Intelligent evaluation and real-time optimization system for prevention and control effect of dynamic disaster in deep fluidized mining

Abstract

This invention relates to the field of evaluation and real-time optimization systems, and discloses an intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized bed mining. The system includes a data acquisition module for real-time acquisition of monitoring data through a multi-source sensor network deployed in the stope, fluidized medium, and surrounding rock; and an intelligent analysis module for constructing a comprehensive evaluation index system based on the monitoring data, encompassing static, dynamic, and stability indicators. This intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized bed mining overcomes the limitations of traditional single-indicator evaluation. By integrating a multi-source sensor network for stress, microseismic, acoustic emission, and displacement, it constructs a comprehensive evaluation index system covering statics, dynamics, and stability. Utilizing a deep learning model to fuse and analyze massive amounts of heterogeneous data, it can reveal multi-dimensional precursory information of disasters.

Description

Technical Field

[0001] This invention relates to the field of evaluation and real-time optimization system technology, and in particular to an intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized bed mining. Background Technology

[0002] With the continuous growth of global demand for mineral resources and the increasing depletion of shallow resources, mineral resource mining inevitably moves towards deeper areas. Deep mining faces an extremely complex and harsh geomechanical environment, characterized by "three highs and one disturbance" (high ground pressure, high ground temperature, high osmotic pressure, and strong mining disturbance). Under such conditions, traditional mining methods are highly susceptible to triggering a series of dynamic disasters, such as rock bursts (rock bursts), large-scale roof pressure, pillar instability, and sudden collapse of goaf areas, which seriously threaten mine safety and the lives of personnel.

[0003] Fluidized bed mining, as a highly efficient mining technology, is being used more and more widely in deep mines. By converting ore or media into a fluidized state for transportation and recycling, mining efficiency can be improved and energy consumption reduced. However, it also introduces new dynamic disaster risks, such as fluidized medium instability, pipeline blockage, and pressure surges, which require real-time monitoring and optimized control.

[0004] Currently, the evaluation of the effectiveness of dynamic disaster prevention and control in fluidized mining is mostly still in the initial stage of "post-event statistics, static assessment, and experience judgment," which has obvious shortcomings. Existing methods often focus on data analysis from a single source, such as monitoring only microseismic events or measuring only stress changes in a certain profile. They lack the fusion analysis of multi-source information (stress, microseismic events, acoustic emission, displacement, etc.), making it difficult to comprehensively and accurately reflect the overall stability of the fluidized medium and surrounding rock system. There is a lack of correlation analysis between various indicators, making it impossible to reveal the multi-faceted precursory characteristics and internal mechanisms of disaster formation. Summary of the Invention

[0005] Given that the existing methods often focus on single-source data analysis, such as monitoring only microseismic events or measuring only stress changes in a certain profile, and lack the fusion analysis of multi-source information (stress, microseismic events, acoustic emission, displacement, etc.), it is difficult to comprehensively and accurately reflect the overall stability state of the fluidized medium and surrounding rock system. Furthermore, there is a lack of correlation analysis between various indicators, which fails to reveal the multi-faceted precursory characteristics and intrinsic mechanisms of disaster formation. Therefore, this invention is proposed.

[0006] Therefore, the purpose of this invention is to provide an intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining. Its purpose is to overcome the limitations of traditional single-index evaluation, and to construct a comprehensive evaluation index system covering statics, dynamics and stability by integrating a multi-source sensor network of stress, microseismic, acoustic emission and displacement. By using a deep learning model to fuse and analyze massive heterogeneous data, it can profoundly reveal the multi-dimensional precursor information of disasters.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized bed mining, comprising:

[0008] The data acquisition module is used to collect monitoring data in real time through a multi-source sensor network deployed in the mining area, fluidized medium and surrounding rock;

[0009] The intelligent analysis module is used to construct a comprehensive evaluation index system that includes static, dynamic and stability indicators based on the monitoring data, and to calculate the dynamic comprehensive index E(t) that characterizes the current prevention and control effect through the intelligent evaluation model.

[0010] The optimization decision module is used to generate corresponding mining or fluidization parameter optimization and control instructions based on the value and trend of the dynamic comprehensive index E(t).

[0011] The execution feedback module is used to execute the optimized control command and evaluate the control effect based on the monitoring data collected after control, thus completing the closed-loop optimization.

[0012] The human-computer interaction and visualization module provides a graphical monitoring interface and multi-dimensional data dashboards for real-time display of monitoring data, dynamic comprehensive index change curves, system early warning information and control command execution status, and supports manual confirmation, parameter adjustment and intervention command issuance.

[0013] As a preferred embodiment of the intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining as described in this invention, the multi-source sensor network includes one or more of the following: stress sensor, microseismic monitor, acoustic emission instrument, displacement sensor, and multi-point displacement meter.

[0014] As a preferred embodiment of the intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining as described in this invention, the comprehensive evaluation index system in the intelligent analysis module specifically includes:

[0015] Static parameters: fluidized medium bearing capacity, stress concentration index, energy storage coefficient;

[0016] Dynamic indicators: energy release rate, event frequency anomaly, and vibration amplitude change rate;

[0017] Stability indicators: displacement convergence rate, plastic zone development index, and crack propagation rate.

[0018] As a preferred embodiment of the intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining as described in this invention, the intelligent evaluation model adopts a deep learning model with an LSTM network or a Transformer architecture; the dynamic comprehensive index E(t) is a continuous value within the range of [0,1], and the closer its value is to 0, the better the prevention and control effect, and the closer it is to 1, the higher the disaster risk.

[0019] As a preferred embodiment of the intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining as described in this invention, the optimization decision module generates optimization control instructions based on the dynamic comprehensive index E(t) in the following specific manner:

[0020] When 0≤E(t)<0.3, the control effect is considered good, and the current mining and fluidization parameters are maintained;

[0021] When 0.3≤E(t)<0.6, a potential risk is identified, an early warning message is generated, and it is recommended to strengthen monitoring.

[0022] When 0.6≤E(t)<0.8, the risk is considered high, and control instructions are generated to adjust the mining intensity or fluidized medium parameters;

[0023] When 0.8≤E(t)≤1, the risk is determined to be extremely high, and an emergency control instruction is generated, including slowing down mining operations and initiating supplementary pressure relief measures.

[0024] As a preferred embodiment of the intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining as described in this invention, the data acquisition module further includes a signal conditioning submodule, a multi-channel synchronous acquisition submodule, an edge computing submodule, and an equipment health monitoring submodule.

[0025] The signal conditioning submodule uses the following formula to calculate the signal quality index (SQI) for evaluating signal quality:

[0026]

[0027] in, The standard deviation of noise. The signal standard deviation is used; when SQI < 0.7, a signal quality alarm is triggered.

[0028] The multi-channel synchronous acquisition submodule uses a timestamp synchronization algorithm to ensure that the acquisition time deviation of each channel is less than 1ms.

[0029] As a preferred embodiment of the intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining as described in this invention, the intelligent analysis module further includes a data preprocessing submodule, a feature engineering submodule, a model management submodule, and an interpretability analysis submodule.

[0030] The feature engineering submodule uses the following formula to calculate the feature importance score:

[0031]

[0032] Where N is the number of samples, and f(x) is the model output function. For the i-th feature;

[0033] The interpretability analysis submodule uses SHAP value analysis to calculate the feature contribution using the following formula:

[0034]

[0035] Where M is the total number of features and S is the subset of features.

[0036] As a preferred embodiment of the intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining as described in this invention, the optimization decision module further includes a multi-objective optimization sub-module, a rule engine sub-module, a contingency plan management sub-module, and a simulation and deduction sub-module.

[0037] The multi-objective optimization submodule employs the following multi-objective optimization function:

[0038] min[ (x), (x), [x] = [Risk Indicators, Cost Indicators, Efficiency Indicators]

[0039] st (x)≤0,i=1,2,...,m

[0040] (x)=0,j=1,2,...,n

[0041] in, (x) (x) (x) represent three optimization objective functions.

[0042] As a preferred embodiment of the intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining as described in this invention, the execution feedback module further includes an instruction safety verification submodule, an actuator interface submodule, an effect evaluation submodule, and an adaptive learning submodule.

[0043] The effect evaluation submodule uses the following comprehensive evaluation function:

[0044]

[0045] Where ΔE is the rate of change of the dynamic composite index. For response time, For resource consumption rate, , , These are the weighting coefficients, and + + =1.

[0046] As a preferred embodiment of the intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining as described in this invention, the human-computer interaction and visualization module further includes a three-dimensional geological visualization sub-module, a real-time alarm sub-module, a decision dashboard sub-module, and an operation log and audit sub-module.

[0047] The real-time alarm submodule uses the following risk warning function:

[0048]

[0049] Where E(t) is the dynamic composite index, Its rate of change, For each individual indicator, α, β, and γ represent the degree of abnormality, and α, β, and γ are the weighting coefficients. The weights of each indicator are given.

[0050] Compared with the prior art, the present invention has at least the following beneficial effects:

[0051] 1. This invention breaks through the limitations of traditional single-index evaluation. By integrating a multi-source sensor network including stress, microseismic, acoustic emission, and displacement, it constructs a comprehensive evaluation index system covering statics, dynamics, and stability. By using a deep learning model to fuse and analyze massive heterogeneous data, it can profoundly reveal multi-dimensional precursory information of disasters, achieve comprehensive perception and accurate judgment of the stability state of fluidized medium surrounding rock systems, and greatly reduce false alarm and missed alarm rates.

[0052] 2. This invention establishes a dynamic evaluation model based on real-time data streams, which can continuously calculate and output the comprehensive index E(t) of prevention and control effectiveness. This completely changes the traditional phased and lagging static evaluation mode. The system can perform "health diagnosis" on the prevention and control effectiveness 24 / 7. Once an abnormal trend is detected, an early warning can be issued in advance, which will buy a valuable "time window" for taking control measures and realize a fundamental transformation from "post-event remediation" to "pre-event warning and in-event control".

[0053] 3. This invention forms a complete intelligent closed loop. The system can automatically generate and execute optimization instructions (such as dynamically adjusting fluidization parameters, conveying rate, or mining intensity) based on real-time evaluation results, and continuously track the effects of regulation to form a feedback closed loop. This data-driven decision-making model reduces excessive reliance on human experience, ensures the scientific nature and timeliness of prevention and control measures, avoids disasters caused by insufficient prevention and control, and prevents resource waste caused by excessive prevention and control, ultimately achieving a balance between safety and efficiency. Attached Figure Description

[0054] Figure 1 This is a schematic diagram of the process framework of the intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining according to the present invention. Detailed Implementation

[0055] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0056] Reference Figure 1 As an embodiment of the present invention, an intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining is provided. This intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining includes a data acquisition module, which is used to collect monitoring data in real time through a multi-source sensor network deployed in the mining area, fluidized medium and surrounding rock. The module has multi-channel synchronous acquisition and signal preprocessing functions, supports setting different sampling frequencies and trigger thresholds, and has a built-in self-diagnosis mechanism to ensure data integrity.

[0057] The intelligent analysis module integrates a time-series database and a real-time computing engine to construct a comprehensive evaluation index system based on monitoring data, including static, dynamic and stability indicators. It uses an intelligent evaluation model based on deep learning algorithms to perform fusion analysis and feature extraction on multi-source heterogeneous monitoring data, calculates the dynamic comprehensive index E(t) that represents the current prevention and control effect, and has the ability to update the model online and perform incremental learning.

[0058] The optimization decision module has a built-in multi-objective optimization algorithm and rule reasoning engine, which is used to automatically generate optimization control instructions for parameters such as mining intensity, fluidization parameters, and pumping rate based on the value, trend and historical sequence pattern of the dynamic comprehensive index E(t), and supports the dynamic selection and simulation of multiple control plans.

[0059] The execution feedback module is equipped with an instruction verification and security interlock mechanism to reliably execute optimization control instructions and evaluate the control effect in real time based on the monitoring data collected after control. By comparing the change and convergence characteristics of the dynamic comprehensive index before and after control, an intelligent closed-loop optimization mechanism of "monitoring-evaluation-control-feedback" is formed.

[0060] The human-computer interaction and visualization module provides a graphical monitoring interface, multi-dimensional data dashboards, and an early warning cockpit for real-time display of monitoring data, dynamic comprehensive index change curves, system early warning information, and the execution status of control commands. It also supports manual confirmation, parameter adjustment, and the issuance of intervention commands.

[0061] Multi-source sensor networks include one or more of the following: stress sensors, microseismic monitors, acoustic emission devices, displacement sensors, and multi-point displacement gauges.

[0062] The comprehensive evaluation index system in the intelligent analysis module specifically includes:

[0063] Static parameters: fluidized medium bearing capacity, stress concentration index, energy storage coefficient;

[0064] Dynamic indicators: energy release rate, event frequency anomaly, and vibration amplitude change rate;

[0065] Stability indicators: displacement convergence rate, plastic zone development index, and crack propagation rate.

[0066] The intelligent evaluation model adopts a deep learning model with LSTM network or Transformer architecture; the dynamic comprehensive index E(t) is a continuous value in the range of [0,1]. The closer its value is to 0, the better the prevention and control effect, and the closer it is to 1, the higher the disaster risk.

[0067] The optimization decision-making module generates optimization control instructions based on the dynamic comprehensive index E(t) in the following specific way:

[0068] When 0≤E(t)<0.3, the control effect is considered good, and the current mining and fluidization parameters are maintained;

[0069] When 0.3≤E(t)<0.6, a potential risk is identified, an early warning message is generated, and it is recommended to strengthen monitoring.

[0070] When 0.6≤E(t)<0.8, the risk is considered high, and control instructions are generated to adjust the mining intensity or fluidized medium parameters;

[0071] When 0.8≤E(t)≤1, the risk is determined to be extremely high, and an emergency control instruction is generated, including slowing down mining operations and initiating supplementary pressure relief measures.

[0072] The data acquisition module also includes a signal conditioning submodule, a multi-channel synchronous acquisition submodule, an edge computing submodule, and a device health monitoring submodule;

[0073] The signal conditioning submodule is used to amplify, filter, and isolate the raw sensor signal to improve the signal-to-noise ratio.

[0074] The multi-channel synchronous acquisition submodule supports high-concurrency data acquisition and adopts GPS / BeiDou dual-mode clock synchronization technology to ensure the timing consistency of multi-source sensors;

[0075] The edge computing submodule has built-in lightweight algorithms that can perform preliminary data processing, anomaly detection, and feature extraction at the data acquisition end, reducing the amount of data transmission.

[0076] The equipment health monitoring submodule diagnoses the working status of sensors and data acquisition devices in real time, identifies sensor drift and disconnection faults through self-diagnostic algorithms, and supports fault self-alarm and redundancy switching.

[0077] The signal conditioning submodule uses the following formula to calculate the Signal Quality Index (SQI) to evaluate signal quality:

[0078]

[0079] in, The standard deviation of noise. The signal standard deviation is used; when SQI < 0.7, a signal quality alarm is triggered.

[0080] The multi-channel synchronous acquisition submodule uses a timestamp synchronization algorithm to ensure that the acquisition time deviation of each channel is less than 1ms.

[0081] The intelligent analysis module also includes a data preprocessing submodule, a feature engineering submodule, a model management submodule, and an interpretability analysis submodule;

[0082] The data preprocessing submodule uses a wavelet transform-Kalman filter joint algorithm for noise filtering, spatiotemporal interpolation to process missing data, and adaptive Z-score normalization to normalize the data.

[0083] The feature engineering submodule automatically extracts time-domain, frequency-domain, and time-frequency-domain features, constructs a high-dimensional feature pool, and uses the random forest algorithm to sort and select features based on their importance.

[0084] The model management submodule supports loading, inference, version management and A / B testing of various machine learning / deep learning models, and provides model performance monitoring and degradation warning functions;

[0085] The interpretability analysis submodule provides SHAP value analysis and decision path visualization, and visualizes the key features that the model focuses on through the attention mechanism, thereby enhancing the model's credibility and interpretability;

[0086] The feature engineering submodule uses the following formula to calculate the feature importance score:

[0087] The feature engineering submodule uses the following formula to calculate the feature importance score:

[0088]

[0089] Where N is the number of samples, and f(x) is the model output function. For the i-th feature;

[0090] The interpretability analysis submodule uses SHAP value analysis to calculate the feature contribution using the following formula:

[0091]

[0092] Where M is the total number of features and S is the subset of features.

[0093] The optimization decision-making module also includes a multi-objective optimization submodule, a rule engine submodule, a contingency plan management submodule, and a simulation and deduction submodule;

[0094] The multi-objective optimization submodule is based on the improved NSGA-II algorithm, which coordinates multi-objective decisions such as safety, efficiency, and cost, and outputs a Pareto optimal solution set;

[0095] The rules engine submodule has a built-in generative rule base based on expert experience and industry standards, supporting a dual-driven decision-making mechanism of "rules + data";

[0096] The contingency plan management submodule stores contingency plans for various typical disaster scenarios and uses case-based reasoning technology to achieve intelligent matching and recommendation of contingency plans;

[0097] The simulation and deduction submodule constructs a virtual model of the mining area based on digital twin technology, performs digital simulation of control commands, and predicts the effects and risks of multiple execution paths;

[0098] The multi-objective optimization submodule uses the following multi-objective optimization function:

[0099] min[ (x), (x), [x] = [Risk Indicators, Cost Indicators, Efficiency Indicators]

[0100] st (x)≤0,i=1,2,...,m

[0101] (x)=0,j=1,2,...,n

[0102] in, (x) (x) (x) represent three optimization objective functions.

[0103] The execution feedback module also includes an instruction safety verification submodule, an executor interface submodule, an effect evaluation submodule, and an adaptive learning submodule;

[0104] The instruction security verification submodule uses formal verification methods to perform logical conflict detection and security boundary judgment on control instructions to ensure instruction security.

[0105] The actuator interface submodule provides a standardized device control interface, supporting multiple industrial protocols such as OPCUA, Modbus, and Profinet, enabling access to devices from multiple brands.

[0106] The effect evaluation submodule quantifies the control effect through key performance indicators, including dimensions such as response time, convergence speed, and resource consumption;

[0107] The adaptive learning submodule uses a reinforcement learning algorithm to automatically adjust control parameters and strategies based on feedback from historical control effects, thereby achieving self-optimization of the control algorithm.

[0108] The effect evaluation submodule uses the following comprehensive evaluation function:

[0109]

[0110] Where ΔE is the rate of change of the dynamic composite index. For response time, For resource consumption rate, , , These are the weighting coefficients, and + + =1.

[0111] The human-computer interaction and visualization module also includes a 3D geological visualization submodule, a real-time alarm submodule, a decision dashboard submodule, and an operation log and audit submodule;

[0112] The 3D geological visualization submodule integrates BIM / GIS technology to realize the 3D dynamic display of the mining area, surrounding rock and fluidized medium, and supports multi-dimensional data fusion display;

[0113] The real-time alarm submodule supports multi-level alarms (prompt, warning, alarm), provides notifications via sound and light, SMS, WeChat and other methods, and realizes intelligent hierarchical push of alarms;

[0114] The decision dashboard submodule provides customized data views and decision support for different users based on role permissions, and supports adaptive display across multiple terminals;

[0115] The operation log and audit submodule records all manual interventions and automated system operations, and uses blockchain technology to ensure the immutability of the logs, meeting security audit requirements;

[0116] The real-time alarm submodule uses the following risk warning function:

[0117]

[0118] Where E(t) is the dynamic composite index, Its rate of change, For each individual indicator, α, β, and γ represent the degree of abnormality, and α, β, and γ are the weighting coefficients. The weights of each indicator are given.

[0119] Example 1

[0120] Step 1: System Deployment and Data Acquisition

[0121] A monitoring network was deployed in the -850m level stope: 12 stress sensors were installed in the fluidized medium, 8 microseismic monitoring instruments were installed on the roof and sidewalls, 6 acoustic emission probes were deployed in key areas, and 15 displacement monitoring points were set up around the roadway. The system continuously collected data at a sampling frequency of 200Hz and transmitted it to the ground data center through the underground industrial ring network.

[0122] Step 2: Data Preprocessing and Feature Extraction

[0123] The data preprocessing submodule processes the raw data;

[0124] Step 3: Intelligent Evaluation Model Construction and Training

[0125] Attention-LSTM model structure

[0126] The model was trained using 3 years of historical data (including records of 12 disaster events). Training parameters: learning rate 0.001, batch size=32, epochs=100.

[0127] Step 4: Real-time evaluation and decision optimization

[0128] The system calculates the dynamic composite index E(t) every minute. When E(t) = 0.72, a level three response is triggered.

[0129] The optimal control parameters are calculated using a multi-objective optimization algorithm.

[0130] Output control instructions: Increase the concentration of fluidized medium from 68% to 72%, and reduce the mining intensity from 3 cycles per day to 2 cycles per day;

[0131] Step 5: Effectiveness Evaluation and Feedback Optimization

[0132] Two hours after the implementation of the control measures, a reassessment showed that E(t) decreased from 0.72 to 0.58. The effect evaluation submodule calculated: ΔE = (0.72 - 0.58) / 0.72 × 100% = 19.4%.

[0133] According to the evaluation rules (ΔE>15% is considered effective), the control measures are deemed effective, and the control parameters for this control measure are added to the successful case database.

[0134] In practice, this invention breaks through the limitations of traditional single-index evaluation. By integrating a multi-source sensor network including stress, microseismic, acoustic emission, and displacement, it constructs a comprehensive evaluation index system covering statics, dynamics, and stability. By using a deep learning model to fuse and analyze massive heterogeneous data, it can profoundly reveal multi-dimensional precursory information of disasters, achieve comprehensive perception and accurate judgment of the stability state of fluidized medium surrounding rock systems, and greatly reduce false alarms and missed alarms.

[0135] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A deep fluidization mining dynamic disaster prevention and control effect intelligent evaluation and real-time optimization system, characterized in that, include: The data acquisition module is used to collect monitoring data in real time through a multi-source sensor network deployed in the mining area, fluidized medium and surrounding rock; The intelligent analysis module is used to construct a comprehensive evaluation index system that includes static, dynamic and stability indicators based on the monitoring data, and to calculate the dynamic comprehensive index E(t) that characterizes the current prevention and control effect through the intelligent evaluation model. The optimization decision module is used to generate corresponding mining or fluidization parameter optimization and control instructions based on the value and trend of the dynamic comprehensive index E(t). The execution feedback module is used to execute the optimized control command and evaluate the control effect based on the monitoring data collected after control, thus completing the closed-loop optimization. The human-computer interaction and visualization module provides a graphical monitoring interface and multi-dimensional data dashboards for real-time display of monitoring data, dynamic comprehensive index change curves, system early warning information and control command execution status, and supports manual confirmation, parameter adjustment and intervention command issuance; The multi-source sensor network includes one or more of the following: stress sensor, micro-vibration monitor, acoustic emission instrument, displacement sensor, and multi-point displacement meter; The comprehensive evaluation index system in the intelligent analysis module specifically includes: Static parameters: fluidized medium bearing capacity, stress concentration index, energy storage coefficient; Dynamic indicators: energy release rate, event frequency anomaly, and vibration amplitude change rate; Stability indicators: displacement convergence rate, plastic zone development index, and crack propagation rate; The intelligent evaluation model adopts a deep learning model with LSTM network or Transformer architecture; the dynamic comprehensive index E(t) is a continuous value in the range of [0,1], and the closer its value is to 0, the better the prevention and control effect, and the closer it is to 1, the higher the disaster risk. The optimization decision-making module generates optimization control instructions based on the dynamic comprehensive index E(t) in the following specific way: When 0≤E(t)<0.3, the control effect is considered good, and the current mining and fluidization parameters are maintained; When 0.3≤E(t)<0.6, a potential risk is identified, an early warning message is generated, and it is recommended to strengthen monitoring. When 0.6≤E(t)<0.8, the risk is considered high, and control instructions are generated to adjust the mining intensity or fluidized medium parameters; When 0.8≤E(t)≤1, the risk is determined to be extremely high, and an emergency control instruction is generated, including slowing down mining operations and initiating supplementary pressure relief measures.

2. The intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining as described in claim 1, characterized in that: The data acquisition module also includes a signal conditioning submodule, a multi-channel synchronous acquisition submodule, an edge computing submodule, and a device health monitoring submodule; The signal conditioning submodule uses the following formula to calculate the signal quality index (SQI) for evaluating signal quality: ； wherein, is the noise standard deviation, is the signal standard deviation; when SQI < 0.7, a signal quality alarm is triggered; The multi-channel synchronous acquisition submodule uses a timestamp synchronization algorithm to ensure that the acquisition time deviation of each channel is less than 1ms.

3. The intelligent evaluation and real-time optimization system for preventing and controlling the effect of deep fluidization mining dynamic disaster according to claim 1, characterized in that: The intelligent analysis module also includes a data preprocessing submodule, a feature engineering submodule, a model management submodule, and an interpretability analysis submodule; The feature engineering submodule uses the following formula to calculate the feature importance score: ； where N is the number of samples, f(x) is the model output function, X i is the i-th feature; The interpretability analysis submodule uses SHAP value analysis to calculate the feature contribution using the following formula: ； Where M is the total number of features and S is the subset of features.

4. The intelligent evaluation and real-time optimization system for preventing and controlling the effect of deep fluidization mining dynamic disaster according to claim 1, characterized in that: The optimization decision-making module also includes a multi-objective optimization sub-module, a rule engine sub-module, a contingency plan management sub-module, and a simulation and deduction sub-module; The multi-objective optimization submodule employs the following multi-objective optimization function: ； F1(x), F2(x), and F3(x) are three optimization objective functions.

5. The intelligent evaluation and real-time optimization system for the prevention and control of dynamic disasters in deep fluidized mining according to claim 1, characterized in that: The execution feedback module also includes an instruction security verification submodule, an executor interface submodule, an effect evaluation submodule, and an adaptive learning submodule; The effect evaluation submodule uses the following comprehensive evaluation function: ； wherein, ΔE is the dynamic comprehensive index change rate, is the response time, is the resource consumption rate, is the weight coefficient, and .

6. The intelligent evaluation and real-time optimization system for preventing and controlling the effect of deep fluidization mining dynamic disasters according to claim 1, characterized in that: The human-computer interaction and visualization module also includes a 3D geological visualization submodule, a real-time alarm submodule, a decision dashboard submodule, and an operation log and audit submodule; The real-time alarm submodule uses the following risk warning function: ； Where E(t) is the dynamic composite index, Its rate of change, For each individual indicator, α, β, and γ represent the degree of abnormality, and α, β, and γ are the weighting coefficients. The weights of each indicator are given.

Images