An intelligent soil-groundwater pollution early warning method and system based on the Internet of Things

Abstract

The application discloses a soil-groundwater pollution intelligent early warning method and system based on Internet of Things, and relates to application of the Internet of Things technology in the field of environmental monitoring and intelligent decision management, and comprises the following steps: S1: collecting soil and groundwater pollution data through an Internet of Things sensor network arranged in a monitoring area.The soil-groundwater pollution intelligent early warning method and system based on the Internet of Things solves the pain points of unclear management subject rights and responsibilities and delayed response in a cross-regional pollution event by constructing a pollution path driven dynamic rights and responsibilities topology network. A geographic fence triggers an automatic reorganization mechanism to realize synchronization of a management grid and pollution diffusion, thereby breaking the rigid constraint of administrative boundaries; and a double verification interlocking design ensures that the early warning decision has both environmental risk control accuracy and cross-regional cooperation feasibility.

Description

Technical Field

[0001] This invention relates to the application of Internet of Things (IoT) technology in the field of environmental monitoring and intelligent decision-making management, specifically to an IoT-based intelligent early warning method and system for soil and groundwater pollution. Background Technology

[0002] In the field of soil and groundwater pollution prevention and control, intelligent early warning systems based on Internet of Things (IoT) technology have become an important tool. These systems acquire environmental data in real time through the deployment of sensor networks and use analytical models for risk assessment and early warning. Such systems significantly improve the timeliness and data coverage of pollution monitoring. However, existing technological solutions have significant shortcomings at the management and supervision level, especially when dealing with complex pollution scenarios involving multiple administrative regions and multi-level management entities. Soil and groundwater pollution has migratory and diffusion characteristics, and pollution source areas and affected areas often span different administrative jurisdictions. Current early warning systems mostly focus on data collection, model optimization, or local alerts within a single region, lacking effective cross-regional information sharing and collaborative response mechanisms. This makes it difficult for upstream pollution risks to be transmitted to downstream areas in a timely manner, creating information barriers. Furthermore, the actual management of pollution incidents involves multiple levels, including on-site personnel, enterprises, local environmental protection departments, and higher-level regulatory agencies, resulting in blurred boundaries of authority and responsibility.

[0003] Existing systems generally fail to effectively integrate management needs at different levels, and cannot dynamically distribute customized early warning information and decision support content based on differences in responsibilities. When sudden pollution occurs or the scope of pollution changes dynamically, the lack of a unified and efficient cross-regional and cross-level collaborative workflow leads to delays in the transformation of early warning information into management actions, difficulty in dynamically optimizing emergency resource allocation based on the pollution situation, and high costs and low efficiency in coordinating responses from all parties, severely weakening the ultimate management effectiveness and practical value of early warnings. Therefore, the current technical problem to be solved is: how to construct an intelligent management framework that supports cross-regional collaborative early warning and multi-level linkage response to address the shortcomings of existing systems in complex pollution scenarios, such as management delays and unclear responsibilities. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides the following technical solution: an intelligent early warning method for soil-groundwater pollution based on the Internet of Things, comprising the following steps:

[0005] S1: Collect soil and groundwater pollution data through an IoT sensor network deployed in the monitoring area;

[0006] S2: Construct a dynamic topology network of rights and responsibilities: Divide continuous spatial responsibility units according to the pollution diffusion prediction path, and form a cross-regional management grid by associating administrative jurisdiction boundaries, pollution source entities and regulatory agency levels;

[0007] S3: Execute graded trigger response: When the pollution spread range reaches the preset geofence, automatically trigger the on-site level, cross-regional level or major event level workflow to push customized early warning information to the corresponding level management terminal;

[0008] S4: Implement dual verification: verify the scientific validity of the pollution diffusion path through the spatial situation credibility verification module, and optimize the multi-party collaboration scheme through the management conflict resolution verification module;

[0009] S5: Based on the verification results, the emergency resource scheduling module dynamically plans the material delivery path.

[0010] Preferably, the construction of the dynamic responsibility topology network includes:

[0011] The pollution plume diffusion path output by the groundwater migration model was used as the segmentation basis;

[0012] Geofencing technology is used to bind spatial responsibility units to administrative boundaries in real time.

[0013] When a pollution plume crosses a geofence, the topology network is automatically reorganized and the list of associated management entities is updated.

[0014] Preferably, the spatial situation credibility verification includes:

[0015] Run a pollution causation-diffusion joint model that couples machine learning and hydrodynamic simulation to generate path cloud maps with labeled confidence regions;

[0016] Real-time sensor data is input into a historical database of similar scenarios for deviation comparison. If the deviation exceeds the set tolerance threshold, a manual review mechanism is activated.

[0017] Preferably, the management conflict resolution verification includes:

[0018] The responsibility weights of each management entity are calculated using the Shapley value allocation model;

[0019] Generate the optimal cost solution for cross-regional cooperation based on the Nash equilibrium principle;

[0020] The stability of the collaborative scheme was tested using Monte Carlo simulation.

[0021] Preferably, the Monte Carlo simulation is injected with random perturbation parameters during execution to test the failure probability of the scheme in a preset number of simulations.

[0022] Preferably, the operation of the emergency resource scheduling module includes:

[0023] Access to a spatiotemporal digital twin containing emergency vehicles, equipment, and personnel;

[0024] An objective function is constructed using the real-time pollution diffusion rate as the independent variable, and the optimal material delivery path is dynamically calculated.

[0025] Preferably, the objective function specifies that the emergency response delay time is less than a fixed coefficient multiple of the pollution diffusion rate.

[0026] Preferred options also include:

[0027] Deploy a decision-making sandbox simulation unit in the management terminal to simulate the pollution recession trend after the execution of disposal instructions;

[0028] Generate a simulation report that requires joint signature and approval from the Space Situation and Management Conflict Verification Module.

[0029] An intelligent early warning system for realizing an IoT-based intelligent early warning method for soil and groundwater pollution includes: an IoT data acquisition module, a dynamic responsibility network construction module, a hierarchical triggering engine, a dual verification module, and a resource scheduling optimization module;

[0030] The IoT data acquisition module collects soil and groundwater pollution data through an IoT sensor network deployed in the monitoring area;

[0031] The dynamic responsibility network construction module constructs a dynamic responsibility topology network: it divides continuous spatial responsibility units according to the pollution diffusion prediction path, and associates administrative jurisdiction boundaries, pollution source owners and regulatory agency levels to form a cross-regional management grid;

[0032] The tiered trigger engine executes tiered trigger responses: when the pollution spread reaches the preset geofence, it automatically triggers on-site, cross-regional, or major event-level workflows to push customized early warning information to the corresponding management terminals.

[0033] The dual verification module implements dual verification: the spatial situation credibility verification module verifies the scientific validity of the pollution diffusion path, and the management conflict resolution verification module optimizes the multi-party collaboration scheme.

[0034] The resource scheduling optimization module drives the emergency resource scheduling module to dynamically plan material delivery routes based on the verification results;

[0035] The hierarchical triggering engine is further configured as follows:

[0036] When an on-site warning is issued, a list of disposal measures containing the coordinates of the contaminated area is pushed to the terminals of local personnel.

[0037] When a cross-regional level early warning is issued, a situational sand table is simultaneously encrypted and sent to relevant parties. The sand table integrates pollution source contribution maps and receptor sensitive point distributions.

[0038] When a major event-level warning is issued, the resource scheduling interface is invoked to output an emergency material allocation plan.

[0039] Preferably, the dual verification module includes:

[0040] The spatial situation verification submodule includes a built-in pollution tracing-diffusion joint model and a historical case database comparison unit.

[0041] The conflict resolution submodule incorporates a power-responsibility game equilibrium algorithm and a conflict scenario simulator.

[0042] This invention provides an intelligent early warning method and system for soil and groundwater pollution based on the Internet of Things (IoT). It has the following beneficial effects:

[0043] This IoT-based intelligent early warning method and system for soil and groundwater pollution addresses the pain points of unclear responsibilities and delayed responses in cross-regional pollution incidents by constructing a dynamic responsibility topology network driven by pollution pathways. A geofence-triggered automatic reorganization mechanism synchronizes the management grid with pollution spread, overcoming the rigid constraints of administrative boundaries. A dual-verification interlocking design ensures that early warning decisions combine the precision of environmental risk control with the feasibility of cross-regional collaboration. A tiered triggering engine, coupled with blockchain-based entity-specific evidence storage, enables seamless upgrades from local response to cross-provincial coordination, improving the efficiency of handling major incidents and reducing management and coordination costs.

[0044] This IoT-based intelligent early warning method and system for soil and groundwater pollution utilizes a geologically adaptive resource scheduling model that transforms pollution diffusion speed into dynamic constraint coefficients, establishing a precise matching mechanism between response time and pollution behavior. A differentiated strategy of pre-deployment in sandy soil layers and high-frequency delivery to clay layers addresses resource misallocation issues in complex geological scenarios. A triple-circuit breaker mechanism in decision-making sandbox simulation proactively blocks execution risks, and combined with pollutant characteristic arbitration rules, improves the pass rate of high-risk pollution treatment plans. Monte Carlo triple perturbation testing ensures the collaborative plan's resilience to extreme risks, and a full-process traceability system supported by resource digital twins enhances emergency resource utilization. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the module interaction of an intelligent early warning system for soil and groundwater pollution based on the Internet of Things according to the present invention.

[0046] Figure 2 This is a flowchart illustrating an intelligent early warning method for soil and groundwater pollution based on the Internet of Things according to the present invention. Detailed Implementation

[0047] 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.

[0048] Please see Figure 1 and Figure 2 This invention provides a technical solution: an intelligent early warning method for soil-groundwater pollution based on the Internet of Things, comprising the following steps:

[0049] S1: Collect soil and groundwater pollution data through an IoT sensor network deployed in the monitoring area;

[0050] S2: Construct a dynamic topology network of rights and responsibilities: Divide continuous spatial responsibility units according to the pollution diffusion prediction path, and form a cross-regional management grid by associating administrative jurisdiction boundaries, pollution source entities and regulatory agency levels;

[0051] S3: Execute graded trigger response: When the pollution spread range reaches the preset geofence, automatically trigger the on-site level, cross-regional level or major event level workflow to push customized early warning information to the corresponding level management terminal;

[0052] S4: Implement dual verification: verify the scientific validity of the pollution diffusion path through the spatial situation credibility verification module, and optimize the multi-party collaboration scheme through the management conflict resolution verification module;

[0053] S5: Based on the verification results, the emergency resource scheduling module dynamically plans the material delivery path.

[0054] It should be further explained that, in the specific implementation process, firstly, data on soil moisture content, pollutant concentration, and groundwater flow velocity in the monitoring area are collected in real time through an Internet of Things sensor network; then, a dynamic responsibility topology network is constructed, including: dividing the continuous polluted area into multiple spatial responsibility units based on the pollution plume diffusion path output by the groundwater migration model; and binding each unit to its respective administrative jurisdiction boundary, polluting enterprise, and regulatory agency level in real time through geofencing technology to form an initial management grid.

[0055] When a pollution plume is detected crossing a preset geofence in real time, such as spreading from City A to the border area of ​​County B, the system automatically reorganizes the topology network, including: adding responsible units for the affected area; updating the list of associated management entities, such as adding the Environmental Protection Bureau of County B as a response entity; and reallocating cross-regional collaborative tasks.

[0056] Then, the response is triggered in a tiered manner, including: if the pollution range is limited to a single responsible unit, a field-level workflow is triggered: pushing pollution coordinates, disposal list, and real-time monitoring links to the mobile terminals of local personnel; if the pollution plume crosses the geofence and enters an adjacent unit, a cross-regional workflow is triggered: automatically generating an encrypted situational sandbox, including a heat map of pollution source contribution and the distribution of receptor sensitive points; synchronizing sandbox data to the management terminals of all related parties and locking collaborative permissions; if the pollution diffusion rate exceeds the critical value or affects drinking water sources, a major event-level workflow is triggered: activating the provincial resource scheduling interface.

[0057] The dual verification mechanism operates, including: Spatial credibility verification: calling the pollution tracing-diffusion joint model to output a probability cloud map of the core pollution path, with high-confidence areas marked in red; real-time comparison with a historical similar scenario case library, automatically freezing the warning command and requesting expert review when the path deviation exceeds the tolerance threshold; Management conflict resolution verification: collecting the disposal resource list and responsibility statement from each management entity; calculating the responsibility weight through the Shapley value allocation model, and generating a cross-regional cooperation plan by combining it with Nash equilibrium; injecting random perturbation parameters to conduct 10,000 Monte Carlo simulations, which only take effect when the failure probability of the plan is lower than the set threshold.

[0058] Resource scheduling optimization: Access the spatiotemporal resource digital twin to obtain the real-time location of emergency vehicles and equipment; use the pollution diffusion speed as the benchmark variable to dynamically calculate the material delivery path, with the goal that the response delay time is always less than a fixed coefficient multiple of the diffusion speed.

[0059] The construction of a dynamic responsibility-based topology network includes:

[0060] The pollution plume diffusion path output by the groundwater migration model was used as the segmentation basis;

[0061] Geofencing technology is used to bind spatial responsibility units to administrative boundaries in real time.

[0062] When a pollution plume crosses a geofence, the topology network is automatically reorganized and the list of associated management entities is updated.

[0063] It should be further explained that, in the specific implementation process, the construction of the dynamic responsibility topology network begins with the spatial analysis of the pollution diffusion path: First, the pollution plume diffusion path output by the groundwater migration model is used as the segmentation benchmark, and this path is updated driven by real-time sensor data; then, based on the position and migration direction of the pollution plume front, the continuous polluted area is divided into multiple spatial responsibility units, and the coverage length of each unit is dynamically adjusted according to the groundwater permeability coefficient to ensure that the unit boundary coincides with the pollutant transport front. Next, geofencing binding is initiated: the unit boundary coordinates are extracted and overlaid with the electronic administrative map, automatically associating the administrative jurisdiction boundaries, the registered location of the enterprise to which the pollution source belongs, and the corresponding regulatory agency level contained within the unit, forming an initial management grid database, where the administrative jurisdiction boundaries include city and county boundaries, and the regulatory agency levels include local and provincial levels.

[0064] When the peak concentration of the pollution plume dissolution crosses a preset geofence, for example, spreading from Unit 3 in City A to Unit 5 in County B, the system immediately triggers a topology network reconfiguration: First, it scans the list of administrative entities and enterprises within the newly added unit and incorporates them into the overall management entity table; second, it reallocates collaborative tasks based on pollution contribution, such as City A being responsible for source control and County B for downstream monitoring; finally, it pushes the access key to the management terminal of the newly added entity and synchronously updates the sharing range of the encrypted situational awareness sandbox. If the pollution plume simultaneously intrudes into multiple units, a task priority queue is generated according to the intrusion time. During the reconfiguration process, the system continuously compares the pollution spread prediction with the measured data. When the boundary offset exceeds the model tolerance, the reconfiguration command is automatically frozen and manual confirmation is requested to avoid network misreconstruction due to sensor anomalies.

[0065] Space situation credibility verification includes:

[0066] Run a pollution causation-diffusion joint model that couples machine learning and hydrodynamic simulation to generate path cloud maps with labeled confidence regions;

[0067] Real-time sensor data is input into a historical database of similar scenarios for deviation comparison. If the deviation exceeds the set tolerance threshold, a manual review mechanism is activated.

[0068] It should be further explained that, in the specific implementation process, after the spatial situation credibility verification is initiated, the pollution tracing-diffusion joint model is first invoked: Soil moisture content, pollutant concentration, and groundwater flow velocity data collected by the real-time sensor network are simultaneously input into the machine learning prediction module and the hydrodynamic simulation module. The machine learning module analyzes historical pollution migration patterns based on long short-term memory networks and outputs a heatmap of the probability distribution of pollution sources; the hydrodynamic module constructs a groundwater flow field based on Darcy's law to simulate pollutant transport paths. The outputs of the two modules are weighted and superimposed in the fusion layer to generate a probability cloud map of the core pollution path, where high-confidence areas are marked in red and low-confidence areas are marked in yellow.

[0069] The system then enters the historical case database comparison phase: It extracts key features such as the current pollution diffusion rate, geological permeability coefficient, and pollutant type, and matches similar scenarios in the historical case database. If a similar case is found, the spatial offset between the real-time path and the historical path is calculated. When the offset exceeds a preset tolerance threshold, for example, if the pollution plume front deviates from the predicted path by more than a set distance, a three-level response is automatically triggered: Level 1 response suspends the issuance of the current warning command; Level 2 response activates backup hydrogeological parameters for remodeling; and Level 3 response sends a manual review request along with the offset data packet to the provincial monitoring platform.

[0070] If no similar cases are found, an expert consultation mode is activated: local hydrogeological experts are convened via video conferencing system to mark the corrected path nodes on the digital twin sand table. The corrected path data is automatically fed back to the model training library to enhance learning samples. Throughout the verification process, when the pollutant type is a highly toxic substance or the diffusion rate exceeds the safety threshold, the tolerance threshold level is forcibly increased to reduce the risk of misjudgment.

[0071] Management conflict resolution verification includes:

[0072] The responsibility weights of each management entity are calculated using the Shapley value allocation model;

[0073] Generate the optimal cost solution for cross-regional cooperation based on the Nash equilibrium principle;

[0074] The stability of the collaborative scheme was tested using Monte Carlo simulation.

[0075] It should be further explained that, in the specific implementation process, when a pollution incident involves multiple management entities, the system initiates conflict resolution verification: First, it collects the disposal resource lists, rights and responsibilities statements, and cost constraint data submitted by each entity through its management terminal. Based on the pollution source contribution analysis results, a Shapley value allocation model is used to calculate the responsibility weights, including: extracting the total pollutant emissions, historical treatment records, and enterprise size factors within each entity's jurisdiction; quantifying the marginal responsibility contribution value of each entity to the pollution incident through a coalition game algorithm; and generating an initial responsibility weight sequence.

[0076] The process then enters the Nash equilibrium optimization phase: Minimizing the total cost of cross-regional collaboration is the objective function, with responsibility weights input as constraints into the multi-agent negotiation model. This model simulates a three-way game scenario, iteratively solving for resource allocation ratios and task division schemes acceptable to all parties. The three-way game scenario includes the pollution source, the diffusion path, and the receptor-sensitive area. If the negotiation reaches a stalemate, such as multiple consecutive rounds of rejected solutions, a backup rule base is activated to inject forced equilibrium parameters.

[0077] After the initial plan is approved, a Monte Carlo simulation stress test is initiated: three typical disturbances are injected into the digital twin environment, including: temporary resource shortage, new sudden pollution sources, and non-response by the management entity. Temporary resource shortage involves randomly removing a set number of emergency vehicles; new sudden pollution sources involve adding new pollution points along the diffusion path; and non-response by the management entity involves marking a certain entity's task status as failed. A set number of random combination disturbance tests are executed, and the failure indicators of the collaborative plan in each test are recorded, such as pollution diffusion exceeding safety limits or cost exceeding thresholds. Finally, only when the number of failures in the disturbance tests is less than the set threshold is an effective command output and a legally binding electronic liability confirmation automatically generated. If the test fails, the process returns to the Nash equilibrium stage to regenerate the plan and apply punitive constraints.

[0078] The Monte Carlo simulation injects random perturbation parameters during execution to test the failure probability of the scheme in a preset number of simulations. It should be further explained that, in the specific implementation process, after obtaining the initial collaborative scheme, the system initiates a Monte Carlo simulation stress test: First, a digital twin environment is constructed, loading the current pollution diffusion model, emergency resource distribution map, and management entity response rule base. During the test, the simulation is executed cyclically a set number of times, with three types of coupled perturbation parameters randomly injected in each cycle: a temporary resource shortage perturbation that randomly disables a set proportion of emergency vehicles or equipment and marks failure points on the map; a new sudden pollution source perturbation that generates secondary pollution points on the pollution plume diffusion path, with their locations intelligently selected based on the geological vulnerability map; and a management entity refusing to respond perturbation that randomly designates a responsible entity into a state of disobedience, freezing its task execution permissions.

[0079] During each test run, the system tracks four key failure indicators: whether the pollution front has crossed the preset geofence boundary, whether the emergency response delay exceeds a set multiple of the pollution spread rate, whether the collaboration cost exceeds the budget threshold, and whether the pollutant concentration in the key receptor sensitive area exceeds the standard. If any of these failure indicators is triggered, the test result is recorded as a failure.

[0080] After completing all test cycles, the total number of failures is counted: if the number of failures is below the set safety threshold, the solution is automatically signed and becomes effective; if the number of failures is in the warning range, buffer constraints are added to the solution, such as adding a backup resource pool; if the number of failures exceeds the critical value, an intelligent re-optimization mechanism is immediately activated, including: based on the clustering analysis results of failure scenarios, locating high-frequency failure links, such as a persistent resource gap at the border of an administrative region, forcibly adjusting the responsibility weight allocation of that link and shortening its response time requirement, and then returning to the Nash equilibrium stage to generate a new solution for retesting. The re-optimization process executes a maximum of a set number of rounds; if the limit is exceeded, manual emergency takeover is initiated.

[0081] The operation of the emergency resource dispatch module includes:

[0082] Access to a spatiotemporal digital twin containing emergency vehicles, equipment, and personnel;

[0083] An objective function is constructed using the real-time pollution diffusion rate as the independent variable, and the optimal material delivery path is dynamically calculated.

[0084] It should be further explained that, in the specific implementation process, when the emergency resource dispatch module is activated, it first connects to the spatiotemporal resource digital twin: real-time acquisition of emergency vehicle GPS coordinates, on-board equipment operating status, professional personnel qualification certification information, and material warehouse inventory, constructing a dynamic resource topology network. The core dispatch logic uses the pollution diffusion rate as the benchmark variable: when the sensor detects an increase in the diffusion rate, the system automatically compresses the response time window and triggers an accelerated dispatch mode.

[0085] The dynamic solution process is implemented in three steps, including:

[0086] The first step is to establish a path competition model, which converts the movement speed of the pollution front into a time pressure coefficient. For example, for every unit increase in the diffusion speed, the target response delay threshold is reduced by a set percentage.

[0087] The second step is to launch a multi-objective optimization engine to simultaneously optimize transportation costs and resource coverage while satisfying the core constraint that "response delay time is less than a fixed coefficient multiple of diffusion speed".

[0088] The third step is to implement anti-interference path correction: When real-time traffic data indicates congestion on a certain road segment or a sudden failure at a resource point, the geological risk compensation mechanism is automatically activated, including: if the pollution diffusion area is located in a high-permeability sand layer, a set number of backup resource vehicles are allowed to be deployed in advance to the predicted diffusion path points; if it is located in a low-permeability clay layer, the precise delivery mode is switched to reduce the amount of each transport but increase the frequency of transport.

[0089] In extreme conditions, such as torrential rain causing a surge in the spread rate, the system activates a cascading dispatch protocol: directly calling upon mobile processing equipment from the provincial emergency resource pool and delivering it to the core pollution area via helicopter, while simultaneously freezing resource allocation requests from non-critical areas. After each dispatch command is issued, the system uses a digital twin to compare the arrival time of materials with the location of the pollution front in real time, and automatically triggers a switch to a backup route when the deviation approaches a critical value.

[0090] The objective function stipulates that the emergency response delay time must be less than a fixed coefficient multiple of the pollution diffusion rate. It should be further explained that, in practical implementation, the core constraint of the objective function is set as "the emergency response delay time must be less than a fixed coefficient multiple of the pollution diffusion rate," and this coefficient is dynamically adjusted according to the geological risk level: in the early stages of pollution diffusion, the system uses a basic coefficient value as a benchmark constraint; when the diffusion rate is continuously monitored to increase in real time, a coefficient compression mechanism is activated, including: if the diffusion rate increases by a set value per unit time, the coefficient value is reduced by a set amount according to a preset step.

[0091] Differentiated coefficient management is implemented for special geological environments: if the polluted area is in a highly permeable sandy soil layer, the system automatically activates the geological acceleration compensation factor to reduce the basic coefficient by a set ratio to offset the risk of rapid diffusion; if it is in a low-permeability clay layer, the coefficient constraints are allowed to be temporarily relaxed within a set time window.

[0092] When encountering extreme weather conditions, such as a red alert for heavy rain, the system activates a disaster amplification compensation: Rainfall intensity is obtained through a meteorological data interface; if each unit of rainfall exceeds a critical value, the coefficient value is additionally lowered by a set amount. A dual safety check is implemented during coefficient adjustment: first, it verifies whether the adjusted response time exceeds human limits; if it is less than a set number of minutes, the adjustment is frozen; second, it checks whether the resource pool can meet the accelerated dispatch requirements; if the number of available helicopters is insufficient, the original coefficient is maintained. Before finally outputting the objective function, the system performs coefficient conflict resolution: when multiple adjustment rules are in effect simultaneously, they are executed in order of priority: geological risk > meteorological disaster > diffusion trend, ensuring that core constraints always remain within the implementable range. All coefficient adjustment records are synchronized to the decision sandbox in real time, allowing management to trace the path of constraint changes.

[0093] Also includes:

[0094] Deploy a decision-making sandbox simulation unit in the management terminal to simulate the pollution recession trend after the execution of disposal instructions;

[0095] Generate a simulation report that requires joint signature and approval from the Space Situation and Management Conflict Verification Module.

[0096] It should be further explained that, in the specific implementation process, the decision sandbox simulation system starts immediately after receiving the disposal instruction at the management terminal: first, it loads the current pollution diffusion model, the digital twin of resource distribution, and real-time snapshots of environmental parameters. The simulation process adopts adaptive time step technology, including: using a fine time step when the initial pollution concentration is high, such as setting a simulation at the minute level, and switching to a coarse-grained time step, such as setting a simulation at the hour level, after the diffusion stabilizes.

[0097] The system dynamically simulates the pollution reduction curve within a set time after the execution of a disposal command: when the command includes pollution source blocking measures, the system gradually reduces the source emission intensity in the model; if the command involves adding monitoring wells downstream, the system updates the monitoring data of the receptor sensitive points. It also synchronously tracks resource consumption dynamics: calculating fuel consumption in real time based on material transportation distance, calculating the wear and tear of consumables such as filter elements based on equipment operating time, and generating labor costs by linking the regional human resource scheduling table.

[0098] Three key triggering conditions are set during the simulation: when the simulated pollution front crosses the preset geographical fence, the simulation is immediately paused and spatial risks are marked; when resource consumption exceeds the budget threshold, a cost circuit breaker mechanism is activated to freeze subsequent instructions; and when the concentration in sensitive areas continues to fail to decrease, previous instruction nodes are automatically traced back. When generating a feasibility report, the spatial situation verification submodule needs to confirm the rationality of the pollution reduction model, and the management conflict resolution submodule needs to sign off on the compliance of resource consumption. The two are interlocked through digital signatures to take effect.

[0099] If the two modules disagree, such as space verification passing but management verification rejecting it, the system will initiate intelligent arbitration: the handling of highly toxic pollutants will be based on space verification, and cross-regional collaboration will prioritize management verification. The final report includes a holographic replay function of the simulation process, supporting the separate display of operational traces according to the responsible party.

[0100] An intelligent early warning system for realizing an IoT-based intelligent early warning method for soil and groundwater pollution includes: an IoT data acquisition module, a dynamic responsibility network construction module, a hierarchical triggering engine, a dual verification module, and a resource scheduling optimization module;

[0101] The IoT data acquisition module collects soil and groundwater pollution data through an IoT sensor network deployed in the monitoring area;

[0102] The dynamic responsibility network construction module constructs a dynamic responsibility topology network: it divides continuous spatial responsibility units according to the pollution diffusion prediction path, and associates administrative jurisdiction boundaries, pollution source owners and regulatory agency levels to form a cross-regional management grid;

[0103] The tiered trigger engine executes tiered trigger responses: when the pollution spread reaches the preset geofence, it automatically triggers on-site, cross-regional, or major event-level workflows to push customized early warning information to the corresponding management terminals.

[0104] The dual verification module implements dual verification: the spatial situation credibility verification module verifies the scientific validity of the pollution diffusion path, and the management conflict resolution verification module optimizes the multi-party collaboration scheme.

[0105] The resource scheduling optimization module drives the emergency resource scheduling module to dynamically plan material delivery routes based on the verification results;

[0106] The tiered triggering engine can be further configured as follows:

[0107] When an on-site warning is issued, a list of disposal measures containing the coordinates of the contaminated area is pushed to the terminals of local personnel.

[0108] When a cross-regional level early warning is issued, a situational awareness sand table is simultaneously encrypted and sent to relevant parties. The sand table integrates pollution source contribution maps and receptor sensitive point distribution.

[0109] When a major event-level warning is issued, the resource scheduling interface is invoked to output an emergency material allocation plan.

[0110] It should be further explained that, in the specific implementation process, this intelligent early warning system acquires real-time data on soil pore water pressure, pollutant ion concentration, and groundwater depth through the distributed sensor nodes of the IoT data acquisition module. After preprocessing by the edge computing gateway, the data is transmitted to the dynamic responsibility network construction module. This module calls upon the geographic information system base map and, when dividing the pollution diffusion path into spatial responsibility units, uses real-time hydrogeological parameters to dynamically calibrate the unit boundaries, including automatically shrinking the unit area to improve management accuracy when the permeability coefficient changes abruptly.

[0111] The tiered triggering engine initiates response logic within a set timeframe after the pollution plume contacts the geofence: At the on-site level, the engine extracts skill tags from the local personnel database, such as "Certified Heavy Metal Processing Personnel," and pushes a customized disposal list containing pollution coordinates to the matching personnel's terminals; at the moment of a cross-regional level alert, the engine activates the encrypted generation protocol for the situation sand table, automatically marking the main responsible parties based on the pollution source contribution map, while simultaneously synchronizing sand table data to all related party management terminals and applying operation permission locks, allowing only the responsible party to modify the source area disposal plan; after a major event-level alert is triggered, the engine calls upon the emergency material digital twin from the resource scheduling optimization module, combining it with the pollution front movement vector to generate a multi-path delivery plan.

[0112] During the operation of the dual verification module, the spatial situation verification submodule performs a similarity scan of the historical case library at set time intervals, while the conflict resolution management submodule monitors feedback signals from various entities in real time. When a cross-regional response fails to reach a consensus within a set duration, the system automatically escalates the warning level and unlocks the authority for mandatory intervention by higher-level regulators.

[0113] The two-factor authentication module includes:

[0114] The spatial situation verification submodule includes a built-in pollution tracing-diffusion joint model and a historical case database comparison unit.

[0115] The conflict resolution submodule incorporates a power-responsibility game equilibrium algorithm and a conflict scenario simulator.

[0116] It should be further explained that during the specific implementation process, when the dual verification module is running, the spatial situation verification submodule first loads the pollution tracing-diffusion joint model and dynamically corrects the prediction path by assimilating the sensor data stream in real time: when the deviation between the measured value and the predicted value of a certain monitoring point continues to exceed the set threshold, the parameter inversion mechanism is automatically triggered, including: adjusting the permeability coefficient or adsorption coefficient in the hydrogeological parameter database until the prediction path converges to the measured trajectory.

[0117] The historical case database comparison unit employs a multi-feature weighted matching strategy: assigning differential weights to three key indicators—pollutant toxicity level, formation permeability, and diffusion rate—prioritizing matching historical scenarios where all three indicators match perfectly; if no perfectly matching case exists, a fuzzy matching mode is activated, relaxing the threshold for individual indicators. The responsibility game equilibrium algorithm embedded in the conflict resolution submodule dynamically adjusts responsibility weights after receiving declarations from various stakeholders: when a stakeholder provides false resource data that is detected by the system, its weight is automatically downgraded and a punitive resource quota is added.

[0118] During the runtime of the conflict scenario simulator, three types of progressive disturbances are injected, including: primary disturbance, which randomly removes a set number of resource vehicles; intermediate disturbance, which superimposes secondary pollution points on the pollution path; and advanced disturbance, which forces a designated subject to enter a disobedience state. After each round of simulation, the root cause of failure is analyzed. If the failure is caused by resource shortage, the area is marked as a vulnerable unit, and its resource allocation priority is automatically increased during subsequent scheduling.

[0119] When space-based verification requires rapid source containment while management verification tends towards phased disposal, the system uses intelligent arbitration based on the pollutant's half-life: persistent pollutants are prioritized for space-based verification, while easily degradable pollutants are handled using a phased strategy. All verification processes generate blockchain evidence, with operational traces segmented and stored according to the responsible party for audit traceability.

[0120] It should be further explained that, in the specific implementation process, during the deployment phase of the IoT sensor network, the monitoring area is divided into several basic units based on hydrogeological characteristics. Each unit is equipped with a soil moisture sensor, a pollutant concentration detector, and a groundwater flow velocity probe. Sensor data is transmitted to edge computing nodes via a low-power wide-area network. The nodes have built-in data cleaning algorithms to eliminate outliers. For example, when a probe's readings exceed a set multiple for adjacent points multiple times consecutively, it is automatically marked as suspicious data, and a backup sensor is activated. The cleaned data stream is input into the central processing platform in real time.

[0121] The construction of the dynamic responsibility-based topology network begins with the spatial analysis of pollution diffusion paths. The central processing platform utilizes a groundwater migration model, which is based on Darcy's law and convection-diffusion equations, and calculates the pollution plume's migration trajectory using real-time hydrological parameters. The pollution front line output by the model serves as the benchmark for dividing spatial responsibility units, and the unit boundaries are dynamically adjusted according to the permeability coefficient: when a sudden increase in the permeability coefficient of a certain area is detected, the unit area of ​​that area is automatically reduced to improve management accuracy.

[0122] The association between units and administrative jurisdictions is achieved through electronic map fencing technology. The system automatically matches the administrative division codes and enterprise registration information covered by the unit. The criterion for determining when a pollution plume crosses the fence is: when the concentration of multiple consecutive monitoring points at the boundary of a unit exceeds a set threshold and the diffusion vector points to adjacent units, network reorganization is triggered. The reorganization process first scans the management entities within the newly added unit and adds them to the collaborative response queue; then, it reallocates task weights based on the contribution of pollution sources. For example, upstream enterprises that contribute more than a set proportion of pollutants must bear the primary responsibility for disposal.

[0123] The response logic of the graded triggering engine is divided into a three-layer progressive structure. The on-site response activation condition is that the pollution range does not exceed a single responsible unit. The system automatically selects local personnel with the corresponding disposal qualifications, such as technicians holding heavy metal treatment certification, and pushes an electronic work order containing three-dimensional location of pollution coordinates, video guidance on disposal operation, and real-time monitoring links to their mobile terminals.

[0124] The cross-regional response is initiated when the pollution plume touches the geofence. The system generates an encrypted situational sandbox: a heat map of the contribution of each administrative region is calculated based on the emission intensity and migration path of the pollution source, and a comprehensive risk map is formed by overlaying the distribution data of sensitive points such as drinking water wells.

[0125] The sand table is synchronized to all relevant management terminals via the government intranet, with differentiated editing permissions: the primary responsible party, i.e., the entity with the highest contribution, has the right to modify the source control plan; secondary responsible parties can edit the path monitoring plan; other units only have viewing permissions. Major event-level responses are triggered by two conditions: a continuous doubling of the pollution diffusion rate or impact on provincial water sources. In this case, the system directly calls the provincial emergency resource database to generate a cascading dispatch plan including helicopter delivery routes.

[0126] The spatial situation credibility verification adopts a dual-model collaborative mechanism. In the pollution causation-diffusion joint model, the machine learning module analyzes the migration patterns of historical pollution events to generate probability predictions, while the hydrodynamic module simulates the physical diffusion process based on real-time hydrological data. The outputs of the two models are dynamically weighted in the fusion layer: when the completeness of geological exploration data is high, the weight of the hydrodynamic model is increased; when data is missing, the weight of the machine learning model is increased.

[0127] In the generated path probability cloud map, high-confidence areas are marked in dark colors, while low-confidence areas are highlighted in light colors. A three-tiered matching strategy is employed for comparison with the historical case database: first, cases with pollutant types and formation structures completely identical are retrieved; if no match is found, the search is broadened to scenarios with similar permeability coefficients; if no results are found, an expert consultation is initiated. A progressive response is used to handle comparison deviations: model parameters are automatically calibrated for slight path deviations; warning commands are frozen and backup hydrological models are activated for moderate deviations; and a review request along with the original data packet is sent to the provincial platform for severe deviations.

[0128] The system verifies and implements dynamic adjustment of responsibility weights to manage conflict resolution. After collecting resource lists submitted by various entities, the system verifies their authenticity through digital twins of resources: if a county claims to be able to mobilize a set number of emergency vehicles but GPS shows that it has insufficient vehicles in use, the system automatically lowers its responsibility weight and adds virtual resource penalty quotas.

[0129] The Shapley value allocation model incorporates three key factors in its calculation: real-time emissions, historical records of violations, and company asset size, ensuring that large, non-compliant companies receive a higher weight. The Nash equilibrium optimization phase simulates a three-way game: polluting companies aim to minimize production stoppage losses, migrating companies demand pollution compensation, and recipient-sensitive areas prioritize water supply security.

[0130] When negotiations reach an impasse, the system enforces a balance based on preset rules, including: prioritizing the source of highly toxic pollutants and treating conventional pollutants according to the sensitive area priority. Monte Carlo simulations inject three types of coupled disturbances: randomly disabling some emergency resource vehicles to simulate shortages; adding secondary pollution points in fault zones; and randomly designating an entity to refuse to perform a task. Each simulation monitors four failure indicators: pollution exceeding the red line, response timeout, cost exceeding limits, and sensitive area exceeding standards. After tens of thousands of simulations, the number of failures is tallied; low-failure solutions take effect immediately, medium-failure solutions receive additional buffer resources, and high-failure solutions trigger a redistribution of responsibility weights.

[0131] Emergency resource dispatch establishes dynamic constraints on pollution diffusion speed and response time. Basic constraint coefficients are initialized based on geological type: smaller coefficients are used for sandy soil layers to offset the risk of rapid diffusion, while larger coefficients are used for clay layers. A speed monitoring and feedback mechanism is implemented during operation: as the diffusion speed increases by a set value per unit time, the coefficient is adjusted downwards in a step-wise manner. Special operating conditions trigger compensation rules: during a red alert for heavy rain, the coefficient is further reduced based on the rainfall intensity; during nighttime responses, the coefficient setting ratio is relaxed to extend the reasonable dispatch window.

[0132] The scheduling path optimization is set with three objectives: the core constraint is to ensure that the response delay is less than the diffusion rate multiplied by the dynamic coefficient; the secondary objective is to minimize the total transportation mileage; and the auxiliary objective is to maximize the coverage of sensitive areas. The geological adaptive strategy is manifested as follows: in high-permeability areas, a set proportion of resource vehicles are deployed to the predicted locations in advance, while in low-permeability areas, a high-frequency, low-batch precision delivery mode is adopted.

[0133] The decision-making sandbox simulation employs variable-step simulation technology. During peak pollution concentration periods, a fine time step is used to simulate minute-level changes, while during periods of stable diffusion, an hourly time step is used to improve efficiency. Action commands are translated into model parameters: source-blocking commands correspond to a linear decay function of pollution emission intensity, and commands for adding monitoring wells update the receptor database.

[0134] The simulation process incorporates three types of circuit breaker mechanisms: a spatial circuit breaker is activated when pollution breaches the dynamic fence, freezing subsequent command flows; a cost circuit breaker is activated when resource consumption exceeds the budget; and a safety circuit breaker triggers operational backtracking for abnormal concentrations in sensitive areas. The feasibility report requires joint signatures from two verification modules: the spatial verification submodule confirms the rationality of the pollution decline curve, and the management verification submodule reviews the compliance of resource consumption. In case of conflicting conclusions, arbitration is based on pollutant characteristics: the spatial verification conclusion is adopted for persistent pollutants, while easily degradable pollutants receive priority in management. The report is generated and stored on the blockchain, supporting the tracing of operational traces by administrative division.

[0135] The system's operational safeguards include: access control with time limits, penalties for false data, and vulnerability unit labeling. Access control with time limits automatically transfers out-of-time cross-regional collaborations to higher-level regulators. Penalties for false data reduce the decision-making weight of entities providing false resource information. Vulnerability unit labeling upgrades resource allocation levels in areas with high resource shortages. All operational data is stored on blockchain nodes, segmented by administrative entity, supporting targeted auditing and cross-regional accountability.

[0136] By constructing a dynamic responsibility topology network driven by pollution pathways, this project addresses the pain points of unclear responsibilities and delayed responses among management entities in cross-regional pollution incidents. A geofencing-triggered automatic reorganization mechanism synchronizes the management grid with pollution spread, overcoming the rigid constraints of administrative boundaries. A dual-verification interlocking design ensures that early warning decisions combine the precision of environmental risk control with the feasibility of cross-regional collaboration. A tiered triggering engine, coupled with blockchain-based entity-specific evidence storage, enables seamless upgrades from local response to cross-provincial coordination, improving the efficiency of handling major incidents and reducing management and coordination costs.

[0137] The geological adaptive resource scheduling model transforms pollution diffusion speed into dynamic constraint coefficients, establishing a precise matching mechanism between response time and pollution behavior. A differentiated strategy of pre-deployment in sand layers and high-frequency delivery in clay layers addresses resource misallocation issues in complex geological scenarios. A triple-circuit breaker mechanism in decision-making sandbox simulation proactively blocks execution risks, and combined with pollutant characteristic arbitration rules, improves the pass rate of high-risk pollution treatment plans. Monte Carlo triple perturbation testing ensures the collaborative plan's resilience to extreme risks, and a full-process traceability system supported by resource digital twins enhances emergency resource utilization.

[0138] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0139] 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 claims and their equivalents.

Claims

1. A smart early warning method for soil-groundwater pollution based on the Internet of Things, characterized in that, Includes the following steps: S1: Collect soil and groundwater pollution data through an IoT sensor network deployed in the monitoring area; S2: Construct a dynamic topology network of rights and responsibilities: Divide continuous spatial responsibility units according to the pollution diffusion prediction path, and form a cross-regional management grid by associating administrative jurisdiction boundaries, pollution source entities and regulatory agency levels; The construction of the dynamic responsibility-based topology network includes: The pollution plume diffusion path output by the groundwater migration model was used as the segmentation basis; Geofencing technology is used to bind spatial responsibility units to administrative boundaries in real time. When a pollution plume crosses a geofence, the topology network is automatically reorganized and the list of associated management entities is updated. S3: Execute graded trigger response: When the pollution spread range reaches the preset geofence, automatically trigger the on-site level, cross-regional level or major event level workflow to push customized early warning information to the corresponding level management terminal; S4: Implement dual verification: verify the scientific validity of the pollution diffusion path through the spatial situation credibility verification module, and optimize the multi-party collaboration scheme through the management conflict resolution verification module; S5: Based on the verification results, the emergency resource scheduling module dynamically plans the material delivery path.

2. The intelligent early warning method for soil-groundwater pollution based on the Internet of Things according to claim 1, characterized in that: The spatial situation credibility verification includes: Run a pollution causation-diffusion joint model that couples machine learning and hydrodynamic simulation to generate path cloud maps with labeled confidence regions; Real-time sensor data is input into a historical database of similar scenarios for deviation comparison. If the deviation exceeds the set tolerance threshold, a manual review mechanism is activated.

3. The intelligent early warning method for soil-groundwater pollution based on the Internet of Things according to claim 1, characterized in that: The management conflict resolution verification includes: The responsibility weights of each management entity are calculated using the Shapley value allocation model; Generate the optimal cost solution for cross-regional cooperation based on the Nash equilibrium principle; The stability of the collaborative scheme was tested using Monte Carlo simulation.

4. The intelligent early warning method for soil-groundwater pollution based on the Internet of Things according to claim 3, characterized in that: The Monte Carlo simulation is injected with random perturbation parameters during execution to test the failure probability of the scheme in a preset number of simulations.

5. The intelligent early warning method for soil-groundwater pollution based on the Internet of Things according to claim 1, characterized in that: The operation of the emergency resource scheduling module includes: Access to a spatiotemporal digital twin containing emergency vehicles, equipment, and personnel; An objective function is constructed using the real-time pollution diffusion rate as the independent variable, and the optimal material delivery path is dynamically calculated.

6. The intelligent early warning method for soil-groundwater pollution based on the Internet of Things according to claim 5, characterized in that: The objective function specifies that the emergency response delay time is less than a fixed coefficient multiple of the pollution diffusion rate.

7. The intelligent early warning method for soil-groundwater pollution based on the Internet of Things according to claim 1, characterized in that: Also includes: Deploy a decision-making sandbox simulation unit in the management terminal to simulate the pollution recession trend after the execution of disposal instructions; Generate a simulation report that requires joint signature and approval from the Space Situation and Management Conflict Verification Module.

8. An intelligent early warning system implementing any one of claims 1-7, characterized in that, include: The module includes an IoT data acquisition module, a dynamic responsibility network construction module, a hierarchical triggering engine, a dual verification module, and a resource scheduling optimization module. The IoT data acquisition module collects soil and groundwater pollution data through an IoT sensor network deployed in the monitoring area; The dynamic responsibility network construction module constructs a dynamic responsibility topology network: it divides continuous spatial responsibility units according to the pollution diffusion prediction path, and associates administrative jurisdiction boundaries, pollution source owners and regulatory agency levels to form a cross-regional management grid; The construction of the dynamic responsibility-based topology network includes: The pollution plume diffusion path output by the groundwater migration model was used as the segmentation basis; Geofencing technology is used to bind spatial responsibility units to administrative boundaries in real time. When a pollution plume crosses a geofence, the topology network is automatically reorganized and the list of associated management entities is updated. The tiered trigger engine executes tiered trigger responses: when the pollution spread reaches the preset geofence, it automatically triggers on-site, cross-regional, or major event-level workflows to push customized early warning information to the corresponding management terminals. The dual verification module implements dual verification: the spatial situation credibility verification module verifies the scientific validity of the pollution diffusion path, and the management conflict resolution verification module optimizes the multi-party collaboration scheme. The resource scheduling optimization module drives the emergency resource scheduling module to dynamically plan material delivery routes based on the verification results; The hierarchical triggering engine is further configured as follows: When an on-site warning is issued, a list of disposal measures containing the coordinates of the contaminated area is pushed to the terminals of local personnel. When a cross-regional level early warning is issued, a situational sand table is simultaneously encrypted and sent to relevant parties. The sand table integrates pollution source contribution maps and receptor sensitive point distributions. When a major event-level warning is issued, the resource scheduling interface is invoked to output an emergency material allocation plan.

9. The intelligent early warning system according to claim 8, characterized in that: The dual verification module includes: The spatial situation verification submodule includes a built-in pollution tracing-diffusion joint model and a historical case database comparison unit. The conflict resolution submodule incorporates a power-responsibility game equilibrium algorithm and a conflict scenario simulator.

Images