CN-121993259-A - Hydraulic support space-time pressure monitoring and early warning system and method for gob-side entry retaining

CN121993259ACN 121993259 ACN121993259 ACN 121993259ACN-121993259-A

Abstract

The application provides a hydraulic support space-time pressure monitoring and early warning system and method for gob-side entry retaining, and belongs to the field of mine pressure monitoring; the method comprises the following steps of multi-source data acquisition and preprocessing, space-time feature matrix construction, space-time matrix dynamic calibration based on real-time positioning, space-time correlation feature extraction, prediction of the incoming pressure of a dynamic graph neural network model based on physical constraints, input of the extracted space-time correlation feature as an input feature matrix into the dynamic graph neural network model for incoming pressure prediction, and simultaneously introduction of physical unsatisfiability penalty into a loss function of the dynamic graph neural network model to enable the incoming pressure predicted by the dynamic graph neural network model to conform to the physical constraints between the working resistance of a bracket and the sinking amount of a roof, and risk identification and dynamic early warning based on space-time attention weight.

Inventors

ZHANG XIN
WANG LIANG
LI XUHONG
LI XIN
Jia Cunhe
ZHAO YONG
LIU JINGJING
MENG QINGYUAN
LI JIANJIA
ZHENG FENGBIN

Assignees

太原向明智控科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260402

Claims (10)

1. A hydraulic support space-time pressure monitoring and early warning method for gob-side entry retaining is characterized by comprising the following steps: Step S1, multi-source data acquisition and preprocessing, namely acquiring the pressure of a lower cavity of a stand column of each hydraulic support and the real-time space position of each hydraulic support, which are supported on a gob-side entry retaining way, and preprocessing the acquired data; Step S2, constructing a space-time feature matrix and extracting features, wherein the method comprises the following steps: s2.1, constructing a space-time feature matrix; s2.2, dynamically calibrating a space-time matrix based on real-time positioning; S2.3, extracting space-time correlation characteristics, including time characteristics and space characteristics, wherein the time characteristics comprise cycle end resistance, time weighted average working resistance, end maximum resistance increasing rate and resistance variation coefficient, and the space characteristics comprise pressure gradient among brackets, local pressure correlation and space pressure distribution entropy; s3, based on the coming pressure prediction of the dynamic graph neural network model of physical constraint, the space-time correlation features extracted in the step S2.3 are used as input feature matrixes to be input into the dynamic graph neural network model for coming pressure prediction, and meanwhile, physical unsatisfiability penalty is introduced into a loss function of the dynamic graph neural network model, so that the coming pressure predicted by the dynamic graph neural network model accords with the physical constraint between the working resistance of the bracket and the sinking amount of the top plate; Step S4, risk identification and dynamic early warning based on space-time attention weight, comprising the following steps: s4.1, calculating a node risk index, namely after obtaining the predicted pressure output by the dynamic graph neural network model based on physical constraint, defining the node risk index by combining the risk propagation probability calculated by the graph attention layer; And S4.2, dividing physical early warning thresholds based on the breaking steps.
2. The method for monitoring and early warning the space-time pressure of the hydraulic support for gob-side entry retaining according to claim 1, further comprising the step S5 of optimizing a support moving strategy, taking a current space-time characteristic matrix as a state, optionally moving a target position as an action, taking an average risk reduction value in a period of time after moving as a reward, training a strategy network by adopting a DQN or PPO algorithm, and directly outputting an optimal decision suggestion of moving a certain hydraulic support under the current pressure distribution through the strategy network.
3. The method for monitoring and early warning the space-time pressure of the hydraulic support for gob-side entry retaining according to claim 2, further comprising the step S6 of updating and self-learning in real time, automatically updating support position information after support moving, adding new pressure-position data into a training set by the system, and adopting a migration learning strategy to finely tune a model so as to adapt to new working conditions.
4. The hydraulic support space-time pressure monitoring and early warning method for gob-side entry retaining according to any one of claims 1 to 3 is characterized in that the acquisition frequency of the pressure data of the lower column cavity of the hydraulic support in step S1 is adaptively adjusted by adopting an adaptive sampling frequency control algorithm, and the sampling frequency is adjusted by comparing the change rate of the pressure data with a basic sampling frequency.
5. A hydraulic support space-time pressure monitoring and early warning method for gob-side entry retaining according to any one of claims 1-3, characterized in that in step S2.1, a plurality of hydraulic supports are formed into a regional block to The hydraulic support is of a space scale, And the secondary working cycle is a time scale, and a three-dimensional space-time characteristic matrix is constructed.
6. The method for monitoring and early warning the space-time pressure of the hydraulic support for gob-side entry retaining according to claim 5, wherein in step S2.2, the following calibration procedure is automatically triggered each time the completion of the support movement and the entering into a stable working state is detected by means of a support-position mapping table updated in real time: Step S2.2.1, reconstructing the spatial relationship according to the latest spatial coordinates of all the hydraulic supports Recalculating the space sequence of the tunnel along the trend of the tunnel; step S2.2.2, moving data separation, when the first detection is detected The stage hydraulic support is at moment When the moving is completed, the pressure data is divided into two parts, wherein the data before moving is included in the original position history, and the data after moving is bound to a new position; And S2.2.3, reconstructing the space-time feature matrix according to the updated logic index.
7. The hydraulic support space-time pressure monitoring and early warning method for gob-side entry retaining according to claim 6, characterized in that the time weighted average working resistance in step S2.3 is a pressure average value calculated by giving more weight to recent data, and is used for filtering historical interference and highlighting recent trends of roof movement; the maximum resistance increasing rate of the tail end is the maximum rising rate of the pressure of the hydraulic support in unit time at the tail end of the working cycle, and is used for judging the intensity of the top plate pressing and the support bearing state; the pressure gradient between the brackets is the space change rate of pressure along the trend of the roadway, namely the change amplitude of pressure in unit distance, and is used for positioning a pressure concentration area or a pressure abnormal area; The local pressure correlation is used for measuring the synchronous degree and linear correlation relation of the pressure change of the adjacent brackets; the spatial pressure distribution entropy is used to characterize the degree of uniformity or confusion of the pressure over the spatial distribution.
8. The hydraulic support space-time pressure monitoring and early warning method for gob-side entry retaining according to claim 7, wherein the step S3 comprises the following steps: Step S3.1, constructing a dynamic diagram topology, and constructing a dynamic diagram structure according to the real-time hydraulic support coordinates calibrated in the step S2.2 Wherein the node Representing hydraulic supports, edges According to the actual distance between the hydraulic supports Dynamic connection, the structure of the graph is updated in real time along with the moving of the bracket; s3.2, extracting spatial coupling features among brackets by adopting a graph roll-up network layer; s3.3, constructing physical information constraint, and constructing an equation of physical mechanics according to a key layer theory Wherein In order to achieve the distance between the suspended roofs, For the working resistance of the bracket, For roof subsidence, the physical unsatisfiability penalty is expressed as: ; Wherein, the A physical constraint intensity coefficient for controlling the weight of the physical constraint in the whole loss function, For the physical deviation metric function, for calculating the degree of deviation between the model predicted value and the physical theoretical value, Future cycle end resistance predicted by a dynamic graph neural network model based on physical constraint; The total loss function is: ; Wherein, the The loss is fitted to the data.
9. The hydraulic support space-time pressure monitoring and early warning method for gob-side entry retaining according to claim 7, wherein the expression of the node risk index is: ; Wherein, the Representing support The greater the value, the higher the risk; For the predicted pressure of the stent itself, Is a safe threshold value for the pressure of the stent itself, The ratio of the predicted pressure of the bracket to the safety threshold value is smaller than 1, which means that the predicted pressure is lower than the threshold value and the risk of the bracket is lower, otherwise, the risk of the bracket is higher; Representing the risk contribution of the neighbors, Is attention weight, representing neighbor stent Opposite bracket Risk influence degree of (2); Representing neighbor risk magnification factors.
10. A hydraulic support space-time pressure monitoring and early warning system for gob-side entry retaining is characterized by comprising the following modules: the sensing module is used for acquiring the pressure and the real-time space position of the lower cavity of the upright post of each hydraulic bracket through the wireless pressure sensor and the positioning tag card; the network transmission module is used for uploading the data of the sensing module to the upper module through the wireless equipment; The upper module is used for carrying out dynamic early warning of pressure prediction, risk identification and breaking step distance based on the hydraulic support space-time pressure monitoring early warning method for gob-side entry retaining according to any one of claims 6 to 9.

Description

Hydraulic support space-time pressure monitoring and early warning system and method for gob-side entry retaining Technical Field The application relates to the technical field of mine pressure monitoring and early warning, in particular to a hydraulic support space-time pressure monitoring and early warning system and method for gob-side entry retaining. Background The gob-side entry retaining technology is one of the core technologies of non-pillar mining, namely, after the mining of the previous working face, a roadway is reserved to be used as a stoping roadway of the next working face through roadside support. In the process, the roadway needs to bear severe mining influence, the roof moves frequently, and surrounding rock stress is redistributed. If the stress state of the support (especially the roadside support) and the relative position relation between the support and the mining space cannot be mastered in time, safety accidents such as pressing frames and roof falling easily occur. The prior art has the following main problems: 1. the traditional hydraulic support pressure monitoring system mostly adopts a wired connection mode, a large number of cables are required to be laid in a roadway, and in a dynamic change environment such as gob-side entry retaining, the cables are easy to damage and difficult to maintain, and potential safety hazards exist. The power consumption problem of the wireless sensor can influence the service life of the sensor. 2. Most of the existing systems only focus on the acquisition and display of pressure data, and lack accurate perception of the spatial position of a hydraulic support. After the bracket is moved, the position information is not updated in time, so that the pressure data cannot be effectively related to the space position, and the pressure analysis of the space dimension is difficult to carry out. 3. The existing system mostly adopts a simple threshold alarming or trend analysis method, lacks depth excavation of pressure data time-space characteristics, cannot accurately predict complex mine pressure appearance rules, and has low early warning accuracy and high false alarm rate. 4. In the aspect of space-time pressure analysis, the prior art is mostly developed aiming at the pressure condition of a working face. In practice, the movement form of the hydraulic support of the working face is different from that of the hydraulic support of the roadway, the hydraulic support of the working face is wholly and orderly moved forward in a row, and irregular movement of the hydraulic support of the roadway occurs. Therefore, a need exists for a gob-side entry retaining hydraulic support pressure monitoring system and method that can integrate temporal and spatial information, and that can achieve accurate prediction and intelligent early warning. Disclosure of Invention The application provides a hydraulic support space-time pressure monitoring and early warning system and method for gob-side entry retaining, which are used for solving the problems that the existing gob-side entry retaining pressure monitoring is single in dimension, early warning is lagged and cannot adapt to dynamic environments, and can adapt to support dynamic movement and realize high-precision space-time pressure monitoring and intelligent early warning. The technical scheme adopted by the application is that the hydraulic support space-time pressure monitoring and early warning method for gob-side entry retaining comprises the following steps: Step S1, multi-source data acquisition and preprocessing, namely acquiring the pressure of a lower cavity of a stand column of each hydraulic support and the real-time space position of each hydraulic support, which are supported on a gob-side entry retaining way, and preprocessing the acquired data; Step S2, constructing a space-time feature matrix and extracting features, wherein the method comprises the following steps: s2.1, constructing a space-time feature matrix; s2.2, dynamically calibrating a space-time matrix based on real-time positioning; S2.3, extracting space-time correlation characteristics, including time characteristics and space characteristics, wherein the time characteristics comprise cycle end resistance, time weighted average working resistance, end maximum resistance increasing rate and resistance variation coefficient, and the space characteristics comprise pressure gradient among brackets, local pressure correlation and space pressure distribution entropy; s3, based on the coming pressure prediction of the dynamic graph neural network model of physical constraint, the space-time correlation features extracted in the step S2.3 are used as input feature matrixes to be input into the dynamic graph neural network model for coming pressure prediction, and meanwhile, physical unsatisfiability penalty is introduced into a loss function of the dynamic graph neural network model, so that the coming pressure predicted by the dynamic g