CN112199836A - Early warning method, device and equipment for periodic pressure on working face based on deep learning - Google Patents
Early warning method, device and equipment for periodic pressure on working face based on deep learning Download PDFInfo
- Publication number
- CN112199836A CN112199836A CN202011061015.0A CN202011061015A CN112199836A CN 112199836 A CN112199836 A CN 112199836A CN 202011061015 A CN202011061015 A CN 202011061015A CN 112199836 A CN112199836 A CN 112199836A
- Authority
- CN
- China
- Prior art keywords
- pressure
- periodic
- prediction model
- working face
- periodic pressure
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000000737 periodic effect Effects 0.000 title claims abstract description 162
- 238000000034 method Methods 0.000 title claims abstract description 59
- 238000013135 deep learning Methods 0.000 title claims abstract description 43
- 239000003245 coal Substances 0.000 claims abstract description 88
- 238000012549 training Methods 0.000 claims abstract description 79
- 230000006870 function Effects 0.000 claims description 97
- 238000005520 cutting process Methods 0.000 claims description 75
- 238000000265 homogenisation Methods 0.000 claims description 60
- 238000003860 storage Methods 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 6
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000005065 mining Methods 0.000 abstract description 11
- 125000004122 cyclic group Chemical group 0.000 description 14
- 239000011435 rock Substances 0.000 description 7
- 238000004088 simulation Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 4
- 238000012885 constant function Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 4
- 230000001965 increasing effect Effects 0.000 description 4
- 238000003062 neural network model Methods 0.000 description 4
- 238000012886 linear function Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 230000003014 reinforcing effect Effects 0.000 description 2
- 239000000725 suspension Substances 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 239000000839 emulsion Substances 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000007620 mathematical function Methods 0.000 description 1
- 239000002243 precursor Substances 0.000 description 1
- 238000003825 pressing Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses an early warning method, device and equipment for periodic pressure of a working face based on deep learning, relates to the technical field of coal mine well mining, and can solve the problems that the existing periodic pressure of the working face is low in prediction efficiency, low in accuracy and incapable of accurately realizing early warning. The method comprises the following steps: training a periodic pressure prediction model based on historical stent pressure data so that the periodic pressure prediction model meets a preset training standard; acquiring pressure data of each support in a working face in real time; and inputting the pressure data into a periodic pressure prediction model meeting the preset training standard, and obtaining a periodic pressure prediction result of the working face. The method and the device are suitable for predicting the periodic pressure of the working face.
Description
Technical Field
The application relates to the technical field of coal mine underground mining, in particular to a working face periodic pressure early warning method, device and equipment based on deep learning.
Background
The existing underground coal mining basically adopts a longwall stoping process, namely a working face mining area adopts equipment such as a hydraulic support to support a top plate, a non-support area is arranged behind the working face, and the top plate of the working face mining area automatically collapses after coal is mined and fills a goaf. The caving condition of the top plate behind the working face directly influences the safety of the front working face, if the lower top plate behind the working face is behind, the upper top plate can form an arch-like giant rock block occlusion structure, the pressure of the hydraulic support of the front working face can be reduced to a certain extent, and the safety of a recovery space is facilitated; if the upper top plate belongs to a hard top plate and a large-area suspended ceiling does not collapse, the upper top plate can be suddenly broken and destabilized after reaching a certain suspended area, so that the working surface is impacted by 'mine earthquake' or 'hurricane', and property loss and casualties are caused; and the other possibility is that the arch-like occlusion structure of the upper top plate is suddenly unstable, so that the pressure of the hydraulic support on the working face is increased suddenly to form periodic pressure, the coal wall caving and roof caving of the working face are induced if the pressure is light, the top plate of the working face is cut along the coal wall if the pressure is heavy, the whole hydraulic support is crushed in a large area, and roof disastrous accidents such as roof cutting and frame pressing occur. After stoping of the working face, the goaf behind the hydraulic support is closed, the goaf belongs to an unsafe space and is difficult for personnel to enter, the suspension condition of the roof and the stability of an arch-like structure are difficult to detect by naked eyes and instruments, and the problem is caused for prediction and prevention of roof disasters.
In the prior art, the stress condition of a hydraulic support is analyzed by monitoring the pressure of an upright column of the hydraulic support, and the periodic instability or collapse of an arch-like structure, namely the periodic pressure on a working face, is calculated through the fluctuation of the pressure of the hydraulic support. However, the current bottleneck limits the accurate acquisition and successful application of the periodic pressure index, and the main problems include: (1) each working face is provided with 100 and 200 hydraulic supports, the lifting frame and the forward movement of each support have a certain sequence and are not synchronous, so that the pressure fluctuation between the supports has certain asynchronization, and the suspension condition of the top plate of the whole working face and the formation and instability of an arch-like structure are difficult to analyze on a time scale; (2) because of the asynchronous motion of the supports, the prior art analyzes the pressure fluctuation of each support independently, and the ordinary 'pressure rise' is often inferred as the occurrence of periodic incoming pressure, so that the actual reasons for the pressure rise of the supports are many, and the sudden rise of the pressure of individual supports is possible due to the fact that the reasons for the pressure rise of the supports are many, long-time production stoppage, coal cutter coal cutting disturbance, support moving motion, non-stress of adjacent supports and poor support position, so that the analysis result often causes two adjacent supports in the same period, one generates periodic incoming pressure and the other does not generate a contradiction phenomenon, and the authenticity of periodic incoming pressure data is influenced; (3) under the same coal bed condition, the collapse and instability of the top plate have certain periodicity, regularity and repeatability, but the existing mine pressure monitoring analysis only carries out routine and simple analysis on the pressure data of the existing support, has certain hysteresis, and lacks of the early warning function for the possibility of disaster accidents of the next cycle of coming pressure, the strength and even the top plate depending on the existing data form and development trend.
The periodic incoming pressure analysis methods all need to manually analyze the pressure fluctuation data of the single support one by one, and have the problems of high working strength, low efficiency and low accuracy.
Disclosure of Invention
In view of this, the application provides a method, a device and equipment for early warning of periodic pressure of a working face based on deep learning, and mainly solves the problems that the existing method for predicting the periodic pressure of the working face is low in efficiency and accuracy and cannot accurately realize early warning.
According to one aspect of the application, a method for early warning of periodic pressure on a working face based on deep learning is provided, and the method comprises the following steps:
training a periodic pressure prediction model based on historical stent pressure data so that the periodic pressure prediction model meets a preset training standard;
acquiring pressure data of each support in a working face in real time;
and inputting the pressure data into a periodic pressure prediction model meeting the preset training standard to obtain a periodic pressure prediction result of the working face.
According to another aspect of the application, a warning device for periodic pressure on a working face based on deep learning is provided, and the device comprises:
the training module is used for training a periodic pressure prediction model based on historical stent pressure data so as to enable the periodic pressure prediction model to meet a preset training standard;
the acquisition module is used for acquiring pressure data of each support in a working surface in real time;
and the input module is used for inputting the pressure data into a periodic pressure prediction model meeting the preset training standard to obtain a periodic pressure prediction result of the working face.
According to yet another aspect of the present application, there is provided a non-transitory readable storage medium having stored thereon a computer program that, when executed by a processor, implements the above-described method for warning of periodic pressure on a work surface based on deep learning.
According to still another aspect of the present application, a computer device is provided, which includes a non-volatile readable storage medium, a processor, and a computer program stored on the non-volatile readable storage medium and executable on the processor, wherein the processor executes the program to implement the above-mentioned early warning method for periodic pressure on a working face based on deep learning.
By means of the technical scheme, compared with the current periodic pressure prediction mode of the working face, the periodic pressure warning method, the periodic pressure warning device and the periodic pressure prediction equipment of the working face based on deep learning can be used for firstly training based on historical support pressure data to obtain a periodic pressure prediction model meeting the preset training standard, and further can be used for acquiring pressure data of each support in the working face in real time; and inputting the pressure data into a periodic pressure prediction model meeting the preset training standard, and further obtaining a periodic pressure prediction result of the working face. In the application, a periodic pressure prediction model and a neural network model are fused, a deep learning tool is used for carrying out simulation training on samples, key indexes in each homogenization cyclic function are learned, accurate sensing results of the current activity state and the stress state of the top plate are obtained in an output layer through reasoning, whether periodic pressure occurs in the next cycle or not is predicted, and the purposes of accurate early warning and prediction can be achieved. In the process of controlling the stope rock stratum, once the early warning indexes show that the working face is in the early stage or the initial stage of the incoming pressure, an instruction can be sent to the electro-hydraulic control system, the stope area reduces the stoping speed, measures such as supporting and the like are enhanced, and roof accidents are avoided. In addition, in the application, the pressure data of each bracket in the working face can be imported into the periodic pressure prediction model in real time, so that the real-time performance of periodic pressure prediction of the working face can be ensured.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:
fig. 1 is a schematic flowchart illustrating an early warning method for periodic pressure on a working face based on deep learning according to an embodiment of the present disclosure;
FIG. 2 is a schematic flow chart illustrating another early warning method for periodic pressure on a working face based on deep learning according to an embodiment of the present disclosure;
fig. 3 is a schematic structural diagram illustrating an early warning system for periodic pressure on a working face based on deep learning according to an embodiment of the present disclosure;
fig. 4 is a schematic structural diagram illustrating an early warning device for periodic pressure on a working face based on deep learning according to an embodiment of the present disclosure;
fig. 5 shows a schematic structural diagram of another early warning device for periodic pressure on a working face based on deep learning according to an embodiment of the present application.
Detailed Description
The present application will be described in detail below with reference to the accompanying drawings in conjunction with embodiments. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
Aiming at the problems that the conventional working face periodic pressure-bearing prediction method is low in efficiency and accuracy and cannot accurately realize early warning, the embodiment of the application provides a working face periodic pressure-bearing early warning method based on deep learning, and as shown in fig. 1, the method comprises the following steps:
101. and training a periodic pressure prediction model based on historical stent pressure data so as to enable the periodic pressure prediction model to meet a preset training standard.
For the embodiment, in a specific application scenario, when a prediction model is trained for a periodic pressure, relevant indexes of a uniform cyclic function of the change of the pressure of the support before and after the periodic pressure in the working face cyclic mining process can be collected as training samples and stored in a 'sample knowledge base'; carrying out simulation training on the sample through deep learning, and learning the relation between the resistance increasing characteristic of the pressure homogenization circulation function of the working face support and the periodic incoming pressure in each coal cutting cycle; taking the actually measured column pressure change homogenization circulation function key index in the circulation coal cutting process as an input layer neuron, then carrying out deep learning and calculation on a sample library, and obtaining the predicted pressure evolution trend, the circulation end resistance value and the roof activity state index of the next coal cutting circulation support in an output layer through reasoning; in the process of controlling the rock stratum of the stope, once the early warning indexes show that the working face is in the early stage or the initial stage of the incoming pressure, an instruction can be sent to the electro-hydraulic control system, measures such as reducing the extraction speed and strengthening the support are taken, and roof accidents are avoided.
The implementation subject of the application can be a prediction system of the periodic pressure of the working face, a periodic pressure prediction model is embedded in the prediction system, when the periodic pressure prediction is carried out, the periodic pressure prediction model can be trained firstly to enable the periodic pressure prediction model to reach a preset training standard, then pressure data of each support in the working face can be obtained in real time, the pressure data are input into the trained periodic pressure prediction model, and a periodic pressure prediction result of the working face is further obtained.
102. Pressure data for each of the supports in the working face is acquired in real time.
In a specific application scenario, when working face extraction is carried out, along with coal mining machine circular coal cutting, a pressure sensor on a hydraulic support stand column uninterruptedly monitors the pressure change of emulsion liquid in the stand column, collected pressure data are transmitted back to a system host, the pressure data of each support are respectively numbered according to coal cutting circulation, the pressure data of any support stand column in a single coal cutting circulation is s, and the pressure data can be defined as:
s[(p1，t1)，(p2，t2)，…,(pn，tn)]
where s-represents certain stent pressure data; p-is the support pressure corresponding to the time t; t-is the time corresponding to the support pressure value; and n-is the pressure data volume collected in the cycle.
103. And inputting the pressure data into a periodic pressure prediction model meeting a preset training standard to obtain a periodic pressure prediction result of the working face.
For the embodiment, the pressure data acquired in real time can be input into the periodic pressure prediction model meeting the preset training standard, so that the periodic pressure prediction result of the working face in the future preset time period can be acquired.
According to the early warning method for the periodic pressure of the working face based on the deep learning in the embodiment, a periodic pressure prediction model meeting a preset training standard can be obtained based on historical support pressure data training, and further, pressure data of each support in the working face can be obtained in real time; and inputting the pressure data into a periodic pressure prediction model meeting a preset training standard, and further obtaining a periodic pressure prediction result of the working face. In the application, a periodic pressure prediction model and a neural network model are fused, a deep learning tool is used for carrying out simulation training on samples, key indexes in each homogenization cyclic function are learned, accurate sensing results of the current activity state and the stress state of the top plate are obtained in an output layer through reasoning, whether periodic pressure occurs in the next cycle or not is predicted, and the purposes of accurate early warning and prediction can be achieved. In the process of controlling the stope rock stratum, once the early warning indexes show that the working face is in the early stage or the initial stage of the incoming pressure, an instruction can be sent to the electro-hydraulic control system, the stope area reduces the stoping speed, measures such as supporting and the like are enhanced, and roof accidents are avoided. In addition, in the application, the pressure data of each bracket in the working face can be imported into the periodic pressure prediction model in real time, so that the real-time performance of periodic pressure prediction of the working face can be ensured.
Further, as a refinement and an extension of the specific implementation of the above embodiment, in order to fully explain the specific implementation process in this embodiment, another early warning method for periodic pressure on a working face based on deep learning is provided, as shown in fig. 2, the method includes:
201. and determining the pressure data of each coal cutting cycle and each bracket in each coal cutting cycle according to the historical bracket pressure data.
In a specific application scene, when coal mining is carried out, a working face mainly supports a top plate by a jack upright post of a hydraulic support, and the actions of lifting and lowering the hydraulic upright post are realized by a control valve group according to instruction operation; and when the support moves forwards, the pulling frame jack lags the coal mining machine by a certain distance to move forwards sequentially according to the instruction of the control valve group. The process of racking itself represents the end of the previous coal cutting cycle and the beginning of the next coal cutting cycle. The coal cutting cycle is accurately divided based on the frame pulling action of each support, and the pressure data of any support stand column in the cycle can be acquired.
202. And determining a target fitting function of each bracket in the coal cutting cycle by fitting the pressure data.
For the present embodiment, in a specific application scenario, the embodiment step 202 may specifically include: respectively calculating a fitting function corresponding to each bracket based on the pressure data; and determining the fitting function with the highest goodness of fit as the target fitting function corresponding to the support.
Wherein for pressure data s it is defined as: s [ (p)1，t1)，(p2，t2)，…,(pn，tn)]Wherein p isnIs tnThe corresponding support pressure at any moment; defining all the support pressure data d in one coal cutting cycle as d [ s ]1，s2，…,sm]Wherein s ismSupport pressure data in one cycle for the support numbered m; according to different working face coal bed conditions and support resistance increasing characteristics, the support pressure time change curve function is basically distributed in a linear function, a logarithmic function, an exponential function and a near constant function, so that a plurality of fitting functions such as the linear function, the logarithmic function, the exponential function and the near constant function can be set, and a specific fitting function formula is as follows:
exponential function: f (t) ═ p0+BeCt
Logarithmic function: f (t) ═ p0+Cln(1+t)
Linear function: f (t) ═ p0+Ct
Near constant function: f (t) ═ p0+C
Wherein F (t) -is a pressure-time fitting function; p is a radical of0-is the initial force; b and C are fitting coefficients; t-is time.
And fitting the pressure data of each support from initial force to cycle end resistance in the coal cutting cycle according to the four functions, and determining the fitting function with the highest fitting goodness as a target fitting function F (t) corresponding to the support.
203. And homogenizing the target fitting function in the same coal cutting cycle to obtain a homogenizing circulation function corresponding to each coal cutting cycle.
For this embodiment, in a specific application scenario, step 203 of the embodiment may specifically include: and homogenizing the target fitting function in the same coal cutting cycle according to a preset formula, and calculating to obtain a homogenizing cycle function corresponding to each coal cutting cycle.
Wherein, the preset formula can be:
in the formula:
-is a scaffold pressure equalization cyclic function; t-is time; n is a radical ofi-the number of exponential, logarithmic, linear, near-constant functions, etc. respectively, fitted; n-is the total number of fitting functions;
-the arithmetic mean of the coefficients of said functions, respectively;
-mean initial force of the above functions, respectively.
The homogenization circulation function is used for representing the stress condition and the bearing characteristic of all the supports of the working face in a coal cutting circulation. By analogy, the support pressure change homogenizing circulation function of each coal cutting circulation is obtained according to the sequence of 1-n
And the like. The cyclic homogenization function is used for representing the stress condition and the bearing characteristic of all the supports of the working face in a coal cutting cycle and is also used for representing the change trend of the support pressure of each coal cutting cycle before and after the periodic pressure on the working face.
Thus, the resistance increasing characteristic of the pressure change of the support in the working surface propelling process is obtained in the form of a fitting function; abnormal mine pressure phenomena such as roof caving, arch structure formation, instability and the like in the working face extraction process can be deduced through the variation of the key indexes of the homogenization circulation function. After a certain data sample is accumulated, the repeated mine pressure rule of periodic pressure comes can be represented in a mathematical function form, the precursor information before and after the pressure comes can be accurately quantized by homogenizing the change and development trend of the index of the cyclic function, and the sample is subjected to simulation training through a deep learning tool, so that the purposes of early warning and prediction are achieved.
204. And training the periodic pressure prediction model by using the homogenization loop function so as to enable the periodic pressure prediction model to meet the preset training standard.
For this embodiment, the step 204 of the embodiment may specifically include: uploading the homogenization circulation functions of all coal cutting cycles to a periodic incoming pressure prediction model according to the coal cutting sequence, so that the homogenization circulation end resistance value of the next coal cutting cycle is output by the cyclic training incoming pressure prediction model based on the preset evaluation index of the homogenization circulation functions; comparing the homogenization end-of-cycle resistance value with a pre-calibrated end-of-cycle resistance value, and calculating an error rate; if the error rate is smaller than the preset error threshold value, judging that the periodic incoming pressure prediction model meets the preset training standard; if the error rate is larger than or equal to the preset error threshold value, the periodic pressure-bearing prediction model is repeatedly trained by using the homogenization circulation function of the coal cutting circulation so as to enable the periodic pressure-bearing prediction model to meet the preset training standard.
In a specific application scenario, in the process of circular coal cutting, pressure data of all supports on a working face are transmitted back to a system host in real time, the system obtains a circular homogenization circulation function through composite operation, and the like in turn, so that a plurality of coal cutting circular homogenization circulation functions are formed and stored in a sample knowledge base; carrying out simulation training on the samples through deep learning of the system, and learning preset evaluation indexes in each homogenization circulation function, such as
The relation between the constant coefficient and the periodic pressure and the pressure variation trend of the bracket.
The input layer in the model corresponds to a series of characteristic values of a homogenization circulation function of a coal cutting circulation, and k evaluation indexes can be set according to requirements, such as
Isoparametric, output layer predictiveAnd (3) taking the resistance value R at the end of the homogenization cycle of the next coal cutting cycle as an index for judging whether the pressure comes or not in a period, and determining the number of neurons in the hidden layer according to the actual requirement. When the model is applied to calculation, the output layer value is the homogenization cycle end resistance value R. Learning samples are built for deep learning according to the measured data, as shown in table 1.
TABLE 1 periodic prediction of incoming pressure for deep learning of measured data samples
After the nth coal cutting cycle is finished, the characteristic indexes of the cyclic functions are homogenized
The optimal predicted value is obtained by inputting from the system input node and adopting a deep learning algorithm, so that the optimal predicted value is learned to average the regularity between the characteristic index change of the cyclic function and the periodic incoming pressure. And inputting the characteristic value of the homogenization circulation function in the (n + 1) th coal cutting circulation process through an input node, and then outputting a predicted value sample of the periodic incoming pressure characteristic index by the system.
Wherein G is the relation between the characteristic indexes of the homogenization circulation function of the nth coal cutting circulation and the previous n-1 coal cutting circulation, namely
G：Rn→R1 
The relationship F is learned to approximate G, i.e., Table 2 holds.
F：Rn→R1 
And (3) forecasting the variation characteristic of the subsequent coal cutting circulation homogenization circulation function by using learned knowledge, and giving the input and the reasoning output of the subsequent coal cutting circulation function variation characteristic sequence sample in the system forecasting process in a table 3.
TABLE 2 memory of learning samples by the System
Note that ε is the minimum value of deviation
TABLE 3 input and inference output of subsequent samples
In a specific training process, for example, a fully mechanized coal mining face of a certain coal mine has 116 supports, column pressure change data of all the supports of the face in 85 complete coal cutting cycles are collected, and then a homogenization cycle function of each coal cutting cycle is obtained through fitting operation. Based on the organized samples, the first 1-20 cycle data samples can be used for deep learning, and the last 21-85 cycle data samples can be used for reasoning verification, so as to compare the system's reasoning forecast value with the measured value. The determination of the input sample is carried out by adopting a self-adaptive method, the input sample consists of the characteristic values of the homogenization circulation function of 20 coal cutting circulation, and the number of the formed system input nodes is 5, namely the number of the homogenization circulation function
Characteristic values are equal; the number of output nodes of the network is 1, and the cyclic number of the generation period pressure comes, and the like. There may be 1 or more hidden layers between the input layer and the output layer, and the hidden layers and the number of nodes in each hidden layer are determined by an adaptive method. The determined network hidden layer in this example is layer 1, the number of hidden nodes is 8, and the network structure designed by this is 116 → 8 → 1.
After learning, the network performs inductive output on each group of output samples. Table 4 shows the network forecast results of the periodic incoming pressure of the 21 st to 45 th coal cutting cycles in this example.
TABLE 4 periodic pressure prediction according to the pressure variation evaluation index of the stand column of the stand
As can be seen from Table 4, the accuracy of predicting the periodic pressure by the adaptive mode recognition method of deep learning can reach more than 90%, and the relative error is less than 5%.
205. The method comprises the steps of acquiring pressure data of each support in a working face in real time, inputting the pressure data into a periodic pressure incoming prediction model meeting a preset training standard, determining a target homogenization circulation function of a current coal cutting circulation according to the pressure data in the periodic pressure incoming prediction model, and outputting a predicted homogenization circulation end resistance value according to the target homogenization circulation function.
For this embodiment, in a specific application scenario, after pressure data of each support in a working plane is obtained in real time, the pressure data may be input into a periodic incoming pressure prediction model meeting a preset training standard, the periodic incoming pressure prediction model may generate a target homogenization circulation function of a current coal cutting circulation according to the pressure data, and then a homogenization circulation end resistance value in a future preset time period may be predicted according to the target homogenization circulation function.
206. And determining the prediction result of the periodic incoming pressure of the working face in the future preset time period based on the predicted resistance value at the end of the homogenization cycle.
Further, for the embodiment, when the predicted value of the resistance value at the end of the homogenization cycle in the future preset time period is determined, the prediction result of the periodic pressure of the working face in the future preset time period can be determined according to the training and learning process.
207. And outputting early warning information according to the periodic incoming pressure prediction result of the working face.
The early warning information can comprise one or more of text prompt information, picture prompt information, audio prompt information, video prompt information, light prompt information, vibration prompt information and other prompt information, and is used for prompting that the working face is pressed periodically to play a role in early warning prompt. In the process of controlling the stope rock stratum, once the early warning indexes show that the working face is in the early stage or the initial stage of the incoming pressure, an instruction can be sent to the electro-hydraulic control system, the stope area reduces the stoping speed, measures such as supporting and reinforcing are taken, and roof accidents are avoided.
By means of the early warning method for the periodic pressure of the working face based on the deep learning, a periodic pressure prediction model meeting a preset training standard can be obtained based on historical support pressure data training, and further pressure data of each support in the working face can be obtained in real time; and inputting the pressure data into a periodic pressure prediction model meeting a preset training standard, and further obtaining a periodic pressure prediction result of the working face. In the application, a periodic pressure prediction model and a neural network model are fused, a deep learning tool is used for carrying out simulation training on samples, key indexes in each homogenization cyclic function are learned, accurate sensing results of the current activity state and the stress state of the top plate are obtained in an output layer through reasoning, whether periodic pressure occurs in the next cycle or not is predicted, and the purposes of accurate early warning and prediction can be achieved. In the process of controlling the stope rock stratum, once the early warning indexes show that the working face is in the early stage or the initial stage of the incoming pressure, an instruction can be sent to the electro-hydraulic control system, the stope area reduces the stoping speed, measures such as supporting and the like are enhanced, and roof accidents are avoided. In addition, in the application, the pressure data of each bracket in the working face can be imported into the periodic pressure prediction model in real time, so that the real-time performance of periodic pressure prediction of the working face can be ensured.
For the application, the structural schematic diagram of the early warning system for the periodic pressure of the working face based on the deep learning can be as shown in fig. 3, and the prediction system can include a sensor module, a pressure data analysis module, a deep learning module, and a periodic pressure prediction module. The method comprises the steps of firstly, acquiring pressure data in real time based on a support stand column pressure sensor, uploading the pressure data to a pressure data analysis module, sequentially carrying out coal cutting cycle division, data fitting, generation of a homogenization circulation function and extraction of key indexes in the homogenization circulation function, inputting the key indexes into a deep learning module trained by a sample knowledge base module, and obtaining a periodic incoming pressure prediction result. Furthermore, measures such as reinforcing support or reducing mining strength can be adopted according to the prediction result.
Further, as a specific implementation of the method shown in fig. 1 and fig. 2, an embodiment of the present application provides an early warning device for periodic pressure on a working surface based on deep learning, as shown in fig. 4, the device includes: a training module 31, an acquisition module 32, and an input module 33;
the training module 31 is used for training a periodic pressure prediction model based on historical stent pressure data so as to enable the periodic pressure prediction model to meet a preset training standard;
an acquisition module 32, which is used for acquiring pressure data of each support in the working face in real time;
the input module 33 may be configured to input the pressure data into a periodic pressure prediction model meeting a preset training standard, so as to obtain a periodic pressure prediction result of the working surface.
In a specific application scenario, when training the prediction model in a training cycle, as shown in fig. 5, the training module 31 may specifically include: a first determining unit 311, a second determining unit 312, a processing unit 313 and a training unit 314;
a first determining unit 311, operable to determine each coal cutting cycle, and pressure data of each rack within each coal cutting cycle, from the historical rack pressure data;
a second determining unit 312, configured to determine a target fitting function of each rack in the coal cutting cycle by fitting the pressure data;
the processing unit 313 can be used for carrying out homogenization treatment on the target fitting function in the same coal cutting cycle to obtain a homogenization circulation function corresponding to each coal cutting cycle;
the training unit 314 is configured to train the periodic pressure prediction model using the equalization circulation function, so that the periodic pressure prediction model meets a predetermined training standard.
Accordingly, in order to determine and obtain the target fitting function of each rack in the coal cutting cycle, the second determining unit 312 is specifically configured to: respectively calculating a fitting function corresponding to each bracket based on the pressure data; and determining the fitting function with the highest goodness of fit as the target fitting function corresponding to the support.
In a specific application scenario, when the homogenization treatment is performed on the target fitting function in the same coal cutting cycle to obtain the homogenization circulation function corresponding to each coal cutting cycle, the processing unit 313 may be specifically configured to: and homogenizing the target fitting function in the same coal cutting cycle according to a preset formula, and calculating to obtain a homogenizing cycle function corresponding to each coal cutting cycle.
Accordingly, in order to obtain the periodic pressure prediction model meeting the preset training standard by using the equalization circulation function, the training unit 314 is specifically configured to: uploading the homogenization circulation functions of all coal cutting cycles to a periodic incoming pressure prediction model according to the coal cutting sequence, so that the homogenization circulation end resistance value of the next coal cutting cycle is output by the cyclic training incoming pressure prediction model based on the preset evaluation index of the homogenization circulation functions; comparing the homogenization end-of-cycle resistance value with a pre-calibrated end-of-cycle resistance value, and calculating an error rate; if the error rate is smaller than the preset error threshold value, judging that the periodic incoming pressure prediction model meets the preset training standard; if the error rate is larger than or equal to the preset error threshold value, the periodic pressure-bearing prediction model is repeatedly trained by using the homogenization circulation function of the coal cutting circulation so as to enable the periodic pressure-bearing prediction model to meet the preset training standard.
In a specific application scenario, when the pressure data acquired in real time is input into the periodic pressure prediction model meeting the preset training standard to obtain the periodic pressure prediction result of the working surface, as shown in fig. 5, the input module 33 may specifically include: an input unit 331, a third determination unit 332;
the input unit 331 is configured to input the pressure data acquired in real time into a periodic incoming pressure prediction model meeting a preset training standard, so as to determine a target homogenization circulation function of a current coal cutting circulation according to the pressure data in the periodic incoming pressure prediction model, and output a predicted homogenization circulation end resistance value according to the target homogenization circulation function;
a third determining unit 332, operable to determine a working face period coming pressure prediction result in a future preset time period based on the predicted homogenization cycle end resistance value.
Correspondingly, as shown in fig. 5, the apparatus further includes: an output module 34;
and the output module 34 is configured to output the warning information according to the prediction result of the periodic pressure of the working surface.
It should be noted that other corresponding descriptions of the functional units related to the early warning device for the periodic pressure on the working face based on the deep learning provided in this embodiment may refer to the corresponding descriptions in fig. 1 to fig. 2, and are not repeated herein.
Based on the above-mentioned methods as shown in fig. 1 to fig. 2, correspondingly, the present embodiment further provides a non-volatile storage medium, on which computer readable instructions are stored, and the readable instructions, when executed by a processor, implement the above-mentioned early warning method for periodic pressure increase of a working surface based on deep learning as shown in fig. 1 to fig. 2.
Based on such understanding, the technical solution of the present application may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (which may be a CD-ROM, a usb disk, a removable hard disk, etc.), and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method of the embodiments of the present application.
Based on the method shown in fig. 1 to fig. 2 and the virtual device embodiments shown in fig. 4 and fig. 5, in order to achieve the above object, the present embodiment further provides a computer device, where the computer device includes a storage medium and a processor; a nonvolatile storage medium for storing a computer program; and a processor for executing a computer program to implement the above-mentioned early warning method based on the periodic pressure of the working face based on deep learning shown in fig. 1 to 2.
Optionally, the computer device may further include a user interface, a network interface, a camera, Radio Frequency (RF) circuitry, a sensor, audio circuitry, a WI-FI module, and so forth. The user interface may include a Display screen (Display), an input unit such as a keypad (Keyboard), etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface), etc.
It will be understood by those skilled in the art that the present embodiment provides a computer device structure that is not limited to the physical device, and may include more or less components, or some components in combination, or a different arrangement of components.
The nonvolatile storage medium can also comprise an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the computer device described above, supporting the operation of information handling programs and other software and/or programs. The network communication module is used for realizing communication among components in the nonvolatile storage medium and communication with other hardware and software in the information processing entity device.
Through the above description of the embodiments, those skilled in the art will clearly understand that the present application can be implemented by software plus a necessary general hardware platform, and can also be implemented by hardware.
By applying the technical scheme, compared with the prior art, the method and the device can firstly train based on historical support pressure data to obtain a periodic pressure prediction model meeting the preset training standard, and further can acquire the pressure data of each support in a working face in real time; and inputting the pressure data into a periodic pressure prediction model meeting a preset training standard, and further obtaining a periodic pressure prediction result of the working face. In the application, a periodic pressure prediction model and a neural network model are fused, a deep learning tool is used for carrying out simulation training on samples, key indexes in each homogenization cyclic function are learned, accurate sensing results of the current activity state and the stress state of the top plate are obtained in an output layer through reasoning, whether periodic pressure occurs in the next cycle or not is predicted, and the purposes of accurate early warning and prediction can be achieved. In the process of controlling the stope rock stratum, once the early warning indexes show that the working face is in the early stage or the initial stage of the incoming pressure, an instruction can be sent to the electro-hydraulic control system, the stope area reduces the stoping speed, measures such as supporting and the like are enhanced, and roof accidents are avoided. In addition, in the application, the pressure data of each bracket in the working face can be imported into the periodic pressure prediction model in real time, so that the real-time performance of periodic pressure prediction of the working face can be ensured.
Those skilled in the art will appreciate that the figures are merely schematic representations of one preferred implementation scenario and that the blocks or flow diagrams in the figures are not necessarily required to practice the present application. Those skilled in the art will appreciate that the modules in the devices in the implementation scenario may be distributed in the devices in the implementation scenario according to the description of the implementation scenario, or may be located in one or more devices different from the present implementation scenario with corresponding changes. The modules of the implementation scenario may be combined into one module, or may be further split into a plurality of sub-modules.
The above application serial numbers are for description purposes only and do not represent the superiority or inferiority of the implementation scenarios. The above disclosure is only a few specific implementation scenarios of the present application, but the present application is not limited thereto, and any variations that can be made by those skilled in the art are intended to fall within the scope of the present application.
Claims (10)
1. A working face periodic pressure early warning method based on deep learning is characterized by comprising the following steps:
training a periodic pressure prediction model based on historical stent pressure data so that the periodic pressure prediction model meets a preset training standard;
acquiring pressure data of each support in a working face in real time;
and inputting the pressure data into a periodic pressure prediction model meeting the preset training standard to obtain a periodic pressure prediction result of the working face.
2. The method according to claim 1, wherein training a periodic pressure prediction model based on historical stent pressure data to make the periodic pressure prediction model meet a preset training standard specifically comprises:
determining each coal cutting cycle and pressure data of each support in each coal cutting cycle according to historical support pressure data;
determining a target fitting function for each said stent within said coal cutting cycle by fitting said pressure data;
homogenizing the target fitting function in the same coal cutting cycle to obtain a homogenizing circulation function corresponding to each coal cutting cycle;
and training a periodic pressure prediction model by using the homogenization circulation function so as to enable the periodic pressure prediction model to meet a preset training standard.
3. The method of claim 2, wherein said determining a target fit function for each said stent within said coal cutting cycle by fitting said pressure data comprises:
respectively calculating a fitting function corresponding to each bracket based on the pressure data;
and determining the fitting function with the highest goodness of fit as the target fitting function corresponding to the support.
4. The method according to claim 3, wherein the homogenizing the target fitting function in the same coal cutting cycle to obtain the homogenizing circulation function corresponding to each coal cutting cycle comprises:
and homogenizing the target fitting function in the same coal cutting cycle according to a preset formula, and calculating to obtain a homogenizing cycle function corresponding to each coal cutting cycle.
5. The method according to claim 4, wherein the training of the periodic pressure prediction model using the equalization circulation function to make the periodic pressure prediction model meet a preset training standard comprises:
uploading the homogenization circulation functions of the coal cutting cycles to a periodic incoming pressure prediction model according to the coal cutting sequence, so that the incoming pressure prediction model is trained in a circulating mode to output a homogenization circulation end resistance value of the next coal cutting cycle based on preset evaluation indexes of the homogenization circulation functions;
comparing the homogenization resistance value at the end of the cycle with a pre-calibrated resistance value at the end of the cycle, and calculating an error rate;
if the error rate is smaller than a preset error threshold value, judging that the periodic incoming pressure prediction model meets a preset training standard;
if the error rate is larger than or equal to the preset error threshold value, repeatedly training the periodic incoming pressure prediction model by using a homogenization loop function of the coal cutting loop so as to enable the periodic incoming pressure prediction model to meet a preset training standard.
6. The method according to claim 5, wherein the inputting the pressure data into the periodic pressure prediction model meeting the preset training standard to obtain the periodic pressure prediction result of the working face comprises:
inputting the pressure data acquired in real time into a periodic pressure incoming prediction model meeting the preset training standard so as to determine a target homogenization circulation function of the current coal cutting circulation according to the pressure data in the periodic pressure incoming prediction model and output a predicted homogenization circulation end resistance value according to the target homogenization circulation function;
and determining a working face periodic incoming pressure prediction result in a future preset time period based on the predicted homogenization cycle end resistance value.
7. The method of claim 6, further comprising:
and outputting early warning information according to the working face periodic incoming pressure prediction result.
8. The utility model provides a working face comes early warning device of pressure in cycle based on deep learning which characterized in that includes:
the training module is used for training a periodic pressure prediction model based on historical stent pressure data so as to enable the periodic pressure prediction model to meet a preset training standard;
the acquisition module is used for acquiring pressure data of each support in a working surface in real time;
and the input module is used for inputting the pressure data into a periodic pressure prediction model meeting the preset training standard to obtain a periodic pressure prediction result of the working face.
9. A non-transitory readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method for warning of periodic pressure on a working surface based on deep learning according to any one of claims 1 to 7.
10. A computer device comprising a non-volatile readable storage medium, a processor, and a computer program stored on the non-volatile readable storage medium and executable on the processor, wherein the processor when executing the program implements the method for early warning of periodic pressure on a deep learning based work surface of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011061015.0A CN112199836B (en) | 2020-09-30 | 2020-09-30 | Working face periodic pressure-entering early warning method, device and equipment based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011061015.0A CN112199836B (en) | 2020-09-30 | 2020-09-30 | Working face periodic pressure-entering early warning method, device and equipment based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112199836A true CN112199836A (en) | 2021-01-08 |
| CN112199836B CN112199836B (en) | 2024-01-05 |
Family
ID=74012493
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011061015.0A Active CN112199836B (en) | 2020-09-30 | 2020-09-30 | Working face periodic pressure-entering early warning method, device and equipment based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112199836B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112836355A (en) * | 2021-01-14 | 2021-05-25 | 西安科技大学 | Method for predicting coal face roof pressure probability |
| CN113775363A (en) * | 2021-09-07 | 2021-12-10 | 中煤科工开采研究院有限公司 | Intelligent sensing system for roof condition of coal mine working face |
| CN115909046A (en) * | 2022-10-09 | 2023-04-04 | 山东科技大学 | Mining pressure profile cloud picture large incoming pressure early warning method and system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103226739A (en) * | 2013-04-24 | 2013-07-31 | 辽宁工程技术大学 | Periodic weighting prediction method based on functional network |
| CN109269556A (en) * | 2018-09-06 | 2019-01-25 | 深圳市中电数通智慧安全科技股份有限公司 | A kind of equipment Risk method for early warning, device, terminal device and storage medium |
-
2020
- 2020-09-30 CN CN202011061015.0A patent/CN112199836B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103226739A (en) * | 2013-04-24 | 2013-07-31 | 辽宁工程技术大学 | Periodic weighting prediction method based on functional network |
| CN109269556A (en) * | 2018-09-06 | 2019-01-25 | 深圳市中电数通智慧安全科技股份有限公司 | A kind of equipment Risk method for early warning, device, terminal device and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 赵毅鑫等: "基于深度学习的大采高工作面矿压预测分析及模型泛化", 《煤炭学报》, vol. 45, no. 1, pages 54 - 65 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112836355A (en) * | 2021-01-14 | 2021-05-25 | 西安科技大学 | Method for predicting coal face roof pressure probability |
| CN112836355B (en) * | 2021-01-14 | 2023-04-18 | 西安科技大学 | Method for predicting coal face roof pressure probability |
| CN113775363A (en) * | 2021-09-07 | 2021-12-10 | 中煤科工开采研究院有限公司 | Intelligent sensing system for roof condition of coal mine working face |
| CN115909046A (en) * | 2022-10-09 | 2023-04-04 | 山东科技大学 | Mining pressure profile cloud picture large incoming pressure early warning method and system |
| CN115909046B (en) * | 2022-10-09 | 2023-10-20 | 山东科技大学 | Large-pressure-coming early warning method and system for mine pressure profile cloud picture |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112199836B (en) | 2024-01-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112199836A (en) | Early warning method, device and equipment for periodic pressure on working face based on deep learning | |
| Zhang et al. | A machine learning framework for assessing post-earthquake structural safety | |
| Vazirizade et al. | Seismic reliability assessment of structures using artificial neural network | |
| Pirizadeh et al. | Probabilistic seismic performance evaluation of non-geometric vertically irregular steel buildings | |
| Bijelić et al. | Validation of the SCEC Broadband Platform simulations for tall building risk assessments considering spectral shape and duration of the ground motion | |
| CN109934398A (en) | A kind of drill bursting construction tunnel gas danger classes prediction technique and device | |
| CN112031839B (en) | Mine pressure space-time bi-periodic prediction method, device and equipment under limited data condition | |
| CN115270635A (en) | Bayes-neural network high-rise building earthquake demand and vulnerability prediction method | |
| Li et al. | A data-driven building’s seismic response estimation method using a deep convolutional neural network | |
| CN103729798A (en) | Coal mine safety evaluation system based on improved k-means clustering | |
| Asgarkhani et al. | Machine learning-based prediction of residual drift and seismic risk assessment of steel moment-resisting frames considering soil-structure interaction | |
| CN107368928A (en) | A kind of combination forecasting method and system of ancient building sedimentation | |
| Li et al. | Vertical ground motion model for the NGA-West2 database using deep learning method | |
| Fomychov et al. | Determination of parameters of non-uniform fractured rock massif in computing experiment | |
| CN116575985A (en) | Informationized monitoring and measuring system for soft rock large deformation tunnel and construction method | |
| Alkroosh et al. | Soft computing for modeling punching shear of reinforced concrete flat slabs | |
| JP2003132296A (en) | Earthquake damage predicting method for building, optimum condition determination processing method used therein and their programs | |
| WO2023194857A1 (en) | Method of and system for predicting strata-related risk in an underground environment | |
| KR101890708B1 (en) | System and method for damage prediction of structures | |
| Zahra et al. | The importance of hazard-consistency when estimating seismic residual drifts in steel moment frames | |
| CN113946790A (en) | Method, system, equipment and terminal for predicting height of water flowing fractured zone | |
| CN114329935A (en) | Method and device for testing three-dimensional mining stress of surrounding rock of coal mine tunnel | |
| Zhang | Post-earthquake performance assessment and decision-making for tall buildings: integrating statistical modeling, machine learning, stochastic simulation and optimization | |
| Mahsuli | Resilience of civil infrastructure by optimal risk mitigation | |
| Worku et al. | An improved first-mode-based pushover analytical procedure for assessing seismic performance of special moment resisting frame building structures |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |