Underground water overdraft area water level early warning method and system

Abstract

The invention provides an underground water overdraft area water level early warning method and system, and relates to the technical field of water quality dynamic monitoring, and the method comprises the steps: building an underground water overdraft area water level early warning model based on the corrected overdraft area range and overdraft degree; inputting the water level change data monitored in real time into the underground water overdraft area water level early warning model, and performing water level early warning to obtain a preliminary early warning result; water level change data of the underground water overdraft area are monitored in real time, related dynamic influence factor data are recorded, and the dynamic influence factor data comprise the rainfall capacity, the evaporation capacity and the man-made mining quantity; calculating a dynamic factor according to the dynamic influence factor data; and correcting the preliminary early warning result through the dynamic factor to obtain a final early warning result. According to the final early warning result, when it is judged that the groundwater level is about to drop to an early warning threshold value, an early warning signal is sent out. According to the invention, the accuracy and timeliness of early warning are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of water quality dynamic monitoring, and in particular to a method and system for early warning of water level in a groundwater over-exploitation zone. Background Art

[0002] Overexploitation of groundwater will not only lead to a drop in groundwater levels, but also cause a series of environmental problems, such as land subsidence and water quality deterioration. Therefore, it is very important to provide timely and accurate water level warnings for groundwater overexploitation areas.

[0003] Traditional methods for warning water levels in groundwater overexploitation areas have many defects. First, traditional methods usually use static data for warning, ignoring the dynamic changes of the groundwater system and the combined effects of multiple influencing factors. This affects the accuracy and timeliness of the warning results, and sometimes cannot meet the needs of modern water resources management.

[0004] Secondly, when determining the scope and degree of over-exploitation, some traditional methods only rely on limited geological and hydrological data, lacking comprehensive data support and scientific verification process. Therefore, this may lead to inaccurate basic data of the early warning model, thereby affecting the reliability of the early warning results. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a method and system for early warning of water level in groundwater over-exploitation areas, thereby improving the accuracy and timeliness of early warning.

[0006] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0007] In a first aspect, a method for early warning of water level in a groundwater overexploitation area is provided, the method comprising:

[0008] Step 1, collect basic data of groundwater overexploitation areas;

[0009] Step 2: Conduct a preliminary assessment of the groundwater overexploitation area to determine the preliminary scope and degree of overexploitation; verify the preliminary scope and degree of overexploitation, and obtain the revised scope and degree of overexploitation by calculating the correlation between each correlation factor and groundwater overexploitation;

[0010] Step 3, based on the revised over-exploitation area range and over-exploitation degree, establish a groundwater over-exploitation area water level early warning model;

[0011] Step 4: input the real-time monitored water level change data into the groundwater over-exploitation area water level early warning model to conduct water level early warning to obtain preliminary early warning results;

[0012] Step 5: Real-time monitoring of water level change data in the groundwater over-exploitation area, and recording of relevant dynamic influencing factor data, including rainfall, evaporation and artificial exploitation; calculating dynamic factors based on the dynamic influencing factor data; and correcting the preliminary warning results through the dynamic factors to obtain the final warning results;

[0013] Step 5: According to the final warning result, when it is predicted that the groundwater level will drop to the warning threshold, a warning signal is issued.

[0014] Furthermore, a preliminary assessment of groundwater overexploitation areas is conducted to determine the initial scope and degree of overexploitation, including:

[0015] Preprocessing the basic data of the groundwater overexploitation area to obtain preprocessed basic data of the groundwater overexploitation area;

[0016] Based on the pre-processed basic data of groundwater over-exploitation areas, the extraction coefficient, i.e. the ratio of actual extraction to exploitable volume, is calculated for each assessment unit;

[0017] Based on historical data, the mining coefficient threshold is determined to judge the degree of over-exploitation;

[0018] The mining coefficient is encoded as the gene of the genetic algorithm using binary coding; an initial population consisting of multiple individuals is randomly generated, and each individual represents a mining coefficient distribution;

[0019] Define a fitness function to evaluate the quality of each individual; select the corresponding individual to enter the next generation according to the fitness function value; randomly select two individuals and exchange some genes to generate new individuals; randomly change the genes in the individuals, repeat the selection, crossover, and mutation operations until the termination condition is met. After the genetic algorithm terminates, the corresponding individual is output as the final solution, which represents a final mining coefficient distribution;

[0020] According to the final mining coefficient distribution and the mining coefficient threshold, the preliminary over-exploitation area and degree of over-exploitation are determined.

[0021] Furthermore, the preliminary over-exploitation area and over-exploitation degree were verified, and the correlation between each correlation factor and groundwater over-exploitation was calculated to obtain the revised over-exploitation area and over-exploitation degree, including:

[0022] The degree of groundwater overexploitation is taken as the reference sequence, i.e., the parent sequence; the absolute difference between each correlation factor sequence and the reference sequence is calculated to form a difference sequence; the maximum and minimum values ​​in the difference sequence are found and recorded as Δ max and Δ min ;

[0023] The correlation coefficient between each correlation factor and groundwater overexploitation is calculated. The correlation coefficient calculation formula is as follows:

[0024]

[0025] Among them, ξ ij (k) is the correlation coefficient between the i-th correlation factor and the j-th overexploitation index at the k-th point; Δ ij (k) is the absolute difference between the i-th correlation factor and the j-th overexploitation index at the k-th point; ρ is the resolution coefficient, which takes a value between (0, 1); ω ij is the weight of the i-th correlation factor on the j-th overexploitation indicator; x i (k) and x j (k) are the values ​​of the i-th correlation factor and the j-th overexploitation index at the k-th point; σ ij is the standard deviation between the ith correlation factor and the jth overexploitation indicator;

[0026] The correlation coefficients of each correlation factor at each time point are averaged to obtain the correlation between each correlation factor and groundwater overdraft;

[0027] According to the magnitude of the correlation, the correlation factors are ranked, and the influence of each correlation factor on groundwater overexploitation is determined to obtain the correlation ranking; according to the correlation ranking, the specific influence of each correlation factor on groundwater overexploitation is analyzed to obtain the result of the correlation analysis;

[0028] According to the results of correlation analysis, the initially assessed over-exploited area and over-exploited degree are revised to obtain the revised over-exploited area and over-exploited degree.

[0029] Furthermore, based on the revised over-exploitation area range and over-exploitation degree, a groundwater level early warning model for over-exploitation areas is established, including:

[0030] Integrate revised data on the extent and degree of overexploitation, and collect data on rainfall, evaporation, and anthropogenic extraction in relation to groundwater level changes;

[0031] Determine the neural network structure, including the number of neurons in the input layer, hidden layer, and output layer, and select the corresponding ReLU function for each layer of the neural network; use the random initialization method to initialize the weights and biases of the neural network;

[0032] The corrected over-exploitation area and over-exploitation degree data as well as rainfall, evaporation and anthropogenic extraction related to groundwater level changes are divided into training set, validation set and test set;

[0033] The mean square error is set as the loss function and Adam optimizer, the neural network is trained using the training set data, and the neural network parameters are adjusted through the back propagation algorithm; the performance of the neural network model is evaluated on the validation set, and the performance of the neural network model is evaluated using the test set data to obtain the evaluation results; the neural network model is adjusted according to the evaluation results to obtain the trained neural network model, wherein the trained neural network model is a water level warning model for groundwater over-exploitation areas.

[0034] Furthermore, the corrected over-exploitation area and over-exploitation degree data, as well as rainfall, evaporation and anthropogenic extraction related to groundwater level changes are divided into training sets, validation sets and test sets, including:

[0035] Compile and amend the data on the extent and degree of overexploitation, as well as rainfall, evaporation and anthropogenic extraction;

[0036] Define an evaluation function to evaluate the performance of the neural network model under different data partitioning schemes;

[0037] Set the parameters of the differential evolution algorithm, including population size NP, crossover factor, scaling factor, and maximum number of iterations; initialize the population, that is, generate NP random data partitioning schemes as initial solutions;

[0038] For each individual in the population, use the training set to train the neural network model and evaluate the performance on the validation set; calculate the evaluation value of each individual according to the evaluation function; enter the iterative process, for each iteration:

[0039] Randomly select three different individuals and generate a new test individual according to the rules of the differential evolution algorithm; evaluate the test individual, that is, use the new training set to train the model and evaluate the performance on the validation set; if the evaluation value of the test individual is better than the evaluation value of the target individual, replace the target individual and update the individuals in the population until the maximum number of iterations is reached;

[0040] According to the final individuals found by the differential evolution algorithm, the final training set, validation set, and test set division scheme is determined.

[0041] Furthermore, the neural network parameters are adjusted through the back propagation algorithm, including:

[0042] Initialize the parameters of the neural network, including weights and biases;

[0043] Input a sample from the training data set and perform forward propagation calculations through the network, that is, calculate the activation values ​​of each layer of neurons according to the current weights and biases until the output of the network is obtained;

[0044] Calculate the mean square error between the network output and the actual label;

[0045] Starting from the output layer, the gradient of the mean square error value to the activation value of each layer of neurons, i.e. the partial derivative, is calculated by the chain rule;

[0046] According to the calculated gradient, the gradient descent algorithm is used to update the weights and biases of each layer of neurons until the stopping condition is met. The validation data set is used to evaluate the performance of the model during training, and the learning rate is adjusted as needed, and the number of network layers is increased or decreased to continuously adjust the parameters of the neural network.

[0047] Furthermore, the basic data of groundwater overexploitation areas include groundwater extraction volume, exploitable volume, water level dynamic change data and data related to geological environmental issues.

[0048] In a second aspect, a groundwater overexploitation area water level early warning system comprises:

[0049] Collection module, used to collect basic data of groundwater overexploitation areas;

[0050] The assessment module is used to make a preliminary assessment of the groundwater overexploitation area, determine the preliminary overexploitation area range and overexploitation degree; verify the preliminary overexploitation area range and overexploitation degree, and obtain the revised overexploitation area range and overexploitation degree by calculating the correlation between each correlation factor and groundwater overexploitation;

[0051] Establish a module for establishing a groundwater overexploitation area water level early warning model based on the revised overexploitation area range and overexploitation degree;

[0052] The early warning module is used to input the real-time monitored water level change data into the groundwater over-exploitation area water level early warning model to carry out water level early warning and obtain preliminary early warning results;

[0053] The calculation module is used to monitor the water level change data of the groundwater over-exploitation area in real time and record the relevant dynamic influencing factor data, including rainfall, evaporation and artificial exploitation; calculate the dynamic factor based on the dynamic influencing factor data; and modify the preliminary warning result through the dynamic factor to obtain the final warning result;

[0054] The judgment module is used to determine, based on the final warning result, when it is predicted that the groundwater level will drop to the warning threshold and issue a warning signal.

[0055] According to a third aspect, a computing device includes:

[0056] one or more processors;

[0057] The storage device is used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method described.

[0058] In a fourth aspect, a computer-readable storage medium stores a program, and when the program is executed by a processor, the method described is implemented.

[0059] The above solution of the present invention includes at least the following beneficial effects:

[0060] By collecting basic information on groundwater overexploitation areas, accurate data support can be provided for subsequent steps. These data include but are not limited to geological structure, hydrogeological conditions, historical mining data, etc., which are crucial for accurately assessing the scope and degree of overexploitation.

[0061] By conducting a preliminary assessment of the groundwater overexploitation area, the scope and extent of overexploitation can be roughly determined. The verification process further improves the accuracy of the assessment by calculating the correlation between various related factors and groundwater overexploitation, ensuring that the early warning model is built on more accurate data.

[0062] By establishing an early warning model based on the corrected over-exploitation area and degree of over-exploitation, the changing trend of groundwater levels can be predicted more accurately. Such a model can comprehensively consider multiple factors, including geological conditions, mining history, etc., thereby improving the accuracy and reliability of the early warning.

[0063] By monitoring the water level change data in real time, we can timely understand the dynamic changes of the groundwater level. By inputting these data into the early warning model, we can quickly obtain preliminary early warning results and provide information support for timely response to possible water level drop problems.

[0064] Real-time monitoring and recording of dynamic influencing factor data such as rainfall, evaporation and human exploitation can provide a more comprehensive understanding of the factors affecting groundwater level changes. By calculating dynamic factors and revising the initial warning results, the accuracy and real-time nature of the warning can be further improved, ensuring that the warning results are closer to the actual situation.

[0065] When it is predicted that the groundwater level will drop to the warning threshold, timely issuing of warning signals can remind relevant departments and personnel to take measures to prevent the groundwater level from further dropping, thereby protecting groundwater resources and reducing environmental problems caused by over-exploitation. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is a flow chart of a method for early warning of water level in groundwater over-exploitation areas provided by an embodiment of the present invention.

[0067] Figure 2 It is a schematic diagram of a groundwater over-exploitation area water level early warning system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0068] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0069] like Figure 1 As shown, an embodiment of the present invention provides a groundwater over-exploitation area water level early warning method, the method comprising the following steps:

[0070] Step 1: Collect basic data on groundwater overexploitation areas, including groundwater extraction volume, exploitable volume, water level dynamic change data, and data related to geological environmental issues;

[0071] Step 2: Conduct a preliminary assessment of the groundwater overexploitation area to determine the preliminary scope and degree of overexploitation; verify the preliminary scope and degree of overexploitation, and obtain the revised scope and degree of overexploitation by calculating the correlation between each correlation factor and groundwater overexploitation;

[0072] Step 3, based on the revised over-exploitation area range and over-exploitation degree, establish a groundwater over-exploitation area water level early warning model;

[0073] Step 4: input the real-time monitored water level change data into the groundwater over-exploitation area water level early warning model to conduct water level early warning to obtain preliminary early warning results;

[0074] Step 5: Real-time monitoring of water level change data in the groundwater over-exploitation area, and recording of relevant dynamic influencing factor data, including rainfall, evaporation and artificial exploitation; calculating dynamic factors based on the dynamic influencing factor data; and correcting the preliminary warning results through the dynamic factors to obtain the final warning results;

[0075] Step 5: According to the final warning result, when it is predicted that the groundwater level will drop to the warning threshold, a warning signal is issued.

[0076] In an embodiment of the present invention, by comprehensively collecting detailed information on groundwater over-exploitation areas, including the amount of extraction, the amount that can be extracted, the data on the dynamic change of water levels, and relevant information on geological environmental issues, these data can help to more accurately understand the current situation and problems of the over-exploitation areas, so as to formulate more effective early warning and management strategies. Preliminary assessment of the scope and degree of over-exploitation helps to quickly identify problem areas. Through the verification process, the preliminary assessment results are corrected by correlation calculation, which improves the accuracy and reliability of the assessment, which helps to ensure the accuracy of subsequent early warning models and reduce the possibility of false alarms and omissions. An early warning model is established based on the corrected scope of over-exploitation areas and the degree of over-exploitation, which can more accurately predict and evaluate the dynamic changes of groundwater levels and help to respond to potential water level drop risks in a timely manner. By monitoring water level changes in real time and inputting data into the early warning model, preliminary early warning results can be quickly obtained. This real-time performance is crucial for timely response and prevention of excessive groundwater exploitation, and helps to protect groundwater resources and the ecological environment. According to dynamic factors such as rainfall, evaporation and human exploitation, the actual situation of groundwater level changes can be more comprehensively reflected. By calculating dynamic factors and correcting the initial warning results, the accuracy and pertinence of the warning are further improved, which helps managers make more informed decisions to cope with changing environmental conditions. When it is predicted that the groundwater level will drop to the warning threshold, the timely issuance of warning signals can remind relevant departments and personnel to take prompt action, which helps prevent excessive decline in groundwater levels, protect groundwater resources, and reduce the environmental problems caused by them. A timely and effective warning system is an important guarantee for water resources management and ecological environmental protection.

[0077] Step 1: Comprehensively and systematically collect basic data on groundwater overexploitation areas. The following is a detailed explanation of this step:

[0078] Groundwater extraction data:

[0079] This data records the historical and current exploitation situation in groundwater over-exploitation areas, including the location of the exploitation wells, exploitation time, and exploitation volume. Understanding the exploitation volume will help analyze the rate of water resource consumption in over-exploitation areas and determine whether the exploitation activities exceed the renewable capacity of groundwater.

[0080] Minable volume data:

[0081] The exploitable volume refers to the maximum amount of water that can be extracted from groundwater without affecting the sustainable use of groundwater and the ecological environment. Collecting this information will help assess whether the current exploitation activities exceed the safety threshold, thereby determining the potential risk of over-exploitation.

[0082] Water level dynamic change data:

[0083] The water level dynamic change data records the temporal and spatial changes of the groundwater level, including seasonal fluctuations and interannual changes.

[0084] Related information on geological environment issues:

[0085] Geological environmental problems such as ground subsidence and water quality deterioration are closely related to groundwater overexploitation. Collecting this information will help us understand the impact of overexploitation on the geological environment.

[0086] In summary, step 1 provides solid data support for subsequent water level warning work by collecting various basic data on groundwater over-exploitation areas. The accuracy and completeness of these data directly affect the reliability of the warning model and the accuracy of the warning results.

[0087] In a preferred embodiment of the present invention, a preliminary assessment is conducted on the groundwater overexploitation area to determine the preliminary scope and degree of overexploitation, including:

[0088] The basic data of groundwater over-exploitation areas are preprocessed to obtain the preprocessed basic data of groundwater over-exploitation areas, specifically including: comprehensive collection of basic data of groundwater over-exploitation areas, including extraction volume, recoverable volume, water level changes, etc.; removal of duplicate, erroneous or incomplete data to ensure the accuracy and consistency of the data; conversion of data into a unified format for subsequent analysis and processing, and interpolation, regression and other methods are used to fill in missing data, and data are standardized to eliminate dimensional differences and facilitate comparison between different indicators.

[0089] Based on the pre-processed basic data of groundwater over-exploitation areas, the exploitation coefficient is calculated for each assessment unit, that is, the ratio of actual exploitation volume to exploitable volume. Specifically, the groundwater over-exploitation area is divided into several assessment units, and each unit is taken as an independent analysis object. For each assessment unit, its actual exploitation volume and exploitable volume are counted, and the exploitation coefficient of each assessment unit is calculated, that is, the ratio of actual exploitation volume to exploitable volume. This ratio reflects the exploitation intensity of the unit.

[0090] Based on historical data, the mining coefficient threshold is determined to judge the degree of over-exploitation, which includes: analyzing historical mining data to understand the impact of different mining coefficients on the groundwater system; based on historical data, setting one or more mining coefficient thresholds to judge the degree of over-exploitation. For example, the mining coefficient can be set to be over-exploited if it is greater than 1, and non-over-exploited if it is less than or equal to 1.

[0091] The mining coefficient is encoded as a gene of a genetic algorithm using a binary coding method; an initial population including multiple individuals is randomly generated, and each individual represents a mining coefficient distribution, specifically including: the mining coefficient is encoded as a gene of a genetic algorithm using a binary coding method, and each gene represents a specific mining coefficient value; an initial population including multiple individuals is randomly generated, and each individual consists of a group of genes (i.e., mining coefficients) and represents a possible mining coefficient distribution.

[0092] Define the fitness function to evaluate the quality of each individual; select the corresponding individual to enter the next generation according to the fitness function value; randomly select two individuals and exchange some genes to generate new individuals; randomly change the genes in the individuals, repeat the selection, crossover, and mutation operations until the termination condition is met. After the genetic algorithm terminates, the corresponding individual is output as the final solution. The final solution represents a final mining coefficient distribution, which includes:

[0093] Determine the evaluation criteria, and based on the evaluation criteria, define a fitness function to calculate the fitness value of each individual. This function should be able to quantify the pros and cons of individuals (i.e., the distribution of mining coefficients); for each individual in the population, use the fitness function to calculate its fitness value; sort all individuals according to their fitness values, and use a selection strategy (such as roulette selection) to select the corresponding individuals to enter the next generation. During the selection process, individuals with higher fitness values ​​should have a greater probability of being selected; according to the selection strategy, select a certain number of individuals from the current population to form a new generation of population; randomly select two individuals from the new generation of population for pairing, and randomly select one or more crossover points, which will determine which genes will be exchanged; at the selected crossover point, exchange the genes of the two individuals to generate Two new individuals, this process simulates the natural cross-breeding of organisms; set a mutation probability, which determines the possibility of each gene mutation; for each individual in the population, randomly select a part of the genes for mutation according to its gene length and mutation probability; randomly change the selected genes, in binary coding, this usually means changing 0 to 1 or 1 to 0, repeat the selection, crossover, and mutation operations until the termination condition is met, and a new population will be generated after each round of operation. After the algorithm terminates, the individual with the highest fitness value is selected from the last generation of the population as the final solution; decode the genes of the final solution (i.e., the mining coefficient of the binary code) into the actual mining coefficient distribution, and output the mining coefficient distribution of the final solution, which represents the optimized groundwater mining strategy. Among them, the calculation formula of the fitness function is:

[0094]

[0095] Where N is the total number of mining coefficients; c iis the exploitation coefficient of the ith assessment unit; is the average value of all mining coefficients; ∈ is a small constant used to avoid division by zero errors; max i (c i ) represents the maximum value of all mining coefficients; T is the preset maximum allowable mining coefficient threshold; α and β are weight coefficients.

[0096] According to the final mining coefficient distribution, combined with the mining coefficient threshold, the preliminary over-exploitation area and over-exploitation degree are determined, specifically including: according to the final mining coefficient distribution output by the genetic algorithm, the mining situation of each assessment unit is analyzed, and combined with the mining coefficient threshold, the assessment unit with the mining coefficient exceeding the threshold is demarcated as an over-exploitation area, and the over-exploitation area is classified according to the size of the mining coefficient, and the over-exploitation degree of each area is quantified. For example, the area with a mining coefficient greater than a certain value can be classified as a severe over-exploitation area, and the area with a mining coefficient less than another value can be classified as a mild over-exploitation area, etc.

[0097] In the embodiment of the present invention, the quality and availability of data are improved by cleaning, integrating and standardizing the original data; the calculation of the mining coefficient, as a key indicator for measuring the intensity of groundwater exploitation, helps to intuitively understand the mining status of each assessment unit and provides a quantitative basis for determining the degree of overexploitation; the mining coefficient threshold set based on historical data provides a scientific standard for defining overexploitation areas and non-overexploitation areas, making the determination of the degree of overexploitation more objective and accurate; through binary coding, the complex mining coefficient problem is simplified to a gene form that can be processed by the genetic algorithm, which is convenient for the efficient operation of the algorithm, and the randomly generated initial population increases the diversity of the search space, which helps the algorithm to find the optimal solution globally; the design of the fitness function can quantitatively evaluate the pros and cons of each individual and provide clear guidance for the selection operation; the selection, crossover and mutation operations simulate the process of natural evolution, and through continuous iterative optimization, the individuals in the population gradually approach the optimal solution, and the application of the genetic algorithm can efficiently and accurately search for the optimal mining coefficient distribution; combined with the final mining coefficient distribution and the mining coefficient threshold, the scope of the overexploitation area can be accurately delineated, and the degree of overexploitation in each area can be quantified.

[0098] In a preferred embodiment of the present invention, the preliminary over-exploitation area and over-exploitation degree are verified, and the correlation between each correlation factor and groundwater over-exploitation is calculated to obtain the revised over-exploitation area and over-exploitation degree, including:

[0099] The degree of groundwater overexploitation is taken as the reference sequence, i.e., the parent sequence; the absolute difference between each correlation factor sequence and the reference sequence is calculated to form a difference sequence; the maximum and minimum values ​​in the difference sequence are found and recorded as Δ max and Δ minSpecifically, it includes: determining the mother sequence, clarifying the data on groundwater overexploitation, which will be the reference sequence, also called the mother sequence. Assuming that there are groundwater overexploitation data at n time points, it can be expressed as a sequence M = {m1, m2, ..., m n}, where m n represents the degree of groundwater overexploitation at the nth time point. Next, determine the data sequence of each factor that may be related to the degree of groundwater overexploitation. Assuming there are p related factors, each factor has data at n time points, then the data of each related factor can be represented as a sequence F j ={f j1 , f j2 , …, f jn}, where j = 1, 2, ... p represents the jth correlation factor, f jn represents the value of the jth correlation factor at the nth time point; for each correlation factor sequence Fj, calculate the absolute difference between it and the parent sequence M to form a new difference sequence D j ={d j1 , d j2 , …, d jn}, where d ji =|f ji -m i |, represents the absolute difference between the jth correlation factor and the degree of groundwater overexploitation at the ith time point.

[0100] In all difference sequences D j Find the maximum value Δ max and the minimum value Δ min , these two values ​​will be used to calculate the correlation coefficient later. Specifically, we can traverse all the elements in the difference sequence and find the maximum value as Δ max , and similarly find the minimum value as Δ min .

[0101] The correlation coefficient between each correlation factor and groundwater overexploitation is calculated. The correlation coefficient calculation formula is as follows:

[0102]

[0103] Among them, ξ ij (k) is the correlation coefficient between the i-th correlation factor and the j-th overexploitation index at the k-th point; Δ ij (k) is the absolute difference between the i-th correlation factor and the j-th overexploitation index at the k-th point; ρ is the resolution coefficient, which takes a value between (0, 1); ω ij is the weight of the i-th correlation factor on the j-th overexploitation indicator; x i (k) and x j(k) are the values ​​of the i-th correlation factor and the j-th overexploitation index at the k-th point; σ ij is the standard deviation between the ith correlation factor and the jth overexploitation indicator;

[0104] The correlation coefficients of each correlation factor at each time point are averaged to obtain the correlation between each correlation factor and groundwater overexploitation, including:

[0105] For each correlation factor, calculate the average of its correlation coefficients at all time points, and calculate the correlation coefficient ξ of each correlation factor at each time point according to the previous step. ij (k), where i represents the index of the correlation factor, j represents the index of the overexploitation indicator, and k represents the index of the time point. The calculation formula for the correlation degree is:

[0106]

[0107] Where n is the total number of time points, R i represents the correlation between the ith correlation factor and groundwater overexploitation; t k is the kth time point, μ is the time center point, and σ is the parameter that controls the width of the weight distribution; s ik is the data standardized value of the i-th associated factor at the k-th time point; log e (s ik +1) is a logarithmic function.

[0108] According to the size of the correlation, the correlation factors are sorted to determine the degree of influence of each correlation factor on groundwater overexploitation to obtain the correlation ranking; according to the correlation ranking, the specific influence of each correlation factor on groundwater overexploitation is analyzed to obtain the results of the correlation analysis, which specifically include: after calculating the correlation of all correlation factors, the correlation factors can be sorted according to the size of the correlation. The sorting can be performed from large to small, so as to more intuitively see which factors have the greatest impact on groundwater overexploitation. The sorted results can be represented as an ordered list, in which each element contains the name of the correlation factor and its corresponding correlation; according to the sorting results of the correlation, the specific influence of each correlation factor on groundwater overexploitation can be analyzed. Factors with high correlation indicate that they have a stronger correlation with groundwater overexploitation and may be the main cause of overexploitation. For each factor with high correlation, the reasons and mechanisms behind it are further analyzed to more accurately understand how they affect groundwater overexploitation.

[0109] According to the results of the correlation analysis, the initially assessed over-exploitation area and over-exploitation degree are revised to obtain the revised over-exploitation area and over-exploitation degree, including: Based on the results of the correlation analysis, the initially assessed over-exploitation area and over-exploitation degree can be revised, especially for those factors that are highly correlated with groundwater over-exploitation, it is necessary to re-examine and adjust the relevant over-exploitation area delineation standards and over-exploitation degree assessment methods. For example, if it is found that agricultural irrigation activities in a certain area are highly correlated with groundwater over-exploitation, then more stringent standards may be required when delineating over-exploitation areas in the area, and corresponding management measures may be considered to reduce the excessive exploitation of groundwater resources by irrigation; the revised over-exploitation area and over-exploitation degree should more scientifically and accurately reflect the actual situation, and provide a stronger basis for subsequent groundwater management and protection work.

[0110] In an embodiment of the present invention, by comprehensively considering the impact of multiple related factors on groundwater overexploitation and using the correlation degree for quantitative analysis, the scope of the overexploitation area and the degree of overexploitation can be more accurately delineated. The method takes into account a variety of related factors that may affect groundwater overexploitation. By calculating the correlation between each factor and groundwater overexploitation, the contribution of each factor to groundwater overexploitation can be more comprehensively understood, thereby formulating more comprehensive management and protection measures. By sorting the related factors, it can be clear which factors have the greatest impact on groundwater overexploitation, so that managers can take targeted measures to reduce overexploitation, optimize resource allocation, and improve management efficiency. The method allows the weights of the related factors to be adjusted according to actual conditions to adapt to the specific conditions of different regions and different time points. This flexibility enables the method to be widely used in groundwater overexploitation assessments under various geological and environmental conditions. By correcting the scope of the overexploitation area and the degree of overexploitation, it helps to formulate a more reasonable groundwater exploitation plan, thereby promoting the sustainable use of groundwater resources.

[0111] In a preferred embodiment of the present invention, based on the corrected over-exploitation area range and over-exploitation degree, a groundwater over-exploitation area water level early warning model is established, including:

[0112] Integrate the corrected over-exploitation area and over-exploitation degree data, and collect rainfall, evaporation, and anthropogenic extraction related to groundwater level changes, including: Integrate the corrected over-exploitation area data with the over-exploitation degree data to ensure the consistency of the two in geographic space and time scales; Clean the integrated data to remove outliers, missing values, or duplicate data to ensure data quality; Collect rainfall data that is closely related to groundwater level changes, which may include historical rainfall records obtained from meteorological stations, satellite remote sensing, or public data sets. Collect evaporation data, which can be obtained through meteorological observation stations or relevant research institutions. Collect anthropogenic extraction data, including extraction records for agricultural, industrial, and residential water use, which may come from water management departments or relevant statistical data sets.

[0113] Determine the neural network structure, including the number of neurons in the input layer, hidden layer, and output layer, and select the corresponding ReLU function for each layer of the neural network; use a random initialization method to initialize the weights and biases of the neural network, specifically including: determining the number of layers of the neural network (including the input layer, hidden layer, and output layer), and determining the number of neurons in each layer, which requires trial and adjustment based on experience to find the best network structure; select a ReLU (Rectified Linear Unit) activation function for each layer of the neural network to increase the nonlinear expression ability of the network; use a random initialization method (such as He initialization, Xavier initialization, etc.) to assign initial values ​​to the weights and biases of the neural network, and ensure that the initialized weights and biases have a suitable distribution and range to avoid gradient vanishing or exploding problems during training.

[0114] The corrected over-exploitation area and over-exploitation degree data as well as rainfall, evaporation and anthropogenic extraction related to groundwater level changes are divided into training set, validation set and test set;

[0115] The mean square error is set as the loss function and the Adam optimizer, the training set data is used to train the neural network, and the neural network parameters are adjusted by the back propagation algorithm; the performance of the neural network model is evaluated on the validation set, and the performance of the neural network model is evaluated using the test set data to obtain the evaluation results; the neural network model is adjusted according to the evaluation results to obtain the trained neural network model, wherein the trained neural network model is a groundwater over-exploitation area water level warning model, specifically including:

[0116] The mean squared error (MSE) is selected as the loss function to measure the difference between the model prediction value and the true value; the Adam optimizer is selected as the parameter update algorithm in the training process. It combines the ideas of AdaGrad and RMSProp and can adaptively adjust the learning rate. The neural network is trained using the training set data, and the prediction value and loss function value are calculated through forward propagation. Then, the weights and biases of the network are updated through the back propagation algorithm. During the training process, the validation set data can be used to evaluate the performance of the model regularly, and the training parameters such as the learning rate and batch size can be adjusted as needed. The performance of the neural network is evaluated on the validation set, and the trend of the loss function value and the prediction accuracy of the model are observed. The trained neural network model is finally evaluated using the test set data to obtain the generalization performance evaluation results of the model. According to the evaluation results obtained, the neural network model is adjusted and optimized, which includes adjusting the network structure (increasing or decreasing the number of layers, the number of neurons, etc.), adjusting the activation function type, modifying the learning rate strategy, etc. The training and evaluation process is repeated until satisfactory model performance is obtained. The final trained neural network model is the groundwater over-exploitation area water level warning model.

[0117] In an embodiment of the present invention, by integrating the corrected over-exploitation area range and over-exploitation degree data, and collecting multi-dimensional information such as rainfall, evaporation and artificial exploitation that are closely related to groundwater level changes, the model can more comprehensively reflect the dynamic changes of the groundwater system, thereby improving the accuracy of early warning. A neural network structure is adopted, and a large amount of actual data is used for training and optimization, so that the model can learn the complex nonlinear relationship between data. This data-driven method provides a scientific basis for groundwater management decisions, which helps to achieve the rational use of resources and environmental protection. Through accurate early warning, the risk area of ​​groundwater over-exploitation can be discovered in time, thereby guiding relevant departments to reasonably allocate and adjust water resource exploitation plans, avoid ecological problems caused by over-exploitation, and ensure the sustainable use of groundwater. The early warning model can quickly respond to abnormal changes in groundwater levels, provide timely information support for emergency management departments, and help improve the ability and efficiency of responding to groundwater crises. Groundwater is an important part of the ecosystem. By establishing a scientific early warning model, it is helpful to protect groundwater resources, maintain ecological balance, and promote the in-depth development of ecological civilization construction. By dividing the model into training sets, validation sets, and test sets, and using mean square error as the loss function and Adam optimizer for training, we can effectively prevent the model from overfitting and enhance its generalization ability in different scenarios. At the same time, adjusting the model according to the evaluation results can further optimize the performance of the model.

[0118] In a preferred embodiment of the present invention, the corrected over-exploitation area range and over-exploitation degree data as well as rainfall, evaporation and artificial exploitation related to groundwater level changes are divided into a training set, a validation set and a test set, including:

[0119] Organize the corrected over-exploitation area range and over-exploitation degree data, as well as rainfall, evaporation and anthropogenic exploitation data, including: Collect the corrected over-exploitation area range and over-exploitation degree data, as well as related rainfall, evaporation and anthropogenic exploitation data, and ensure that these data are consistent in time and space. Clean the data, remove outliers, missing values ​​or duplicate values, standardize or normalize the data to eliminate dimensional differences between different features; convert the data into a format suitable for neural network model input, for example, organize the data into a two-dimensional array or tensor, and save it in an appropriate file format (such as CSV, NumPy array, etc.).

[0120] An evaluation function is defined to evaluate the performance of the neural network model under different data partitioning schemes, where the evaluation function is the mean square error.

[0121] Set the parameters of the differential evolution algorithm, including the population size NP, crossover factor, scaling factor and maximum number of iterations; initialize the population, that is, generate NP random data partitioning schemes as the initial solution, specifically including: set the parameters of the differential evolution algorithm, including the population size (NP, i.e. the number of individuals), crossover factor (used to control the generation method of new individuals), scaling factor (used to control the scaling degree of the differential vector) and maximum number of iterations (the maximum number of running rounds of the algorithm); according to the population size, randomly generate NP data partitioning schemes as the initial solution, each scheme should contain the partition ratio or specific index of the training set, validation set and test set.

[0122] For each individual in the population, the neural network model is trained using the training set and the performance is evaluated on the validation set. The evaluation value of each individual is calculated according to the evaluation function. The iterative process is entered. For each iteration, the following is specifically included:

[0123] Three different individuals are randomly selected, and a new test individual is generated according to the rules of the differential evolution algorithm; the test individuals are evaluated, that is, the model is trained using the new training set, and the performance is evaluated on the validation set; if the evaluation value of the test individual is better than the evaluation value of the target individual, the target individual is replaced, and the individuals in the population are updated until the maximum number of iterations is reached, specifically including: for each individual in the population (i.e., the data partitioning scheme), the neural network model is trained using the training set, and its performance is evaluated on the validation set. The evaluation value of each individual is calculated according to the evaluation function; in each iteration, three different individuals (recorded as X1, X2, X3) are randomly selected, and a new test individual is generated according to the rules of the differential evolution algorithm (such as DE / rand / 1 / bin strategy), which usually involves differential operations and crossover operations on the selected individuals; the generated test individuals are evaluated, that is, the neural network model is trained using the new training set, and its performance is evaluated on the validation set, and if the evaluation value of the test individual is better than the evaluation value of the target individual (i.e., the performance is better), the target individual is replaced with the test individual. After one round of iteration, update the individuals in the population, retain the individuals with better performance, eliminate the individuals with poor performance, and determine whether the maximum number of iterations has been reached. If so, stop the iteration; otherwise, return to step 4 to continue the iteration process.

[0124] According to the final individuals found by the differential evolution algorithm, the final training set, validation set and test set partitioning scheme is determined, specifically including: after the iteration, the individual with the best evaluation value is selected from the population as the final data partitioning scheme; according to the final data partitioning scheme, the original data set is divided into training set, validation set and test set, ensuring that the three sets are consistent in data distribution and independent of each other, and saving the final data partitioning scheme and the corresponding training set, validation set and test set data.

[0125] In an embodiment of the present invention, by carefully dividing the data set, it can be ensured that the training set contains sufficiently diverse data samples so that the neural network model can fully learn the inherent laws of the data. At the same time, the effective division of the validation set and the test set helps to accurately evaluate the generalization ability of the model and avoid overfitting or underfitting problems. As an efficient optimization algorithm, the differential evolution algorithm can automatically find the best data partitioning scheme in the search space. Through continuous iteration and evolution, the algorithm can gradually approach the global optimal solution, thereby significantly improving the performance of the neural network model. In the data partitioning process, various factors related to groundwater level changes (such as rainfall, evaporation, and artificial mining) are taken into account, which enables the trained neural network model to better adapt to complex and changeable actual conditions. Therefore, the model has stronger robustness and predictive ability when facing new data. Traditional data partitioning methods often rely on the experience and subjective judgment of experts, while this process uses an automated algorithm to perform data partitioning, reducing the impact of human factors on model performance, which not only improves work efficiency, but also ensures the objectivity and consistency of data partitioning.

[0126] In a preferred embodiment of the present invention, the neural network parameters are adjusted by a back propagation algorithm, including:

[0127] Initialize the parameters of the neural network, including weights and biases, specifically: randomly initialize the weight matrix for each layer of the neural network (except the output layer). Common initialization methods include random decimal initialization (such as sampling from a normal distribution with a mean of 0 and a variance of 1), Xavier initialization (automatically adjusting the variance according to the number of input and output neurons), or He initialization (initialization method optimized for the ReLU activation function and its variants); initialize the bias to a decimal of 0 or close to 0, because the main function of the bias is to adjust the activation threshold of the neuron, rather than introducing additional feature transformations.

[0128] Input a sample from the training data set and perform forward propagation calculations through the network, that is, according to the current weights and biases, calculate the activation values ​​of each layer of neurons until the output of the network is obtained, specifically including: receiving a training sample as input and passing it to the first hidden layer; for each hidden layer, calculate the activation value of each neuron in the current layer according to the weights and biases of the current layer and the activation value of the previous layer. This is usually done through matrix multiplication and addition operations, and then apply an activation function (such as ReLU, Sigmoid or Tanh) to introduce nonlinearity; the last layer (usually the output layer) is calculated in a similar way to the hidden layer, but no activation function needs to be applied (this depends on the specific task, for example, it may not be required in regression tasks), and the activation value of the output layer is the predicted output of the network.

[0129] Calculate the mean square error between the network output and the actual label;

[0130] Starting from the output layer, the chain rule is used to calculate the gradient of the mean square error value with respect to the activation value of each layer of neurons, that is, the partial derivative, specifically including: calculating the gradient of the output layer activation value with respect to the mean square error, which is usually completed through a derivative operation to obtain the error gradient corresponding to each output neuron; then, starting from the output layer, the gradient of the hidden layer neurons is calculated back layer by layer, which is achieved through the chain rule, that is, the gradient passed down from the previous layer and the weight of the current layer are used to calculate the gradient of the neurons in the current layer; in the back propagation process, the gradient values ​​corresponding to each weight and bias need to be accumulated for subsequent parameter updates.

[0131] According to the calculated gradient, the weights and biases of each layer of neurons are updated using the gradient descent algorithm until the stopping condition is met. The validation data set is used to evaluate the performance of the model during the training process, and the learning rate is adjusted as needed, and the number of network layers is increased or decreased to continuously adjust the parameters of the neural network, including: according to the calculated gradient value and the set learning rate, the weights and biases of each layer of neurons are updated using the gradient descent algorithm; a stopping condition is set, such as reaching the maximum number of iterations, the loss function value converges below a certain threshold, or the performance of the validation set is no longer improved, etc. When the stopping condition is met, the iteration is stopped and the current network parameters are saved. During the training process, the validation data set is used regularly to evaluate the performance of the model. This helps to detect whether the model is overfitting or underfitting, and adjusts the learning rate, the number of network layers and other hyperparameters accordingly.

[0132] In an embodiment of the present invention, the back propagation algorithm calculates the gradient and updates the network parameters so that the prediction error of the model on the training data is gradually reduced. This helps to improve the model's ability to fit the training data, thereby optimizing its generalization performance on unseen data. Through the gradient descent algorithm, the neural network can automatically adjust its weights and biases to adapt to different training samples. This adaptive learning ability enables the neural network to handle complex nonlinear problems and continuously improve its performance during the learning process. The back propagation algorithm is an efficient optimization method that uses the chain rule to transmit error signals layer by layer, avoiding the direct calculation of the gradient of the entire network for all parameters, thereby reducing the computational complexity, which enables the neural network to achieve rapid learning and optimization under limited computing resources. During the training process, the structure of the neural network can be adjusted according to the performance evaluation results of the verification data set, such as increasing or decreasing the number of network layers. This flexibility enables the neural network to be customized for different problems and data sets. Optimization, to further improve model performance. The learning rate is an important parameter in the gradient descent algorithm. It determines the step size of the parameter update. By dynamically adjusting the learning rate according to the performance changes during the training process, the model can converge quickly in the early stage and make more refined adjustments in the later stage to achieve better optimization effects.

[0133] In a preferred embodiment of the present invention, step 4, inputting the real-time monitored water level change data into the groundwater over-exploitation area water level early warning model to perform water level early warning to obtain a preliminary early warning result, including:

[0134] Use sensor networks to collect water level change data in groundwater overexploitation areas in real time; preprocess the collected data, including noise removal, outlier detection and processing, data format conversion, etc., to ensure data quality and consistency. Convert the preprocessed data into a format suitable for neural network model input, for example, organize the data into a two-dimensional array or tensor. Load the trained groundwater overexploitation area water level warning model; input the preprocessed real-time monitoring data into the warning model, which usually involves passing the data to the input layer of the model; after the model receives the input data, it will perform forward propagation calculations. During the forward propagation process, the model will calculate the activation values ​​of neurons layer by layer according to the current weights and biases until it reaches the output layer; at the output layer, the model will output a specific value as a preliminary warning result, which represents the model's prediction of the current water level change trend, and whether there is an overexploitation risk based on the specific value.

[0135] In a preferred embodiment of the present invention, step 5 monitors the water level change data of the groundwater over-exploitation area in real time and records the relevant dynamic influencing factor data, the dynamic influencing factor data includes rainfall, evaporation and artificial exploitation, specifically including:

[0136] The water level in the groundwater over-exploitation area is continuously and automatically monitored by using sensor networks, water level monitoring stations or other monitoring equipment. These equipment can transmit water level data to the data processing center or early warning system regularly (such as every hour, every day) or in real time. The received real-time water level data needs to be preprocessed, including noise removal, outlier detection and processing, etc., to ensure the accuracy and reliability of the data. In addition to the water level change data, some dynamic factors affecting the groundwater level change need to be recorded, including rainfall, evaporation and human exploitation. Rainfall is an important source of groundwater recharge. Changes in rainfall will directly affect the height of the groundwater level. Historical rainfall records can be obtained through meteorological stations, satellite remote sensing or public data sets, and current rainfall can be monitored in real time. Evaporation is an important way of groundwater consumption. Increased evaporation will cause the groundwater level to drop. Evaporation data can be obtained through meteorological observation stations or relevant research institutions. The amount of groundwater exploitation by human activities (such as agricultural irrigation, industrial water use and residential water use) is also an important factor affecting the groundwater level. These data usually come from water management departments or relevant statistical data sets. For the above dynamic influencing factor data, a corresponding monitoring and recording system needs to be established. This can be achieved by installing sensors, collecting public data, or cooperating with other departments. The collected data needs to be preprocessed, including removing outliers, processing missing values, and converting data formats to ensure the quality and consistency of the data. The collected dynamic influencing factor data needs to be stored in a database or file for subsequent analysis and use. At the same time, a data management system needs to be established to ensure the integrity, security, and accessibility of the data. By real-time monitoring of water level change data and recording related dynamic influencing factor data, we can have a more comprehensive understanding of the changing trends and influencing factors of the groundwater level, thereby improving the accuracy of the early warning system. Through timely monitoring and early warning, excessive exploitation of groundwater can be prevented.

[0137] In a preferred embodiment of the present invention, in step 5, the dynamic factor is calculated according to the dynamic influencing factor data, wherein the calculation formula of the dynamic factor D is:

[0138]

[0139] Among them, R represents the current rainfall (unit: mm / day); Historical average rainfall (same unit as current rainfall); σ R represents the standard deviation of rainfall, which is used to measure the volatility of rainfall; E represents the current evaporation (unit: mm / day); represents the historical average evaporation (the unit is the same as the current evaporation); σ E represents the standard deviation of evaporation, which is used to measure the volatility of evaporation; X represents the current amount of artificial mining (unit: m 3 / sky); represents the historical average anthropogenic mining volume (the unit is the same as the current mining volume); σ X It represents the standard deviation of artificial mining volume, which is used to measure the volatility of mining volume; a, b, c, d, e and f are weights and exponential coefficients.

[0140] In a preferred embodiment of the present invention, in step 5, the preliminary warning result is corrected by a dynamic factor to obtain a final warning result, wherein after the dynamic factor is calculated, it can be used to correct the preliminary warning result, and the correction formula can be:

[0141]

[0142] Among them, t represents the current time; T represents the cycle; represents the phase shift; n represents the number of time points used to calculate the moving average; P i represents the warning result at the i-th time point in the past; R, E and X represent the current rainfall, evaporation and artificial exploitation respectively; Respectively represent the historical average rainfall, evaporation, and artificial mining; σ R , σ E , σ X represent the standard deviation of rainfall, evaporation and artificial exploitation respectively; A, B and C represent weight coefficients; F w It represents the final warning result, which is a value corrected by dynamic factors; W represents the preliminary warning result, which is a value indicating the risk of overexploitation.

[0143] In a preferred embodiment of the present invention, step 5, judging when the underground is predicted according to the final warning result, specifically includes:

[0144] Collect and analyze basic data on groundwater overexploitation areas, including:

[0145] Geological and hydrogeological data, including the distribution, thickness, permeability, and storage capacity of groundwater layers, which help understand the natural recharge and discharge conditions of groundwater.

[0146] Historical water level data, collect long-term water level monitoring data in groundwater over-exploitation areas, including interannual changes, seasonal fluctuations, etc. These data are the basis for determining early warning thresholds.

[0147] Extraction data records extraction activities in groundwater overexploitation areas, including extraction volume, location of extraction wells, extraction time, etc. These data help assess the impact of extraction activities on groundwater levels.

[0148] Environmental data: Collect environmental data related to groundwater overexploitation, such as rainfall, evaporation, surface water recharge, etc. These data help to understand the source of groundwater recharge and the way it is consumed.

[0149] Reference standards for determining warning thresholds include:

[0150] The safe water level is determined based on the principle of sustainable utilization of groundwater resources. It is the lowest water level allowed without affecting the sustainable utilization of groundwater and the ecological environment.

[0151] Historical lowest water level: Analyze historical water level data to find the lowest point of the groundwater level as one of the references for the early warning threshold.

[0152] Based on the safe water level and in combination with the specific conditions of the groundwater over-exploitation area, an early warning threshold slightly lower than the safe water level is determined so that an early warning can be issued when the groundwater level approaches a dangerous level. If the historical minimum water level is significantly lower than the safe water level, the early warning threshold can be considered to be set between the historical minimum water level and the safe water level to reflect the actual situation of the groundwater level and the historical impact of exploitation activities.

[0153] The calculated final warning result is compared with the pre-set warning threshold. If the final warning result exceeds the warning threshold, it means that the groundwater level is at risk of dropping to a dangerous level. The level of the warning is determined based on the gap between the final warning result and the warning threshold. For example, multiple warning levels (such as level 1 warning, level 2 warning, level 3 warning, etc.) can be set, and each level corresponds to different warning thresholds and response measures. When it is determined that a warning is necessary, a warning signal is sent to relevant departments, enterprises and the public through appropriate channels (such as text messages, emails, online platforms, alarm systems, etc.) according to the warning level. The warning signal contains information such as the warning level, warning area, warning reason, and recommended response measures, so that the recipient can take appropriate actions in a timely manner.

[0154] like Figure 2 As shown, an embodiment of the present invention further provides a groundwater over-exploitation area water level early warning system, comprising:

[0155] Collection module, used to collect basic data of groundwater overexploitation areas;

[0156] The assessment module is used to make a preliminary assessment of the groundwater overexploitation area, determine the preliminary overexploitation area range and overexploitation degree; verify the preliminary overexploitation area range and overexploitation degree, and obtain the revised overexploitation area range and overexploitation degree by calculating the correlation between each correlation factor and groundwater overexploitation;

[0157] Establish a module for establishing a groundwater overexploitation area water level early warning model based on the revised overexploitation area range and overexploitation degree;

[0158] The early warning module is used to input the real-time monitored water level change data into the groundwater over-exploitation area water level early warning model to carry out water level early warning and obtain preliminary early warning results;

[0159] The calculation module is used to monitor the water level change data of the groundwater over-exploitation area in real time and record the relevant dynamic influencing factor data, including rainfall, evaporation and artificial exploitation; calculate the dynamic factor based on the dynamic influencing factor data; and modify the preliminary warning result through the dynamic factor to obtain the final warning result;

[0160] The judgment module is used to determine, based on the final warning result, when it is predicted that the groundwater level will drop to the warning threshold and issue a warning signal.

[0161] It should be noted that the system is a system corresponding to the above method, and all implementation methods in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0162] The embodiment of the present invention further provides a computing device, comprising: a processor, a memory storing a computer program, wherein when the computer program is executed by the processor, the method described above is executed. All implementations in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0163] The embodiment of the present invention also provides a computer-readable storage medium storing instructions, which, when executed on a computer, enable the computer to execute the method described above. All implementations in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

Claims

1. A groundwater over-exploitation area water level early warning method, characterized in that: The method comprises: Step 1, collect basic data of groundwater overexploitation areas; Step 2: Conduct a preliminary assessment of the groundwater overexploitation area to determine the preliminary scope and degree of overexploitation; verify the preliminary scope and degree of overexploitation, and obtain the revised scope and degree of overexploitation by calculating the correlation between each correlation factor and groundwater overexploitation; Step 3, based on the revised over-exploitation area range and over-exploitation degree, establish a groundwater over-exploitation area water level early warning model; Step 4: input the real-time monitored water level change data into the groundwater over-exploitation area water level early warning model to conduct water level early warning to obtain preliminary early warning results; Step 5: Real-time monitoring of water level change data in the groundwater over-exploitation area, and recording of relevant dynamic influencing factor data, including rainfall, evaporation and artificial exploitation; calculating dynamic factors based on the dynamic influencing factor data; and correcting the preliminary warning results through the dynamic factors to obtain the final warning results; Step 5: According to the final warning result, when it is predicted that the groundwater level will drop to the warning threshold, a warning signal is issued.

2. A groundwater over-exploitation area water level early warning method according to claim 1, characterized in that: Conduct a preliminary assessment of groundwater overexploitation areas and determine the initial scope and extent of overexploitation, including: Preprocessing the basic data of the groundwater overexploitation area to obtain preprocessed basic data of the groundwater overexploitation area; Based on the pre-processed basic data of groundwater over-exploitation areas, the extraction coefficient, i.e. the ratio of actual extraction to exploitable volume, is calculated for each assessment unit; Based on historical data, the mining coefficient threshold is determined to judge the degree of over-exploitation; The mining coefficient is encoded as the gene of the genetic algorithm using binary coding; an initial population consisting of multiple individuals is randomly generated, and each individual represents a mining coefficient distribution; Define a fitness function to evaluate the quality of each individual; select the corresponding individual to enter the next generation according to the fitness function value; randomly select two individuals and exchange some genes to generate new individuals; randomly change the genes in the individuals, repeat the selection, crossover, and mutation operations until the termination condition is met. After the genetic algorithm terminates, the corresponding individual is output as the final solution, which represents a final mining coefficient distribution; According to the final mining coefficient distribution and the mining coefficient threshold, the preliminary over-exploitation area and degree of over-exploitation are determined.

3. A groundwater over-exploitation area water level early warning method according to claim 2, characterized in that: The preliminary over-exploitation area and over-exploitation degree were verified, and the revised over-exploitation area and over-exploitation degree were obtained by calculating the correlation between each correlation factor and groundwater over-exploitation, including: The degree of groundwater overexploitation is taken as the reference sequence, i.e., the parent sequence; the absolute difference between each correlation factor sequence and the reference sequence is calculated to form a difference sequence; the maximum and minimum values ​​in the difference sequence are found and recorded as Δ max and Δ min ; Calculate the correlation coefficient between each correlation factor and groundwater overexploitation; The correlation coefficients of each correlation factor at each time point are averaged to obtain the correlation between each correlation factor and groundwater overdraft; According to the magnitude of the correlation, the correlation factors are ranked, and the influence of each correlation factor on groundwater overexploitation is determined to obtain the correlation ranking; according to the correlation ranking, the specific influence of each correlation factor on groundwater overexploitation is analyzed to obtain the result of the correlation analysis; According to the results of correlation analysis, the initially assessed over-exploited area and over-exploited degree are revised to obtain the revised over-exploited area and over-exploited degree.

4. A groundwater over-exploitation area water level early warning method according to claim 3, characterized in that: Based on the revised over-exploitation area range and over-exploitation degree, a groundwater level early warning model for over-exploitation areas is established, including: Integrate revised data on the extent and degree of overexploitation, and collect data on rainfall, evaporation, and anthropogenic extraction in relation to groundwater level changes; Determine the neural network structure, including the number of neurons in the input layer, hidden layer, and output layer, and select the corresponding ReLU function for each layer of the neural network; use the random initialization method to initialize the weights and biases of the neural network; The corrected over-exploitation area and over-exploitation degree data as well as rainfall, evaporation and anthropogenic extraction related to groundwater level changes are divided into training set, validation set and test set; The mean square error is set as the loss function and Adam optimizer, the neural network is trained using the training set data, and the neural network parameters are adjusted through the back propagation algorithm; the performance of the neural network model is evaluated on the validation set, and the performance of the neural network model is evaluated using the test set data to obtain the evaluation results; the neural network model is adjusted according to the evaluation results to obtain the trained neural network model, wherein the trained neural network model is a water level warning model for groundwater over-exploitation areas.

5. A groundwater over-exploitation area water level early warning method according to claim 4, characterized in that: The corrected over-exploitation area and over-exploitation degree data, as well as rainfall, evaporation, and anthropogenic extraction related to groundwater level changes are divided into training, validation, and test sets, including: Compile and amend the data on the extent and degree of overexploitation, as well as rainfall, evaporation and anthropogenic extraction; Define an evaluation function to evaluate the performance of the neural network model under different data partitioning schemes; Set the parameters of the differential evolution algorithm, including population size NP, crossover factor, scaling factor, and maximum number of iterations; initialize the population, that is, generate NP random data partitioning schemes as initial solutions; For each individual in the population, use the training set to train the neural network model and evaluate the performance on the validation set; calculate the evaluation value of each individual according to the evaluation function; enter the iterative process, for each iteration: Randomly select three different individuals and generate a new test individual according to the rules of the differential evolution algorithm; evaluate the test individual, that is, use the new training set to train the model and evaluate the performance on the validation set; if the evaluation value of the test individual is better than the evaluation value of the target individual, replace the target individual and update the individuals in the population until the maximum number of iterations is reached; According to the final individuals found by the differential evolution algorithm, the final training set, validation set, and test set division scheme is determined.

6. A groundwater over-exploitation area water level early warning method according to claim 5, characterized in that: The neural network parameters are adjusted through the back-propagation algorithm, including: Initialize the parameters of the neural network, including weights and biases; Input a sample from the training data set and perform forward propagation calculations through the network, that is, calculate the activation values ​​of each layer of neurons based on the current weights and biases until the output of the network is obtained; Calculate the mean square error between the network output and the actual label; Starting from the output layer, the gradient of the mean square error value to the activation value of each layer of neurons, i.e. the partial derivative, is calculated by the chain rule; According to the calculated gradient, the gradient descent algorithm is used to update the weights and biases of each layer of neurons until the stopping condition is met. The validation data set is used to evaluate the performance of the model during training, and the learning rate is adjusted as needed, and the number of network layers is increased or decreased to continuously adjust the parameters of the neural network.

7. A groundwater over-exploitation area water level early warning method according to claim 6, characterized in that: The basic data of groundwater overexploitation areas include groundwater extraction volume, exploitable volume, water level dynamic change data and relevant data on geological environment issues.

8. A groundwater over-exploitation area water level early warning system, characterized in that: Applied to the method according to any one of claims 1 to 7, comprising: Collection module, used to collect basic data of groundwater overexploitation areas; The assessment module is used to make a preliminary assessment of the groundwater overexploitation area, determine the preliminary overexploitation area range and overexploitation degree; verify the preliminary overexploitation area range and overexploitation degree, and obtain the revised overexploitation area range and overexploitation degree by calculating the correlation between each correlation factor and groundwater overexploitation; Establish a module for establishing a groundwater overexploitation area water level early warning model based on the revised overexploitation area range and overexploitation degree; The early warning module is used to input the real-time monitored water level change data into the groundwater over-exploitation area water level early warning model to carry out water level early warning and obtain preliminary early warning results; The calculation module is used to monitor the water level change data of the groundwater over-exploitation area in real time and record the relevant dynamic influencing factor data, including rainfall, evaporation and artificial exploitation; calculate the dynamic factor based on the dynamic influencing factor data; and modify the preliminary warning result through the dynamic factor to obtain the final warning result; The judgment module is used to determine, based on the final warning result, when it is predicted that the groundwater level will drop to the warning threshold and issue a warning signal.

9. A computing device, characterized in that include: one or more processors; A storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the method as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program, which, when executed by a processor, implements the method according to any one of claims 1 to 7.

Images