A method and device for subarea caving of extra-thick hard coal

Abstract

The present application provides a kind of extra-thick hard coal sub-region caving coal dynamic control method and device.The method collects top coal migration state data;Multi-source data processing is carried out using data level, feature level and decision level fusion algorithm, to generate top coal fragmentation degree evaluation results, caving state identification results and roof stability analysis results;According to top coal fragmentation degree evaluation results, caving state identification results and roof stability analysis results, multi-model collaborative reasoning is carried out using D-S evidence theory, fuzzy logic and Bayesian network, to generate the opening and closing time prediction and control instruction of each area caving gate;The control instruction is issued to hydraulic support through electro-hydraulic control system, to realize the differential opening and closing control of sub-region caving gate, and dynamically adjust the control strategy according to real-time feedback data.The present application realizes the sub-region dynamic precise control of extra-thick hard coal seam caving coal process, significantly improves the top coal recovery rate and reduces the gangue mixing rate and support instability risk, improves the mining efficiency and safety.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of process control of fully mechanized top coal caving mining in extra-thick hard coal seams, and in particular to a method and device for dynamic control of regional top coal caving in extra-thick hard coal seams. BACKGROUND

[0002] As an important object of coal resource mining in China, extra-thick hard coal seams are widely distributed in large coal bases such as Shaanxi, Inner Mongolia and Xinjiang, and their reserves account for about 50% of thick coal seam resources, which is the key support for realizing high-yield and efficient mining in coal mines. In related technologies, the top coal caving mining process cooperates with the coal cutting by the coal mining machine, the support by the hydraulic support and the control of the coal caving mouth to build a whole-process control system from coal body breakage, migration monitoring to coal caving execution. Specifically, this technology covers key links such as data acquisition, state analysis and control decision, among which the top coal migration state monitoring, coal caving timing control and electro-hydraulic control system constitute the core support. With the improvement of the intelligentization and automation level of coal mines, the traditional top coal caving process gradually evolves towards multi-sensor fusion, dynamic prediction and intelligent control to cope with the mining challenges under complex geological conditions.

[0003] However, in the existing top coal caving control method, a unified coal caving command is directly used to control all coal caving mouths, which lacks real-time perception and dynamic response mechanism for the occurrence conditions and broken state of the top coal, which may cause over-caving or under-caving problems in the coal caving process, or cause serious safety hazards due to coal wall spalling and support instability. Specifically, when the mining and caving height increases, the top coal fragmentation degree and migration law show significant spatial heterogeneity, and the traditional electro-hydraulic control system cannot perform regional and differentiated control according to multi-source monitoring data, resulting in large fluctuations in coal caving efficiency, and it is difficult to stabilize the top coal recovery rate above 90%, and the gangue mixing rate generally exceeds 15%. Based on this, the existing technology is difficult to realize the coordinated optimization of high efficiency, safety and environmental protection under complex coal seam conditions, and it is urgent to build a new top coal caving process system with dynamic prediction and adaptive control capability. SUMMARY

[0004] The present application aims to at least partially solve one of the technical problems in the related art.

[0005] In order to improve the precision of coal caving control and the effect of coal caving, in view of the over-caving and under-caving problems in the coal caving process, the present application proposes a method for dynamic control of regional top coal caving in extra-thick hard coal seams. Through the data returned by the real-time monitoring system for the migration state of top coal installed inside the working face and on the hydraulic support, the top coal caving process is monitored and analyzed. When the occurrence conditions of the top coal above the working face change simultaneously at multiple coal caving mouths, the system will automatically adjust the parameters of the change at the same time, thereby improving the coal caving effect.

[0006] Another object of the present application is to propose a device for dynamic control of regional top coal caving in extra-thick hard coal seams.

[0007] To achieve the above object, the application provides a special thick hard coal sub-regional caving dynamic control method, which comprises the following steps:

[0008] S1, collecting top coal migration state data through a plurality of sensors arranged on hydraulic supports and a working face;

[0009] S2, based on the top coal migration state data, performing multi-source data processing by using data level, feature level and decision level fusion algorithms to generate top coal fragmentation degree evaluation results, caving state identification results and roof stability analysis results;

[0010] S3, according to the top coal fragmentation degree evaluation results, caving state identification results and roof stability analysis results, performing multi-model collaborative reasoning by using D-S evidence theory, fuzzy logic and Bayesian network to generate opening and closing time prediction and control instructions of each regional caving gate;

[0011] S4, issuing the control instructions to the hydraulic supports through an electro-hydraulic control system to realize differential opening and closing control of the regional caving gates, and dynamically adjusting the control strategy according to real-time feedback data.

[0012] The special thick hard coal sub-regional caving dynamic control method can further have the following additional technical features:

[0013] In an embodiment of the application, the sensors include pressure displacement integrated sensors, inclination sensors, height measuring sensors, laser displacement sensors, microseismic sensors and vibration acceleration sensors; the collecting of the top coal migration state data through the plurality of sensors arranged on the hydraulic supports and the working face further comprises:

[0014] S11, the pressure displacement integrated sensors are pre-installed in the hydraulic cylinders to collect the support working resistance and displacement in real time;

[0015] S12, the microseismic sensors are installed behind the roof anchor rods and the coal wall to monitor the top coal crack propagation state.

[0016] In an embodiment of the application, the multi-source data processing based on the top coal migration state data by using the data level, feature level and decision level fusion algorithms further comprises:

[0017] S21, in the data level fusion stage, the wavelet threshold denoising is used to process the hydraulic impact noise, and the IEEE 1588 protocol is used to realize time synchronization of the sensor data;

[0018] S22, in the feature level fusion stage, the cubic spline interpolation is used to generate a continuous deformation curve of the roof displacement, and the curvature extreme value and the change rate thereof are extracted as feature parameters.

[0019] In an embodiment of the present application, the multi-model collaborative reasoning using D-S evidence theory, fuzzy logic and Bayesian network further comprises:

[0020] S31, the D-S evidence theory is used for fusing the confidence of the strength of the pressure-vibration model output, the probability of gangue mixing of the coal body vibration recognition model output and the risk level of the roof stability model output;

[0021] S32, the Bayesian network dynamically updates the prior probability according to real-time data, and adjusts the decision weight to realize risk adaptive balance.

[0022] In an embodiment of the present application, further comprising:

[0023] S5, when the rear scraper conveyor exists pose offset or is not pushed to the position, the system automatically switches to the manual intervention mode, and pushes the coal deposit operation suggestion and the current area top coal state information to the operator;

[0024] S6, in the manual intervention mode, the system generates a corrected coal deposit opening and closing time control strategy according to the coal deposit instruction input by the operator, in combination with the top coal fragmentation degree and the deposit state of the current area.

[0025] To achieve the above purpose, a second aspect embodiment of the present application provides a special thick hard coal subarea top coal deposit dynamic control device, comprising: a data acquisition module for acquiring top coal migration state data through a plurality of sensors arranged on hydraulic supports and working faces, the sensors including pressure displacement integrated sensors, inclination sensors, height measuring sensors, laser displacement sensors, microseismic sensors and vibration acceleration sensors; a multi-source data processing module for processing multi-source data based on the top coal migration state data by using data level, feature level and decision level fusion algorithms to generate top coal fragmentation degree evaluation results, deposit state recognition results and roof stability analysis results; a multi-model collaborative reasoning module for generating opening and closing time prediction and control instructions of each area coal deposit according to the evaluation results and analysis results by using D-S evidence theory, fuzzy logic and Bayesian network for multi-model collaborative reasoning; an electro-hydraulic control execution module for issuing the control instructions to the hydraulic support through the electro-hydraulic control system to realize differential opening and closing control of the subarea coal deposit, and dynamically adjusting the control strategy according to real-time feedback data.

[0026] The embodiment of the present application designs a top coal migration state real-time monitoring system based on multi-sensor perception, constructs a top coal fragmentation degree dynamic evaluation model, analyzes the top coal thickness and caving law, formulates an automatic coal caving control mechanism of top coal caving timing control and manual intervention coordinated control, optimizes the opening and closing timing of the coal caving mouth, carries out regional accurate coal caving according to the top coal fragmentation degree, and reduces the gangue mixing and top coal loss. In order to improve the coal caving recovery rate and reduce resource waste, a nonlinear coal caving process is proposed, the coal caving sequence is adjusted according to the real-time coal caving state, and the regional top coal caving is realized through the electro-hydraulic control system of the hydraulic support. The method is beneficial to realize the intelligent control of the top coal caving process in the super-thick hard coal seam, and improve the coal caving efficiency and resource recovery rate.

[0027] The embodiment of the present application provides a regional top coal caving mining process method for super-thick hard coal, which comprises a top coal migration state real-time monitoring system, a coal caving best time dynamic prediction system and a coal caving mining process control system, and each system cooperates to realize efficient and accurate control of super-thick hard coal seam mining.

[0028] In order to improve the coal caving control accuracy and effect, a top coal caving best time dynamic prediction method and system are proposed to solve the over-caving and under-caving problems in the coal caving process. Through the data returned by the top coal migration state real-time monitoring system installed in the working face and the hydraulic support, the top coal caving process is monitored and analyzed. When the top coal occurrence conditions above the working face change simultaneously at multiple coal caving mouths, the system will automatically adjust the changed parameters at the same time, and improve the coal caving effect.

[0029] According to a specific implementation manner of the embodiment of the present application, the top coal migration state real-time monitoring system is composed of two types of sensors, namely support top coal state monitoring sensors and coal mining process monitoring sensors, and a data processing module. The two types of sensors are arranged in the working face, and the data collected by each sensor is transmitted to the crossheading working face centralized control center. The support top coal state monitoring sensors mainly include pressure displacement integrated sensors, inclination sensors, height measuring sensors and laser displacement sensors, which monitor the key data such as support working resistance, hydraulic support posture, roof subsidence and the like. The pressure displacement integrated sensors are pre-installed in the hydraulic cylinder during the manufacturing of the support, and the inclination sensors, height measuring sensors and laser displacement sensors are installed on the hydraulic support structure. The above-mentioned sensors are used to measure the support and mining state data in the mining process. The coal mining process monitoring sensors mainly include microseismic sensors and vibration acceleration sensors. The microseismic sensors are installed at the back of the roof anchor and the coal wall to measure the top coal crack propagation state in real time. The vibration acceleration sensors are installed at the web of the tail beam of the hydraulic support to detect the vibration data generated by the impact of the top coal and gangue falling on the upper part of the hydraulic support on the tail beam structure, and the dynamic data of the top coal caving are calculated and analyzed according to the vibration data.

[0030] According to a specific implementation manner of an embodiment of the present application, after the two categories of sensors, i.e., the support top coal state monitoring sensor and the coal mining process monitoring sensor, collect working face data, the working face data are transmitted to a data processing module of a crossheading working face centralized control center for processing, and through a multi-dimensional, multi-physical field data fusion algorithm, the working face data are processed and analyzed in cooperation, so that accurate perception and decision optimization of a top coal caving state, surrounding rock stability and coal caving efficiency are realized. The data fusion algorithm is pushed forward layer by layer from three levels of data level, feature level and decision level, and the specific implementation technical route and steps are as follows:

[0031] Data level fusion:

[0032] In the data level fusion stage, firstly, the original data of different data source sensors are preprocessed and time-space aligned. The pressure-displacement integrated sensor collects displacement and pressure fluctuation signals of the support hydraulic cylinder in real time, and the hydraulic impact noise is eliminated through wavelet threshold denoising; the vibration sensor monitors the vibration spectrum caused by coal body caving, and the characteristic frequency band of the vibration caused by the coal body is extracted after Fourier transform; the laser displacement sensor records the roof subsidence trajectory, and the abnormal jump is smoothed through Kalman filtering. After the data are preliminarily processed, time synchronization is realized based on IEEE 1588, PTP accurate time and other protocols, and the ICP point cloud registration algorithm is used to map to a unified three-dimensional coordinate system to eliminate spatial deviation. After preprocessing, the data are standardized and normalized to form a time-space aligned multi-modal data set.

[0033] Feature level fusion:

[0034] In the feature level fusion stage, the algorithm extracts high-discrimination physical features and pattern features from multi-source data. For pressure-vibration time series data, time domain statistics (mean, variance, kurtosis) and frequency domain indicators (main frequency, energy entropy, spectral kurtosis) are calculated, and a multi-dimensional joint feature vector is constructed; the roof displacement data are processed through cubic spline interpolation to generate a continuous deformation curve, and the curvature extreme value and change rate are extracted. Subsequently, the data dimension reduction method is used for processing, and finally a compact and information-rich fusion feature vector is generated, which significantly improves the efficiency of subsequent classification and regression models.

[0035] Decision level fusion:

[0036] In the decision level fusion stage, the system realizes multi-model collaborative reasoning based on D-S evidence theory, fuzzy logic and Bayesian network. Firstly, for the identification of top coal caving state, the pressure-vibration characteristics are input into the support vector machine, and the caving intensity confidence (0~1) is output. The coal body caving vibration characteristics are calculated by the random forest classifier to obtain the gangue mixing probability (0~100%). The roof displacement data are input into the neural network model to predict the roof stability risk level (low / medium / high). The D-S evidence theory is used to fuse the outputs of each model: the basic probability assignment function (BPA) is defined, the confidence of the pressure-vibration model on "normal caving", the confidence of the coal body caving vibration identification model on "excessive gangue" and the confidence of the working face support state are processed synchronously, and the joint confidence is calculated to solve the uncertainty problem of a single sensor. At the same time, a fuzzy logic controller is constructed, the control rule base of the analysis conclusion of the roof subsidence rate, the gangue content, the top coal caving state and the instruction of closing the caving gate, changing the caving gate opening degree is formulated, the continuous variables are converted into fuzzy language variables, and the accurate control instruction is output after processing. Finally, the Bayesian network is used to dynamically update the prior probability, and the decision weight is adjusted according to the real-time data, for example, when the roof monitoring data shows that the roof fracture risk rises, the weight of the safety factor is automatically increased, and the risk adaptive balance is realized.

[0037] In order to optimize the precision and effect of multi-source data fusion processing, the system introduces an online learning mechanism: the reinforcement learning algorithm is used to dynamically adjust the hyperparameters of feature extraction and decision model. A three-dimensional reward function is constituted by the recovery rate, the gangue rate and the safety risk, the network weight is updated through the policy gradient, so that the model continuously optimizes the decision ability and accuracy under the feedback of real data.

[0038] According to a specific implementation mode of the embodiment of the application, the caving optimal time dynamic prediction system is used to pre-set or real-time adjust the time point and sequence of caving according to the geological conditions of the coal seam, the thickness of the coal seam, the inclination of the coal seam, the top coal crushing degree and the mining progress and other factors. The system obtains the real-time monitoring of the change of the coal seam, the distribution of the coal block and the dynamic data in the caving process through the sensors installed on the coal mining face, and generates the optimal caving time sequence scheme based on the analysis and processing of the above data.

[0039] According to the random medium ore drawing theory, the broken top coal is regarded as a continuous flowing random medium, according to the curve characteristics of the moving speed law of the bulk top coal and the mutual position of the bottom end of the curve and the center line of the ore drawing opening, the bulk top coal moving area is divided into a plurality of sections from top to bottom with the height characteristics as the reference, when the hydraulic support draws the coal, the broken coal will move along the movement curve to the ore drawing opening, the relationship between the drawing amount and the coal position transformation can be obtained by calculating the relationship between the time and the moving track line of the coal and the gangue, the mathematical model of the size of the ore drawing opening, the opening time and the coal drawing amount is derived, the ore drawing time control model is established based on the above model, the multi-objective optimization control function is combined to establish a multi-objective function including the recovery rate (≥ 90%), the gangue content (≤ 15%) and the energy consumption efficiency, the ore drawing time is automatically adjusted according to the ore drawing state change of the continuous multiple ore drawing openings, so that the ore drawing control state is always in the state of meeting the current regional ore drawing working condition, and the whole process regional automatic top coal drawing can be realized by cooperating with the electro-hydraulic control system of the hydraulic support.

[0040] According to a specific implementation manner of the embodiment of the present application, the ore drawing recovery process control system runs in the workstation of the crossheading control center, reads the decision result output by the ore drawing strategy optimization algorithm, decomposes and transforms the decision instruction into machine control instruction, and according to the priority order of the regional ore drawing process flow, the control instruction is issued to the electro-hydraulic control system through the field bus to form a closed loop of 'perception-fusion-decision-execution'. In the ore drawing process, the control system can automatically control the lifting of the hydraulic support, the running speed of the scraper conveyor and the opening and closing of the ore drawing opening, so that the top coal is released at the best time point, and the waste of resources or safety hazards caused by too early or too late ore drawing is avoided. For example, when the sensor detects that the coal seam is thick, the system will automatically prolong the ore drawing time to ensure that more coal is mined out, and when the coal seam is thin or there is a fault, the system will shorten the ore drawing time to avoid the safety risk caused by over-mining.

[0041] According to a specific implementation manner of the embodiment of the present application, the regional ore drawing process flow is: in the process of top coal drawing, the process flow of 'coal cutting→support moving→pushing the front scraper conveyor→top coal drawing→pulling the rear scraper conveyor→moving the transfer conveyor→next cycle' needs to be executed, for the top coal drawing process, the working face is divided into a plurality of regions, and according to the decision result obtained by the top coal movement state real-time monitoring system and the ore drawing best time dynamic prediction system proposed in the present application, the regional ore drawing control is performed.

[0042] The embodiment of the present application proposes a top coal movement state real-time monitoring system, an ore drawing best time dynamic prediction system and an ore drawing recovery process control system, and the systems work cooperatively to realize efficient and accurate control of the thick hard coal seam mining.

[0043] Compared with the prior art, the present application has the following specific advantages:

[0044] 1) Through accurate time control and dynamic adjustment, automation of coal mining is realized, not only the resource utilization is improved, but also the error rate and safety risk of manual operation are significantly reduced, which provides strong technical support for modern coal mining;

[0045] 2) The caving optimal time dynamic prediction system can effectively improve the caving control compared with the traditional control method.

[0046] The method and device of the embodiment of the present application realize the regional dynamic precise control of the top coal caving process of the super-thick hard coal seam, significantly improve the top coal recovery rate and reduce the gangue mixing rate and safety hazards.

[0047] The additional aspects and advantages of the present application will be partially given in the following description, partially will become obvious from the following description, or will be understood through the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0048] The above and / or additional aspects and advantages of the present application will become apparent and more readily appreciated from the following description of the embodiments, with reference to the following figures, wherein:

[0049] Figure 1 A flowchart of a regional dynamic control method for top coal caving of super-thick hard coal is provided for the embodiment of the present application;

[0050] Figure 2 A structural diagram of a regional dynamic control device for top coal caving of super-thick hard coal is provided for the embodiment of the present application. DETAILED DESCRIPTION

[0051] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0052] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the present application.

[0053] A regional dynamic control method and device for top coal caving of super-thick hard coal according to the embodiments of the present application will be described below with reference to the accompanying drawings.

[0054] Figure 1 is a flowchart of a regional dynamic control method for top coal caving of super-thick hard coal according to the embodiments of the present application, likeFigure 1 as shown, comprising:

[0055] S1, by arranging a plurality of sensors on the hydraulic support and the working face to collect the top coal migration state data.

[0056] Specifically, this step realizes real-time data collection of the top coal migration state by arranging a plurality of sensors on the hydraulic support and the working face, and is the core implementation link of the "top coal migration state real-time monitoring system" in the present application. In some implementations, the system is composed of a support top coal state monitoring sensor and a coal mining process monitoring sensor, and specifically includes a pressure displacement integrated sensor, an inclination sensor, a height measuring sensor, a laser displacement sensor, a microseismic sensor, and a vibration acceleration sensor, which are respectively used to monitor key parameters such as the support state of the hydraulic support, the roof deformation feature, the coal body crack expansion, and the falling dynamic.

[0057] The pressure displacement integrated sensor is usually pre-installed in the hydraulic cylinder inside the hydraulic support column or tail beam, and is used to synchronously collect the displacement amount (precision ±0.1 mm) and the working resistance (range 0~100 MPa, sampling frequency 100 Hz) of the hydraulic cylinder, to provide high-precision support data for the support state of the support. The inclination sensor (range ±15°, resolution 0.01°) and the height measuring sensor (range 0~10 m, precision ±2 mm) are installed on the support structure, and are used to monitor the attitude change of the support and the roof subsidence amount, to ensure the stability of the support system. The laser displacement sensor (measurement range 0~5 m, precision ±0.5 mm) is used to obtain the roof subsidence trajectory in real time, to provide spatial data support for the top coal migration modeling.

[0058] In the coal mining process monitoring sensor, the microseismic sensor (frequency response range 1~1000 Hz, sensitivity 100 mV / g) is arranged behind the roof anchor or the coal wall, and is used to capture the microseismic signals generated by the top coal crack expansion. The vibration acceleration sensor (range ±50g, sampling rate 2000 Hz) is installed at the tail beam web, and can detect the impact vibration of the support structure when the top coal and the gangue fall, so as to identify the dynamic characteristics of the falling coal body.

[0059] The sensor data described above is transmitted to the crossheading working face centralized control center through the CAN bus or the industrial Ethernet, to realize the centralized collection and synchronous processing of multi-source heterogeneous data. This step provides basic data support for the subsequent feature extraction, data fusion, and coal drawing decision, and is the key prerequisite for realizing the regional precise coal drawing control. Through the collaborative perception of multiple sensors, the system can master the top coal breaking state, the migration law, and the surrounding rock stability in real time, to provide a scientific basis for the intelligent opening and closing control of the coal drawing opening, so as to significantly improve the coal drawing efficiency and the resource recovery rate, reduce the gangue mixing rate, and ensure the safe and efficient mining of the working face.

[0060] Further, S1 comprises:

[0061] S11, the pressure displacement integrated sensor is pre-installed in the hydraulic cylinder, and is used for collecting the working resistance and displacement of the support in real time.

[0062] Specifically, the pressure displacement integrated sensor is pre-installed in the hydraulic cylinder, and is used for collecting the working resistance and displacement of the support in real time, which is one of the key components of the real-time monitoring system for the top coal migration state in the application. The sensor realizes the synchronous measurement of the internal hydraulic oil pressure of the hydraulic support column or the jack and the displacement of the piston rod by integrating a high-precision pressure sensor and a linear displacement sensor, so as to reflect the force state and structural deformation of the support in the supporting process.

[0063] In terms of technical implementation, the pressure displacement integrated sensor usually adopts the combination of a MEMS (Micro-Electro-Mechanical System) pressure sensing element and an LVDT (Linear Variable Differential Transformer) or a magnetostrictive displacement sensing technology. The pressure sensing part measures the internal oil pressure of the hydraulic cylinder to calculate the roof pressure borne by the support, and the measurement range is generally 0~100 MPa, and the accuracy can reach ±0.5%FS; the displacement sensing part realizes real-time acquisition of the extension and retraction displacement of the piston rod through a non-contact measurement method, and the measurement accuracy is ±0.1 mm, and the measurement range is usually 0~1000 mm, which meets the monitoring needs of the hydraulic support under different supporting heights. The sensor communicates with the crossheading working face centralized control center through RS485 or CAN bus protocol to ensure the real-time and stability of data transmission.

[0064] The data collected by the sensor includes instantaneous pressure value, average pressure value, pressure fluctuation frequency, displacement change rate, displacement-time curve, etc., which are used to evaluate the crushing degree of the top coal, the supporting state of the support and the subsidence trend of the roof. For example, when the working resistance of the support continuously decreases and the displacement increases, it may indicate that the top coal has been significantly crushed and starts to collapse, and the system can judge that the region enters the caving window period.

[0065] The sensor is deployed in the column or tail beam jack of the hydraulic support, and is suitable for dynamic monitoring of the fully mechanized caving face in the thick and hard coal seam. During the caving process, the system analyzes the resistance and displacement data of multiple supports to identify the starting point and ending point of the top coal caving, provides key input for the dynamic prediction system of the optimal caving time, and realizes intelligent control of regional caving.

[0066] The technical value of this step lies in that through high-precision and real-time force and displacement data collection, reliable basis is provided for subsequent data fusion and decision model, the response speed and accuracy of caving control are effectively improved, and resource waste and safety hazards caused by over-caving or under-caving are reduced, which is an important foundation for realizing intelligent top coal caving mining in thick and hard coal seams.

[0067] S12, the microseismic sensor is installed behind the roof anchor and the coal wall for monitoring the roof coal crack propagation state.

[0068] Specifically, the microseismic sensor is installed behind the roof anchor and the coal wall for monitoring the roof coal crack propagation state, which is an important part of the real-time monitoring system for the roof coal migration state in the present application. The sensor senses the microseismic signals caused by crack propagation, coal body crushing, and roof coal migration in the coal rock mass during mining, thereby achieving dynamic monitoring of the internal structure changes of the roof coal. The microseismic signals are usually high-frequency and low-energy elastic waves, and their propagation characteristics are closely related to the expansion direction, speed, and energy release of the coal body crack.

[0069] In a specific implementation, the microseismic sensor adopts a high-sensitivity piezoelectric ceramic or MEMS (Micro-Electro-Mechanical System) structure, and its frequency response range is usually set to 100 Hz-2000 Hz, and the sampling frequency is not less than 10 kHz, so as to ensure high-precision capture of microseismic events. The sensor installation position needs to avoid direct mechanical vibration interference, and is usually arranged at the end of the roof anchor or within 1-3 meters behind the coal wall, and is fixed by screw or glued to the anchor or coal surface for close coupling to improve signal transmission efficiency. During installation, the relevant standards of the Coal Mine Safety Regulations (GB 3836) and the General Technical Requirements for Coal Mine Underground Monitoring and Control Systems (AQ 1029) need to be followed to ensure stable operation of the sensor in the complex underground environment with high humidity, high dust, and high impact.

[0070] The original data collected by the microseismic sensor are transmitted to the crossheading working face centralized control center through RS485 or CAN bus, and are fused with multi-source data such as pressure displacement sensors and vibration acceleration sensors. In the data level fusion stage, the microseismic signals are subjected to time-frequency analysis by wavelet transform or short-time Fourier transform to extract the characteristic energy distribution and main frequency component of crack propagation, and a dynamic evaluation model of the roof coal crushing degree is constructed in combination with the roof displacement and support posture data. The model can identify the active area of the roof coal crack propagation, and provide key input parameters for the dynamic prediction system of the optimal caving time.

[0071] Through this step, the system can realize real-time sensing of the evolution process of the internal crack of the roof coal, judge whether the roof coal is in the state of imminent collapse or has formed a high-crushing-degree area, thereby providing a scientific basis for regional caving control. In actual application, this technology can effectively reduce the coal loss and gangue mixing caused by improper caving time, improve the roof coal recovery rate to more than 90%, and reduce the safety risks in the mining process, such as coal wall spalling and roof instability. Therefore, this step plays a key role in the present application, and is one of the core supporting technologies for realizing intelligent and precise top coal caving mining in super-thick hard coal seams.

[0072] S2, based on the top coal migration state data, multi-source data processing is carried out by using data level, feature level and decision level fusion algorithm to generate top coal crushing degree evaluation result, caving state recognition result and roof stability analysis result.

[0073] Specifically, based on the top coal migration state data, multi-source data processing is carried out by using data level, feature level and decision level fusion algorithm to generate top coal crushing degree evaluation result, caving state recognition result and roof stability analysis result, which is the core link of realizing intelligent control of regional top coal caving in super-thick hard coal seam.

[0074] Firstly, in the data level fusion stage, the original data from the pressure displacement integrated sensor, the inclination sensor, the height measuring sensor, the laser displacement sensor, the microseismic sensor and the vibration acceleration sensor are preprocessed. The preprocessing includes wavelet threshold denoising, Fourier transform spectrum analysis, Kalman filter trajectory smoothing and other methods to eliminate noise interference and improve data quality. Subsequently, the system aligns the multi-source heterogeneous data in time and space dimensions through the IEEE 1588 time synchronization protocol and the ICP point cloud registration algorithm, and constructs a unified three-dimensional space-time data set, laying a foundation for subsequent feature extraction and decision analysis.

[0075] In the feature level fusion stage, the system extracts time domain and frequency domain features such as mean, variance, kurtosis, main frequency and energy entropy from the preprocessed data, and constructs a multi-dimensional joint feature vector. The curvature extreme value and change rate of the continuous deformation curve of the roof are extracted by cubic spline interpolation. To improve the calculation efficiency, the system uses dimension reduction methods such as principal component analysis (PCA) or t-SNE to generate a compact and information-complete fusion feature vector for subsequent model input.

[0076] In the decision level fusion stage, the system uses D-S evidence theory, fuzzy logic and Bayesian network for multi-model collaborative reasoning. For example, support vector machine (SVM) is used to identify the strength of top coal caving, outputting confidence (0~1); random forest classifier is used to calculate the probability of gangue mixing (0~100%); neural network model is used to predict the roof stability risk level (low / medium / high). Through D-S theory fusion of each model output, a fuzzy control rule base is constructed to convert continuous variables into fuzzy language variables, and finally accurate coal caving control instructions are output.

[0077] In addition, the system introduces an online reinforcement learning mechanism to dynamically adjust model parameters with the optimization objectives of recovery rate (≥90%), gangue content (≤15%) and energy efficiency, realizing continuous optimization of the decision model. This step plays a role in connecting the previous and the next steps in the whole process, converting multi-source perception data into executable coal caving control strategies, significantly improving coal caving efficiency, resource recovery rate and working face safety, and providing key support for intelligent and precise mining of super-thick hard coal seams.

[0078] Further, S2 comprises:

[0079] S21, the data-level fusion stage adopts wavelet threshold denoising to process hydraulic impact noise, and realizes time synchronization of sensor data through IEEE 1588 protocol.

[0080] Specifically, in the data-level fusion stage, the present application adopts wavelet threshold denoising technology to process hydraulic impact noise, so as to improve the signal-to-noise ratio and reliability of the top coal migration state monitoring data. Specifically, the displacement and pressure fluctuation signals collected by the pressure-displacement integrated sensor often contain high-frequency noise caused by hydraulic system impact, mechanical vibration, etc., which will interfere with the accurate identification of the top coal caving state. Therefore, the system uses discrete wavelet transform (DWT) to decompose the original signal into 5-7 layers, usually selects db4 or sym8, etc. with good time-frequency localization characteristics as the mother wavelet basis function, extracts the detail coefficients and approximation coefficients of different frequency subbands. Then, the soft threshold or hard threshold function is applied to the detail coefficients for noise suppression, wherein the threshold selection can be based on the general threshold formula (such as VisuShrink) or the adaptive threshold method (such as Stein’s Unbiased Risk Estimate, SURE), so as to balance the denoising effect and signal fidelity. The denoised signal is reconstructed through inverse wavelet transform to obtain clearer top coal caving dynamic characteristics.

[0081] At the same time, the system realizes time synchronization of multi-sensor data through IEEE 1588 Precision Time Protocol (PTP). IEEE 1588 protocol is based on master-slave clock mechanism, through timestamp exchange and delay compensation algorithm, the local time of each sensor node is synchronized with the master clock, the time synchronization accuracy can reach sub-microsecond level (<1 μs), which meets the high-precision time alignment requirements of multi-source heterogeneous data in complex electromagnetic environment of coal mine underground. In actual deployment, each sensor node needs to be configured with a hardware timestamp module supporting PTP, and the control center acts as the master clock source, periodically sends synchronization messages (Sync Message) and delay request messages (Delay Request Message), to realize accurate calibration of the sampling time of each sensor.

[0082] This step plays a key preprocessing role in the whole system, providing high-quality and high-consistency multi-modal data basis for subsequent feature-level and decision-level fusion. Through the joint processing of wavelet denoising and time synchronization, the system can effectively eliminate noise interference, ensure the accurate alignment of key parameters such as top coal caving state, support posture, and coal and gangue identification in time and space dimensions, thereby improving the real-time and accuracy of coal caving control, and providing reliable data support for intelligent and efficient mining of super-thick hard coal seams.

[0083] S22, the feature level fusion stage generates a roof displacement continuous deformation curve using cubic spline interpolation, and extracts the curvature extreme value and change rate as feature parameters.

[0084] Specifically, in the feature level fusion stage, the present application uses the cubic spline interpolation method to process the roof displacement data, generate a continuous deformation curve, and extract the curvature extreme value and change rate as feature parameters for representing the dynamic response characteristics of the roof during the top coal caving process. The technical implementation of this step is based on the roof displacement time series data collected by multiple sensors, such as laser displacement sensors, altimeter sensors, etc. The original data of these sensors are usually discrete sampling points, which have problems such as inconsistent time intervals and uneven spatial distribution. Therefore, the cubic spline interpolation method is used to interpolate and fit these discrete data, construct a continuous and smooth roof displacement deformation curve, and thus more accurately reflect the overall movement trend of the roof during the coal caving process.

[0085] In the specific implementation, the cubic spline interpolation uses the natural boundary condition (Natural Spline), i.e., the second derivative at the two endpoints of the curve is zero, to ensure the smoothness of the interpolation curve at the boundary. In the interpolation process, a cubic polynomial is fitted between each adjacent sampling point to ensure the continuity of the first and second derivatives at the nodes. The interpolated roof displacement curve has continuity, which can effectively eliminate noise interference and abrupt points in the original data and improve the stability and accuracy of feature extraction.

[0086] In the feature extraction aspect, the system first calculates the curvature function of the roof displacement curve, which is defined as the bending degree of the displacement curve at a certain point, and its mathematical expression is:

[0087]

[0088] where the numerator represents the modulus of the cross product of the velocity vector and the acceleration vector, which is proportional to the curvature of the path at that point, and the denominator is the cube of the modulus of the velocity vector, which is used for normalization to ensure that the curvature calculation is independent of the speed of the path; is the curvature, x'(t) represents the speed of the path in the x direction, i.e., the rate of change of the x position with respect to time t; y'(t) represents the speed of the path in the y direction, i.e., the rate of change of the y position with respect to time t; x''(t) represents the acceleration of the path in the x direction, i.e., the rate of change of the x direction speed with respect to time t; y''(t) represents the acceleration of the path in the y direction, i.e., the rate of change of the y direction speed with respect to time t.

[0089] By traversing the entire curve, the curvature extreme point is extracted as the key feature of the roof deformation or fracture. In addition, the system also calculates the curvature derivative of the roof displacement curve, that is, the change gradient of the curvature with time, to identify the acceleration or deceleration trend of the roof deformation, so as to judge whether the top coal is in the state of imminent collapse or stability.

[0090] The feature parameters extracted in this step will be the key input in the subsequent decision level fusion, combined with pressure-vibration features, gangue recognition results, etc., through D-S evidence theory and fuzzy logic for multi-model collaborative reasoning to realize dynamic optimization control of the opening and closing time of the coal drawing hole. In practical application, the curvature extreme value and the change rate of the roof displacement curve can be used as quantitative indicators of the degree of top coal crushing and the state of drawing, and the typical threshold range is curvature extreme value , change rate . By setting reasonable feature thresholds, the system can realize real-time identification and early warning of the top coal drawing state, significantly improving the accuracy and safety of the coal drawing control.

[0091] S3, according to the evaluation results and analysis results, using D-S evidence theory, fuzzy logic and Bayesian network for multi-model collaborative reasoning to generate the opening and closing time prediction and control instructions of the coal drawing hole in each area.

[0092] Specifically, in some implementations, the step of "according to the evaluation results and analysis results, using D-S evidence theory, fuzzy logic and Bayesian network for multi-model collaborative reasoning to generate the opening and closing time prediction and control instructions of the coal drawing hole in each area" in the method of the present application is the core decision-making link for realizing intelligent control of top coal drawing. This step fuses multi-source heterogeneous data, combines three mainstream uncertainty reasoning methods, and constructs a multi-model collaborative decision-making mechanism to improve the accuracy and adaptability of the coal drawing control.

[0093] In terms of technical implementation, the system first extracts and fuses the multi-dimensional data (such as pressure, displacement, vibration, microseismic, etc.) from the real-time monitoring system of the top coal migration state to form a decision input with physical meaning. Subsequently, the Dempster-Shafer (D-S) evidence theory is used to fuse the outputs of multiple models, define the basic probability assignment function (BPA), combine the confidence levels of each model for "normal falling", "gangue exceeding", "roof instability" and other states, and calculate the joint confidence to solve the uncertainty problem of a single sensor or model under complex working conditions. At the same time, a fuzzy logic controller is constructed to convert continuous variables such as roof subsidence rate (unit: mm / min), gangue content (unit: %), and top coal falling state into fuzzy language variables (such as "high", "medium", "low"), and establish a control rule library based on expert experience and historical data to realize dynamic adjustment of the opening and closing time of the caving mouth. In addition, Bayesian networks are used to dynamically update the prior probability and adjust the decision weight of each model according to real-time monitoring data, for example, when the roof fracture risk rises, the system automatically increases the weight of the safety factor to achieve adaptive balance of risk.

[0094] The system sets the confidence level of top coal falling intensity to [0, 1], the probability of gangue mixing to [0%, 100%], and the roof stability risk level to low, medium, and high. The control command output includes the opening and closing time of the caving mouth (unit: seconds), the opening degree (unit: mm), and the execution priority (0-5 levels). The system generates an optimal control strategy by considering the recovery rate (≥90%), the gangue content (≤15%), and the energy efficiency through a multi-objective optimization function.

[0095] This step is suitable for regional caving control in fully mechanized top coal caving faces in extremely thick and hard coal seams, especially in complex top coal occurrence conditions, large changes in coal seam thickness, and fault development. It can dynamically adjust the caving strategy according to real-time monitoring data to avoid over-caving or under-caving, improve resource recovery rate, and reduce safety risks. For example, in areas with thick coal seams, the system can extend the caving time and increase the opening degree of the caving mouth, while in areas with thin coal seams or faults, it can shorten the caving time and reduce the opening degree to prevent gangue mixing and roof instability.

[0096] The technical effect of this step is that through the multi-model collaborative reasoning mechanism, the intelligent level and dynamic response capability of caving control are effectively improved, solving the problem of unified caving command and inability to adaptively adjust in traditional electro-hydraulic control systems, thereby significantly improving the caving efficiency, resource recovery rate, and safety of the working face, providing key technical support for efficient and green mining of extremely thick and hard coal seams.

[0097] Further, S3 comprises:

[0098] S31, the D-S evidence theory is used to fuse the caving intensity confidence output of the pressure-vibration model, the gangue mixing probability output of the coal body vibration recognition model, and the risk level output of the roof stability model.

[0099] Specifically, in the coal drawing optimal time dynamic prediction system of the application, the D-S evidence theory (Dempster-Shafer Theory) is used to fuse multi-source decision information from the pressure-vibration model, the coal body vibration recognition model, and the roof stability model, thereby improving the intelligent level and decision reliability of the coal drawing control. The core of this step is to generate a comprehensive decision result by weighting and fusing the confidence of the outputs of multiple models through an uncertainty reasoning mechanism, so as to guide the sub-regional coal drawing operation of the hydraulic support.

[0100] In terms of technical implementation, the D-S evidence theory quantitatively processes the outputs of each model by defining a basic probability assignment function (BPA). The caving intensity confidence output of the pressure-vibration model is a continuous value in the interval [0, 1], representing the possibility of top coal caving at the current time; the gangue mixing probability output of the coal body vibration recognition model is a percentage value in the interval [0, 100%], reflecting the risk degree of gangue mixing during caving; the roof stability model outputs a risk level, usually divided into low, medium, and high, corresponding to different safety thresholds. The system maps the outputs of the above three types of models to the BPA of basic propositions, calculates the joint confidence through the orthogonal sum (Dempster's Rule of Combination), and thereby comprehensively judges whether the current region is suitable for coal drawing.

[0101] The system sets the BPA weight coefficients of each model, for example, the weight of the pressure-vibration model is 0.4, the weight of the coal body vibration recognition model is 0.3, and the weight of the roof stability model is 0.3. The weights can be dynamically adjusted according to actual working conditions through an online learning mechanism. In addition, the system sets a confidence threshold, when the joint confidence is higher than 0.75, it is determined as "suitable for coal drawing"; when it is lower than 0.55, it is determined as "not suitable for coal drawing", and in the intermediate interval, it enters the "observation" state, which is further processed by the fuzzy logic controller.

[0102] This fusion step runs in the decision engine of the crossheading working face centralized control center, and real-time receives sensor data from each hydraulic support, combines the current mining progress, coal seam thickness, top coal fragmentation, and other parameters to dynamically evaluate multiple coal drawing regions. For example, when the roof fracture risk rises, the system automatically increases the weight of the roof stability model, thereby giving priority to safety factors in decision-making to avoid roof fall accidents caused by over-caving.

[0103] The technical effect of this step is that the uncertainty problem of single model output is effectively solved by D-S evidence theory, and the robustness and accuracy of the caving decision are improved. In practical application, this fusion mechanism can significantly reduce the gangue mixing rate (controlled below 15%), improve the top coal recovery rate (≥90%), and realize adaptive optimization of the caving process, providing key support for intelligent and efficient mining of ultra-thick hard coal seams.

[0104] S32, the Bayesian network dynamically updates the prior probability according to real-time data and adjusts the decision weight to realize risk adaptive balance.

[0105] Specifically, in some implementations, the Bayesian network dynamically updates the prior probability according to real-time data and adjusts the decision weight to realize risk adaptive balance, which is based on the technical implementation principle of probability reasoning and uncertainty modeling. As a directed acyclic graph model, the Bayesian network describes the causal relationship between key variables such as top coal caving state, surrounding rock stability, and gangue mixing through conditional probability distribution (CPD) between nodes. In the process of top coal caving mining, the system collects multi-source heterogeneous data (such as pressure, displacement, vibration, microseismic, etc.) through real-time monitoring of top coal migration state, inputs them into the Bayesian network model as observation values of evidence nodes, and updates the posterior probability distribution of each node.

[0106] The system first constructs an initial Bayesian network structure based on historical data and expert knowledge, defines key variables such as top coal caving state, roof stability, caving opening degree, and caving time, and their dependency relationships. During operation, the system continuously receives real-time data streams from sensors and dynamically corrects the prior probability through online learning mechanism. For example, when the roof monitoring data shows that the roof fracture risk is rising, the system will automatically adjust the prior probability of the "safety factor" node and increase its proportion in the decision weight, so as to give priority to the safety threshold in caving control and avoid roof fall accidents caused by roof instability.

[0107] The conditional probability table (CPT) of each node in the Bayesian network needs to be calibrated according to the actual working condition. For example, the confidence range of top coal caving state recognition is 0~1, the roof stability risk level is divided into low, medium, and high three grades, and the gangue mixing probability range is 0~100%. The system realizes quantitative control of decision logic by setting threshold values (such as roof stability risk ≥ medium, caving opening degree ≤ 50%).

[0108] This step is deployed in the decision system of the crossheading control center in practical applications, combines the D-S evidence theory with the fuzzy logic controller, and forms a multi-model collaborative reasoning mechanism. The technical effect is that, by dynamically updating the prior probability and adjusting the decision weight, the system can respond to the changes in the top coal occurrence conditions in real time, realize the risk self-adaptive balance of the caving process, improve the caving efficiency and safety, and provide key support for the intelligent mining of the super-thick hard coal seam.

[0109] S4, the control instruction is issued to the hydraulic support through the electro-hydraulic control system to realize the differential opening and closing control of the regional caving mouth, and the control strategy is dynamically adjusted according to the real-time feedback data.

[0110] Specifically, the control instruction is issued to the hydraulic support through the electro-hydraulic control system to realize the differential opening and closing control of the regional caving mouth, and the control strategy is dynamically adjusted according to the real-time feedback data, which is the core execution link of the caving mining process control system in the present application. This step converts the caving control strategy output by the upper decision system into specific hydraulic support control instructions, realizes independent control of multiple caving mouths, and thus improves the caving efficiency and resource recovery rate.

[0111] In terms of technical implementation, this step relies on the distributed control architecture of the electro-hydraulic control system (EHS) of the hydraulic support. The system decomposes the caving strategy into multiple control sub-tasks through the workstation of the crossheading control center, each sub-task corresponding to the hydraulic support caving mouth of a specific area. The control instruction is issued to the local controller (such as PLC or embedded control unit) of each hydraulic support through field bus protocols such as industrial Ethernet or CAN bus with millisecond-level response speed, to control the opening and closing angle and duration of the tail beam plugboard. In some implementations, the system supports multi-channel parallel control, which can simultaneously control the caving actions of multiple supports to ensure the coordination and synchronization of the caving process.

[0112] The system dynamically sets the opening and closing time window (usually 5-30 seconds), opening angle (0°-90° adjustable), and caving priority (such as high, medium, and low) of the caving mouth according to key parameters such as top coal fragmentation, caving rate, and gangue mixing probability. The adjustment of the control strategy is based on a multi-objective optimization function, including the recovery rate (≥90%), the gangue content (≤15%), and the energy efficiency. The system updates the control parameters online through reinforcement learning algorithm to adapt to the dynamic changes of the coal seam occurrence conditions.

[0113] In application scenarios, this step is widely applicable to super-thick hard coal seam fully mechanized caving faces, especially in complex geological conditions where the coal seam thickness is greater than 10m, the top coal is hard, and the caving property is poor. Through regional control, the system can preferentially open the caving mouth in the area with high top coal fragmentation and good caving state, avoiding premature caving when the top coal is not fully fragmented or there are large coal bodies, thereby reducing coal loss and gangue mixing.

[0114] The technical effect of this step is to realize dynamic closed-loop control of the coal drawing process, significantly improving the accuracy and adaptability of coal drawing. Through differential control and real-time feedback adjustment, the system can effectively deal with uneven top coal occurrence, complex coal-rock interface and other problems, improve resource recovery rate and reduce mining risk, providing key support for intelligent mining of ultra-thick hard coal seams.

[0115] Further, it also includes:

[0116] S5, when the rear scraper conveyor exists pose offset or is not pushed to the position, the system automatically switches to manual intervention mode, and pushes the coal drawing operation suggestion and the current area top coal state information to the operator

[0117] Specifically, when the rear scraper conveyor exists pose offset or is not pushed to the position, the system automatically switches to manual intervention mode, and pushes the coal drawing operation suggestion and the current area top coal state information to the operator, which is one of the key mechanisms for realizing safe and efficient control of the coal drawing process in the present application. Based on the cooperation of the hydraulic support electro-hydraulic control system and the real-time monitoring system of the top coal migration state, combined with the preset abnormal state recognition threshold and operation logic, the system realizes smooth transition from automatic control to manual intervention.

[0118] The system collects the pose parameters of the conveyor in real time through the inclination sensors, displacement sensors and laser ranging devices arranged on the tail beam of the hydraulic support, the pushing jack and the rear scraper conveyor. When the lateral offset of the rear scraper conveyor is detected to exceed the set threshold (such as ±50mm) or the longitudinal pushing is not in place (such as the pushing step deviation exceeds ±100mm), the system determines that it is an abnormal state, triggering the manual intervention mode. At this time, the system uploads the current hydraulic support number, top coal fragmentation (calculated by vibration acceleration sensor and microseismic data, range 0~1), top coal thickness (calculated by fusion of laser displacement sensor and height measuring sensor, accuracy ±5mm), coal drawing opening state (0°~90°) and other key parameters to the crossheading working face centralized control center, and pushes the coal drawing operation suggestion to the operator through the human-machine interface (HMI), including whether to suspend coal drawing, adjust the coal drawing sequence, manually control the coal drawing opening degree, etc.

[0119] The system adopts IEEE 1588 time synchronization protocol to ensure the consistency of the timestamps of the data of each sensor, and the error is controlled within 1 ms. The top coal fragmentation evaluation model is constructed based on the vibration frequency spectrum energy entropy and the main frequency deviation. When the energy entropy value exceeds the threshold value 0.75 or the main frequency deviation exceeds ± 10 Hz, the system determines that the top coal fragmentation is high, and the coal drawing strategy needs to be adjusted. At the same time, according to the requirements of the coal mine safety regulations (AQ 1020-2006) on the opening degree of the coal drawing port and the safe distance of the top coal migration, the matching relationship between the opening degree of the coal drawing port and the height of the top coal is set to ensure the coal and gangue separation efficiency and the stability of the roof during the coal drawing process.

[0120] At the application scene level, this step is suitable for the situation that the automatic coal drawing process is blocked due to the abnormal equipment pose in the fully mechanized caving face of the ultra-thick hard coal seam, especially in the working conditions of complex geological structure, high coal seam hardness, and uneven top coal fragmentation. It has significant adaptability and safety advantage. The operator can quickly determine whether to adjust the coal drawing sequence or manually intervene in the action of the coal drawing port according to the real-time data and suggestions pushed by the system, so as to avoid the risk of top coal loss or support instability caused by system misjudgment or equipment abnormality.

[0121] The technical effect of this step is that by introducing the abnormal state recognition and manual intervention mechanism, the robustness and safety of the coal drawing process are effectively improved. When the automatic control fails or there is potential risk, the system can timely switch to the manual mode to ensure the continuity and controllability of the coal drawing operation, and at the same time, by pushing accurate top coal state information, the operator can make scientific decisions to improve resource recovery rate and reduce the probability of safety accidents.

[0122] S6, in the manual intervention mode, the system generates a corrected coal drawing port opening and closing time control strategy according to the coal drawing instruction input by the operator, combined with the top coal fragmentation and drawing state of the current area.

[0123] Specifically, in the manual intervention mode, the system generates a corrected coal drawing port opening and closing time control strategy according to the coal drawing instruction input by the operator, combined with the top coal fragmentation and drawing state of the current area. The core is to realize dynamic response and precise adjustment under the man-machine collaborative control. The technical implementation of this step is based on the collaborative decision mechanism of multi-source sensor fusion analysis results and operator experience instructions. The operator inputs the coal drawing instruction through the human-machine interface (HMI) of the crossheading control center, including the target coal drawing area, the expected coal drawing amount, the coal drawing priority and other parameters. After receiving the instruction, the system first retrieves the real-time monitoring data of the area, including the top coal fragmentation index (CTDI, Coal Top Disintegration Index), the drawing state recognition result (such as drawing intensity confidence, gangue mixing probability, etc.), and dynamically corrects it combined with the pre-set coal drawing control model.

[0124] In some implementations, the system employs a fuzzy logic controller (Fuzzy Logic Controller) to weight and fuse the operator instructions and sensor feedback. For example, when the operator sets the caving opening opening time to 30 seconds, and the system monitors that the top coal fragmentation degree of the area is 0.75 (full scale is 1), and the gangue mixing probability is 12%, the system will correct the original instruction according to the preset control rule base, and may adjust the opening time to 28 seconds to avoid gangue mixing caused by over-caving. In addition, the system also supports multi-model collaborative reasoning based on D-S evidence theory to evaluate the confidence of the top coal caving state, ensuring the scientificity and safety of the correction strategy.

[0125] This step is usually deployed in the middle and rear areas of fully mechanized caving faces in thick hard coal seams in actual applications, especially in areas with complex geological conditions, uneven top coal occurrence, or the presence of faults. Through the combination of manual intervention and automatic control, the system can quickly respond and adjust the caving strategy when unexpected conditions or model prediction deviations are large, thereby improving caving efficiency, reducing resource loss, and effectively controlling safety risks during mining. This technical means significantly enhances the controllability and adaptability of the top coal caving process, and is a key link to realize intelligent and precise caving control.

[0126] The thick hard coal sub-regional top coal caving dynamic control method of the embodiment of the present application realizes dynamic and precise control of sub-regional top coal caving in thick hard coal seams, improves top coal recovery rate and reduces gangue mixing rate, and effectively solves the problem of over-caving or under-caving caused by coal seam changes in traditional top coal caving processes.

[0127] Further, the thick hard coal sub-regional top coal caving dynamic control method of the embodiment of the present application is described in conjunction with the accompanying drawings.

[0128] The main purpose of the present application is to meet the demand for efficient mining technology of thick hard coal seams in coal mines, and to provide a thick hard coal sub-regional top coal caving mining technology method. The system can adapt to the environmental characteristics and control requirements of thick hard coal seam working faces, and can realize intelligent and efficient caving control of thick hard coal seam working face top coal caving. Table 1 shows the sub-regional caving process flow.

[0129] Table 1 Sub-regional caving process flow

[0130]

[0131] A thick hard coal sub-regional top coal caving mining technology method according to an embodiment of the present application is described below.

[0132] The working face A1 uses a double drum coal cutter to cut coal and load coal, a tail beam of a lifting hydraulic support, a shrink tail beam, a top coal plugging plate and a coal loading. The coal is sequentially transported by a front and rear scraper conveyor, a transfer machine, a crusher and a belt conveyor. The hydraulic support supports the roof to realize a fully mechanized coal mining process of the fully mechanized caving face, and the coal caving and mining process is as follows: coal cutting → support moving → front scraper conveyor pushing → top coal plugging → rear scraper conveyor pulling → transfer machine moving → next cycle.

[0133] S101, coal cutting:

[0134] The coal cutter cutting mode is bidirectional coal cutting, two cuts are cut once in a round trip, the drum is self-rotated to break coal by using the cutting tooth, when cutting coal from the head to the tail, the right drum cuts the top coal and the left drum cuts the bottom coal; when cutting coal from the tail to the head, the left drum cuts the top coal and the right drum cuts the bottom coal.

[0135] The coal cutter cuts coal by oblique cutting, when the coal cutter cuts to the head and the tail position, the scraper conveyor behind has moved close to the coal wall, at this time, the drum positions of the coal cutter are exchanged, the end is cut through, then the rear drum is raised, the coal cutter reverses along the curved section of the conveyor to cut into the coal wall, then the scraper conveyor behind is pushed to move close to the coal wall, so that the scraper conveyors of the working face are in a straight line. The front and rear drum positions of the coal cutter are exchanged again, the front drum is lowered and the rear drum is raised, the coal cutter returns to cut coal to the end, after cutting through the coal wall, the coal cutter returns to normally cut coal to the other end.

[0136] S102, support moving:

[0137] The middle support supports in time according to the support moving sequence, the support moving is 5-10 supports (10-20 m) behind the rear drum of the coal cutter in sequence and follows the coal cutter to move. When the transition support group moves, the support is moderately lagged according to the coal caving state and the three machine pose state, and the front scraper conveyor power part is synchronously moved by the pushing jack. The mine pressure monitoring system feeds back in real time, the support pose is adjusted, and the coal caving condition of the transition section is ensured.

[0138] S103, pushing the front scraper conveyor:

[0139] The front scraper conveyor is normally pushed: after the coal cutting by the coal cutter, the front scraper conveyor is automatically pushed in groups at a distance of 10 supports behind the rear drum of the coal cutter, the pushing direction is sequentially pushed along the advancing direction of the coal cutter, and one step is pushed.

[0140] S104, top coal plugging operation:

[0141] The coal discharge port of the support tail beam plug-in board is opened, the coal discharge control system is started to discharge coal and implement the regulation and control of the coal discharge amount, and the coal gangue is stopped when it is mixed. The working face adopts "two knives and one discharge", the coal discharge step is 1.6 m, the top coal is discharged in an alternating operation sequence with the coal cutting of the coal mining machine, that is, when the coal mining machine cuts the coal from the machine head and cuts the triangular coal, the first round of coal discharge is performed according to the top coal state selected in the region, when the coal mining machine cuts the coal from the machine tail and cuts the triangular coal, the second round of coal discharge is performed according to the top coal state selected in the region, and the triangular area of the transition section at both ends of the working face is completed by the transition support group and the rear scraper conveyor. In the process of discharging coal, the top coal falling state data obtained by the sensor system is calculated to judge the current working face middle region coal discharge state, and the coal discharge region is determined and executed according to the priority of the "regional coal discharge process flow", so as to dynamically adjust the coal discharge parameters, realize the adaptive optimization of the coal discharge process, and the coal discharge action of the transition section hydraulic support is decided according to the current top coal falling situation in the region, the coal discharge state of the adjacent middle support and the current action. When there is an abnormal state such as pose offset of the rear conveyor and not pushing to the position, the automatic coal discharge process is switched to manual operation.

[0142] S105, pull the rear conveyor and the transfer conveyor and the belt conveyor:

[0143] The rear conveyor is pulled automatically in groups in sequence by using an electro-hydraulic controller. The rear conveyor of the working face is pulled behind the coal discharge point, the machine head of the rear conveyor is pulled first when the coal mining machine cuts coal from the machine head to the tail, and the rear conveyor is pulled from the machine head to the tail in sequence; the tail of the rear conveyor is pulled first when the coal mining machine cuts coal from the machine tail to the machine head, and the rear conveyor is pulled from the machine tail to the machine head in sequence. When the working face advances by one step, the transfer conveyor is moved once, and when the distance between the machine head of the transfer conveyor and the tail of the working face equipment train is less than a certain distance, the equipment train is pulled to the front working face, the train track is moved forward synchronously, and the cycle is repeated in sequence.

[0144] In order to realize the above-mentioned embodiment, as Figure 2 shown, the embodiment further provides a special thick hard coal regional top coal dynamic control device 10, which comprises:

[0145] A data acquisition module 100 is arranged to acquire top coal movement state data through a plurality of sensors arranged in the hydraulic support and the working face;

[0146] A multi-source data processing module 200 is arranged to perform multi-source data processing based on the top coal movement state data by using a data level, feature level and decision level fusion algorithm to generate a top coal fragmentation degree evaluation result, a falling state identification result and a roof stability analysis result;

[0147] The multi-model collaborative reasoning module 300 is configured to perform multi-model collaborative reasoning by using D-S evidence theory, fuzzy logic and Bayesian network according to the top coal fragmentation evaluation result, the caving state identification result and the roof stability analysis result, and generate the opening and closing time prediction and control instruction of the regional caving gate.

[0148] The electro-hydraulic control execution module 400 is configured to issue the control instruction to the hydraulic support through an electro-hydraulic control system, realize differential opening and closing control of the regional caving gate, and dynamically adjust the control strategy according to real-time feedback data.

[0149] Further, the sensors include a pressure-displacement integrated sensor, an inclination sensor, an altimeter sensor, a laser displacement sensor, a microseismic sensor and a vibration acceleration sensor; and the data acquisition module is further configured to:

[0150] The pressure-displacement integrated sensor is pre-installed in the hydraulic cylinder and is configured to acquire the working resistance and displacement of the support in real time.

[0151] The microseismic sensor is installed behind the roof anchor rod and the coal wall and is configured to monitor the crack propagation state of the top coal.

[0152] Further, the multi-source data processing module is further configured to:

[0153] In the data level fusion stage, the wavelet threshold denoising is used to process the hydraulic impact noise, and the IEEE 1588 protocol is used to realize time synchronization of the sensor data.

[0154] In the feature level fusion stage, the cubic spline interpolation is used to generate a continuous deformation curve of the roof displacement, and the curvature extreme value and the change rate thereof are extracted as feature parameters.

[0155] Further, the multi-model collaborative reasoning module is further configured to:

[0156] The D-S evidence theory is used to fuse the caving intensity confidence output by the pressure-vibration model, the gangue mixing probability output by the coal body vibration identification model and the risk level output by the roof stability model.

[0157] The Bayesian network dynamically updates the prior probability according to real-time data and adjusts the decision weight to realize adaptive balance of the risk.

[0158] Further, the system further comprises:

[0159] The pose monitoring module is configured to monitor the pose state of the rear scraper conveyor and trigger the manual intervention mode when the pose deviation or the non-pushing to the position is detected.

[0160] An artificial intervention module is configured to receive an operator input coal drawing instruction in an artificial intervention mode, and generate a corrected coal drawing opening and closing time control strategy in combination with a current top coal fragmentation degree and a drawing state of a region.

[0161] The thick hard coal sub-region top coal drawing dynamic control device according to the embodiment of the present application is based on a top coal migration state real-time monitoring system, a coal drawing optimal time dynamic prediction system and a coal drawing mining process control system, and each system cooperates to realize efficient and accurate control of thick hard coal seam mining. The present application provides a key technical solution suitable for fully mechanized top coal drawing automatic monitoring, decision-making and control execution in thick hard coal seams, which provides technical support for efficient production control of thick hard coal seam top coal drawing mining.

[0162] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example" or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, different embodiments or examples described in the present specification and the features of different embodiments or examples can be combined and combined by those skilled in the art without contradiction.

[0163] In addition, the terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "a plurality of" is at least two, for example, two, three, etc., unless otherwise specifically limited.

Claims

1. A dynamic control method for top coal caving in different areas of extra-thick hard coal, characterized in that: include: S1, collects top coal movement status data through multiple types of sensors arranged on hydraulic supports and working faces; S2, based on the top coal movement status data, multi-source data processing is performed using data-level, feature-level, and decision-level fusion algorithms to generate top coal fragmentation assessment results, caving status identification results, and roof stability analysis results; S3, based on the top coal fragmentation assessment results, caving state identification results, and roof stability analysis results, multi-model collaborative reasoning is performed using DS evidence theory, fuzzy logic, and Bayesian networks to generate opening and closing time predictions and control instructions for coal caving ports in each area; S4, sending the control instructions to the hydraulic support through the electro-hydraulic control system to achieve differentiated opening and closing control of the coal discharge ports in different areas, and dynamically adjusting the control strategy according to real-time feedback data.

2. The method according to claim 1, wherein The sensors include a pressure-displacement integrated sensor, an inclination sensor, a height sensor, a laser displacement sensor, a microseismic sensor, and a vibration acceleration sensor; the multiple sensors arranged on the hydraulic support and the working face collect top coal movement status data, and also include: S11, the pressure-displacement integrated sensor is pre-installed inside the hydraulic cylinder and is used to collect the working resistance and displacement of the bracket in real time; S12, the microseismic sensor is installed behind the roof anchor and the coal wall to monitor the expansion status of the top coal crack.

3. The method according to claim 1, wherein The multi-source data processing based on the top coal migration status data using data level, feature level and decision level fusion algorithms further includes: S21, wavelet threshold denoising is used to process hydraulic impact noise in the data-level fusion stage, and the time synchronization of sensor data is achieved through the IEEE 1588 protocol; S22, in the feature-level fusion stage, cubic spline interpolation is used to generate the continuous deformation curve of the top plate displacement, and its curvature extreme value and change rate are extracted as feature parameters.

4. The method according to claim 1, wherein The multi-model collaborative reasoning using DS evidence theory, fuzzy logic and Bayesian network also includes: S31, the DS evidence theory is used to integrate the confidence level of the falling strength output by the pressure-vibration model, the probability of the gangue mixing output by the coal body vibration identification model, and the risk level output by the roof stability model; S32, the Bayesian network dynamically updates the prior probability according to the real-time data and adjusts the decision weight to achieve risk adaptive balance.

5. The method according to claim 1, wherein Also includes: S5: When the rear scraper conveyor is offset or not in place, the system automatically switches to manual intervention mode and sends coal placement operation suggestions and current top coal status information to the operator; S6, in the manual intervention mode, the system generates a revised coal caving port opening and closing time control strategy based on the coal caving instruction input by the operator and combined with the top coal fragmentation and caving status in the current area.

6. A dynamic control device for top coal caving in different areas of extra-thick hard coal, characterized in that: include: The data acquisition module is used to collect top coal movement status data through multiple types of sensors arranged on the hydraulic support and working face; A multi-source data processing module is used to process the multi-source data based on the top coal movement status data using a data-level, feature-level, and decision-level fusion algorithm to generate top coal fragmentation assessment results, caving status identification results, and roof stability analysis results; A multi-model collaborative reasoning module is used to perform multi-model collaborative reasoning based on the top coal fragmentation assessment results, caving state identification results, and roof stability analysis results using DS evidence theory, fuzzy logic, and Bayesian networks to generate opening and closing time predictions and control instructions for coal caving ports in each area; The electro-hydraulic control execution module is used to send the control instructions to the hydraulic support through the electro-hydraulic control system, realize the differentiated opening and closing control of the coal venting port in different areas, and dynamically adjust the control strategy according to real-time feedback data.

7. The device according to claim 6, characterized in that The sensors include a pressure-displacement integrated sensor, an inclination sensor, a height sensor, a laser displacement sensor, a microseismic sensor, and a vibration acceleration sensor; the data acquisition module is also used for: The pressure-displacement integrated sensor is pre-installed inside the hydraulic cylinder and is used to collect the working resistance and displacement of the bracket in real time; The microseismic sensors are installed behind the roof anchors and coal walls to monitor the expansion status of top coal cracks.

8. The device according to claim 6, wherein The multi-source data processing module is further configured to: Wavelet threshold denoising is used to process hydraulic impact noise at the data level fusion stage, and the time synchronization of sensor data is achieved through the IEEE 1588 protocol. In the feature-level fusion stage, cubic spline interpolation is used to generate the continuous deformation curve of the roof displacement, and its curvature extreme value and change rate are extracted as feature parameters.

9. The device according to claim 6, wherein The multi-model collaborative reasoning module is also used to: The DS evidence theory is used to integrate the confidence level of the falling strength output by the pressure-vibration model, the probability of the gangue mixing output by the coal body vibration identification model, and the risk level output by the roof stability model; The Bayesian network dynamically updates the prior probability according to real-time data and adjusts the decision weight to achieve risk adaptive balance.

10. The device according to claim 6, wherein Also includes: The posture monitoring module is used to monitor the posture status of the rear scraper conveyor and trigger the manual intervention mode when it detects posture deviation or failure to move into place; The manual intervention module is used to receive the coal caving instructions input by the operator in the manual intervention mode, and generate a revised coal caving port opening and closing time control strategy based on the top coal fragmentation and caving status in the current area.

Images