Monitoring station network optimization method for underground water landing funnel

Abstract

The invention relates to a monitoring station network optimization method for an underground water landing funnel, and the method comprises the steps: selecting a learning region, obtaining the distribution condition of the monitoring station network of the underground water landing funnel in the learning region, and extracting an evaluation index of a monitoring precision partition grade; based on the evaluation indexes of the monitoring precision partition levels, a random forest algorithm is adopted to carry out weight learning, and an optimization model is constructed; and obtaining evaluation index data of a research area, inputting the evaluation index data into the optimization model, outputting a groundwater landing funnel monitoring station network distribution optimization scheme of the research area, and adjusting the current groundwater landing funnel monitoring station network distribution condition based on the optimization scheme to complete groundwater landing funnel monitoring station network optimization of the research area. According to the invention, efficient optimization and dynamic upgrading of the monitoring network can be realized, and the method is suitable for underground water funnel area monitoring precision zoning, station layout adjustment and automatic decision support scenes.

Description

Technical Field

[0001] The present invention relates to cross - technical fields such as groundwater hydrology, spatial data analysis, mathematical modeling, machine learning, etc., and particularly relates to a method for optimizing the monitoring station network of groundwater depression funnels. Background Technique

[0002] Affected by China's natural geography, climate characteristics and human activities, the current waterlogging and drought disasters in China are serious. Groundwater monitoring is a long - term and basic work to master dynamic elements such as groundwater level (depth), water temperature, water quality, water volume, etc., and study their changing laws. By optimizing the layout of the monitoring station network, the dynamic changes of the groundwater system can be captured more accurately, the monitoring blind areas can be reduced, and the spatial coverage rate and representativeness of data can be improved.

[0003] The work of optimizing the monitoring station network generally establishes an index system for evaluating the monitoring network according to the actual situation of the study area, and determines the index weights through methods such as expert scoring and analytic hierarchy process. It is highly subjective, relies on high - quality spatial data and is difficult to adapt to complex hydrogeological conditions. Therefore, it is necessary to explore and construct a method for optimizing the monitoring station network of groundwater depression funnels by combining transfer learning and random forest models. Summary of the Invention

[0004] The purpose of the present invention is to address the problems of the traditional monitoring network layout relying on subjective experience and poor adaptability, and provide a method for optimizing the monitoring station network of groundwater depression funnels, realizing the efficient optimization and dynamic upgrade of the monitoring network, and being applicable to scenarios such as monitoring accuracy zoning, site layout adjustment and automated decision - making support in groundwater funnel areas.

[0005] To achieve the above - mentioned purpose, the present invention provides the following solutions:

[0006] A method for optimizing the monitoring station network of groundwater depression funnels, comprising:

[0007] Select a learning area, obtain the distribution of the monitoring station network for groundwater depression funnels in the learning area, and extract evaluation indexes of the monitoring accuracy zoning level;

[0008] Based on the evaluation indexes of the monitoring accuracy zoning level, use the random forest algorithm for weight learning to construct an optimization model;

[0009] Obtain the evaluation index data of the study area, input it into the optimization model, output the optimized distribution plan of the monitoring station network for groundwater depression funnels in the study area, and adjust the current distribution of the monitoring station network for groundwater depression funnels based on the optimization plan to complete the optimization of the monitoring station network for groundwater depression funnels in the study area.

[0010] Optionally, selecting the learning area includes:

[0011] Select an area where the similarity of the hydrogeological conditions, climatic conditions, and human activity impacts in the research area is within the preset threshold and the optimization classification of the groundwater drawdown funnel monitoring station network has been completed as the learning area.

[0012] Optionally, obtaining the distribution of the groundwater drawdown funnel monitoring station network in the learning area and extracting the evaluation indicators for the monitoring accuracy zoning level includes:

[0013] Extract the monitoring accuracy zoning level based on the distribution of the groundwater drawdown funnel monitoring station network in the learning area, obtain the evaluation indicators used for the monitoring accuracy zoning level, and construct a characteristic variable data set;

[0014] Assign labels to the monitoring accuracy zoning results of the learning area based on the monitoring accuracy zoning level to construct a label data set.

[0015] Optionally, based on the evaluation indicators of the monitoring accuracy zoning level, use the random forest algorithm for weight learning, and the construction of the optimization model includes:

[0016] Discretize the characteristic variable data set and the label data set to obtain discrete data;

[0017] Substitute the discrete data into the Spyder environment of the Python language, use the random forest algorithm for fitting verification, determine the importance of each evaluation indicator, and adjust the accuracy during training and testing to obtain an optimization model.

[0018] Optionally, after obtaining the optimization model, it further includes:

[0019] If the fitting effect reaches the preset effect, save the optimization model; if the fitting effect does not reach the preset effect, re-select the learning area, extract the evaluation indicators of the monitoring accuracy zoning level again and input them into the random forest algorithm for fitting verification until the fitting effect reaches the preset effect to obtain the final optimization model.

[0020] Optionally, obtaining the evaluation index data of the research area and inputting it into the optimization model, and outputting the optimization scheme for the distribution of the groundwater drawdown funnel monitoring station network in the research area includes:

[0021] Collect the evaluation index data of the research area;

[0022] Perform quota assignment on the evaluation index data and then substitute it into the optimization model to output the zoning situation of the groundwater drawdown funnel monitoring station network in the research area, where the quota assignment is based on the preset classification criteria and corresponding characteristic values of each evaluation index.

[0023] The beneficial effects of the present invention are:

[0024] The monitoring network optimization method for groundwater drawdown funnels proposed by the present invention breaks through the technical limitations of traditional methods that rely on subjective experience and static thresholds by integrating transfer learning and random forest models, significantly improving the dynamic adaptability and reliability of monitoring and early warning. Based on multi-source heterogeneous data such as hydrogeological conditions, climate characteristics, and human activities, the present invention constructs a feature set, uses random forest to automatically mine the weights of key indicators, replaces the traditional expert scoring method, effectively reduces subjective bias, and enhances the adaptability to complex geological environments. At the same time, by combining transfer learning to reuse the optimization results of similar regions, an early warning baseline model for the study area is quickly constructed, overcoming the dependence of newly built models on long-term data accumulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0026] Figure 1 Flowchart of a monitoring network optimization method for groundwater drawdown funnels according to an embodiment of the present invention;

[0027] Figure 2 Schematic diagram of the learning target accuracy partition according to an embodiment of the present invention;

[0028] Figure 3 Graph of the training and test accuracy changes of the random forest according to an embodiment of the present invention, where (a) is the accuracy change on the training set and (b) is the accuracy change on the test set;

[0029] Figure 4 Schematic diagram of the importance distribution of feature factors according to an embodiment of the present invention;

[0030] Figure 5 Schematic diagram of the shallow groundwater exploitation modulus partition in the study area according to an embodiment of the present invention;

[0031] Figure 6 Schematic diagram of the shallow groundwater TDS partition in the study area according to an embodiment of the present invention;

[0032] Figure 7 Schematic diagram of the impact area of shallow groundwater ecological recharge in the study area according to an embodiment of the present invention;

[0033] Figure 8 Schematic diagram of the shallow groundwater funnel area in the study area according to an embodiment of the present invention;

[0034] Figure 9 Schematic diagram of the shallow hydraulic gradient partition in the study area according to an embodiment of the present invention;

[0035] Figure 10 Schematic diagram of the urbanization degree zoning in the study area of the embodiment of the present invention;

[0036] Figure 11 Schematic diagram of the major water diversion project zoning in the study area of the embodiment of the present invention;

[0037] Figure 12 Schematic diagram of the major project zoning in the study area of the embodiment of the present invention;

[0038] Figure 13 Schematic diagram of the monitoring accuracy zoning in the study area of the embodiment of the present invention;

[0039] Figure 14 Schematic diagram of the distribution of existing monitoring wells in the study area of the embodiment of the present invention;

[0040] Figure 15 Schematic diagram of the distribution of optimized monitoring wells in the study area of the embodiment of the present invention. Detailed implementation manners

[0041] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0042] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0043] This embodiment provides an optimization method for the monitoring station network of a groundwater depression funnel, as Figure 1 shown, including:

[0044] Select a learning area, obtain the distribution of the monitoring station network of the groundwater depression funnel in the learning area, and extract the evaluation indicators of the monitoring accuracy zoning level;

[0045] Based on the evaluation indicators of the monitoring accuracy zoning level, use the random forest algorithm for weight learning to construct an optimization model;

[0046] Obtain the evaluation index data of the study area, input it into the optimization model, output the optimized distribution plan of the monitoring station network of the groundwater depression funnel in the study area, and adjust the current distribution of the monitoring station network of the groundwater depression funnel based on the optimization plan to complete the optimization of the monitoring station network of the groundwater depression funnel in the study area.

[0047] Specifically, in this embodiment, by integrating transfer learning and random forest model, the technical limitations of traditional methods that rely on subjective experience and static thresholds are broken through, significantly improving the dynamic adaptability and reliability of monitoring and early warning. Based on multi-source heterogeneous data such as hydrogeological conditions, climate characteristics, and human activities, this embodiment uses random forest to automatically mine the weights of key indicators, replacing the traditional expert scoring method, effectively reducing subjective bias and enhancing the adaptability to complex geological environments. At the same time, combined with transfer learning, the optimization results of similar regions are reused to quickly construct an early warning baseline model for the study area, overcoming the dependence of newly built models on long-term data accumulation.

[0048] Further, the selection of the learning area includes:

[0049] Select an area where the similarities of hydrogeological conditions, climate conditions, and human activity impacts with the study area are all within a preset threshold and where the optimization and grading of the groundwater depression cone monitoring station network have been completed as the learning area.

[0050] Specifically, in this embodiment, the selected learning area and the study area are in the same region with similar hydrogeological conditions; the groundwater development and utilization forms are similar to those of the study area; the human activity impacts are similar to those of the study area; and there is support from the latest monitoring network optimization results.

[0051] Further, the evaluation indicators for obtaining the distribution of the groundwater depression cone monitoring station network in the learning area and extracting the monitoring accuracy zoning level include:

[0052] Based on the distribution of the groundwater depression cone monitoring station network in the learning area, extract the monitoring accuracy zoning level, obtain the evaluation indicators used for the monitoring accuracy zoning level, and construct a characteristic variable data set;

[0053] Based on the monitoring accuracy zoning level, assign labels to the monitoring accuracy zoning results of the learning area to construct a label data set.

[0054] Specifically, in this embodiment, according to the research results of the hydrogeological unit near the learning area, several characteristic variables are selected as evaluation indicators; the monitoring accuracy zoning results obtained in the learning area are assigned labels as the label data set, and labels 0, 1, and 2 are given to the high-precision area, medium-precision area, and low-precision area respectively.

[0055] Further, based on the evaluation indicators of the monitoring accuracy zoning level, use the random forest algorithm for weight learning to construct an optimization model, including:

[0056] Discretize the characteristic variable data set and the label data set to obtain discrete data;

[0057] Substitute the discrete data into the Spyder environment of Python language, use the random forest algorithm for fitting verification, determine the importance of each evaluation index, and adjust the accuracy during training and testing to obtain an optimized model.

[0058] After obtaining the optimized model, it further includes:

[0059] If the fitting effect reaches the preset effect, save the optimized model; if the fitting effect does not reach the preset effect, reselect the study area, extract the evaluation indexes of the monitoring accuracy zoning level again and input them into the random forest algorithm for fitting verification until the fitting effect reaches the preset effect to obtain the final optimized model.

[0060] Furthermore, obtaining the evaluation index data of the study area and inputting it into the optimized model, the output of the optimized distribution plan of the groundwater depression funnel monitoring network in the study area includes:

[0061] Collect the evaluation index data of the study area;

[0062] After performing quota assignment on the evaluation index data and substituting it into the optimized model, output the zoning situation of the groundwater depression funnel monitoring network in the study area, where the quota assignment is based on the preset classification criteria and corresponding characteristic values of each evaluation index.

[0063] Specifically, in this embodiment, the evaluation indexes of the study area are integrated, quota assignment is performed on the evaluation indexes according to their respective criteria and then substituted into the optimized model to calculate the monitoring accuracy zoning situation, such as high-precision area, relatively high-precision area, medium-precision area, etc. Optimize the layout of the study area according to the monitoring accuracy zoning, supplement monitoring points in areas that do not meet the requirements, and appropriately reduce the investment in monitoring points in areas that exceed the accuracy requirements.

[0064] The following takes the study area of "Ningbolong groundwater depression funnel" as an example to specifically illustrate a monitoring network optimization method for groundwater depression funnels proposed in this embodiment. The western boundary of the study area is the Taihang Mountains, the northern part is the Hutuo River, and the eastern and southern boundaries are demarcated by the county boundaries affected by the funnel, involving 22 counties and districts in 3 cities (Shijiazhuang City, Hengshui City, Xingtai City), with a total area of 10,900 km 2This area is basically located in the piedmont alluvial-proluvial inclined plain in the middle of the Haihe Plain, with high in the west and low in the east, and the terrain slope is generally between 0.5‰ and 1‰. The average annual precipitation in the region is usually 500-600 mm, and 80% of the precipitation is concentrated in June-September. The interannual variation of precipitation in the area is large, less than 400 mm in dry years and more than 800 mm in wet years. The main rivers in the area are the Hutuo River section at the northern boundary and the subsections of the Fuyang River flowing through the funnel. After 2018, with the implementation of the river and lake water replenishment plan, the river sections in the area gradually received water, forming a certain form of leakage recharge to the shallow groundwater.

[0065] First, obtain the study area. The selection of the study area mainly considers the following aspects:

[0066] 1. Both the study area and the Hutuo River Basin are located in the piedmont area of the Haihe Plain, and the hydrogeological conditions are similar;

[0067] 2. The areas are adjacent, and the forms of groundwater development and utilization are similar;

[0068] 3. There are rivers receiving water replenishment and large-scale water conservancy project facilities in both areas;

[0069] 4. The Hutuo River Basin is supported by the latest optimization results of the monitoring network in 2021.

[0070] After selecting the study area, a characteristic variable data set will be established according to the evaluation indicators used in the study area. According to the research results of nearby hydrogeological units, the selected characteristic variables include: 1. Groundwater exploitable resource modulus zoning; 2. Groundwater TDS zoning; 3. Scope of ecological water replenishment impact zoning; 4. Groundwater funnel area; 5. Hydraulic gradient zoning; 6. Degree of urbanization; 7. Major water transfer project zoning; 8. Major projects.

[0071] Take the accuracy zoning results obtained in the Hutuo River area as the label data set, and assign labels 0, 1, and 2 to the high-precision area, medium-precision area, and low-precision area respectively. The classification effect is shown in Figure 2 , and the monitoring density standards corresponding to each accuracy are shown in Table 1. Take the 8 index sets as the characteristic data set.

[0072] Table 1

[0073]

[0074] Using the discrete data of the characteristic data and label data as samples, substitute them into the Spyder environment of Python language and use RandomForestClassifier for simulation verification. By debugging, the accuracy of the model on the training set and test set both reaches more than 0.96. It can be seen that the model fitting effect is excellent, as shown in Figure 3, where (a) is the accuracy change on the training set and (b) is the accuracy change on the test set, and the model expansion prediction can be utilized.

[0075] In addition, the importance distribution of feature factors is introduced. See Figure 4 , it can be seen that the main feature factors affecting the grading are the funnel area and the situation of ecological water replenishment areas, followed by the hydraulic gradient and the exploitable modulus characteristics. The TDS zoning has a certain influence on the evaluation results, while the distribution of major projects, urbanization characteristics, and major water diversion projects have relatively little influence on the grading results. This is basically consistent with the weight situation determined by the analytic hierarchy process, indicating that the random forest model has well completed the learning of the analytic hierarchy process and played a substitution role. Table 2 presents the weight distribution of feature factors determined by the analytic hierarchy process and the random forest.

[0076] Table 2

[0077]

[0078] After analyzing the weights of the grading indicators of the Hutuo River monitoring network, the established random forest model can be directly applied to the grading work of the monitoring network in typical funnel areas.

[0079] Therefore, it is first necessary to integrate the 8 evaluation indicators in the study area to form a feature dataset. The distribution of each feature factor in the study area is shown in Figures 5 to 12 .

[0080] (1) Groundwater exploitable resource modulus zoning;

[0081] From Figure 5 it can be seen that the exploitation volume of shallow groundwater in the piedmont urban area of the study area is relatively large. In Shijiazhuang City and its nearby suburban counties and the area of Neiqiu - Longyao - Renxian, it is usually 120,000 - 170,000 m 3 / (km 2 ·a); it shows a decreasing trend towards the southeast. The exploitation modulus drops to 100,000 m 3 / (km 2 ·a) along the line of Baixiang - Ningjin - Shenzhou; it drops to 50,000 - 80,000 m 3 / (km 2 ·a) in the Jizhou area on the eastern boundary of the study area. It can be seen that the groundwater exploitable modulus is relatively small in the core area of the Ningbai Long funnel, that is, it is not easy to obtain the same amount of recharge after groundwater exploitation, which causes the groundwater level to drop and thus forms a groundwater funnel.

[0082] (2) Groundwater TDS zoning;

[0083] The TDS value is one of the important metrics for water quality. Generally, aquifers with TDS < 1 g / L are the main water sources to ensure people's production and living. From Figure 6It can be seen that the TDS value in most areas in the northwest of the study area is < 1 g / L. The TDS of shallow groundwater along the line from the southeast of Longyao to the central part of Ningjin, the central part of Xinji, and the northwest of Shenzhou increases to 1 - 2 g / L, showing a trend of developing towards slightly saline water. In the southeast of the study area, the TDS of groundwater increases to more than 3 g / L, and in some local areas, it exceeds 5 g / L, seriously reducing the utilization value of groundwater. From the perspective of exploitable utilization, the shallow groundwater exploitation in the study area is concentrated in the central and western regions. That is to say, compared with the southeastern region, the exploitation and utilization degree of shallow groundwater in the central and western regions is higher.

[0084] (3) Ecological water replenishment influence area;

[0085] The ecological water replenishment influence area mainly refers to the alluvial - proluvial fan and the river channel belt. This area is the specific implementation and development section of the national groundwater recharge strategy. In order to evaluate the influence range and degree of recharge on groundwater recovery, it is necessary to focus on and monitor the groundwater level in this area in detail. For example Figure 7 , the main recharge influence areas in the study area are the alluvial - proluvial fan of the Hutuo River, the Hutuo River channel belt, and the Fuyang River channel belt.

[0086] (4) Groundwater funnel area;

[0087] According to the groundwater level statistical results in 2020, the funnel range enclosed by the 15 m isoline in the study area is as Figure 8 shown. The groundwater funnel is the key object of this study and an important obvious indicator to measure the degree of groundwater over - extraction. Therefore, it is necessary to focus on monitoring the funnel area.

[0088] (5) Hydraulic gradient zoning;

[0089] The hydraulic gradient is an index reflecting the combined action of climate change factors and human activity factors on groundwater level changes. For example Figure 9 , the hydraulic gradient in the study area gradually decreases from the mountain front to the east, dropping from 2 - 5 to about 0.05. At the same time, it can be seen that the hydraulic gradient around the funnel is greater than that inside the funnel, and groundwater moves in the direction of decreasing hydraulic gradient, which reflects the occurrence form of the funnel from the side.

[0090] (6) Urbanization degree;

[0091] Figure 10 The urbanization distribution area of the study area is given. Generally speaking, the exploitation and utilization of groundwater in urban areas is mainly for industrial and domestic water use, while in non - urbanized areas, it is mainly for agricultural exploitation. It can be seen that the urbanization degree in the funnel area is not high.

[0092] (7) Major water diversion project zoning;

[0093] To collect the impact of the water diversion project on the groundwater level changes, it is necessary to conduct key monitoring along the water diversion project. Therefore, this indicator is designed. For example, Figure 11 , major water diversion projects within the study area.

[0094] (8) Major projects;

[0095] This indicator is mainly designed to study the impact of water level changes on major projects. For example, Figure 12 , the major projects within the study area are mainly railways.

[0096] After assigning values to the above 8 indicators according to the standards in Table 3 and substituting them into the model, the zoning situation can be calculated using the model prediction function.

[0097] Table 3

[0098]

[0099] The prediction results of the monitoring accuracy zoning of the Ningbolong funnel area using the random forest model are shown in Figure 13 . It can be seen that the key monitoring area of Ningbolong is mainly composed of three parts: 1. Hutuo River channel belt; 2. Fuyang River channel belt; 3. Funnel distribution area. Secondly, the piedmont part is mainly the secondary monitoring subzone, while the southeastern region is mainly the general monitoring subzone.

[0100] According to the accuracy zoning evaluation results, the study area can be divided into 4 parts: 1. Hutuo River channel belt, located at the northern boundary of the study area, with an area of 240.68 km 2 ; 2. Higher accuracy area in front of the mountain, mainly in Shijiazhuang City and its suburban counties, with an area of 2146.03 km 2 ; 3. High-precision area of the funnel, located in the core area of the Ningbolong funnel, with an area of 5721.69 km 2 ; 4. Medium-precision area, mainly distributed in the Julu-Jizhou area in the southeastern part of the study area, with an area of 2791.6 km 2 . According to the accuracy zoning, optimize the layout of each part of the study area, mainly using average layout. The optimization is reflected in improving the accuracy of the interpolation results. The distribution of the number of wells before and after optimization is shown in Table 4. The distribution of various monitoring wells in the study area before and after optimization is shown in Figure 14 and Figure 15 .

[0101] Table 4

[0102]

[0103] The monitoring density of the Hutuo River channel belt before optimization is 1.25 per 100 km 2 , far less than the lower limit value of 3.75 in the high-precision area. It is recommended to add 8 monitoring wells along the river. After optimization and adjustment, the monitoring density will reach 4.58 per 100 km 2。

[0104] The current layout density in the higher-precision area in front of the mountain is 2.38 per 100 km 2 , slightly less than the layout requirement for the higher-precision area. Therefore, optimization and densification were carried out for Yuanshi County, Luancheng County, and Gaocheng County where the monitoring wells in the area are relatively sparse. After densification, there are a total of 67 monitoring wells, and the monitoring density can reach 3.12 per 100 km 2 。

[0105] The monitoring density in the high-precision area of the funnel is 1.03 per 100 km 2 , far less than the monitoring standard for the high-precision area. Therefore, uniform densification was carried out for the non-channel zone in the area, and key densification was carried out for the channel zone. It is planned to add 171 monitoring wells after optimization, and the monitoring density is increased to 4.02 per 100 km 2 。

[0106] The monitoring density in the medium-precision area in the southeast of the study area is 1.36 per 100 km 2 , and the medium-precision area has met the evaluation requirements, but the distribution is relatively uneven. Therefore, it is recommended to appropriately add 17 wells in the sparse area, and the monitoring density is increased to 1.97 per 100 km 2 。

[0107] The embodiments described above are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for optimizing the monitoring station network of a groundwater drawdown funnel, characterized in that, Including: Select a study area, obtain the distribution of groundwater depression cone monitoring station networks in the study area, and extract evaluation indicators for the monitoring accuracy zoning levels; Based on the evaluation indicators for the monitoring accuracy zoning levels, use the random forest algorithm for weight learning to construct an optimization model; Obtain the evaluation indicator data for the study area, input it into the optimization model, output the optimized scheme for the distribution of groundwater depression cone monitoring station networks in the study area, and adjust the current distribution of groundwater depression cone monitoring station networks based on the optimized scheme to complete the optimization of the groundwater depression cone monitoring station networks in the study area.

2. The method for optimizing the monitoring station network of the groundwater drawdown funnel according to claim 1, wherein, Selecting the study area includes: Select an area where the similarities in hydrogeological conditions, climatic conditions, and human activity impacts with the study area are all within a preset threshold and where the optimization grading of the groundwater depression cone monitoring station network has been completed as the study area.

3. The monitoring station network optimization method for the groundwater depression cone according to claim 1, characterized in that, Obtaining the distribution of groundwater depression cone monitoring station networks in the study area and extracting evaluation indicators for the monitoring accuracy zoning levels includes: Extract the monitoring accuracy zoning levels based on the distribution of groundwater depression cone monitoring station networks in the study area, obtain the evaluation indicators used for the monitoring accuracy zoning levels, and construct a characteristic variable data set; Assign labels to the monitoring accuracy zoning results of the study area based on the monitoring accuracy zoning levels to construct a label data set.

4. The method for optimizing the monitoring station network of the groundwater depression cone according to claim 3, characterized in that Based on the evaluation indicators for the monitoring accuracy zoning levels, using the random forest algorithm for weight learning to construct an optimization model includes: Discretize the characteristic variable data set and the label data set to obtain discrete data; Substitute the discrete data into the Spyder environment of the Python language, use the random forest algorithm for fitting verification, determine the importance of each evaluation indicator, and adjust the accuracy during training and testing to obtain an optimization model.

5. The optimization method of the monitoring station network for the groundwater drawdown funnel according to claim 4, characterized in that, After obtaining the optimization model, it further includes: If the fitting effect reaches the preset effect, save the optimization model; if the fitting effect does not reach the preset effect, reselect the study area, extract the evaluation indicators for the monitoring accuracy zoning levels again and input them into the random forest algorithm for fitting verification until the fitting effect reaches the preset effect to obtain the final optimization model.

6. The method for optimizing the monitoring station network of the groundwater depression cone according to claim 1, characterized in that, Obtaining the evaluation indicator data for the study area and inputting it into the optimization model to output the optimized scheme for the distribution of groundwater depression cone monitoring station networks in the study area includes: Collect the evaluation indicator data for the study area; After performing quota assignment on the evaluation indicator data, substitute it into the optimization model to output the zoning situation of the groundwater depression cone monitoring station networks in the study area, where the quota assignment is based on the preset classification criteria and corresponding characteristic values of each evaluation indicator.

Images