performance during training, helping to determine whether the model is overfitting or underfitting, and adjusting the training process.

[0075] Use the training dataset to train with the gradient descent algorithm to optimize the initial parameters of the depth prediction model. Gradient descent is an optimization algorithm used to minimize the loss function. In deep learning, the loss function measures the gap between the model's predicted values and the actual values. Gradient descent calculates the derivative (gradient) of the loss function with respect to the model parameters (such as the weights and biases of a neural network), and then adjusts the model parameters according to the gradient information, gradually reducing the value of the loss function until it converges to the optimal solution. The training data is divided into input features (such as time features, spatial features, etc.) and target outputs (such as the change value of groundwater storage). These input features and output values will be input into the model as training data. During training, the model will generate predicted values based on the input features and calculate the difference between the predicted values and the actual values. This difference is measured by the loss function. Use the gradient descent algorithm to update the model parameters by calculating the gradient of the loss function with respect to the model parameters. Each time it is updated, the model adjusts the parameters in the negative direction of the gradient, so that the value of the loss function gradually decreases.

[0076] At the initial training, the model parameters are generally randomly initialized. As the training progresses, the model adjusts these parameters through each gradient descent iteration to reduce the prediction error. The initial training parameters are the first batch of relatively reasonable parameters obtained when the model optimizes the training data after multiple iterations. Through multiple iterations of optimization, parameters that can effectively predict the change of groundwater storage are finally obtained. The gradient descent algorithm will continuously adjust the parameters to gradually reduce the loss function. In each iteration, the model parameters will be updated according to the calculated gradient, gradually approaching the optimal solution. The learning rate of gradient descent determines the step size of each update. An overly large learning rate may cause the parameter update to go too far, and an overly small learning rate may lead to a too slow convergence speed.

[0077] Evaluate the initial training parameters using multiple validation data. Apply the initial training parameters to the validation data, use the model to make predictions, and calculate the error between the prediction results and the true values. Calculate the training score based on the prediction error of the model. Common scoring criteria include mean squared error (MSE) and mean absolute error (MAE). The mean squared error can quantify the prediction accuracy of the model by calculating the average of the squared differences between the predicted values and the true values. The smaller the mean squared error, the closer the prediction results of the model are to the true values, and the better the performance of the model.

[0078] Compare the calculated training score with the expected score. If the training score is less than the expected score, it means that the performance of the model is not good enough and needs to be adjusted. The expected score is a performance standard of the model, usually determined in the model design stage and set according to actual needs and business goals. Adjust the training data by adjusting the training dataset or the parameter optimization algorithm to gradually improve the model performance until the training score reaches or exceeds the expected score. Adjust the training data or model parameters according to the current performance of the model. For example, add more sample data, or generate new training data through data augmentation techniques, or adjust the hyperparameters of the model (such as learning rate, number of neural network layers, depth of the tree, etc.) to make the model better fit the training data. Retrain using the adjusted data or parameters, update the model parameters, and evaluate again using the validation data. Through iterative training and evaluation, gradually approach the expected score.

[0079] After multiple iterative adjustments, the training score of the model has reached the expected score. At this time, the model is accurate enough to effectively predict the change in groundwater storage. The model obtained at this time is used as a deep prediction model, which can predict the future change in groundwater storage and be applied to practical problems. Through iterative training and adjustment, the deep prediction model can more accurately predict the change in groundwater storage. The normalization and feature extraction processes enhance the robustness of the model, enabling it to handle complex spatio-temporal data.

[0080] In summary, the intelligent prediction system for groundwater storage change based on spatio-temporal sequence analysis provided by this application has the following technical effects:

[0081] Through a data acquisition module, the data acquisition module is used to collect and analyze data on the groundwater storage in the target area through multiple data sources to obtain a dataset of groundwater storage changes; a time series analysis module, the time series analysis module is used to perform time series analysis based on the dataset of groundwater storage changes to generate characteristics of the groundwater storage change trend; a spatial analysis module, the spatial analysis module is used to perform spatial analysis based on the dataset of groundwater storage changes to generate sensitive factors for groundwater storage changes; a spatio-temporal mapping module, the spatio-temporal mapping module is used to perform spatio-temporal mapping of the characteristics of the groundwater storage change trend and the sensitive factors for groundwater storage changes to construct a spatio-temporal regularity map; a change prediction module, the change prediction module is used to build a depth prediction model according to the spatio-temporal regularity map, synchronize the dataset of groundwater storage changes to the depth prediction model for prediction, and generate predicted data on groundwater storage changes. That is to say, by collecting multiple data, performing time series analysis and spatial analysis on the data of groundwater storage changes, performing spatio-temporal mapping on the time trend characteristics and spatial sensitive factors of groundwater storage changes, constructing a spatio-temporal regularity map reflecting the change law of groundwater storage in the target area, and building a depth prediction model based on this to predict the changes in groundwater storage, the accuracy of predicting the changes in groundwater storage is improved.

[0082] Embodiment 2. Based on the same inventive concept as the intelligent prediction system for groundwater storage changes based on spatio-temporal sequence analysis in the foregoing Embodiment 1, the present application also provides an intelligent prediction method for groundwater storage changes based on spatio-temporal sequence analysis. Please refer to the attached Figure 2 , the intelligent prediction method for groundwater storage changes based on spatio-temporal sequence analysis includes:

[0083] S100: Collect and analyze data on the groundwater storage in the target area through multiple data sources to obtain a dataset of groundwater storage changes; S200: Perform time series analysis based on the dataset of groundwater storage changes to generate characteristics of the groundwater storage change trend; S300: Perform spatial analysis based on the dataset of groundwater storage changes to generate sensitive factors for groundwater storage changes; S400: Perform spatio-temporal mapping of the characteristics of the groundwater storage change trend and the sensitive factors for groundwater storage changes to construct a spatio-temporal regularity map; S500: Build a depth prediction model according to the spatio-temporal regularity map, synchronize the dataset of groundwater storage changes to the depth prediction model for prediction, and generate predicted data on groundwater storage changes.

[0084] Further, the step of collecting and analyzing data on the groundwater storage in the target area through multiple data sources to obtain a dataset of groundwater storage changes includes:

[0085] S110: Collect data on the groundwater storage in the target area based on a remote sensing device group to obtain a multi-source remote sensing data set; S120: Sense the data on the groundwater storage in the target area based on a sensing device group to obtain a ground observation data set; S130: Perform integrated learning on the multi-source remote sensing data set and the ground observation data set to construct a downscaling inversion model; S140: Set a first resolution, and perform change analysis according to the first resolution through the downscaling inversion model to obtain a groundwater storage change data set, and there is a corresponding relationship between the groundwater storage change data set and the first resolution.

[0086] Further, the performing integrated learning on the multi-source remote sensing data set and the ground observation data set to construct a downscaling inversion model includes:

[0087] S131: Extract the spatial features of the target area based on the multi-source remote sensing data set, and extract the meteorological factors of the target area based on the ground observation data set; S132: Map and match the spatial features and the meteorological factors to generate a mapping connection network, and construct a gradient boosting decision tree according to the mapping connection network; S133: Perform cross-validation on the gradient boosting decision tree according to a coarse resolution, prune the gradient boosting decision tree according to the verification result, and construct the downscaling inversion model.

[0088] Further, the performing time series analysis on the groundwater storage change data set to generate groundwater storage change trend features includes:

[0089] S210: Divide the groundwater storage change data set according to the change time series to construct a time series data set; S220: Use a long short-term memory network to capture the data in the time series data set to obtain multiple trend data, and the multiple trend data include long-term trend data and short-term trend data; S230: Perform periodic analysis according to the long-term trend data to obtain groundwater storage change period features; S240: Perform fluctuation analysis according to the short-term trend data to obtain groundwater storage change rate features; S250: Add the groundwater storage change period features and the groundwater storage change rate features to the groundwater storage change trend features.

[0090] Further, the performing spatial analysis on the groundwater storage change data set to generate groundwater storage change sensitive factors includes:

[0091] S310: Perform grid processing on the groundwater storage change dataset according to the geographical information of the target area to generate multiple grid cells; S320: Traverse the multiple grid cells and conduct correlation analysis in combination with the groundwater storage change dataset to generate multiple correlation coefficients; S330: Conduct geographically weighted regression on the multiple grid cells according to the multiple correlation coefficients to generate multiple spatial influence weight values; S340: Arrange the multiple spatial influence weight values in descending order to generate a spatial influence weight sequence, and identify the target area based on the spatial influence weight sequence to generate multiple identified areas; S350: Determine the spatial weight distribution data according to the multiple identified areas, and conduct data sensitivity analysis on the groundwater storage change dataset based on the spatial weight distribution data to generate the groundwater storage change sensitivity factor.

[0092] Further, the step of performing spatio-temporal mapping on the groundwater storage change trend feature and the groundwater storage change sensitivity factor to construct a spatio-temporal regularity map includes:

[0093] S410: Align the groundwater storage change trend feature and the groundwater storage change sensitivity factor in terms of time to obtain a first alignment parameter; S420: Align the groundwater storage change trend feature and the groundwater storage change sensitivity factor in terms of space to obtain a second alignment parameter; S430: Align the groundwater storage change trend feature and the groundwater storage change sensitivity factor spatio-temporally according to the first alignment parameter and the second alignment parameter to construct a spatio-temporal mapping relationship; S440: Divide the target area according to the spatio-temporal mapping relationship to construct spatio-temporal grid data; S450: Conduct dynamic mapping calculation based on the spatio-temporal grid data to obtain a dynamic mapping value, and synchronize the dynamic mapping value to the spatio-temporal regularity map.

[0094] Further, the step of building a depth prediction model according to the spatio-temporal regularity map includes:

[0095] S510: Traverse the spatio-temporal law map to perform time analysis according to the dynamic mapping value, and obtain a time feature set; S520: Traverse the spatio-temporal law map to perform space analysis according to the dynamic mapping value, and obtain a space feature set; S530: Perform normalization processing on the time feature set and the space feature set, and determine data ratio information according to the processing result; S540: Construct multiple training data and multiple verification data according to the data ratio information, perform gradient descent based on the multiple training data, and generate initial training parameters; S550: Evaluate the initial training parameters based on the multiple verification data to generate a training score. When the training score is less than the expected score, adjust the multiple training data, and iterate in this way until the training score is greater than or equal to the expected score, and obtain the deep prediction model.

[0096] In the present specification, each embodiment is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The foregoing Figure 1 The intelligent prediction system for groundwater storage change based on spatio-temporal sequence analysis and specific examples in the first embodiment are equally applicable to the intelligent prediction method for groundwater storage change based on spatio-temporal sequence analysis in this embodiment. Through the foregoing detailed description of the intelligent prediction system for groundwater storage change based on spatio-temporal sequence analysis, those skilled in the art can clearly know the intelligent prediction method for groundwater storage change based on spatio-temporal sequence analysis in this embodiment. Therefore, for the sake of simplicity of the specification, it will not be described in detail here. For the method disclosed in the embodiment, since it corresponds to the system disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the system part.

[0097] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0098] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is also intended to include these changes and variations.

Claims

1. An intelligent prediction system for groundwater reserves changes based on spatiotemporal sequence analysis, characterized in that: include: A data collection module, wherein the data collection module is used to collect and analyze the groundwater reserves in the target area through multiple data sources to obtain a groundwater reserve change data set; A time series analysis module, the time series analysis module is used to perform time series analysis based on the groundwater reserve change data set to generate groundwater reserve change trend characteristics; A spatial analysis module, the spatial analysis module is used to perform spatial analysis based on the groundwater reserve change data set to generate a groundwater reserve change sensitivity factor; A space-time mapping module, which is used to perform space-time mapping on the groundwater reserve change trend characteristics and the groundwater reserve change sensitive factors to construct a space-time regularity map; A change prediction module is used to model according to the spatiotemporal regularity map, construct a depth prediction model, synchronize the groundwater reserve change data set to the depth prediction model for prediction, and generate groundwater reserve change prediction data.

2. The intelligent prediction system for groundwater reserves changes based on spatiotemporal sequence analysis according to claim 1, characterized in that: The data acquisition module comprises: A remote sensing data acquisition unit, used to collect data on groundwater reserves in a target area based on a remote sensing equipment group to obtain a multi-source remote sensing data set; A sensor data acquisition unit, used to perform data sensing on groundwater reserves in a target area based on a sensor device group to obtain a ground observation data set; An integrated learning unit, used for performing integrated learning on the multi-source remote sensing dataset and the ground observation dataset to construct a downscaling inversion model; The change analysis unit is used to set a first resolution, perform change analysis according to the first resolution through the downscaling inversion model, and obtain a groundwater reserve change data set, and there is a corresponding relationship between the groundwater reserve change data set and the first resolution.

3. The intelligent prediction system for groundwater reserves change based on spatiotemporal sequence analysis according to claim 2 is characterized in that: The integrated learning unit comprises: A meteorological factor extraction subunit, configured to extract spatial features of a target area based on the multi-source remote sensing data set, and to extract meteorological factors of the target area based on the ground observation data set; A decision tree construction subunit, used for mapping and matching the spatial features with the meteorological factors, generating a mapping connection network, and constructing a gradient boosting decision tree according to the mapping connection network; The cross-validation subunit is used to perform cross-validation based on the gradient boosting decision tree at a coarse resolution, prune the gradient boosting decision tree according to the validation result, and construct the downscale inversion model.

4. The intelligent prediction system for groundwater reserves change based on spatiotemporal sequence analysis according to claim 1, characterized in that: The timing analysis module comprises: A time series division unit, used to divide the groundwater reserve change data set according to the change time sequence to construct a time series data set; A data capture unit, used to capture the time series data set using a long short-term memory network to obtain a plurality of trend data, wherein the plurality of trend data includes long-term trend data and short-term trend data; A periodic analysis unit, used for performing periodic analysis based on the long-term trend data to obtain periodic characteristics of groundwater reserve changes; A fluctuation analysis unit, used to perform fluctuation analysis based on the short-term trend data to obtain the characteristics of groundwater reserve change rate; The feature integration unit is used to add the groundwater reserve change period feature and the groundwater reserve change rate feature to the groundwater reserve change trend feature.

5. The intelligent prediction system for groundwater reserve changes based on spatiotemporal sequence analysis according to claim 1, characterized in that: The spatial analysis module comprises: A grid processing unit, used for performing grid processing based on the groundwater reserve change data set according to the geographic information of the target area to generate a plurality of grid units; A correlation analysis unit, used for traversing the plurality of grid units and combining the groundwater reserve change data set to perform correlation analysis and generate a plurality of correlation coefficients; A weighted regression unit, used for performing geographical weighted regression on the plurality of grid cells according to the plurality of correlation coefficients to generate a plurality of spatial impact weight values; A region identification unit, configured to generate a spatial influence weight sequence by arranging the multiple spatial influence weight values ​​in descending order, and identify the target region based on the spatial influence weight sequence to generate multiple identified regions; A sensitivity analysis unit is used to determine spatial weight distribution data according to the multiple identified areas, perform data sensitivity analysis on the groundwater reserve change data set based on the spatial weight distribution data, and generate the groundwater reserve change sensitivity factor.

6. The intelligent prediction system for groundwater reserve changes based on spatiotemporal sequence analysis according to claim 1, characterized in that: The space-time mapping module comprises: A time alignment unit, used for time-aligning the groundwater reserve change trend characteristic with the groundwater reserve change sensitivity factor to obtain a first alignment parameter; A spatial alignment unit, used for spatially aligning the groundwater reserve change trend characteristic with the groundwater reserve change sensitivity factor to obtain a second alignment parameter; A spatiotemporal alignment unit, configured to perform spatiotemporal alignment on the groundwater reserve change trend characteristics and the groundwater reserve change sensitivity factor according to the first alignment parameter and the second alignment parameter, and construct a spatiotemporal mapping relationship; A grid division unit, used for dividing the target area according to the time-space mapping relationship to construct time-space grid data; A mapping calculation unit is used to perform dynamic mapping calculation based on the spatiotemporal grid data, obtain dynamic mapping values, and synchronize the dynamic mapping values ​​to the spatiotemporal regularity map.

7. The intelligent prediction system for groundwater reserve changes based on spatiotemporal sequence analysis according to claim 6, characterized in that: The change prediction module is specifically used for: A time analysis unit, used for traversing the spatiotemporal regularity map and performing time analysis according to the dynamic mapping value to obtain a time feature set; A spatial analysis unit, used for traversing the spatiotemporal regularity map and performing spatial analysis according to the dynamic mapping value to obtain a spatial feature set; A normalization processing unit, used to perform normalization processing on the temporal feature set and the spatial feature set, and determine data ratio information according to the processing result; A gradient descent unit, used to construct a plurality of training data and a plurality of verification data according to the data ratio information, and perform gradient descent based on the plurality of training data to generate initial training parameters; A model training unit is used to evaluate the initial training parameters based on the multiple verification data to generate a training score. When the training score is less than the expected score, the multiple training data are adjusted, thereby iterating until the training score is greater than or equal to the expected score, thereby obtaining the depth prediction model.

8. An intelligent prediction method for groundwater reserve changes based on spatiotemporal sequence analysis is characterized by: The method is performed by the intelligent prediction system for groundwater reserve changes based on spatiotemporal sequence analysis according to any one of claims 1 to 7, wherein the intelligent prediction method for groundwater reserve changes based on spatiotemporal sequence analysis comprises: Collect and analyze groundwater reserves in the target area through multiple data sources to obtain a groundwater reserve change dataset; Performing time series analysis based on the groundwater reserve change data set to generate groundwater reserve change trend characteristics; Performing spatial analysis based on the groundwater reserve change data set to generate a groundwater reserve change sensitivity factor; Performing spatiotemporal mapping of the groundwater reserve change trend characteristics and the groundwater reserve change sensitive factors to construct a spatiotemporal regularity map; Modeling is performed according to the spatiotemporal regularity map, a depth prediction model is constructed, the groundwater reserve change data set is synchronized to the depth prediction model for prediction, and groundwater reserve change prediction data is generated.

Images