A soil arsenic pollution dynamic evolution prediction method based on time series deep learning

Abstract

The application discloses a soil arsenic pollution dynamic evolution prediction method based on time sequence deep learning, and relates to the technical field of soil pollution. The method comprises information acquisition: obtaining soil information of a region to be predicted to obtain basic soil information. The application establishes a reference based on previous soil arsenic pollution by setting a reference establishment method, and obtains change information by using a control variable method to explore the changes of different arsenic forms under different conditions. The application completes dynamic evolution prediction of soil arsenic in target region information by using the reference directory and the change information as a benchmark for information matching through a pollution evolution method. The application obtains the final pollution condition by considering emergencies according to actual use requirements and reflecting the emergencies in the basic pollution condition through a redundancy adding method after optimization, and the whole purpose is to predict the arsenic condition in the soil at a future time by using historical observation data and external variables, and to provide a basis for risk early warning, management strategy evaluation and monitoring plan optimization.

Description

Technical Field

[0001] This invention relates to the field of soil pollution technology, specifically to a method for predicting the dynamic evolution of soil arsenic pollution based on temporal deep learning. Background Technology

[0002] Arsenic pollution in soil refers to the concentration or mobility of arsenic and its compounds in the soil exceeding the natural background level. It originates from industrial emissions, arsenic-based pesticides used in agriculture, mining, waste leakage, and other pathways, causing arsenic and its organic / inorganic forms to migrate and transform between soil, water, and organisms, threatening plant health, groundwater safety, and human health. It is necessary to reduce exposure risks through source control, monitoring and assessment, and remediation.

[0003] A method and system for analyzing and providing early warning of heavy metal pollution in soil, with patent publication number CN120298187A, relates to the field of environmental monitoring technology. The method includes: determining the particle uniformity of soil in a target area and triggering a deep pollution collection command based on the particle uniformity; in response to receiving the deep pollution collection command, collecting migration trend data of pollutants at different depths in the target area soil and environmental data of the target area; based on the migration trend data, environmental data, and a pre-trained target pollution dynamic analysis model, determining the migration prediction results of pollutants at different depths in the target area soil; based on the migration trend data, environmental data, and particle uniformity, determining a soil pollution early warning index for the target area, and issuing an early warning for the target area based on the soil pollution early warning index and the migration prediction results. Therefore, this solution can improve the response speed of early warning and achieve accurate analysis of pollutant migration paths.

[0004] Existing soil pollution early warning methods largely rely on single-point monitoring, static thresholds, or post-hoc assessments, lacking the ability to predict future trends. Therefore, they suffer from the following drawbacks: insufficient spatiotemporal coverage, making comprehensive monitoring of large areas difficult; weak understanding of dynamic migration, diffusion, and degradation processes, failing to provide early warning of rapid increases in pollutant concentrations; lack of responsiveness to sudden events, geological anomalies, or climate change; and a lack of quantitative uncertainty and scenario analysis, making it difficult for decision-makers to obtain credible future risk ranges and prioritize response strategies, easily leading to lag effects and reducing the timeliness and accuracy of early warnings. Thus, current methods are more like "post-event alerts," failing to achieve proactive prevention and early intervention. This invention is proposed to address these shortcomings. Summary of the Invention

[0005] The purpose of this invention is to provide a method for predicting the dynamic evolution of soil arsenic pollution based on time-series deep learning, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for predicting the dynamic evolution of soil arsenic pollution based on time-series deep learning, the method comprising:

[0007] Information Acquisition: Obtain basic soil information by acquiring soil information of the area to be predicted;

[0008] Regional division: Based on basic soil information, the region to be predicted is divided using regional analysis methods, and the arsenic speciation and distribution in the basic soil information are added to obtain the target region information;

[0009] Basic reference establishment: The arsenic forms are preset, including organic arsenic and inorganic arsenic. Based on the arsenic forms, the reference establishment method is used to explore the changes of different arsenic forms in soil under different conditions to obtain a reference catalog and change information. The reference catalog includes target arsenic, reference information and correlation.

[0010] Time determination: The required time range is obtained by determining the time range of dynamic evolution through time determination methods;

[0011] Basic pollution evolution: Based on target area information and the required time range, the basic pollution situation is obtained by matching the reference catalog with the pollution evolution method and combining the change information.

[0012] Redundant event addition: An event library is obtained by generating events that affect changes in arsenic pollution through a redundancy generation method. The event library includes redundant events and impact information.

[0013] Target time prediction: Obtain the predicted time point, and based on the predicted time point and the required time range, obtain the experienced time range. Based on the experienced time range and the event library, optimize the basic pollution situation through the redundancy addition method to obtain the final pollution situation.

[0014] Furthermore, the regional analysis method includes: presetting a regional grid, dividing the region to be predicted based on the regional grid to obtain several target regions, numbering the target regions, adding basic soil information to the target regions accordingly to obtain several regional information, extracting the regional information containing arsenic to obtain target regional information, and the target regional information includes target arsenic morphology, target soil information and target number.

[0015] Furthermore, the reference establishment method includes: determining geological conditions based on the area to be predicted; obtaining historical information based on the geological conditions by acquiring historical arsenic evolution data, including historical soil information, historical arsenic state, historical arsenic evolution information, and historical environmental information; using historical soil information, historical environmental information, and historical arsenic evolution information as reference information, and historical arsenic state as target arsenic; establishing a correlation between target arsenic and reference information; classifying target arsenic based on its arsenic forms to obtain classification results; establishing an organic catalog for storing reference information and correlations for target arsenic as organic arsenic and an inorganic catalog for storing reference information and correlations for target arsenic as inorganic arsenic based on the classification results; establishing a reference catalog for storing the organic and inorganic catalogs; and using a neighboring study method with controlled variables to explore the relationship between arsenic changes and soil information based on historical information to obtain change information.

[0016] Furthermore, the adjacent exploration method includes: presetting a fixed time period and fixed factor items, wherein the fixed factor items include one of the following: past soil information, past arsenic evolution information, and past environmental information, and past arsenic state; extracting all information corresponding to items with consistent fixed factor items under the fixed time period based on an organic or inorganic catalog to obtain multiple differential sub-information; extracting the differential parts from the differential sub-information to obtain differential information; recording the fixed time period and fixed factor items of the differential information to obtain basic information; and integrating the basic information and differential information to obtain change information.

[0017] Furthermore, the time determination method includes: determining whether the user has a dynamically evolving time range to obtain a determination result; when the determination result indicates that the user has a dynamically evolving time range, obtaining the dynamically evolving time range to obtain the required time range; when the determination result indicates that the user does not have a dynamically evolving time range, obtaining the real-time time to obtain the starting point; presetting a fixed time; and continuously adding the fixed time based on the starting point until the user reports that the fixed time should be stopped to obtain the required time range.

[0018] Furthermore, the pollution evolution method includes: obtaining the initial state of arsenic based on target area information; searching for information corresponding to the target area information and the initial state of arsenic in a reference directory based on the initial state of arsenic and the target area information to obtain a matching result; when the matching result indicates that there is a target arsenic and reference information in the reference directory that corresponds to the initial state of arsenic and the target area information, extracting the past arsenic evolution information of the target arsenic to obtain the first segment evolution information; extracting the time length of the segment evolution information based on the required time range to obtain the remaining time length; matching information in the reference information based on the remaining time length, the final state of arsenic under the segment evolution information, and the target area information until the arsenic pollution evolution within the required time range is completed to obtain the second segment evolution information; integrating the first segment evolution information and the second segment evolution information to obtain the basic pollution situation; when the matching result indicates that there is no target arsenic and reference information corresponding to the initial state of arsenic and the target area information in the reference directory, selecting the target arsenic and reference information that is closest to the initial state of arsenic and the target area information in the reference directory, and combining it with change information to obtain the first segment evolution information.

[0019] Furthermore, the redundancy generation method includes: pre-setting influencing factors, which are redundant events; obtaining the impact of influencing factors on arsenic pollution based on previous information to obtain impact results, which are impact information; establishing a correspondence between influencing factors and impact results; and establishing an event database to store redundant events, impact information, and correspondence.

[0020] Furthermore, the redundancy addition method is as follows: based on the experienced time range, the basic pollution situation under the required time range is extracted to obtain the final pollution situation. When the user has an event addition requirement, redundant events and impact information are selected based on the event library to obtain the target event and target impact. Addition nodes are set. Based on the initial time of the required time range and the addition nodes, the segment time is obtained. Based on the segment time, the basic pollution situation is extracted to obtain the situation to be optimized. Based on the addition nodes, the target event and target impact are imported into the situation to be optimized. Based on the predicted time point and the addition nodes, the range to be acquired is obtained. Based on the final state of arsenic in the situation to be optimized and the soil information, the arsenic situation under the range to be acquired is obtained in conjunction with the pollution evolution method. The arsenic situation under the range to be acquired is combined with the situation to be optimized to obtain the final pollution situation.

[0021] Compared with the prior art, the beneficial effects of the present invention are:

[0022] This method for predicting the dynamic evolution of soil arsenic pollution based on temporal deep learning establishes a reference based on past soil arsenic pollution data. It then uses a controlled variable method to explore the changes in different arsenic forms under various conditions. Finally, a pollution evolution method is used to match the reference catalog and change information to dynamically predict the evolution of soil arsenic in the target area. A redundancy addition method is employed to consider unforeseen events based on actual needs and incorporates these events into the baseline pollution data for optimization, resulting in the final pollution situation. The overall aim is to use historical observation data and external variables to predict future soil arsenic levels, providing a basis for risk warning, remediation strategy evaluation, and monitoring plan optimization.

[0023] Meanwhile, by using the set regional analysis method, the area to be predicted is divided into multiple regions, so that the pollution dynamic evolution prediction of the arsenic-containing areas can be carried out separately. By using the set redundancy generation method to consider the impact of sudden events, an event library is generated in combination with the impact of sudden events, so that events can be added to the basic pollution situation according to the actual situation, so that the dynamic evolution prediction of soil arsenic pollution is more in line with the actual situation.

[0024] Meanwhile, classifying target arsenic by its speciation simplifies information retrieval during subsequent arsenic pollution evolution analysis. For inorganic arsenic, pollution evolution prediction only requires matching information within the inorganic catalog; similarly, for organic arsenic, matching only within the organic catalog is sufficient. The adjacent exploration method employs a controlled variable approach to investigate the relationship between arsenic variations in the organic and inorganic catalogs and soil information. This provides a foundation for future evolution predictions when soil information differs. The first segment of evolution information includes different segments corresponding to different environmental conditions, facilitating analysis of varying arsenic pollution dynamic evolution predictions under different weather conditions. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the overall process of the present invention;

[0026] Figure 2 This is a schematic diagram of the region division method of the present invention;

[0027] Figure 3 This is a schematic diagram of the arsenic pollution evolution prediction and treatment structure of the present invention;

[0028] Figure 4 This is a schematic diagram of the required time range structure of the present invention. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] Dynamic prediction of soil arsenic pollution is necessary because pollution levels are constantly changing due to the spatiotemporal coupling of multiple driving factors (rainfall, temperature, surface / groundwater dynamics, land use change, pollution source activity, etc.), making single-point or static monitoring insufficient to reveal future trends and exposure risks. Dynamic prediction can capture temporal dependence and spatial correlation, providing assessments of pollution evolution, exposure levels, and uncertainties under future scenarios. This enables early warning, optimization of monitoring and remediation strategies, evaluation of remediation effectiveness, and reduction of potential harm to populations and ecosystems. Through forecast intervals and scenario analysis, decision-makers can make more forward-looking and cost-effective decisions regarding source control, monitoring layout, land use planning, and emergency response.

[0031] like Figures 1-4 As shown, this invention provides a technical solution: a method for predicting the dynamic evolution of soil arsenic pollution based on time-series deep learning, the method comprising:

[0032] Information Acquisition: Obtain basic soil information by acquiring soil information of the area to be predicted;

[0033] It is important to note that the area to be predicted is the area where dynamic evolution prediction of soil arsenic pollution is required. The basic soil information is the soil information of this area. Basic soil information should be collected, such as soil physicochemical properties: pH, organic matter, clay content, particle size distribution, CEC (cation exchange capacity), iron and aluminum oxide content, sulfate / phosphate content, etc.; hydrogeochemical information: redox potential (Eh), water content, groundwater level conditions, soil moisture index; topographic and climatic information: slope, aspect, altitude, precipitation, temperature, evaporation, etc.; arsenic-related background information: total arsenic in the sample, quantitative As(III) / As(V), distribution of organic arsenic, adsorption / desorption characteristics data, and mineral binding state information.

[0034] Regional division: Based on basic soil information, the region to be predicted is divided using regional analysis methods, and the arsenic speciation and distribution in the basic soil information are added to obtain the target region information;

[0035] It is important to note that by using the regional analysis method, the area to be predicted is divided into multiple regions to facilitate the prediction of pollution dynamic evolution in arsenic-containing areas. The target area information is the basic soil information of the divided areas.

[0036] Basic reference establishment: The arsenic forms are preset, including organic arsenic and inorganic arsenic. Based on the arsenic forms, the reference establishment method is used to explore the changes of different arsenic forms in soil under different conditions to obtain a reference catalog and change information. The reference catalog includes target arsenic, reference information and correlation.

[0037] It is important to note that the reference establishment method is used to explore the dynamic evolution of different arsenic forms in soil under different conditions to obtain a reference catalog. Specifically, the reference is established based on the previous soil arsenic pollution, and the control variable method is used to explore the changes of different arsenic forms under different conditions to obtain change information. With the reference catalog and change information, it is convenient to predict the dynamic evolution of soil arsenic pollution in the target area.

[0038] Time determination: The required time range is obtained by determining the time range of dynamic evolution through time determination methods;

[0039] It is important to note that the time determination method is used to determine the specific time range for dynamic evolution, so as to make predictions on the dynamic evolution of soil arsenic pollution over a complete period of time.

[0040] Basic pollution evolution: Based on target area information and the required time range, the basic pollution situation is obtained by matching the reference catalog with the pollution evolution method and combining the change information.

[0041] It should be noted that by using the pollution evolution method set up, information matching is performed based on the reference catalog and change information to complete the dynamic evolution prediction of soil arsenic in the target area information. The basic pollution situation is the dynamic evolution prediction of soil arsenic in the target area information.

[0042] Redundant event addition: An event library is obtained by generating events that affect changes in arsenic pollution through a redundancy generation method. The event library includes redundant events and impact information.

[0043] It is important to note that the impact of unforeseen events is considered through the redundant generation method. An event library is generated based on the impact of unforeseen events, so that events can be added to the basic pollution situation according to the actual situation. This makes the prediction of the dynamic evolution of soil arsenic pollution more in line with the actual situation.

[0044] Target time prediction: Obtain the predicted time point, and based on the predicted time point and the required time range, obtain the experienced time range. Based on the experienced time range and the event library, optimize the basic pollution situation through the redundancy addition method to obtain the final pollution situation.

[0045] It is important to note that the prediction time points are obtained by the user. Specifically, this involves obtaining the time points needed for predicting the dynamic evolution of soil arsenic pollution. The process of obtaining the experienced time range based on the prediction time points and the required time range involves extracting the time range between the starting point of the required time range and the prediction time points. The specific time range needed for predicting the dynamic evolution of soil arsenic pollution is the experienced time range. By using a set redundancy addition method, considering sudden events according to actual usage needs, and reflecting these sudden events in the basic pollution situation, the final pollution situation is obtained. The final pollution situation includes specific soil arsenic migration and concentration conditions, etc.

[0046] The prediction of dynamic evolution of soil arsenic pollution based on temporal deep learning aims to use historical observation data and multiple external variables (such as rainfall, temperature, soil properties, land use, etc.) to predict the future trend of arsenic concentration and pollution level in the soil. Through end-to-end neural networks (such as Transformer, LSTM / GRU, or spatiotemporal models combined with GNN) or hybrid physically constrained neural networks, it takes into account both prediction accuracy and uncertainty assessment, and provides a basis for risk warning, remediation strategy evaluation and monitoring plan optimization.

[0047] like Figure 1 and Figure 2 As shown, the regional analysis method includes: presetting a regional grid, dividing the area to be predicted based on the regional grid to obtain several target areas, numbering the target areas, adding basic soil information to the target areas to obtain several regional information, extracting the regional information where arsenic exists to obtain target regional information, and the target regional information includes the target arsenic form, target soil information and target number.

[0048] It is important to note that the size of the regional grid is determined based on the actual usage. Specifically, the area to be predicted can be divided into regions according to differences in geological or soil conditions. By numbering the regions, staff can easily distinguish the changes in arsenic in each region. The process of adding basic soil information to the target region to obtain several regional information can be achieved by sampling the soil in each region after regional division, or by acquiring the information as a whole and then allocating it to extract the information of the regions containing arsenic to obtain the target region information. In other words, the information of the regions containing arsenic pollution is selected as the target region information.

[0049] like Figure 1As shown, the reference establishment method includes: determining the geological conditions based on the area to be predicted; obtaining historical information based on the geological conditions by acquiring historical arsenic evolution data, including historical soil information, historical arsenic state, historical arsenic evolution information, and historical environmental information; using historical soil information, environmental information, and historical arsenic evolution information as reference information, and historical arsenic state as target arsenic; establishing the correlation between target arsenic and reference information; classifying target arsenic based on its arsenic forms to obtain classification results; establishing an organic catalog to store reference information and correlations for target arsenic as organic arsenic and an inorganic catalog to store reference information and correlations for target arsenic as inorganic arsenic based on the classification results; establishing a reference catalog to store the organic and inorganic catalogs; and using the adjacent exploration method with controlled variables to explore the relationship between arsenic changes and soil information based on historical information to obtain change information.

[0050] It is important to note that the geological conditions refer to the geological conditions of the area to be predicted. Specific geological conditions include the slope of the ground in the area to be predicted. By obtaining historical arsenic evolution data under the same geological conditions as a reference, subsequent arsenic evolution predictions can be facilitated. The specific items in the historical soil information are consistent with those in the basic soil information mentioned above. The historical arsenic state refers to the specific state of arsenic, including its concentration information and specific distribution location. The historical arsenic evolution information is a collection of data on the migration and other actions of arsenic under the influence of historical soil and environmental information. A reference directory is established to store the above information. Using prior information as a reference provides a foundation for subsequent predictions of soil arsenic pollution evolution. The process of classifying target arsenic by arsenic speciation can reduce the difficulty of information retrieval when conducting subsequent arsenic pollution evolution predictions. When predicting the pollution evolution of inorganic arsenic, information matching is only required in the inorganic catalog, and when predicting the pollution evolution of organic arsenic, information matching is only required in the organic catalog. By using the adjacent exploration method and the controlled variable method, the relationship between arsenic changes in the organic and inorganic catalogs and soil information is explored, so as to provide a foundation for subsequent evolution predictions when there are differences in soil information.

[0051] The adjacent investigation method includes: setting a fixed time period and fixed factor items. The fixed factor items include one of the following: past soil information, past arsenic evolution information, and past environmental information, as well as past arsenic state. Based on the organic or inorganic catalog, all information corresponding to the items with consistent fixed factor items under the fixed time period is extracted to obtain multiple differential sub-information. The differential parts in the differential sub-information are extracted to obtain differential information. The fixed time period and fixed factor items of the differential information are recorded to obtain basic information. The basic information and differential information are integrated to obtain change information.

[0052] It is important to note that the fixed time period is determined based on actual usage. It represents a time length, specifically the difference information caused by the same fixed factor within that time length. The fixed factor is one of the following: past soil information, past arsenic evolution information, and past environmental information, along with past arsenic status. When the fixed factor is past soil information, multiple pieces of information corresponding to the item with consistent past soil information and past arsenic status within a given time length are extracted from the organic or inorganic catalog. These multiple pieces of information can be understood as a single piece of past information, including past soil information, past arsenic status, past arsenic evolution information, and past environmental information. By eliminating past soil information and past arsenic status, the influence of past arsenic evolution information and past environmental information is determined. The difference information is obtained by extracting the influence, so that it can be directly used in reverse after obtaining such differences later. The process of recording the fixed time period and fixed factor of the difference information to obtain basic information is the recording and reference process, ensuring the application scenario of the difference information. Based on the above, the wider the scope of past information, the higher the accuracy of subsequent soil arsenic pollution prediction.

[0053] like Figure 1 As shown, the time determination method includes: determining whether the user has a dynamically evolving time range to obtain a judgment result; when the judgment result indicates that the user has a dynamically evolving time range, obtaining the dynamically evolving time range to obtain the required time range; when the judgment result indicates that the user does not have a dynamically evolving time range, obtaining the real-time time to obtain the starting point; presetting a fixed time; and continuously adding the fixed time based on the starting point until the user reports that the fixed time will stop being added to obtain the required time range.

[0054] It is important to note that by setting the time determination method, that is, obtaining the time range for predicting the dynamic evolution of soil arsenic pollution in the area to be predicted, so that the dynamic evolution prediction has a time relationship, the starting point is obtained by obtaining the real time, a fixed time is preset, and the fixed time is continuously superimposed based on the starting point until the user feedback stops the superposition of the fixed time to obtain the required time range. In other words, the time range is actively generated when the user does not provide a time range for predicting the dynamic evolution of soil arsenic pollution.

[0055] like Figure 1 and Figure 3As shown, the pollution evolution method includes: obtaining the initial state of arsenic based on target area information; searching for information corresponding to the target area information and the initial state of arsenic in a reference catalog based on the initial state of arsenic and the target area information to obtain matching results; when the matching result indicates that there is target arsenic and reference information in the reference catalog that corresponds to the initial state of arsenic and the target area information, extracting the past arsenic evolution information of the target arsenic to obtain the first segment of evolution information; extracting the time length of the segment of evolution information based on the required time range to obtain the remaining time length; matching information in the reference information based on the remaining time length, the final state of arsenic under the segment of evolution information, and the target area information until the arsenic pollution evolution within the required time range is completed to obtain the second segment of evolution information; integrating the first segment of evolution information and the second segment of evolution information to obtain the basic pollution situation; when the matching result indicates that there is no target arsenic and reference information corresponding to the initial state of arsenic and the target area information in the reference catalog, selecting the target arsenic and reference information that is closest to the initial state of arsenic and the target area information in the reference catalog, and combining it with change information to obtain the first segment of evolution information.

[0056] It is important to note that the process of obtaining the initial state of arsenic based on target area information requires recording the arsenic state under the initial conditions to facilitate dynamic evolution. Arsenic evolution information is obtained by searching for corresponding information in a reference catalog using the initial arsenic state and target area information, and then extracting the evolution information of the target arsenic. Specifically, the principle is to use the reference catalog as a benchmark to match the arsenic state corresponding to the initial arsenic state with the evolution information obtained from soil information. It is important to note that the first segment of evolution information includes different segments of evolution information corresponding to different environmental conditions, so that staff can analyze different predictions of arsenic pollution dynamic evolution under different weather conditions. When no information corresponding to the initial arsenic state and target area information is found in the reference catalog, the process of selecting the target arsenic with the closest initial arsenic state and target area information, and combining it with change information to obtain the first segment of evolution information involves first selecting information with the same initial arsenic state in the reference catalog, obtaining the difference between the selected information and the target area information, and adding the influence of the change information on the previous arsenic evolution information under the selected information to obtain the first segment of evolution information. The subsequent acquisition of the second segment of evolution information is also done in this way when information cannot be matched in the reference catalog. Figure 3 As shown, based on the state of arsenic in the target area, the corresponding catalog and the change information generated by the corresponding catalog are used to perform pollution evolution prediction processing on the arsenic in the target area.

[0057] like Figure 1As shown, the redundancy generation method includes: pre-setting influencing factors, which are redundant events; obtaining the impact of influencing factors on arsenic pollution based on previous information to obtain impact results, which are impact information; establishing the correspondence between influencing factors and impact results; and establishing an event database to store redundant events, impact information, and correspondence.

[0058] It is important to note that influencing factors can be added or obtained by staff through their own experiments. These factors include pesticides and biopesticides; certain arsenic compounds used as insecticides, herbicides, or preservatives may lead to arsenic accumulation or redistribution in the soil with long-term use; water management and irrigation, such as irrigation with water sources with high arsenic content or groundwater extraction causing rise / recharge processes, altering Eh / pH and promoting changes in arsenic speciation and migration; soil improvement and fertilization, such as adding arsenic-containing organic fertilizers, phosphate fertilizers, or amendments, changing the competitive relationship of adsorption sites, increasing free arsenic or altering its bound state; and waste treatment and reuse of contaminated sites, such as industrial waste residue and arsenic-containing waste. Improper treatment of sludge and leachate from domestic waste can lead to arsenic entering the soil-water system. By simulating these events under known arsenic pollution evolution scenarios, the impact of these events on the arsenic pollution evolution can be obtained, i.e., the impact results can be obtained. By integrating the above events and results into an event database, a redundant event database for sudden events can be established. This allows for the addition of sudden events to the dynamic evolution prediction of arsenic as needed during actual prediction, resulting in more realistic predictions. Specific prediction units can be arsenic concentration and arsenic migration and transformation, etc. Therefore, the impact results of the above events correspond to arsenic concentration and arsenic migration and transformation.

[0059] like Figure 1 and Figure 4 As shown, the redundancy addition method is as follows: Based on the experienced time range, the basic pollution situation under the required time range is extracted to obtain the final pollution situation. When the user has an event addition requirement, redundant events and impact information are selected from the event library to obtain the target event and target impact. Addition nodes are set, and the segment time is obtained based on the initial time of the required time range and the addition node. The basic pollution situation is extracted based on the segment time to obtain the situation to be optimized. The target event and target impact are imported into the situation to be optimized based on the addition node. The range to be acquired is obtained based on the predicted time point and the addition node. The arsenic situation under the range to be acquired is obtained based on the final state of arsenic in the situation to be optimized and the soil information, combined with the pollution evolution method. The arsenic situation under the range to be acquired is combined with the situation to be optimized to obtain the final pollution situation.

[0060] It is important to note that, such as Figure 4 As shown, Figure 4The text expresses the relationship between the required time range and the predicted time point. The predicted time point and the required time range differ, representing the arsenic pollution situation at a single point in time and the arsenic pollution situation over a period of time. Through a redundant addition method, when a user has a need to add sudden events, the user sets an addition node, and the basic pollution situation at the initial time and the addition node is obtained to determine the situation to be optimized. Based on the arsenic pollution situation in the optimization situation, known events are added, and their impact on arsenic is added to obtain the information on the arsenic state. Then, based on the information on the arsenic state and target area information, combined with a pollution evolution method, the arsenic situation within the target range is obtained. By splicing the two types of arsenic situations according to time, the final pollution situation is obtained.

[0061] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended embodiments and their equivalents.

Claims

1. A method for predicting the dynamic evolution of soil arsenic pollution based on temporal deep learning, the method comprising: Information Acquisition: Obtain basic soil information by acquiring soil information of the area to be predicted; The method is characterized in that it further includes: Regional division: Based on basic soil information, the region to be predicted is divided using regional analysis methods, and the arsenic speciation and distribution in the basic soil information are added to obtain the target region information; Basic reference establishment: The arsenic forms are preset, including organic arsenic and inorganic arsenic. Based on the arsenic forms, the reference establishment method is used to explore the changes of different arsenic forms in soil under different conditions to obtain a reference catalog and change information. The reference catalog includes target arsenic, reference information and correlation. Time determination: The required time range is obtained by determining the time range of dynamic evolution through time determination methods; Basic pollution evolution: Based on target area information and the required time range, the basic pollution situation is obtained by matching the reference catalog with the pollution evolution method and combining the change information. Redundant event addition: An event library is obtained by generating events that affect changes in arsenic pollution through a redundancy generation method. The event library includes redundant events and impact information. Target time prediction: Obtain the predicted time point, and based on the predicted time point and the required time range, obtain the experienced time range. Based on the experienced time range and the event library, optimize the basic pollution situation through the redundancy addition method to obtain the final pollution situation.

2. The method for predicting the dynamic evolution of soil arsenic pollution based on time-series deep learning according to claim 1, characterized in that: The regional analysis method includes: presetting a regional grid, dividing the area to be predicted based on the regional grid to obtain several target areas, numbering the target areas, adding basic soil information to the target areas to obtain several regional information, extracting the regional information containing arsenic to obtain target regional information, and the target regional information includes target arsenic morphology, target soil information and target number.

3. The method for predicting the dynamic evolution of soil arsenic pollution based on time-series deep learning according to claim 1, characterized in that: The reference establishment method includes: determining geological conditions based on the area to be predicted; obtaining historical information based on the geological conditions, including historical soil information, historical arsenic state, historical arsenic evolution information, and historical environmental information; using historical soil information, historical environmental information, and historical arsenic evolution information as reference information, and historical arsenic state as target arsenic; establishing a correlation between target arsenic and reference information; classifying target arsenic based on its arsenic forms to obtain classification results; establishing an organic catalog for storing reference information and correlations for target arsenic as organic arsenic and an inorganic catalog for storing reference information and correlations for target arsenic as inorganic arsenic based on the classification results; establishing a reference catalog for storing the organic and inorganic catalogs; and using a neighboring study method with controlled variables to explore the relationship between arsenic changes and soil information based on historical information to obtain change information.

4. The method for predicting the dynamic evolution of soil arsenic pollution based on time-series deep learning according to claim 3, characterized in that: The adjacent exploration method includes: presetting a fixed time period and fixed factor items. The fixed factor items include one of the following: past soil information, past arsenic evolution information, and past environmental information, as well as past arsenic state. Based on an organic or inorganic catalog, all information corresponding to items with consistent fixed factor items under the fixed time period is extracted to obtain multiple differential sub-information. The differential parts in the differential sub-information are extracted to obtain differential information. The fixed time period and fixed factor items of the differential information are recorded to obtain basic information. The basic information and differential information are integrated to obtain change information.

5. The method for predicting the dynamic evolution of soil arsenic pollution based on time-series deep learning according to claim 1, characterized in that: The time determination method includes: determining whether the user has a dynamically evolving time range to obtain a judgment result; when the judgment result indicates that the user has a dynamically evolving time range, obtaining the dynamically evolving time range to obtain the required time range; when the judgment result indicates that the user does not have a dynamically evolving time range, obtaining the real-time time to obtain the starting point; presetting a fixed time; and continuously adding the fixed time based on the starting point until the user reports that the fixed time should be stopped to obtain the required time range.

6. The method for predicting the dynamic evolution of soil arsenic pollution based on time-series deep learning according to claim 1, characterized in that: The pollution evolution method includes: obtaining the initial state of arsenic based on target area information; searching for information corresponding to the target area information and the initial state of arsenic in a reference directory based on the initial state of arsenic and the target area information to obtain a matching result; when the matching result indicates that there is target arsenic and reference information in the reference directory that corresponds to the initial state of arsenic and the target area information, extracting the past arsenic evolution information of the target arsenic to obtain the first segment evolution information; extracting the time length of the segment evolution information based on the required time range to obtain the remaining time length; matching information in the reference information based on the remaining time length, the final state of arsenic under the segment evolution information, and the target area information until the arsenic pollution evolution within the required time range is completed to obtain the second segment evolution information; integrating the first segment evolution information and the second segment evolution information to obtain the basic pollution situation; when the matching result indicates that there is no target arsenic and reference information corresponding to the initial state of arsenic and the target area information in the reference directory, selecting the target arsenic and reference information that is closest to the initial state of arsenic and the target area information in the reference directory, and combining it with change information to obtain the first segment evolution information.

7. The method for predicting the dynamic evolution of soil arsenic pollution based on time-series deep learning according to claim 3, characterized in that: The redundancy generation method includes: pre-setting influencing factors, which are redundant events; obtaining the impact of influencing factors on arsenic pollution based on previous information to obtain impact results, which are impact information; establishing a correspondence between influencing factors and impact results; and establishing an event database to store redundant events, impact information, and correspondence.

8. The method for predicting the dynamic evolution of soil arsenic pollution based on time-series deep learning according to claim 1, characterized in that: The redundancy addition method is as follows: Based on the basic pollution situation under the required time range, the final pollution situation is obtained. When the user has an event addition requirement, redundant events and impact information are selected from the event library to obtain the target event and target impact. Addition nodes are set. Based on the initial time of the required time range and the addition nodes, the segment time is obtained. Based on the segment time, the basic pollution situation is extracted to obtain the situation to be optimized. Based on the addition nodes, the target event and target impact are imported into the situation to be optimized. Based on the predicted time point and the addition nodes, the range to be acquired is obtained. Based on the final state of arsenic in the situation to be optimized and the soil information, the arsenic situation under the range to be acquired is obtained in conjunction with the pollution evolution method. The arsenic situation under the range to be acquired is combined with the situation to be optimized to obtain the final pollution situation.

Images