Coal mine drainage monitoring safety early warning control method and system based on sensor big data information

Abstract

The present application relates to the technical field of coal mine drainage, and more particularly to a coal mine drainage monitoring safety early warning control method and system based on sensor big data information, which first analyzes multi-source sensing data to generate multi-dimensional constraint data containing interface coordinates, rate weights and scene types; secondly, a dynamic deposition potential field matrix is constructed using a spatial interpolation decay function, and a deposition dynamic prediction model containing morphological evolution and sediment diffusion terms is established accordingly; then, the synergy coefficient is calculated by combining the water sump volume evaluation value and the potential field intensity, the local dredging allocation component is generated and fused with the prediction model to form a working condition adaptive monitoring control field; finally, based on the control field and the scene type, a gangue discharge strategy vector field is constructed, a smart linkage early warning scheme is generated through dynamic probability calculation, and is converted into real-time control instructions. The method realizes accurate perception, trend prediction and closed-loop management and control of the deposition form, effectively improving the safety and intelligent level of the drainage system.

Description

Technical Field

[0001] This invention relates to the field of coal mine drainage technology, and in particular to a coal mine drainage monitoring, safety early warning and control method and system based on sensor big data information. Background Technology

[0002] The underground drainage system in coal mines is a crucial link in ensuring safe production. Water sump, as the main water storage and sedimentation facility, accumulates coal dust, rock debris, and suspended particles from the water flowing into the roadways over long periods, forming coal slime. As the slime thickness increases, the effective volume of the water sump gradually decreases, reducing the drainage system's regulating capacity and easily causing blockages in the suction wells, pump wear, and even cavitation damage. In severe cases, this can lead to drainage failure, threatening mine safety. Currently, most coal mines still rely on manual, periodic sludge removal, which suffers from long monitoring cycles, data lag, and subjective judgment, making it difficult to accurately grasp the spatiotemporal distribution and development trend of coal slime accumulation inside the water sump.

[0003] While some mines have attempted to use ultrasonic or liquid level sensors for simple monitoring, these methods are limited by their reliance on single sensing technologies. They cannot achieve refined, three-dimensional dynamic reconstruction of complex underwater sedimentation interfaces, and lack intelligent waste rock removal decision support and safety early warning mechanisms based on real-time sedimentation status. Furthermore, existing systems generally lack adaptive response capabilities to different water tank conditions (such as alternating operation of the main and auxiliary tanks, fluctuations in influent sediment content, and interference from dredging operations), resulting in rigid waste rock removal control strategies, low dredging efficiency, and significant resource waste.

[0004] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this invention is to provide a coal mine drainage monitoring and safety early warning control method and system based on sensor big data information. It aims to solve the technical problems of existing coal mine water tank coal slime siltation monitoring methods being single and lagging behind, lacking the ability to finely reconstruct complex underwater siltation morphology and the adaptive intelligent sludge removal decision mechanism based on real-time working conditions, resulting in low sludge removal efficiency and potential drainage safety hazards.

[0006] To achieve the above objectives, the present invention provides a coal mine drainage monitoring and safety early warning control method based on sensor big data information, the method comprising:

[0007] The interface feature analysis and siltation parameter formalization processing of the real-time collected multi-source sensor data of the water tank are performed to generate multi-dimensional coal slime siltation constraint data, which includes siltation interface coordinates, siltation rate weights and water tank operation scenario types.

[0008] Based on the sedimentation interface coordinates and sedimentation rate weights in the multi-dimensional coal slime sedimentation constraint data, a dynamic sedimentation potential field matrix reflecting the distribution of coal slime sedimentation potential is constructed by combining spatial interpolation decay functions.

[0009] Using the dynamic sedimentation potential field matrix as the driving source, a dynamic prediction model for coal slime sedimentation, including sedimentation morphology evolution term and sediment erosion and diffusion term, is established.

[0010] The siltation impact coordination coefficient is calculated based on the assessed volume of the coal mine drainage system water tank and the current location siltation potential field intensity. A local dredging and gangue allocation component is generated and fused with the coal slime siltation dynamic prediction model to obtain the adaptive siltation monitoring and control field for the water tank working condition.

[0011] Based on the adaptive siltation monitoring and control field of the water tank operating conditions and the water tank operation scenario type, a waste rock removal operation strategy vector field is constructed. An intelligent linkage waste rock removal early warning scheme is generated through dynamic probability calculation and converted into real-time waste rock removal monitoring and control commands to complete closed-loop siltation management and scheduling.

[0012] Optionally, the step of performing interface feature analysis and formal processing of the real-time collected multi-source sensor data of the water tank to generate multi-dimensional coal slime sedimentation constraint data includes:

[0013] Feature extraction and semantic annotation are performed on the multi-source sensor data of the water tank to obtain a set of siltation elements containing siltation interface identifier, siltation intensity descriptor and operation constraint descriptor;

[0014] The set of sedimentation elements is input into a pre-trained coal mine water tank sedimentation knowledge base for mapping and calibration to obtain a standardized combination of sedimentation parameters.

[0015] Based on the standardized siltation parameter combination, the coordinates of the siltation interface and the siltation rate weight are extracted respectively to obtain the coordinate values ​​and weight values.

[0016] The water tank operation scenario is classified according to the siltation interface identifier to obtain the water tank operation scenario type, and then combined with the siltation interface coordinates and siltation rate weight to obtain the multi-dimensional coal slime siltation constraint data.

[0017] Optionally, the step of constructing a dynamic sedimentation potential field matrix reflecting the distribution of coal slime sedimentation potential by combining the sedimentation interface coordinates and sedimentation rate weights in the multi-dimensional coal slime sedimentation constraint data through spatial interpolation decay functions includes:

[0018] Based on the coordinates of the sedimentation interface in the multi-dimensional coal slime sedimentation constraint data, a spatial attenuation sedimentation contribution function is constructed for each sedimentation interface point to obtain the sedimentation potential field distribution of a single-point region.

[0019] Based on the sedimentation rate weight, the single-point area sedimentation potential field distribution of each sedimentation interface point is weighted by sedimentation priority to obtain the coal slime sedimentation priority potential field.

[0020] The multiple coal slime deposition priority potential fields are superimposed and combined according to the spatial location of the water tank to obtain the coal slime deposition potential distribution of the entire water tank.

[0021] Based on the water tank operation scenario type, the attenuation range parameter is set and the distribution of coal slime siltation potential in the full water tank is matrixed to obtain the dynamic siltation potential field matrix.

[0022] Optionally, the step of using the dynamic sedimentation potential field matrix as the driving source to establish a dynamic prediction model for coal slime sedimentation that includes sedimentation morphology evolution terms and sediment erosion and diffusion terms includes:

[0023] The sedimentation gradient components are calculated based on the dynamic sedimentation potential field matrix to obtain the sedimentation driving force vector pointing to the region of high sedimentation intensity.

[0024] Gradient calculation of the sedimentation field is performed based on the distribution density of sediment content in the inflow water in different areas of the reservoir to obtain the sedimentation morphology evolution term that prevents local sedimentation estimation bias.

[0025] Based on the requirement of continuity in the coal slime deposition process, the coal slime transportation is subjected to diffusion treatment to obtain the sediment erosion diffusion term that suppresses the fluctuation of the deposition morphology estimation.

[0026] The mass conservation equation is used to solve the siltation driving force vector, siltation morphology evolution term, and sediment erosion and diffusion term to obtain the dynamic prediction model of coal slime siltation describing the dynamic deposition of coal slime under the complex working conditions of a water tank.

[0027] Optionally, the step of calculating the siltation impact coordination coefficient based on the coal mine drainage system water tank volume assessment value and the current location siltation potential field intensity, generating a local dredging and waste rock allocation component, and fusing it with the coal slime siltation dynamic prediction model to obtain the water tank operating condition adaptive siltation monitoring and control field includes:

[0028] The siltation impact synergy coefficient for each water tank zone is obtained by performing a weighted product calculation based on the water tank volume assessment value and the current location siltation potential field intensity.

[0029] Based on the siltation impact synergy coefficient and the dredging operation radius, a local waste rock allocation field is constructed to obtain the siltation distribution of each water tank zone;

[0030] The distribution of the siltation volume is adjusted and calculated to obtain the local siltation and waste rock removal adjustment component to prevent excessive or untimely waste rock removal operations.

[0031] The local dredging and waste rock allocation component is superimposed and fused with the coal slime siltation dynamic prediction model to obtain the water tank condition adaptive siltation monitoring and control field, which has both global morphological prediction and local operation coordination functions.

[0032] Optionally, the step of constructing a local waste rock disposal and allocation field based on the siltation impact synergy coefficient and the dredging operation radius to obtain the siltation distribution of each water storage zone includes:

[0033] Based on the siltation impact synergy coefficient, the radial distance of the water tank space is normalized to obtain the distance synergy factor;

[0034] The operational radius of the dredging operation is used to set the waste rock allocation range, and the power-law attenuation calculation is performed on the distance coordination factor to obtain the spatial siltation attenuation coefficient.

[0035] The siltation influence synergy coefficient and the spatial siltation attenuation coefficient are multiplied to obtain the siltation intensity value at each spatial location of the water tank;

[0036] Based on the siltation intensity value, the radial and tangential siltation components of the surrounding space are decomposed to obtain the siltation distribution that describes the dynamic distribution of coal slime siltation in the entire water tank.

[0037] Optionally, the step of constructing a waste rock removal operation strategy vector field based on the adaptive siltation monitoring and control field of the water tank operating conditions and the water tank operation scenario type, generating an intelligent linkage waste rock removal early warning scheme through dynamic probability calculation, and converting it into real-time waste rock removal monitoring and control commands to complete closed-loop siltation management and scheduling includes:

[0038] Based on the adaptive siltation monitoring and control field of the water tank and the water tank operation scenario type, feature extraction is performed to obtain the siltation operation feature vector;

[0039] Based on the siltation operation feature vector, the priority weight of each water tank zone in the waste rock removal process is calculated to obtain the waste rock removal operation strategy vector field;

[0040] The effective volume of the current water tank and the operating status of the water pump in the coal mine drainage system are input into the waste rock removal operation strategy vector field for probability matrix optimization to obtain the early warning trigger probability value of each water tank partition corresponding to the waste rock removal stage;

[0041] Based on the early warning trigger probability value, the siltation risk level is determined and control instructions are generated to obtain the intelligent linkage silt removal early warning scheme, which includes graded early warning levels, timing of silt removal operations, and path of dredging equipment. This scheme is then converted into the real-time silt removal monitoring and control instructions to complete the closed-loop siltation management and scheduling.

[0042] Furthermore, to achieve the above objectives, the present invention also provides a coal mine drainage monitoring and safety early warning control system based on sensor big data information, the system comprising:

[0043] The data parsing module is used to perform interface feature parsing and formal processing of siltation parameters on the real-time collected multi-source sensor data of the water tank, and generate multi-dimensional coal slime siltation constraint data. The multi-dimensional coal slime siltation constraint data includes siltation interface coordinates, siltation rate weights and water tank operation scenario types.

[0044] The potential field construction module is used to construct a dynamic siltation potential field matrix that reflects the distribution of coal slime siltation potential based on the siltation interface coordinates and siltation rate weights in the multi-dimensional coal slime siltation constraint data and by combining spatial interpolation decay functions.

[0045] The model prediction module is used to establish a dynamic prediction model for coal slime deposition, which includes the dynamic sedimentation potential field matrix as the driving source and includes sedimentation morphology evolution term and sediment erosion and diffusion term.

[0046] The adaptive control module is used to calculate the siltation impact coordination coefficient based on the coal mine drainage system water tank volume assessment value and the current location siltation potential field intensity, generate local dredging and gangue allocation components and integrate them with the coal slime siltation dynamic prediction model to obtain the water tank working condition adaptive siltation monitoring and control field.

[0047] The closed-loop scheduling module is used to construct a waste rock removal operation strategy vector field based on the adaptive siltation monitoring and control field of the water tank operating conditions and the water tank operation scenario type. It generates an intelligent linkage waste rock removal early warning scheme through dynamic probability calculation and converts it into real-time waste rock removal monitoring and control commands to complete the closed-loop siltation management and scheduling.

[0048] Furthermore, to achieve the above objectives, the present invention also provides a coal mine drainage monitoring and safety early warning control device based on sensor big data information. The device includes: a memory, a processor, and a coal mine drainage monitoring and safety early warning control program based on sensor big data information stored in the memory and executable on the processor. The coal mine drainage monitoring and safety early warning control program based on sensor big data information is configured to implement the steps of the coal mine drainage monitoring and safety early warning control method based on sensor big data information as described above.

[0049] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing a coal mine drainage monitoring and safety early warning control program based on sensor big data information. When the coal mine drainage monitoring and safety early warning control program based on sensor big data information is executed by a processor, it implements the steps of the coal mine drainage monitoring and safety early warning control method based on sensor big data information as described above.

[0050] This invention provides a coal mine drainage monitoring and safety early warning control method based on sensor big data information. This method analyzes interface features and formalizes parameters from multi-source sensor data, constructing multi-dimensional constrained data by combining sedimentation interface coordinates, sedimentation rate weights, and operational scenario types. This overcomes the limitations of traditional single-point detection methods, achieving high-precision, real-time dynamic monitoring of the spatial distribution and evolution trend of coal slime sedimentation within water tanks. Furthermore, it introduces a dynamic sedimentation potential field matrix based on a spatial interpolation attenuation function and integrates sedimentation morphology evolution terms and sediment erosion and diffusion terms to establish a dynamic prediction model. This model can accurately simulate the coal slime deposition process under complex working conditions and effectively predict future sedimentation hotspots. By analyzing morphological changes, a scientific basis for early intervention is provided. By calculating the synergistic coefficient of siltation impact and generating local dredging and waste rock allocation components, global prediction and local response are combined to construct a siltation monitoring and control field that is adaptive to the working conditions of the reservoir. This enables the system to have intelligent response capabilities to different operating states (such as high sediment content water and alternating dredging operations), improving the flexibility and applicability of the control strategy. Based on the control field and operating scenarios, a waste rock disposal operation strategy vector field is constructed. Combined with dynamic probability calculation, a graded early warning scheme is generated and transformed into real-time control commands. This realizes the transformation from passive dredging to active early warning, intelligent linkage, and closed-loop control, significantly improving the timeliness and resource utilization efficiency of waste rock disposal operations. Attached Figure Description

[0051] Figure 1 This is a flowchart illustrating an embodiment of the coal mine drainage monitoring and safety early warning control method based on sensor big data information of the present invention.

[0052] Figure 2 This is a structural block diagram of an embodiment of the coal mine drainage monitoring and safety early warning control system based on sensor big data information of the present invention.

[0053] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0054] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0055] Reference Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the coal mine drainage monitoring and safety early warning control method based on sensor big data information of the present invention.

[0056] In one embodiment, the coal mine drainage monitoring and safety early warning control method based on sensor big data information includes:

[0057] Step S100: The interface feature analysis and siltation parameter formalization processing of the real-time collected multi-source sensor data of the water tank are performed to generate multi-dimensional coal slime siltation constraint data. The multi-dimensional coal slime siltation constraint data includes siltation interface coordinates, siltation rate weights and water tank operation scenario types.

[0058] The multi-source sensor data of the water tank can be synchronously collected data from multiple sensors (such as ultrasound, pressure, turbidity, liquid level, etc.) on the internal state of the water tank, which can be used to provide raw observation information to support interface identification and parameter extraction. For example, the multi-source sensor data of the water tank can include one or more of acoustic echo data, hydrostatic pressure data, and optical turbidity data. Interface feature analysis is the process of identifying and extracting the spatial features of the coal slime-water interface from the multi-source sensor data, which can be used to solve the problem of insufficient resolution of a single sensor in ambiguous interface regions. Formal processing of sedimentation parameters can be the operation of converting interface features into standardized numerical parameters (such as coordinates, rate weights), which can be used to transform unstructured observation data into structured inputs that can be used for modeling. Furthermore, formal processing of sedimentation parameters can include formalization of geometric parameters, formalization of dynamic parameters, formalization of statistical parameters, etc. Sedimentation interface coordinates can be a set of coordinates describing the position of the coal slime-water interface in the water tank space, which can be used as a spatial reference point for potential field construction. The sedimentation rate weight can be a normalized index reflecting the trend of sedimentation thickness change at a certain location per unit time. It can be used as a driving factor for potential field strength, reflecting the degree of sedimentation activity. For example, the sedimentation rate weight can include historical moving average rate weight, real-time differential rate weight, and predicted correction rate weight. The water tank operation scenario type can be a classification label describing the current operation mode of the water tank, such as main tank operation, auxiliary tank dredging, and high sediment content water inflow. It can be used to constrain the applicable boundaries of prediction models and control strategies. Multi-dimensional coal slime sedimentation constraint data can be a structured dataset composed of sedimentation interface coordinates, sedimentation rate weights, and water tank operation scenario type. It can be used to characterize the multi-dimensional features of coal slime sedimentation state within the water tank.

[0059] The process of analyzing interface features and formalizing sedimentation parameters from real-time collected multi-source sensor data of the water tank involves simultaneously receiving data from multiple sensors, identifying the coal slime-water interface using signal fusion algorithms, and converting interface features into structured parameters such as coordinates and rates. Furthermore, this operation can be achieved by employing deep learning networks for multimodal signal fusion analysis and traditional image processing methods combining edge detection and threshold segmentation, thereby enabling refined extraction of complex underwater interfaces and improving parameter accuracy. Generating multi-dimensional coal slime sedimentation constraint data can be achieved by packaging the formalized sedimentation interface coordinates, sedimentation rate weights, and the identified water tank operation scenario type into a unified data structure. Further, this operation can be implemented by organizing three types of parameters in JSON format and embedding spatiotemporal condition labels as tensors, thus constructing a sedimentation state representation system with spatial, temporal, and conditional dimensions.

[0060] Step S200: Based on the sedimentation interface coordinates and sedimentation rate weights in the multi-dimensional coal slime sedimentation constraint data, a dynamic sedimentation potential field matrix reflecting the distribution of coal slime sedimentation potential is constructed by combining spatial interpolation attenuation functions.

[0061] The spatial interpolation attenuation function combination can be a set of mathematical functions with spatial attenuation characteristics, which can be used to interpolate between known points to obtain a continuous field. In a specific embodiment, the spatial interpolation attenuation function combination can include exponential attenuation interpolation functions, Gaussian kernel interpolation functions, power-law attenuation interpolation functions, etc. The dynamic sedimentation potential field matrix can be a two-dimensional or three-dimensional matrix generated based on the spatial interpolation attenuation function combination, which can be used to reflect the distribution intensity of coal slime sedimentation potential at various locations within the water reservoir. Further, the dynamic sedimentation potential field matrix can be reconstructed using the sedimentation interface coordinates as spatial anchor points, combined with sedimentation rate weights as source strength, and through spatial interpolation attenuation functions. For example, the dynamic sedimentation potential field matrix can include radial basis function potential fields, inverse distance weighted potential fields, kriging interpolation potential fields, etc. Constructing the dynamic sedimentation potential field matrix through the spatial interpolation attenuation function combination can be achieved by using the sedimentation interface coordinates as spatial nodes, sedimentation rate weights as source strength, and applying spatial interpolation attenuation functions to generate a continuous potential field throughout the entire reservoir domain. Furthermore, this operation can be achieved by using radial basis functions for global interpolation and by employing local inverse distance weighted interpolation combined with boundary constraints, thereby obtaining a siltation potential distribution with physical interpretability and spatiotemporal smoothness.

[0062] Step S300: Using the dynamic sedimentation potential field matrix as the driving source, establish a dynamic prediction model for coal slime sedimentation that includes sedimentation morphology evolution terms and sediment erosion and diffusion terms.

[0063] The sedimentation morphology evolution term can be a dynamic term in the model describing how historical sedimentation morphologies influence future depositional trends. It can be used to preserve the memory of the sedimentation process and improve long-term prediction stability. In an exemplary embodiment, the sedimentation morphology evolution term may include topographic-dependent evolution terms, sedimentary sequence evolution terms, and interface curvature feedback evolution terms. The sediment flushing and diffusion term can be a physical process term in the model simulating the resuspension, migration, and redeposition of sediment caused by water flow disturbance. It can be used to enhance the model's responsiveness to dynamic inflow conditions. The coal slime sedimentation dynamic prediction model can be a mathematical model that integrates the sedimentation morphology evolution term and the sediment flushing and diffusion term. It can be used to simulate the deposition and migration of coal slime within the water reservoir in future periods. For example, the coal slime sedimentation dynamic prediction model may include a continuous medium model based on partial differential equations, a discrete phase model based on particle tracking, or a hybrid-driven coupling model. Establishing a dynamic prediction model for coal slime deposition that includes terms on depositional morphology evolution and sediment erosion and diffusion can be achieved by using the dynamic depositional potential field matrix as the initial field and iteratively solving two dynamic equations describing the continuation of historical morphology and the migration of water flow disturbances. Furthermore, this operation can be implemented by solving partial differential equations using the finite difference method and simulating sediment movement using Lagrange particle tracking, thus accurately simulating depositional evolution paths under different inflow conditions.

[0064] Step S400: Calculate the siltation impact coordination coefficient based on the coal mine drainage system water tank volume assessment value and the current location siltation potential field intensity, generate a local dredging and gangue allocation component and integrate it with the coal slime siltation dynamic prediction model to obtain the water tank working condition adaptive siltation monitoring and control field.

[0065] The coal mine drainage system's water sump volume assessment value can be a quantitative indicator of the current effective water storage capacity of the sump, representing the net usable volume after considering siltation. This value can be used as a key input for calculating the siltation impact synergy coefficient. The current location's siltation potential field intensity can be a numerical value at a specific location in the dynamic siltation potential field matrix, representing the level of siltation potential at that point and reflecting the local siltation risk level, used for synergy coefficient calculation. Furthermore, the current location's siltation potential field intensity can include observed driving intensity, predicted enhancement intensity, and historical cumulative intensity. The siltation impact synergy coefficient can be a dimensionless index calculated by combining the volume assessment value and the potential field intensity, used to characterize the comprehensive impact of current siltation on system safety and function. For example, the siltation impact synergy coefficient can include volume compression synergy coefficient, suction port blockage risk synergy coefficient, and pump wear risk synergy coefficient. The local dredging and wastewater disposal allocation component can be a dredging priority and resource allocation suggestion for a specific area, generated based on the synergy coefficient. This can be used to achieve rapid response to high-risk areas and supplement local details of global prediction. In one exemplary embodiment, the local dredging and waste rock disposal allocation component may include equipment scheduling components, energy consumption optimization components, and operation sequence components. The adaptive siltation monitoring and control field for the water reservoir can be a comprehensive field that integrates the local dredging and waste rock disposal allocation component with the output of the coal slime siltation dynamic prediction model. It can be used to achieve coordination between global siltation prediction and local emergency dredging needs, forming a control benchmark that can be dynamically adjusted according to changes in operating conditions. Furthermore, the adaptive siltation monitoring and control field for the water reservoir can include a volume-constrained control field, a high-sand-concentration impact control field, and a dredging operation interference control field.

[0066] The synergy coefficient of siltation impact is calculated based on the assessed volume of the coal mine drainage system's water sump and the current location's siltation potential field intensity. This can be achieved by substituting the assessed volume and potential field intensity into a preset function (such as a weighted product or nonlinear mapping) to calculate a comprehensive impact index. Furthermore, this operation can be implemented by using fuzzy rule reasoning to calculate the synergy coefficient and by using a neural network regression model to fit the multidimensional input-output relationship, thereby quantifying the combined impact of siltation on system function and safety. Generating local dredging and waste rock allocation components and fusing them with the coal slime siltation dynamic prediction model can be achieved by determining the dredging priority of high-risk areas based on the synergy coefficient and injecting this information as a correction term into the prediction model output. Further, this operation can be achieved by superimposing Gaussian pulse-type allocation components into the prediction field and strengthening the weight of local areas through a masking mechanism, thereby achieving coupling between global prediction and local emergency response. The adaptive siltation monitoring and control field for the water sump operating conditions is obtained, which can be formed by integrating the fused prediction results and allocation components to create the final control reference field. Furthermore, this operation can be achieved by storing the control field in the form of a raster map and encoding the dredging direction and intensity in the form of a vector field, thereby constructing a monitoring and control basis that can be dynamically adjusted according to the operating status.

[0067] Step S500: Based on the adaptive siltation monitoring and control field of the water tank working condition and the water tank operation scenario type, construct the waste rock removal operation strategy vector field, generate an intelligent linkage waste rock removal early warning scheme through dynamic probability calculation, and convert it into real-time waste rock removal monitoring and control instructions to complete the closed-loop siltation management and scheduling.

[0068] The waste rock removal operation strategy vector field can be a directional field constructed based on the control field and the operation scenario type. Each location contains information on dredging priority and execution direction, which can be used to provide spatial guidance and task allocation basis for waste rock removal equipment, supporting hierarchical early warning and command generation. In a specific embodiment, the waste rock removal operation strategy vector field may include an efficiency priority vector field, a safety priority vector field, a resource balance vector field, etc. Dynamic probability calculation can be a probability model that calculates the probability of triggering different early warning levels based on the control field strength and scenario type. It can be used to support the generation of hierarchical early warning schemes and avoid false alarms and missed alarms. The intelligent linkage waste rock removal early warning scheme can be a structured plan that includes multi-level early warning thresholds, response measures, and linkage equipment, which can be used to provide differentiated response strategies for different risk levels. For example, the intelligent linkage waste rock removal early warning scheme may include linkage schemes corresponding to yellow attention warnings, orange intervention warnings, and red emergency warnings. Real-time waste rock removal monitoring and control commands can be specific operation commands converted from early warning schemes that can directly drive waste rock removal equipment to execute, which can be used to complete closed-loop control from decision-making to execution. In one exemplary embodiment, real-time wastewater discharge monitoring and control commands may include pump start / stop commands, valve opening adjustment commands, and dredging robot path commands.

[0069] Constructing a waste rock removal operation strategy vector field based on the adaptive siltation monitoring and control field of the water tank and the water tank operation scenario type can be achieved by mapping the spatial intensity and direction information of the control field to specific operation strategy vectors in combination with the scenario type. Furthermore, this operation can be implemented by using gradient descent direction as the siltation path guide and cluster centers as anchor points for waste rock removal operations, thus providing spatial guidance and task allocation basis for waste rock removal equipment. Generating an intelligent linkage waste rock removal early warning scheme through dynamic probability calculation can be based on the intensity distribution of the strategy vector field, using a probability model to calculate the trigger probability of different early warning levels, and matching preset response measures. Further, this operation can be achieved by training a Bayesian network based on historical data for probability inference and using a sliding window to statistically analyze abnormal frequencies to generate early warnings, thus achieving hierarchical and accurate early warning output. Converting these into real-time waste rock removal monitoring and control commands to complete closed-loop siltation management and scheduling can be achieved by parsing the measures in the early warning scheme into a sequence of commands executable by specific equipment and issuing them to the execution unit. Furthermore, this operation can be achieved by issuing industrial control commands through the OPCUA protocol and pushing device control commands through the MQTT message queue, thereby realizing closed-loop control from sensing to execution.

[0070] Taking a high-sand-concentration water inflow event during the alternating operation of the main and auxiliary storage compartments as an example, the coal mine drainage monitoring safety early warning control method based on sensor big data information in this embodiment can be as follows: When the main storage compartment is draining water while the auxiliary storage compartment is in standby mode, a sudden increase in the sand content of the roadway water inflow occurs, and multiple source sensors simultaneously detect the increase in turbidity and abnormal fluctuations in liquid level in the entrance area of ​​the auxiliary storage compartment; the system identifies the new siltation interface at the bottom of the auxiliary storage compartment through interface feature analysis, and generates multi-dimensional constraint data in combination with high-sand-concentration scenario labels; the dynamic siltation potential field matrix shows that the potential field intensity in the entrance area is rapidly increasing; the prediction model integrates the sediment erosion and diffusion term, predicting that the siltation will expand to the center within the next 2 hours; the decrease in the volume assessment value triggers the calculation of the coordination coefficient, generating a local dredging allocation component, prioritizing the area around the auxiliary storage compartment's water intake well; the control field is adjusted accordingly, and the strategy vector field points to the high-risk area; the dynamic probability calculation determines an orange warning, and automatically generates control commands to start the auxiliary storage compartment dredging pump group and adjust the diversion valve, realizing early intervention.

[0071] In one embodiment, the real-time collected multi-source sensor data of the water tank is subjected to interface feature analysis and formal processing of sedimentation parameters to generate multi-dimensional coal slime sedimentation constraint data, including:

[0072] Feature extraction and semantic annotation were performed on the multi-source sensor data of the water tank to obtain a set of siltation elements containing siltation interface identifiers, siltation intensity descriptors, and operation constraint descriptors.

[0073] Feature extraction can be the process of identifying and separating signal patterns or statistics related to siltation status from multi-source sensor data of a water tank. This can provide basic input for semantic annotation, improving the accuracy of subsequent semantic recognition. Semantic annotation can be the operation of assigning engineering semantic labels to the extracted features, which can be used to establish a mapping relationship between physical observations and engineering concepts. Further, semantic annotation can include interface semantic annotation, intensity level annotation, and working condition constraint annotation. A siltation interface identifier can be a semantic label used to uniquely identify the type or state of the coal slime-water interface, which can be used to support subsequent operational scenario classification and knowledge base matching. For example, a siltation interface identifier can include a clear layered interface identifier, a diffuse transition interface identifier, and a disturbed redistribution interface identifier. A siltation intensity descriptor can be a semantic indicator characterizing the degree of siltation activity or density in a local area, which can be used as a semantic basis for generating siltation rate weights. In a specific embodiment, the siltation intensity descriptor can include a deposition rate intensity descriptor, a particle concentration intensity descriptor, and a compaction density intensity descriptor. Job constraint descriptors can be semantic tags that reflect the limitations or impacts of current operating conditions on dredging operations. They can be used to assist in determining the type of water tank operation and setting the boundaries of control strategies. Furthermore, job constraint descriptors can include equipment availability constraint descriptors, water flow stability constraint descriptors, safety threshold constraint descriptors, etc.

[0074] The siltation element set can be a structured data unit composed of siltation interface identifiers, siltation intensity descriptors, and operational constraint descriptors, which can be used to express engineering semantic information in sensor data. In this embodiment, the siltation element set can transform the original physical quantity into a semantic unit with computability and reasoning ability, supporting subsequent knowledge base mapping and parameter standardization. For example, the siltation element set can include point-like siltation elements, region-continuous siltation elements, and dynamically disturbed siltation elements. Feature extraction and semantic annotation of multi-source sensor data from a water tank yields a siltation element set containing siltation interface identifiers, siltation intensity descriptors, and operational constraint descriptors. This can be achieved by performing pattern recognition and semantic assignment on the original sensor signals, generating three types of descriptors and combining them into a structured element set. Furthermore, this operation can be implemented by automatically extracting features using a convolutional neural network and attaching a classification head for semantic annotation, or by manually designing features and combining them with decision tree rules for semantic annotation. This enables the transformation from original physical quantities to engineering semantic units, improving data interpretability and robustness of subsequent processing.

[0075] The set of sedimentation elements is input into the pre-trained coal mine water sump sedimentation knowledge base for mapping and calibration to obtain a standardized combination of sedimentation parameters.

[0076] The pre-trained coal mine water sump siltation knowledge base can be a semantic mapping database constructed by integrating historical dredging records, geological and hydrological characteristics, and equipment operating parameters. It can be used to calibrate and standardize siltation elements. In one specific embodiment, the pre-trained coal mine water sump siltation knowledge base can collect long-term operating data from multiple mines offline and organize semantic relationships using knowledge graphs or embedded vectors. Furthermore, the pre-trained coal mine water sump siltation knowledge base can solve the data incomparability problem caused by sensor heterogeneity and environmental noise, achieving data semantic alignment across operating conditions and equipment. For example, the pre-trained coal mine water sump siltation knowledge base can include rule-driven knowledge bases, vector embedding knowledge bases, graph neural network knowledge bases, etc. Mapping calibration can be a process of semantically aligning and parameter correcting the set of siltation elements with the pre-trained coal mine water sump siltation knowledge base. It can be used to eliminate sensing bias and output parameter combinations that conform to a unified standard. In an exemplary embodiment, mapping calibration can include nearest neighbor semantic matching calibration, embedded spatial projection calibration, rule reasoning calibration, etc.

[0077] Standardized sedimentation parameter combinations can be a set of parameters with unified dimensions and semantic consistency, output after knowledge base mapping and calibration. This set can serve as a reliable source for extracting coordinate and weight values. Furthermore, standardized sedimentation parameter combinations can include geometric-dynamic coupling parameter combinations, spatiotemporal-working-condition joint parameter combinations, and multi-scale fusion parameter combinations. Inputting the sedimentation element set into a pre-trained coal mine water sump sedimentation knowledge base for mapping and calibration yields standardized sedimentation parameter combinations. This can be achieved by matching or projecting semantic descriptors with historical patterns in the knowledge base, correcting parameter biases, and outputting standardized results. Further, this operation can be implemented through semantic matching calibration via knowledge graph entity links, or by calculating similarity using a pre-trained embedding model and weighted averaging to output standardized parameters. This can address the data drift problem caused by sensor heterogeneity and environmental interference.

[0078] Based on the standardized combination of sedimentation parameters, the coordinates of the sedimentation interface and the weight of the sedimentation rate are extracted to obtain the coordinate values ​​and weight values.

[0079] The coordinate values ​​can be specific spatial location values ​​resolved from the standardized sedimentation parameter combination, and can be used to construct the sedimentation interface coordinates. The weight values ​​can be normalized values ​​representing the sedimentation rate trend resolved from the standardized sedimentation parameter combination, and can be used to construct sedimentation rate weights. Furthermore, the weight values ​​can include weight values ​​based on intensity descriptor mapping, weight values ​​derived from historical change rates, weight values ​​adjusted based on prediction confidence, etc. Extracting the sedimentation interface coordinates and sedimentation rate weights based on the standardized sedimentation parameter combination yields coordinate values ​​and weight values, which can be specific values ​​of the corresponding spatial location and temporal evolution indicators resolved from the standardized parameters. In an exemplary embodiment, this operation can be achieved by directly reading the coordinate and weight values ​​from the parameter fields, or by decoupling the coordinates and weights from the composite parameters through a regression model, thereby obtaining accurate input parameters that can be used for potential field construction.

[0080] The water tank operation scenario is classified based on the siltation interface identifier to obtain the water tank operation scenario type, and then combined with the siltation interface coordinates and siltation rate weight to obtain multi-dimensional coal slime siltation constraint data.

[0081] The water sump operation scenario classification processing can be an operation based on the siltation interface identifier to determine the current water sump operation mode, which can be used to output the water sump operation scenario type for constructing multi-dimensional constraint data. In a specific embodiment, the water sump operation scenario classification processing can include rule engine classification processing, support vector machine classification processing, graph neural network classification processing, etc. Combining the water sump operation scenario type with the siltation interface coordinates and siltation rate weights to obtain multi-dimensional coal slime siltation constraint data can be achieved by integrating the three types of parameters into a unified data object according to a preset structure. Furthermore, this operation can be implemented by encapsulating the three types of parameters in the form of key-value pairs or encoding the three types of information separately in the form of tensor channels, thereby forming a high semantic density input with spatial, temporal, and operational dimensions.

[0082] Taking a high-noise environment with heterogeneous deployment of multiple sensors as an example, the coal mine drainage monitoring and safety early warning control method based on sensor big data information in this embodiment can be as follows: A mine water tank is simultaneously deployed with an ultrasonic array, a turbidity meter, and a pressure gradient sensor. Due to roadway vibration, some ultrasonic echoes are distorted. The system first extracts features from the multi-source data, identifies the turbidity abrupt change zone and the pressure abnormal gradient zone, and generates a diffusion transition interface identifier, a high concentration intensity descriptor, and an equipment vibration constraint descriptor through semantic annotation. This set of sedimentation elements is input into a pre-trained coal mine water tank sedimentation knowledge base. Based on historical records of similar working conditions, the knowledge base calibrates the distorted ultrasonic data to a reliable interface position and outputs a standardized parameter combination. Accurate coordinate values ​​and weight values ​​are extracted from it, and the current scenario is determined to be a high-sediment-laden water scenario based on the interface identifier. Finally, the three are combined to form multi-dimensional coal slime sedimentation constraint data to support the subsequent construction of a high-precision potential field.

[0083] In one embodiment, based on the sedimentation interface coordinates and sedimentation rate weights in multi-dimensional coal slime sedimentation constraint data, a dynamic sedimentation potential field matrix reflecting the distribution of coal slime sedimentation potential is constructed by combining spatial interpolation decay functions, including:

[0084] Based on the coordinates of the sedimentation interface in the multi-dimensional coal slime sedimentation constraint data, a spatial attenuation sedimentation contribution function is constructed for each sedimentation interface point to obtain the sedimentation potential field distribution of a single-point region.

[0085] In this system, each sedimentation interface point can be a discrete sedimentation location within the water reservoir, determined by the coordinates of the sedimentation interface. These points can serve as the source points for the spatially decaying sedimentation contribution function, driving the generation of a local potential field. The spatially decaying sedimentation contribution function can be a mathematical function centered on a single sedimentation interface point, decreasing with spatial distance. It can be used to characterize the range and intensity of the point's influence on the sedimentation potential of the surrounding area. Furthermore, the spatially decaying sedimentation contribution function can simulate the natural sedimentation and diffusion of coal slime particles in still water. The sedimentation potential field distribution in a single-point area can be a local influence field generated by a single sedimentation interface point through the spatially decaying sedimentation contribution function. This can be used to describe the potential influence range and intensity of single-point sedimentation on the adjacent area. Constructing the spatially decaying sedimentation contribution function for each sedimentation interface point based on the sedimentation interface coordinates in the multi-dimensional coal slime sedimentation constraint data can be achieved by using each sedimentation interface coordinate as the source point and selecting a function form with spatial decay characteristics to construct a local influence model. Furthermore, this operation can be achieved by constructing contribution functions at each point using a Gaussian kernel function and an inverse distance squared decay function, thus enabling discrete observations to possess physically reasonable spatial extrapolation capabilities. Obtaining the sedimentation potential field distribution in a single-point region can be achieved by calculating the function value distribution of the spatially decaying sedimentation contribution function within its local neighborhood. Furthermore, this operation can be achieved by sampling within a preset radius to generate a potential field image and directly outputting the function value field through an analytical expression, thereby forming a continuous local potential field centered on a single point.

[0086] Based on the sedimentation rate weight, the single-point area sedimentation potential field distribution of each sedimentation interface point is weighted by sedimentation priority to obtain the coal slime sedimentation priority potential field.

[0087] The sedimentation priority weighting process can be used to modulate the amplitude of the sedimentation potential field distribution in a single-point region using sedimentation rate weights. This can be used to enable high sedimentation rate regions to obtain a higher response in the potential field and enhance the sensitivity to time evolution. The coal slime sedimentation priority potential field can be the single-point region sedimentation potential field distribution after weighting by sedimentation rate weights, which can be used to reflect the differences in activity levels of each sedimentation point in the time dimension. Furthermore, the coal slime sedimentation priority potential field can be the priority modulated potential field corresponding to each sedimentation point after weighting. Performing sedimentation priority weighting on the single-point region sedimentation potential field distribution at each sedimentation interface point according to the sedimentation rate weights can be done by using the sedimentation rate weights as multiplicative factors to scale the amplitude of the single-point potential field distribution. Further, this operation can be achieved by using linear multiplication for weighting or by using nonlinear mapping with a saturation upper limit, thereby enhancing the performance of high sedimentation rate regions in the overall potential field. The coal slime sedimentation priority potential field can be obtained by outputting the priority modulated potential field corresponding to each sedimentation point after weighting. Furthermore, this operation can be achieved by storing the priority potential fields of each point in tensor form and retaining only the salient regions in sparse field form, thereby generating a local potential field containing time evolution information.

[0088] The multiple coal slime deposition priority potential fields are superimposed and combined according to the spatial location of the water tank to obtain the coal slime deposition potential distribution of the entire water tank.

[0089] The spatial location of the water reservoir can be the geometric positioning information of each point inside the water reservoir in a unified coordinate system, which can be used as a spatial alignment reference for the superposition of multiple priority potential fields. The distribution of coal slime deposition potential in the entire water reservoir can be a continuous potential field covering the entire water reservoir space, formed by superimposing all coal slime deposition priority potential fields. This can be used to achieve a fusion expression from local perception to global deposition risk distribution, providing a foundation for subsequent matrixing. Superimposing and combining multiple coal slime deposition priority potential fields according to the spatial location of the water reservoir can be done by superimposing all priority potential fields point by point in a unified spatial coordinate system. Furthermore, this operation can be achieved by using algebraic summation superposition or maximum value coverage superposition, thereby fusing local information to form a global potential distribution. Obtaining the coal slime deposition potential distribution in the entire water reservoir can be achieved by outputting the superimposed continuous full-field distribution result. Furthermore, this operation can be achieved by outputting in the form of a floating-point image or in the form of a three-dimensional scalar field, thereby obtaining a continuous deposition potential field covering the entire water reservoir.

[0090] Based on the water tank operation scenario type, the attenuation range parameter is set and the distribution of coal slime siltation potential in the full water tank is matrixed to obtain the dynamic siltation potential field matrix.

[0091] The attenuation range parameter can be an adjustable parameter controlling the radius of action of the spatial attenuation sedimentation contribution function. Its value is affected by the type of water sump operation scenario and can be used to dynamically adjust the potential field influence range according to the working conditions, such as expanding during main sump operation and shrinking during dredging. Matrixing can be an operation that maps the continuous full-sump coal slime sedimentation potential distribution into a numerical matrix on a regular grid. It can be used to generate a structured data format, facilitating subsequent model input and calculation. Setting the attenuation range parameter based on the water sump operation scenario type and performing matrixing on the full-sump coal slime sedimentation potential distribution can be achieved by selecting the corresponding attenuation range parameter according to the identified operation scenario (such as main sump operation or dredging operation) and discretizing the continuous field into a regular matrix. Furthermore, this operation can be achieved by using a fixed-resolution raster for matrixing and adaptively adjusting the grid density according to the potential field gradient, thereby realizing a working condition-adaptive structured expression of the potential field. The resulting dynamic sedimentation potential field matrix can be the final structured matrix output as input to the subsequent prediction model. Furthermore, this operation can be achieved by storing the planar potential field in the form of a two-dimensional array and the three-dimensional potential field in the form of a three-dimensional voxel matrix, thereby generating standard input data with spatial continuity, time sensitivity and operating condition adaptability.

[0092] For example, in a scenario where the secondary storage compartment is in standby mode and about to switch to the primary storage compartment, the coal mine drainage monitoring safety early warning control method based on sensor big data information in this embodiment can be as follows: The system detects multiple siltation interface points at the bottom of the secondary storage compartment, which have low siltation rate weights but are widely distributed; according to the standby storage compartment scenario type, a small attenuation range parameter is set to focus on local disturbances; after each point generates a spatial attenuation siltation contribution function, a weak priority potential field is formed by low-weighting; after superposition, the potential distribution of the entire storage compartment is obtained, showing that although there are no high-intensity hot spots around the water intake well, there is a uniform thin layer of siltation; after matrix processing, a dynamic siltation potential field matrix is ​​generated, which is used by the prediction model to determine whether secondary deposition will be caused by water flow disturbance after the switch, thereby deciding whether to start local dredging in advance.

[0093] In one embodiment, the dynamic sedimentation potential field matrix is ​​used as the driving source to establish a dynamic prediction model for coal slime sedimentation that includes sedimentation morphology evolution terms and sediment erosion and diffusion terms, including:

[0094] The sedimentation gradient components are calculated based on the dynamic sedimentation potential field matrix to obtain the sedimentation driving force vector pointing to the region of high sedimentation intensity.

[0095] The sedimentation gradient component can be a vector component composed of the first-order partial derivatives of the dynamic sedimentation potential field matrix in various spatial directions. It reflects the spatial rate of change of sedimentation potential and can be used to construct the driving force direction pointing to areas of high sedimentation intensity, guiding the prediction model to focus on areas with significant sedimentation growth trends. In this embodiment, the sedimentation gradient component can be obtained by performing spatial gradient calculations on the dynamic sedimentation potential field matrix. A high sedimentation intensity area can be a local region in the dynamic sedimentation potential field matrix where the value is significantly higher than the neighborhood, representing a location prone to future sedimentation hotspots. It can be used as the target direction of the sedimentation driving force vector, guiding the model to focus its resources. For example, a high sedimentation intensity area can include inlet impact sedimentation areas, areas of rapid velocity drop sedimentation, and backflow vortex accumulation areas. Calculating the sedimentation gradient component based on the dynamic sedimentation potential field matrix to obtain the sedimentation driving force vector pointing to high sedimentation intensity areas can be achieved by performing spatial gradient calculations on the dynamic sedimentation potential field matrix, extracting the gradient direction and magnitude at each point, and constructing a driving force vector field. Furthermore, this operation can be achieved by using the Sobel operator to perform discrete gradient calculations and using the finite difference method to solve the continuous gradient field, thereby introducing a spatial gradient-driven mechanism that enables the model to actively identify regions with significant siltation growth trends.

[0096] Gradient calculation of the sedimentation field is performed based on the distribution density of sediment content in different areas of the reservoir to obtain the sedimentation morphology evolution term to prevent local sedimentation estimation bias.

[0097] The distribution density of sediment content in inflow water in different areas of the reservoir can be a spatial distribution function describing the sediment concentration flowing into each inlet point or zone of the reservoir per unit time. This can be used as a basis for correcting the sedimentation morphology evolution term, compensating for deviations in deposition rate caused by differences in water source. In an exemplary embodiment, the distribution density of sediment content in inflow water in different areas of the reservoir can be constructed using real-time monitoring data from multi-source sensors. Furthermore, the distribution density of sediment content in inflow water in different areas of the reservoir can include the sediment content density of water flowing into the main tunnel, the sediment content density of water flowing into branch tunnels, and the sediment content density of dredging return water. Based on the distribution density of sediment content in inflow water in different areas of the reservoir, gradient calculation is performed on the sedimentation field to obtain a sedimentation morphology evolution term that prevents local sedimentation estimation errors. This can be achieved by using the inflow sediment content distribution density as a spatial weight to perform a weighted gradient calculation on the sedimentation field, generating the corrected evolution term. Furthermore, this operation can be achieved by using convolution kernels to fuse sediment density and sedimentation field for local gradient correction, and by embedding sediment density gradient terms into variable coefficient partial differential equations. This can compensate for the deposition rate estimation bias caused by differences in local water flow conditions and improve the accuracy of spatial modeling.

[0098] Based on the requirement of continuity in the coal slime deposition process, the coal slime transport is subjected to diffusion treatment to obtain a sediment erosion diffusion term that suppresses fluctuations in the estimation of deposition morphology.

[0099] The continuity requirement of the coal slime deposition process can be a physical constraint that the mass and motion state of coal slime must not change abruptly during deposition. This can be used to ensure that sediment transport simulation conforms to the basic laws of fluid mechanics and avoids non-physical oscillations. In a specific embodiment, the continuity requirement of the coal slime deposition process is manifested as a constraint on the smoothness of the spatiotemporal changes in the concentration field. Coal slime transport can be the migration process of coal slime particles in water bodies driven by water flow, used as the modeling object for sediment scouring and diffusion terms, reflecting sediment resuspension and diffusion behavior. For example, coal slime transport can include suspended sediment transport, bedload transport, mixed phase transport, etc. Diffusion processing can be an operation that applies mathematical diffusion operators to the coal slime transport process to simulate particle dispersion effects. This can be used to smooth high-frequency noise in the prediction results and suppress morphological estimation fluctuations. Further, diffusion processing can include Fickian diffusion processing, turbulent eddy diffusion processing, random walk diffusion processing, etc. To ensure the continuity of the coal slime deposition process, a diffusion treatment is applied to the coal slime transport, resulting in a sediment erosion diffusion term that suppresses fluctuations in the estimated deposition morphology. This can be achieved by introducing a diffusion operator into the coal slime transport equation to simulate the dispersion behavior of particles under water flow disturbance. Furthermore, this operation can be further implemented by adding a Laplacian diffusion term to the transport equation and employing an artificial viscosity term to achieve numerical diffusion stabilization. This suppresses morphological oscillations in the prediction results caused by sampling noise or discretization errors, thereby enhancing temporal stability.

[0100] By solving the mass conservation equation for the sedimentation driving force vector, sedimentation morphology evolution term, and sediment erosion and diffusion term, a dynamic prediction model for coal slime sedimentation describing the dynamic deposition of coal slime under complex working conditions in a water tank is obtained.

[0101] The mass conservation equation can be a partial differential equation describing the constant mass of coal slime in the spatiotemporal domain. It takes the form that the sum of the partial derivative of concentration with respect to time and the concentration flux divergence equals the source-sink term. This equation can be used as a coupled solution framework to ensure that the deposition driving force, evolution term, and diffusion term work synergistically under unified physical constraints. In a specific embodiment, the mass conservation equation is numerically approximated using a discrete grid system. The dynamic deposition of coal slime under complex conditions in a water-filled reservoir can be considered as the spatiotemporal deposition behavior of coal slime under the influence of multiple factors such as sediment concentration fluctuations, water flow disturbances, and dredging operations. This can be used as the final output target of the prediction model to characterize the deposition evolution process under real-world conditions. For example, the dynamic deposition of coal slime under complex conditions in a water-filled reservoir can include high-sediment-concentration steady-flow deposition, intermittent pulsed water flow deposition, and recovery deposition after dredging disturbances. By solving the mass conservation equations for the sedimentation driving force vector, sedimentation morphology evolution term, and sediment erosion and diffusion term, a dynamic prediction model for coal slime deposition under complex water-filled conditions is obtained. This model can be achieved by substituting the three terms as source or flux terms into the mass conservation equations and solving the spatiotemporal evolution process numerically. Furthermore, this operation can be achieved by discretizing the conservation equations using the finite volume method and simulating the multiphase transport process using the lattice Boltzmann method, thus ensuring the model's physical consistency and accurately characterizing the entire coal slime migration-deposition process under complex conditions.

[0102] Taking the sudden high-sediment-content water inrush at the main silt inlet as an example, the dynamic prediction model for coal slime deposition in this embodiment can be as follows: When the sediment content at the main silt inlet increases sharply due to geological activity, multi-source sensor data show a surge in turbidity in the area; the dynamic deposition potential field matrix generated by the system shows a high-intensity peak near the inlet; by calculating the deposition gradient component, the deposition driving force vector pointing to the area is obtained; at the same time, based on the measured sediment content distribution density of the incoming water, the deposition field is weighted gradient corrected to generate a deposition morphology evolution term that is closer to reality; diffusion processing smooths out the prediction peak caused by the instantaneous high concentration; the three are jointly input into the mass conservation equation for solution, and the prediction model accurately predicts the process of sediment diffusion and gradual sedimentation along the water flow direction, providing early warning of the secondary deposition zone that may form around the downstream intake well.

[0103] In one embodiment, a siltation impact synergy coefficient is calculated based on the assessed volume of the coal mine drainage system's water sump and the current location's siltation potential field intensity. A local dredging and waste rock allocation component is generated and fused with the coal slime siltation dynamic prediction model to obtain an adaptive siltation monitoring and control field for the water sump's operating conditions, including:

[0104] The siltation impact coordination coefficient of each water tank zone is obtained by weighted multiplication calculation based on the water tank volume assessment value and the current location siltation potential field intensity.

[0105] The weighted product operation can be a mathematical operation that multiplies the assessed volume of the water tank by the current location's sedimentation potential field intensity according to preset weights. This can be used to quantify the comprehensive impact of sedimentation in each zone on the overall safety and function of the system. In this embodiment, the weighted product operation can be performed on each water tank zone, using its assessed volume and the corresponding location's sedimentation potential field intensity as input, according to preset weight parameters. The sedimentation impact coordination coefficient of each water tank zone can be a dimensionless index obtained through the weighted product operation, characterizing the comprehensive risk of sedimentation in a single water tank zone, and can be used as a priority basis for the allocation of local wastewater discharge resources. For example, the sedimentation impact coordination coefficient of each water tank zone can reflect the potential threat level of that zone to the system's operational safety under the current operating conditions. In an exemplary embodiment, the sedimentation impact coordination coefficient of each water tank zone can include, but is not limited to, a volume compression-dominated coordination coefficient, a suction port blockage-dominated coordination coefficient, and a pump wear-dominated coordination coefficient. The siltation impact coordination coefficient for each water tank zone is obtained by performing a weighted product calculation based on the assessed volume of the water tank and the potential field intensity at the current location. This can be achieved by multiplying the assessed volume of each water tank zone by the potential field intensity at the corresponding location according to a preset weight. Furthermore, this operation can be implemented using a fixed weight α × assessed volume × (1 - β × potential field intensity), where α is the weight coefficient for the volume holding capacity and β is the siltation risk sensitivity coefficient, or by dynamically outputting weight parameters using a working condition identification module. This allows for the coupling of the system's remaining capacity with local siltation risk, quantifying the overall threat level.

[0106] Based on the synergy coefficient of siltation impact and the radius of action of dredging operations, a local waste rock disposal and allocation field is constructed to obtain the siltation distribution of each water storage zone;

[0107] The dredging operation radius can be the radius of the spatial range that a single dredging operation can effectively cover. It can be used to constrain the construction boundary of the local waste rock disposal site, ensuring the physical accessibility of resource scheduling. In a specific embodiment, the dredging operation radius, together with the siltation impact synergy coefficient, determines the spatial expansion range of the local waste rock disposal site. For example, the dredging operation radius can include, but is not limited to, the dredging radius of pump suction, mechanical scraper dredging, and hydraulic flushing dredging. The local waste rock disposal site can be a spatial field generated based on the synergy coefficient and the radius of action, describing the intensity of dredging demand in each zone. It can be used to transform abstract risk indicators into an operation guidance field with a spatial range of action. Furthermore, the local waste rock disposal site can be a continuous field distribution formed by spatial diffusion within its radius of action, with the synergy coefficient as the source intensity. In an exemplary embodiment, the local waste rock disposal site can include, but is not limited to, a Gaussian kernel convolutional disposal field, a circular mask superposition disposal field, and a Voronoi partitioned disposal field. The distribution of sediment accumulation in each water reservoir zone can be mapped from the local waste rock disposal and allocation field, reflecting the spatial distribution of sediment accumulation that needs to be processed in each zone. This can be used to provide quantitative input for the calculation of allocation components. In this embodiment, the distribution of sediment accumulation in each water reservoir zone can be obtained by normalizing or scaling the local waste rock disposal and allocation field. For example, the distribution of sediment accumulation in each water reservoir zone can include, but is not limited to, sediment accumulation calculated from a prediction model, sediment accumulation normalized by potential field intensity, and sediment accumulation calculated from volumetric loss.

[0108] Based on the siltation impact synergy coefficient and the dredging operation radius, a local wastewater allocation field is constructed to obtain the siltation distribution in each water storage zone. This can be achieved by using the synergy coefficient as the source intensity and its radius as the influence range, generating a continuous allocation field through spatial diffusion or masking expansion. Furthermore, this operation can be implemented by using a circular Gaussian kernel for convolution expansion at the synergy coefficient location, or by using polygonal influence regions for Boolean superposition, thereby transforming discrete risk indicators into spatially accessible operation guidance distributions.

[0109] The distribution of silt was adjusted and the allocation of local silt removal and waste rock disposal was calculated to prevent excessive or untimely waste rock disposal operations.

[0110] The allocation component calculation can be an operation that normalizes, thresholds, or smooths the distribution of siltation to generate executable dredging commands. This can be used to prevent excessive or untimely waste removal operations and achieve precise resource allocation. In one specific embodiment, the allocation component calculation can suppress ineffective fluctuations by applying upper and lower limit constraints or dynamic adjustment factors to the siltation distribution. For example, the allocation component calculation can include, but is not limited to, energy consumption minimization allocation calculation, timeliness maximization allocation calculation, and equipment lifespan balancing allocation calculation. The local dredging and waste removal allocation component to prevent excessive or untimely waste removal operations can be a dredging intensity vector output after allocation component calculation, possessing anti-over-dredging and anti-lag characteristics. This can be used to balance dredging efficiency and resource consumption, avoiding ineffective or delayed intervention. In an exemplary embodiment, the local dredging and waste removal allocation component to prevent excessive or untimely waste removal operations can include, but is not limited to, hysteresis control type allocation components, sliding mode control type allocation components, and fuzzy rule type allocation components.

[0111] The distribution of silt accumulation is analyzed to calculate localized dredging and wastewater removal allocation components that prevent excessive or untimely wastewater removal operations. This can be achieved by applying upper and lower thresholds, smoothing filters, or dynamic adjustment factors to the silt accumulation distribution, generating a final executable allocation intensity. Furthermore, this operation can be implemented by introducing a hysteresis controller to set start / stop thresholds or by using an exponential decay function to suppress high-frequency fluctuations, thereby avoiding uncontrolled silt accumulation due to insufficient dredging or resource waste due to excessive dredging.

[0112] By superimposing and fusing the local dredging and waste rock allocation components with the coal slime siltation dynamic prediction model, an adaptive siltation monitoring and control field for water tank conditions with both global morphological prediction and local operation coordination functions is obtained.

[0113] The siltation vector superposition and fusion can be an operation that superimposes local dredging and waste rock allocation components into the output field of the coal slime siltation dynamic prediction model in vector form. This can be used to inject local operation feedback while preserving the global prediction morphology. In this embodiment, the siltation vector superposition and fusion can be a point-by-point vector synthesis of the prediction field and allocation components in the same coordinate system. For example, the siltation vector superposition and fusion can include, but is not limited to, linear weighted superposition, nonlinear gated superposition, and spatiotemporally aligned superposition. The global morphology prediction can be a forward-looking simulation result of the coal slime siltation dynamic prediction model on the overall siltation evolution trend of the water tank, which can be used to provide long-term, large-scale siltation development prediction. Furthermore, the global morphology prediction can be the output of a time-series model driven by historical siltation data. The local operation coordination function can be an embedded capability in the control field that can respond to real-time dredging needs and adjust local strategies. This can be used to achieve dynamic coupling between prediction and execution, and improve the system's closed-loop control capability. For example, the local operation coordination function can be reflected by the injection of local dredging and waste rock allocation components. In one exemplary embodiment, the local operation collaboration function may include, but is not limited to, equipment status feedback collaboration, operation progress tracking collaboration, and real-time energy consumption optimization collaboration.

[0114] By superimposing and fusing the local dredging and waste rock allocation component with the dynamic prediction model of coal slime siltation, an adaptive siltation monitoring and control field for water tank conditions is obtained, which combines global morphological prediction and local operation coordination functions. This can be achieved by using the allocation component as a correction vector and superimposing it point-by-point with the siltation field output by the prediction model in the same coordinate system. Furthermore, this operation can be achieved by using a weighted average to fuse the prediction field and the allocation field, or by covering the prediction value with the allocation component in key areas, thereby achieving an organic unity between global prediction and local response, forming an adaptive control benchmark for operating conditions.

[0115] For example, in a scenario where the main silo experiences a sudden influx of water with high sediment content during the dredging of the auxiliary silo, the coal mine drainage monitoring and safety early warning control method based on sensor big data information in this embodiment can be as follows: The system detects a sudden increase in the sediment content of the water entering the main silo, and the dynamic prediction model shows that the central part of the main silo will rapidly accumulate sediment in the future; at the same time, the auxiliary silo is undergoing manual dredging, and its volume assessment value is low, but the potential field strength has decreased. For each zone of the main silo, the system performs a weighted product based on the volume assessment value (still relatively high) and the high potential field strength to obtain a high coordination coefficient; combined with the 3-meter radius of action of the dredging pump group, a local waste rock allocation field is constructed to identify the 5-meter × 5-meter area that needs to be prioritized; after calculating the allocation component, a medium-intensity dredging command is output to avoid energy waste caused by full-power operation; this allocation component is superimposed with the output of the prediction model to form a new control field, automatically increasing the priority of the main silo dredging, while the auxiliary silo is temporarily downgraded due to its low coordination coefficient, realizing dynamic resource reallocation.

[0116] In one embodiment, a local waste rock allocation field is constructed based on the siltation impact synergy coefficient and the dredging operation radius to obtain the siltation distribution of each water storage zone, including:

[0117] The radial distance of the water tank space is normalized based on the siltation effect synergy coefficient to obtain the distance synergy factor;

[0118] The spatial radial distance of the water tank can be the Euclidean distance from any location within the water tank to a designated dredging center point. This distance can be used as a basic geometric parameter for normalization, reflecting the relative distance between the spatial location and the work source point. Furthermore, the spatial radial distance of the water tank can include, but is not limited to, one or more of the following: the radial distance from the main pump inlet, the radial distance from the current position of the dredging robot, and the radial distance from the center of the inlet. Normalization can be a mathematical transformation that scales the spatial radial distance of the water tank to the [0, 1] interval according to the maximum effective range or tank size. This can be used to eliminate the influence of water tank size differences on the calculation of the synergy factor and improve the model's generalization ability. In an exemplary embodiment, normalization can be implemented using methods such as normalization of the effective radius, normalization of the tank diameter, and normalization of the number of grid cells. The distance synergy factor can be a spatial weighting factor generated by coupling the normalized radial distance with the siltation impact synergy coefficient. This can be used to characterize the effective risk contribution of a location after considering geometric distance. For example, the distance synergy factor can include, but is not limited to, one or more of the following: linear modulation distance synergy factor, exponential modulation distance synergy factor, and piecewise threshold distance synergy factor. Based on the siltation impact synergy coefficient, the radial distance of the water tank space is normalized to obtain the distance synergy factor. This can be obtained by dividing the radial distance of each point by the radius of the dredging operation or the characteristic scale of the tank body, and combining it with the synergy coefficient to generate a normalized weight. This allows the risk assessment of water tanks of different sizes to have scale invariance.

[0119] The spatial siltation attenuation coefficient is obtained by setting the waste rock allocation range based on the dredging operation radius and performing power-law attenuation calculation on the distance coordination factor.

[0120] The waste rock disposal and allocation domain can be a spatial area where the dredging equipment can effectively operate, defined by the radius of the dredging operation. This domain can constrain the effective range of the power-law attenuation calculation, preventing ineffective areas from participating in the allocation. In a specific embodiment, the waste rock disposal and allocation domain can be one or more of the following: a spherical domain, a cylindrical domain, and a fan-shaped directional domain. The power-law attenuation calculation can be a mathematical operation that attenuates the distance coordination factor by a negative power of the distance. This can be used to simulate the engineering law that dredging efficiency naturally decreases with increasing distance. For example, the power-law attenuation calculation can be one or more of the following: inverse square attenuation, inverse cubic attenuation, and adaptive exponential attenuation. The spatial siltation attenuation coefficient can be a coefficient obtained after power-law attenuation calculation, reflecting the modulation effect of dredging accessibility on siltation intensity. This coefficient can be used to reflect the spatial non-uniformity of the equipment's operating capacity. The spatial siltation attenuation coefficient is obtained by setting the sludge distribution range based on the dredging operation radius and performing power-law attenuation calculations on the distance coordination factor. This coefficient can be calculated by applying an attenuation function of the form r^-α to the normalized distance within the radius of operation, setting it to zero beyond the radius. Here, r is the normalized distance variable, representing the relative distance between the target location (siltation point) and the operation center point of the dredging equipment; α is the attenuation exponent factor, representing the rate or sensitivity of dredging efficiency decreasing with increasing distance, i.e., the exponent in the power-law function. Furthermore, this operation can be achieved by using a fixed exponent α = 2 with inverse square attenuation, dynamically adjusting the α value according to the equipment type. This simulates the physical reality of dredging efficiency decreasing with distance, improving the rationality of coordination.

[0121] The siltation intensity value at each spatial location of the reservoir is obtained by multiplying the siltation impact synergy coefficient and the spatial siltation attenuation coefficient.

[0122] The siltation intensity value at each spatial location of the reservoir can be a scalar field value obtained by multiplying the siltation impact synergy coefficient and the spatial siltation attenuation coefficient. This value can be used to integrate system risk and spatial accessibility, forming a physically meaningful siltation potential index. The siltation intensity value at each spatial location is obtained by multiplying the siltation impact synergy coefficient and the spatial siltation attenuation coefficient for each spatial point. Furthermore, this operation can be achieved through direct scalar multiplication and the introduction of a nonlinear activation function to enhance the contrast of high-risk areas, thereby integrating system-level risk and spatial accessibility to generate an engineering-interpretable siltation potential field.

[0123] Based on the siltation intensity value, the radial and tangential siltation quantities of the surrounding spatial partitions are decomposed to obtain the siltation quantity distribution describing the dynamic distribution of coal slime siltation in the entire water tank.

[0124] The surrounding spatial partition can be a local neighborhood grid or a continuous micro-element region centered on the current calculation point, which can be used as the calculation unit for radial and tangential component decomposition. The radial sedimentation quantity can be the projection component of the sedimentation intensity value in the direction pointing towards the dredging center, which can be used to reflect the accumulation trend perpendicular to the equipment's axis of action and guide the priority of deep dredging. The tangential sedimentation quantity can be the projection component of the sedimentation intensity value on a circumferential plane perpendicular to the radial direction, which can be used to capture lateral sedimentation characteristics caused by water flow asymmetry or deposition offset. The dynamic distribution of coal slime sedimentation in the entire water tank can be a vector field composed of radial and tangential sedimentation quantities at all spatial locations, which can be used to fully describe the accumulation morphology and potential migration tendency of coal slime in three-dimensional space. Based on the sedimentation intensity value, the radial and tangential sedimentation quantities are decomposed into the surrounding spatial partition to obtain the sedimentation quantity distribution describing the dynamic distribution of coal slime sedimentation in the entire water tank. This can be achieved by converting the scalar sedimentation intensity into vector components in two orthogonal directions (radial and tangential) in a local coordinate system. Furthermore, this operation can be achieved by directly decomposing the data into radial and tangential components in polar coordinates and then projecting the decomposition after extracting the principal direction using the gradient operator. This allows the data to be upgraded from a scalar field to a direction-sensitive vector field, supporting refined waste rock disposal path planning.

[0125] Taking the coordinated dredging of a circular water tank as an example, the coal mine drainage monitoring and safety early warning control method based on sensor big data information in this embodiment can be as follows: a main drainage pump is arranged in the center of the water tank, and two mobile dredging pumps are set up around it. The system calculates that the dredging influence coordination coefficient is the highest in the northeast quadrant; for each point in this area, the radial distance to the nearest dredging pump is normalized to obtain the distance coordination factor; within a 3-meter radius of action, r^-2 power law attenuation is applied to generate the spatial dredging attenuation coefficient; after multiplication, the distribution of dredging intensity values ​​is obtained; further decomposed into radial components (pointing to the pump inlet) and tangential components (reflecting the dredging offset along the tank wall); the results show that there is a strong tangential component in the northeast corner, indicating that the silt accumulates along the wall in an arc-shaped band; based on this, the system schedules the dredging pumps to advance along the tangential path, rather than just in a straight line towards the center, significantly improving the dredging coverage rate.

[0126] In one embodiment, a waste rock removal operation strategy vector field is constructed based on the adaptive siltation monitoring and control field of the water tank operating conditions and the water tank operation scenario type. An intelligent linkage waste rock removal early warning scheme is generated through dynamic probability calculation and converted into real-time waste rock removal monitoring and control commands to complete closed-loop siltation management and scheduling, including:

[0127] Based on the adaptive siltation monitoring and control field of the water tank and the water tank operation scenario type, feature extraction is performed to obtain the siltation operation feature vector;

[0128] The siltation operation feature vector can be a structured numerical vector extracted from the adaptive siltation monitoring and control field of the reservoir and the reservoir operation scenario type, used to characterize the current siltation state and operation requirements. It can be used as input for calculating the priority of waste rock removal, achieving compression and pattern recognition of high-dimensional state information. In this embodiment, the siltation operation feature vector is generated as a low-dimensional dense vector by jointly encoding the spatial distribution of the control field and the semantic labels of the scenario type. Furthermore, the siltation operation feature vector can be a spatial distribution feature sub-vector, a condition label embedding sub-vector, a dynamic evolution trend sub-vector, etc. Based on the adaptive siltation monitoring and control field of the reservoir and the reservoir operation scenario type, feature extraction is performed to obtain the siltation operation feature vector, which can be generated by jointly encoding the spatial distribution of the control field and the semantic labels of the scenario type. Further, this operation can be achieved by using a graph neural network to aggregate and encode the control field grid, using an embedding layer to map the scenario type into a vector, and concatenating it with the statistical features of the control field, thereby achieving compression of high-dimensional states into computable features and supporting subsequent priority modeling.

[0129] Based on the siltation operation feature vector, the priority weight of each water tank zone in the waste rock removal process is calculated to obtain the waste rock removal operation strategy vector field;

[0130] The water tank zoning can be a process of dividing the internal space of the water tank into several independent management units according to function or geometric structure. This can be used to support zoning-level priority assessment of wastewater removal and equipment path planning. For example, water tank zoning can be the suction well influence zone, the main sedimentation zone, or the peripheral slow-flow zone. Wastewater removal priority weight calculation can be a quantification process of allocating dredging task priorities to each water tank zoning based on the siltation operation feature vector. This can be used to optimize resource allocation in space and avoid uniform or random dredging. In an exemplary embodiment, the wastewater removal priority weight calculation inputs the feature vector into the weight calculation module, outputs a scalar priority for each water tank zoning, and combines it with spatial location to form a vector field. Furthermore, the wastewater removal priority weight calculation can be risk-driven weight, volume recovery benefit weight, or equipment accessibility correction weight, etc. Calculating the wastewater removal priority weight for each water tank zoning during the wastewater removal process based on the siltation operation feature vector yields a wastewater removal operation strategy vector field. This can be achieved by inputting the feature vector into the weight calculation module, outputting a scalar priority for each water tank zoning, and combining it with spatial location to form a vector field. Furthermore, this operation can be achieved by regressing partition weights through a fully connected neural network, matching feature patterns based on a rule base, and assigning weights by looking up a table. This allows for the construction of a task-oriented spatial strategy representation to guide precise resource allocation.

[0131] The effective volume of the current water tank and the operating status of the water pump in the coal mine drainage system are input into the waste rock removal operation strategy vector field for probability matrix optimization, so as to obtain the early warning trigger probability value of each water tank zone corresponding to the waste rock removal stage;

[0132] The effective volume of the current water tank in the coal mine drainage system can be the net volume that can actually be used to regulate and store inrush water after considering the current siltation. This volume can be used as a constraint in probability matrix optimization, reflecting the system's safety margin. In one specific embodiment, the effective volume of the current water tank in the coal mine drainage system is obtained based on three-dimensional reconstruction, level-area integration, or historical correction methods. The pump operating status can describe the operational health and load level reflected by the current pump operating parameters (such as flow rate, head, vibration, and current). This can be used to determine whether the system is approaching cavitation or overload critical points, affecting the early warning trigger threshold. For example, the pump operating status can be hydraulic performance status, mechanical wear status, electrical load status, etc. Probability matrix optimization can be a process of probabilistically calibrating the initial wastewater discharge strategy vector field using the effective volume and pump operating status as external constraints. This can be used to ensure that the early warning trigger probability value simultaneously reflects siltation risk and the system's operational safety boundary. In this embodiment, probability matrix optimization uses the effective volume and pump status as constraints to probabilistically recalibrate the initial strategy vector field, outputting a zone-level early warning probability. The effective volume of the current water sump and the operating status of the pumps in the coal mine drainage system are input into the waste rock removal operation strategy vector field for probability matrix optimization. This yields the early warning trigger probability value for each water sump zone during the corresponding waste rock removal stage. This can be achieved by recalibrating the initial strategy vector field with constraints based on the effective volume and pump status, outputting zone-level early warning probabilities. Furthermore, this operation can be implemented by constructing a conditional probability table for lookup optimization and using an online learning model to dynamically adjust the probability mapping relationship. This ensures that early warnings not only depend on the degree of siltation but also are coupled with the system's operational safety boundaries, improving early warning reliability. The early warning trigger probability value can be a quantified probability indicator of each water sump zone triggering an early warning during a specific waste rock removal stage, after probability matrix optimization. This can be used as a direct basis for risk level determination, supporting continuous rather than binary early warnings. Based on the early warning trigger probability value, siltation risk level determination and control command generation are performed, resulting in an intelligent linkage waste rock removal early warning scheme that includes graded early warning levels, waste rock removal operation timing, and sludge removal equipment paths. This scheme is then converted into real-time waste rock removal monitoring and control commands to complete closed-loop siltation management and scheduling.

[0133] The siltation risk level determination can be a classification process based on the probability value of early warning triggering, which can be used to match differentiated response strategies for different risk levels. In this embodiment, the siltation risk level determination classifies risk levels according to a preset probability threshold and calls the contingency plan library to generate a complete plan. Control command generation can be a conversion process that maps risk levels to specific execution actions, which can be used to output an executable plan that includes the timing of the operation and the equipment path. The graded early warning level can be a multi-level early warning label based on the degree of risk, such as attention, intervention, and emergency, which can be used to guide the initiation of response measures of different intensities. The timing of the dredging operation can be a suggested time window or triggering condition for initiating the dredging operation, which can be used to avoid wasting resources too early or causing system failure too late. The dredging equipment path can be a spatial movement trajectory planned for the dredging robot or mobile pump group, which can be used to improve dredging efficiency and avoid obstacles or sensitive areas. Based on the probability value of early warning triggers, the risk level of siltation is determined and control commands are generated, resulting in an intelligent linkage silt removal early warning scheme that includes graded early warning levels, timing of silt removal operations, and paths for dredging equipment. This can be achieved by classifying risk levels according to preset probability thresholds and generating a complete scheme including timing and path from a contingency plan library. Furthermore, this operation can use decision trees for level determination and command matching, and generate silt removal paths and timing sequences that meet multiple objectives through an optimized solver, thus realizing the transformation from probability output to structured decision schemes and supporting multi-level responses. This is converted into real-time silt removal monitoring and control commands to complete closed-loop siltation management and scheduling. This can be achieved by associating the silt removal equipment path and timing in the early warning scheme with specific equipment control protocols and issuing execution commands. Furthermore, this operation can issue pump and valve control commands through an industrial PLC interface and publish silt removal robot path objectives through the ROS navigation stack, thus completing the closed loop from decision to execution and ensuring the scheme's implementation.

[0134] Taking the zonal dredging scheduling when the water pump is near the cavitation critical point as an example, the coal mine drainage monitoring safety early warning control method based on sensor big data information in this embodiment can be as follows: The system monitors that the effective volume of the main chamber has dropped to 35% of the design value, and the water pump vibration and current fluctuation show that it is close to the cavitation threshold; at this time, the adaptive siltation monitoring control field of the water chamber shows that the potential field intensity of the zone around the suction well is high; the feature extraction module integrates the control field with the high sand-containing water scene label to generate a siltation operation feature vector; the priority calculation module assigns the highest weight to the suction well influence area accordingly; the probability matrix optimization module combines the low effective volume and the critical state of the water pump to significantly improve the probability value of triggering the early warning in this zone; the risk level judgment module outputs a red emergency warning; the control command generation module determines to start dredging immediately, prioritize cleaning within a 5-meter radius of the suction well, and plan the dredging robot to operate along a circular path to avoid disturbing the water flow; the command is sent to the robot and the backup pump group in real time to complete the closed-loop intervention.

[0135] In addition, refer to Figure 2 To achieve the above objectives, the present invention also provides a coal mine drainage monitoring and safety early warning control system based on sensor big data information, the system comprising:

[0136] Data parsing module 10 is used to perform interface feature parsing and siltation parameter formalization processing on the real-time collected multi-source sensor data of the water tank, and generate multi-dimensional coal slime siltation constraint data. The multi-dimensional coal slime siltation constraint data includes siltation interface coordinates, siltation rate weights and water tank operation scenario types.

[0137] The potential field construction module 20 is used to construct a dynamic siltation potential field matrix that reflects the distribution of coal slime siltation potential based on the siltation interface coordinates and siltation rate weights in the multi-dimensional coal slime siltation constraint data and by combining spatial interpolation attenuation functions.

[0138] Model prediction module 30 is used to establish a dynamic prediction model for coal slime deposition, which includes deposition morphology evolution term and sediment erosion and diffusion term, by using the dynamic deposition potential field matrix as the driving source.

[0139] The adaptive control module 40 is used to calculate the siltation influence coordination coefficient based on the coal mine drainage system water tank volume assessment value and the current location siltation potential field intensity, generate a local dredging and gangue allocation component and integrate it with the coal slime siltation dynamic prediction model to obtain the water tank working condition adaptive siltation monitoring and control field.

[0140] The closed-loop scheduling module 50 is used to construct a waste rock removal operation strategy vector field based on the adaptive siltation monitoring and control field of the water tank operating conditions and the water tank operation scenario type, generate an intelligent linkage waste rock removal early warning scheme through dynamic probability calculation, and convert it into real-time waste rock removal monitoring and control commands to complete the closed-loop siltation management and scheduling.

[0141] Other embodiments or specific implementations of the coal mine drainage monitoring and safety early warning control system based on sensor big data information described in this invention can be referred to the above-mentioned method embodiments, and will not be repeated here.

[0142] Furthermore, to achieve the above objectives, the present invention also provides a coal mine drainage monitoring and safety early warning control device based on sensor big data information. The device includes: a memory, a processor, and a coal mine drainage monitoring and safety early warning control program based on sensor big data information stored in the memory and executable on the processor. The coal mine drainage monitoring and safety early warning control program based on sensor big data information is configured to implement the steps of the coal mine drainage monitoring and safety early warning control method based on sensor big data information as described above.

[0143] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing a coal mine drainage monitoring and safety early warning control program based on sensor big data information. When the coal mine drainage monitoring and safety early warning control program based on sensor big data information is executed by a processor, it implements the steps of the coal mine drainage monitoring and safety early warning control method based on sensor big data information as described above.

[0144] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A coal mine drainage monitoring safety early warning control method based on sensor big data information, characterized in that, The method includes: The interface feature analysis and siltation parameter formalization processing of the real-time collected multi-source sensor data of the water tank are performed to generate multi-dimensional coal slime siltation constraint data, which includes siltation interface coordinates, siltation rate weights and water tank operation scenario types. Based on the sedimentation interface coordinates and sedimentation rate weights in the multi-dimensional coal slime sedimentation constraint data, a dynamic sedimentation potential field matrix reflecting the distribution of coal slime sedimentation potential is constructed by combining spatial interpolation decay functions. Using the dynamic sedimentation potential field matrix as the driving source, a dynamic prediction model for coal slime sedimentation, including sedimentation morphology evolution term and sediment erosion and diffusion term, is established. The siltation impact coordination coefficient is calculated based on the assessed volume of the coal mine drainage system water tank and the current location siltation potential field intensity. A local dredging and gangue allocation component is generated and fused with the coal slime siltation dynamic prediction model to obtain the adaptive siltation monitoring and control field for the water tank working condition. Based on the adaptive siltation monitoring and control field of the water tank operating conditions and the water tank operation scenario type, a waste rock removal operation strategy vector field is constructed. An intelligent linkage waste rock removal early warning scheme is generated through dynamic probability calculation and converted into real-time waste rock removal monitoring and control commands to complete closed-loop siltation management and scheduling. 2.The coal mine drainage monitoring safety early warning control method based on sensor big data information according to claim 1, characterized in that, The process involves analyzing interface features and formalizing sedimentation parameters of the real-time collected multi-source sensor data from the water tank to generate multi-dimensional coal slime sedimentation constraint data, including: Feature extraction and semantic annotation are performed on the multi-source sensor data of the water tank to obtain a set of siltation elements containing siltation interface identifier, siltation intensity descriptor and operation constraint descriptor; The set of sedimentation elements is input into a pre-trained coal mine water tank sedimentation knowledge base for mapping and calibration to obtain a standardized combination of sedimentation parameters. Based on the standardized siltation parameter combination, the coordinates of the siltation interface and the siltation rate weight are extracted respectively to obtain the coordinate values ​​and weight values. The water tank operation scenario is classified according to the siltation interface identifier to obtain the water tank operation scenario type, and then combined with the siltation interface coordinates and siltation rate weight to obtain the multi-dimensional coal slime siltation constraint data.

3. The coal mine drainage monitoring and safety early warning control method based on sensor big data information as described in claim 1, characterized in that, The dynamic sedimentation potential field matrix, reflecting the distribution of coal slime sedimentation potential, is constructed by combining the sedimentation interface coordinates and sedimentation rate weights in the multi-dimensional coal slime sedimentation constraint data through spatial interpolation attenuation functions. This includes: Based on the coordinates of the sedimentation interface in the multi-dimensional coal slime sedimentation constraint data, a spatial attenuation sedimentation contribution function is constructed for each sedimentation interface point to obtain the sedimentation potential field distribution of a single-point region. Based on the sedimentation rate weight, the single-point area sedimentation potential field distribution of each sedimentation interface point is weighted by sedimentation priority to obtain the coal slime sedimentation priority potential field. The multiple coal slime deposition priority potential fields are superimposed and combined according to the spatial location of the water tank to obtain the coal slime deposition potential distribution of the entire water tank. Based on the water tank operation scenario type, the attenuation range parameter is set and the distribution of coal slime siltation potential in the full water tank is matrixed to obtain the dynamic siltation potential field matrix.

4. The coal mine drainage monitoring and safety early warning control method based on sensor big data information as described in claim 1, characterized in that, The step of establishing a dynamic prediction model for coal slime deposition, using the dynamic sedimentation potential field matrix as the driving source and including sediment erosion and diffusion terms, includes: The sedimentation gradient components are calculated based on the dynamic sedimentation potential field matrix to obtain the sedimentation driving force vector pointing to the region of high sedimentation intensity. Gradient calculation of the sedimentation field is performed based on the distribution density of sediment content in the inflow water in different areas of the reservoir to obtain the sedimentation morphology evolution term that prevents local sedimentation estimation bias. Based on the requirement of continuity in the coal slime deposition process, the coal slime transportation is subjected to diffusion treatment to obtain the sediment erosion diffusion term that suppresses the fluctuation of the deposition morphology estimation. The mass conservation equation is used to solve the siltation driving force vector, siltation morphology evolution term, and sediment erosion and diffusion term to obtain the dynamic prediction model of coal slime siltation describing the dynamic deposition of coal slime under the complex working conditions of a water tank.

5. The coal mine drainage monitoring and safety early warning control method based on sensor big data information as described in claim 4, characterized in that, The process involves calculating the siltation impact coordination coefficient based on the coal mine drainage system water tank volume assessment value and the current location siltation potential field intensity, generating a local dredging and waste rock allocation component, and fusing it with the coal slime siltation dynamic prediction model to obtain the water tank operating condition adaptive siltation monitoring and control field, including: The siltation impact synergy coefficient for each water tank zone is obtained by performing a weighted product calculation based on the water tank volume assessment value and the current location siltation potential field intensity. Based on the siltation impact synergy coefficient and the dredging operation radius, a local waste rock allocation field is constructed to obtain the siltation distribution of each water tank zone; The distribution of the siltation volume is adjusted and calculated to obtain the local siltation and waste rock removal adjustment component to prevent excessive or untimely waste rock removal operations. The local dredging and waste rock allocation component is superimposed and fused with the coal slime siltation dynamic prediction model to obtain the water tank condition adaptive siltation monitoring and control field, which has both global morphological prediction and local operation coordination functions.

6. The coal mine drainage monitoring and safety early warning control method based on sensor big data information as described in claim 5, characterized in that, The process of constructing a local waste rock disposal and allocation field based on the siltation impact synergy coefficient and the dredging operation radius yields the siltation distribution in each water storage zone, including: Based on the siltation impact synergy coefficient, the radial distance of the water tank space is normalized to obtain the distance synergy factor; The operational radius of the dredging operation is used to set the waste rock allocation range, and the power-law attenuation calculation is performed on the distance coordination factor to obtain the spatial siltation attenuation coefficient. The siltation influence synergy coefficient and the spatial siltation attenuation coefficient are multiplied to obtain the siltation intensity value at each spatial location of the water tank; Based on the siltation intensity value, the radial and tangential siltation components of the surrounding space are decomposed to obtain the siltation distribution that describes the dynamic distribution of coal slime siltation in the entire water tank.

7. The coal mine drainage monitoring and safety early warning control method based on sensor big data information as described in claim 1, characterized in that, The process of constructing a waste rock removal operation strategy vector field based on the adaptive siltation monitoring and control field of the water tank operating conditions and the water tank operation scenario type, generating an intelligent linkage waste rock removal early warning scheme through dynamic probability calculation, and converting it into real-time waste rock removal monitoring and control commands to complete closed-loop siltation management and scheduling includes: Based on the adaptive siltation monitoring and control field of the water tank and the water tank operation scenario type, feature extraction is performed to obtain the siltation operation feature vector; Based on the siltation operation feature vector, the priority weight of each water tank zone in the waste rock removal process is calculated to obtain the waste rock removal operation strategy vector field; The effective volume of the current water tank and the operating status of the water pump in the coal mine drainage system are input into the waste rock removal operation strategy vector field for probability matrix optimization to obtain the early warning trigger probability value of each water tank partition corresponding to the waste rock removal stage; Based on the early warning trigger probability value, the siltation risk level is determined and control instructions are generated to obtain the intelligent linkage silt removal early warning scheme, which includes graded early warning levels, timing of silt removal operations, and path of dredging equipment. This scheme is then converted into the real-time silt removal monitoring and control instructions to complete the closed-loop siltation management and scheduling.

8. A coal mine drainage monitoring and safety early warning control system based on sensor big data information, characterized in that, The system includes: The data parsing module is used to perform interface feature parsing and formal processing of siltation parameters on the real-time collected multi-source sensor data of the water tank, and generate multi-dimensional coal slime siltation constraint data. The multi-dimensional coal slime siltation constraint data includes siltation interface coordinates, siltation rate weights and water tank operation scenario types. The potential field construction module is used to construct a dynamic siltation potential field matrix that reflects the distribution of coal slime siltation potential based on the siltation interface coordinates and siltation rate weights in the multi-dimensional coal slime siltation constraint data and by combining spatial interpolation decay functions. The model prediction module is used to establish a dynamic prediction model for coal slime deposition, which includes the dynamic sedimentation potential field matrix as the driving source and includes sedimentation morphology evolution term and sediment erosion and diffusion term. The adaptive control module is used to calculate the siltation impact coordination coefficient based on the coal mine drainage system water tank volume assessment value and the current location siltation potential field intensity, generate local dredging and gangue allocation components and integrate them with the coal slime siltation dynamic prediction model to obtain the water tank working condition adaptive siltation monitoring and control field. The closed-loop scheduling module is used to construct a waste rock removal operation strategy vector field based on the adaptive siltation monitoring and control field of the water tank operating conditions and the water tank operation scenario type. It generates an intelligent linkage waste rock removal early warning scheme through dynamic probability calculation and converts it into real-time waste rock removal monitoring and control commands to complete the closed-loop siltation management and scheduling.

9. A coal mine drainage monitoring and safety early warning control device based on sensor big data information, characterized in that, The device includes: a memory, a processor, and a coal mine drainage monitoring and safety early warning control program based on sensor big data information stored in the memory and executable on the processor. The coal mine drainage monitoring and safety early warning control program based on sensor big data information is configured to implement the steps of the coal mine drainage monitoring and safety early warning control method based on sensor big data information as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a coal mine drainage monitoring and safety early warning control program based on sensor big data information. When the coal mine drainage monitoring and safety early warning control program based on sensor big data information is executed by a processor, it implements the steps of the coal mine drainage monitoring and safety early warning control method based on sensor big data information as described in any one of claims 1 to 7.

Images