CN115017206A - Mine CO abnormal disturbance intelligent identification and coal spontaneous combustion early warning value determination method - Google Patents
Mine CO abnormal disturbance intelligent identification and coal spontaneous combustion early warning value determination method Download PDFInfo
- Publication number
- CN115017206A CN115017206A CN202210537773.8A CN202210537773A CN115017206A CN 115017206 A CN115017206 A CN 115017206A CN 202210537773 A CN202210537773 A CN 202210537773A CN 115017206 A CN115017206 A CN 115017206A
- Authority
- CN
- China
- Prior art keywords
- concentration
- spontaneous combustion
- time
- early warning
- coal
- 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.)
- Pending
Links
- 239000003245 coal Substances 0.000 title claims abstract description 93
- 238000002485 combustion reaction Methods 0.000 title claims abstract description 77
- 230000002269 spontaneous effect Effects 0.000 title claims abstract description 77
- 238000000034 method Methods 0.000 title claims abstract description 45
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 26
- 238000013145 classification model Methods 0.000 claims abstract description 18
- 238000012549 training Methods 0.000 claims abstract description 15
- 238000005065 mining Methods 0.000 claims abstract description 11
- 230000008569 process Effects 0.000 claims description 15
- 238000005516 engineering process Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 7
- 210000002569 neuron Anatomy 0.000 claims description 7
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 6
- 238000004519 manufacturing process Methods 0.000 claims description 6
- 238000002474 experimental method Methods 0.000 claims description 5
- 239000011159 matrix material Substances 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 claims 1
- 230000002265 prevention Effects 0.000 abstract description 6
- 201000004569 Blindness Diseases 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 16
- 230000008859 change Effects 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000005422 blasting Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 210000004027 cell Anatomy 0.000 description 2
- 238000005520 cutting process Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- UGFAIRIUMAVXCW-UHFFFAOYSA-N Carbon monoxide Chemical compound [O+]#[C-] UGFAIRIUMAVXCW-UHFFFAOYSA-N 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 229910002091 carbon monoxide Inorganic materials 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2474—Sequence data queries, e.g. querying versioned data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A50/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE in human health protection, e.g. against extreme weather
- Y02A50/20—Air quality improvement or preservation, e.g. vehicle emission control or emission reduction by using catalytic converters
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Economics (AREA)
- Molecular Biology (AREA)
- Human Resources & Organizations (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- Fuzzy Systems (AREA)
- Databases & Information Systems (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Probability & Statistics with Applications (AREA)
- Quality & Reliability (AREA)
- Agronomy & Crop Science (AREA)
- Animal Husbandry (AREA)
- Marine Sciences & Fisheries (AREA)
- Mining & Mineral Resources (AREA)
- Primary Health Care (AREA)
- Alarm Systems (AREA)
Abstract
The invention discloses a method for intelligently identifying abnormal disturbance of mine CO and determining a coal spontaneous combustion early warning value, which comprises the following steps of: acquiring historical time sequence data of the CO concentration of the upper corner of the goaf in a state without coal spontaneous combustion signs during normal mining of the working face; dividing the acquired data into subsequences, and marking whether each subsequence is an interference sequence; establishing an upper corner CO concentration time series classification model by adopting an LSTM network; establishing a CO concentration time series data reconstruction model in an interference-free state by adopting an LSTM self-encoding network; determining a coal spontaneous combustion state abnormity early warning threshold value of the goaf according to a reconstruction error during the training period of a CO concentration time series data reconstruction model under an interference-free state; and carrying out real-time early warning on the spontaneous combustion state of the coal in the goaf. The method takes the upper corner CO concentration data actually measured on site as the drive, fully considers the correlation of the time dimension, can effectively reduce the false missing report of spontaneous combustion of coal in the goaf, and reduces the blindness of fire prevention and extinguishing strategy in the goaf.
Description
Technical Field
The invention relates to the field of mine coal spontaneous combustion early warning, in particular to a method for intelligently identifying mine CO abnormal disturbance and determining a coal spontaneous combustion early warning value.
Background
Spontaneous combustion of coal in a goaf is one of main disasters faced in the coal mining process of an underground mine, and once the spontaneous combustion occurs, serious casualties and huge economic losses can be caused. Timely and accurate early warning of spontaneous combustion of coal in the goaf is of great significance to guarantee safe and efficient production of coal mines. CO is an index gas for representing the spontaneous combustion of the coal in the goaf, the coal mine fire prevention and extinguishing rules clearly indicate that 24ppm of carbon monoxide cannot be simply used as the index critical value of the natural coal ignition mark gas, and at the present stage, the index gas measurement experiment is usually adopted to determine the CO concentration early warning critical value of the spontaneous combustion of the coal in the goaf. The upper corner of the working face is used as a main gathering area of CO gas in the goaf and is a key monitoring point of spontaneous combustion of coal in the goaf, however, in the coal mining process, the underground condition is complex and changeable, the CO gas at the upper corner is often interfered by non-coal spontaneous combustion factors such as working face pressure and blasting, the concentration of the CO gas at the upper corner is abnormally increased, the actual ignition state of the goaf cannot be effectively reflected by the early warning critical value of the CO concentration obtained through a laboratory small experiment, coal mining operators cannot accurately judge the goaf, the condition of false alarm missing is easily caused, and the coal mining progress is seriously influenced. In addition, the spontaneous combustion development of the coal in the goaf is a gradual evolution rather than mutation process, and the characteristic of the traditional goaf coal spontaneous combustion early warning method adopting the CO concentration early warning critical value cannot be considered, so that the method is obviously not scientific.
Disclosure of Invention
Aiming at the technical defects, the invention aims to provide a mine CO abnormal disturbance intelligent identification and coal spontaneous combustion early warning value determination method, which aims to improve the accuracy of coal spontaneous combustion early warning in a goaf and reduce the blindness of a goaf fire prevention and extinguishing strategy.
In order to solve the technical problems, the invention adopts the following technical scheme: the method comprises the following steps:
step 1: acquiring historical time sequence data of the CO concentration of the upper corner of the goaf in a state without coal spontaneous combustion signs during normal mining of the working face;
step 2: dividing the historical time sequence data of the upper corner CO concentration obtained in the step (1) into a section of subsequence by adopting a sliding window technology, and marking each subsequence type as an interference sequence and a non-interference sequence one by one according to whether a corresponding time interval without interference of non-coal spontaneous combustion factors exists in the actual production period of a coal mine; the non-coal spontaneous combustion factors comprise working face pressure, blasting, coal cutting and trackless rubber-tyred vehicle gathering;
and step 3: taking all the subsequences obtained in the step 2 as training samples, and establishing an upper corner CO concentration time sequence classification model by adopting an LSTM network;
and 4, step 4: taking the interference-free type subsequence obtained in the step 2 as a training sample, and establishing a CO concentration time series data reconstruction model in an interference-free state by adopting an LSTM self-coding network;
and 5: determining a coal spontaneous combustion abnormal early warning threshold value of the goaf according to the reconstruction error obtained in the step 4 during the CO concentration time series data reconstruction model training period in the non-interference state;
step 6: and (3) collecting the CO concentration data of the upper corner of the working face in real time by using a CO sensor, generating a CO concentration time sequence by adopting a sliding window technology which is the same as that in the step (2), and performing the spontaneous combustion real-time early warning of the coal in the goaf by using the data obtained in the steps (3) to (5).
Preferably, in step 1, the time intervals of the historical time-series data of the upper corner CO concentration are equal to or less than 15 min.
Preferably, in step 2, the sliding window width of the sliding window technique is greater than or equal to 20, that is, at least 20 data points are contained in the sliding window, and the moving time step of the sliding window technique is 1 time point.
Preferably, in step 2, as long as a data point in the subsequence is in the non-coal spontaneous combustion factor interference period, the subsequence is marked as an interfered sequence, and all data points in the subsequence are not in the non-coal spontaneous combustion factor interference period, the subsequence is marked as a non-interfered sequence.
Preferably, in step 3, the LSTM network includes an input layer, a hidden layer, and an output layer, and the overall structure is built by the LSTM loop unit;
the input layer neuron nodes correspond to the CO concentration values of the time sequence at all moments, the number of hidden layer layers and the number of neuron nodes are determined by adopting an experimental method, and a sigmoid function is selected as an output layer classifier;
in the training process, the minimum cross entropy loss function is taken as a target, the Adam self-adaptive algorithm is adopted to optimize the cross entropy loss function until the cross entropy loss function is converged, and an upper corner CO concentration time series classification model is obtained.
Preferably, the circulation unit of the LSTM includes a forgetting gate, an input gate, an output gate, and a unit state;
the operation of the circulation unit of the LSTM is as follows: firstly according to the input x of the current time t And the external state h of the previous moment t-1 Calculating forgetting door f t And input gate i t And an output gate o t And candidate state c t ；
Then forget the door f t And input gate i t And internal state c of the previous time t-1 Combined with the internal state c of the current time t Updating is carried out;
finally, the internal state c of the current moment is compared t And output gate o t Combining to obtain the external state h of the current time t ；
The calculation formulas of the forgetting gate, the input gate, the output gate and the unit state are as follows:
forget the door: f. of t ＝σ(W f *[h t-1 ,x t ]+b f )
An input gate: i.e. i t ＝σ(W i *[h t-1 ,x t ]+b i )
An output gate: o t ＝σ(W o *[h t-1 ,x t ]+b o )
Candidate states:c t ＝tanh(W c *[h t-1 ,x t ]+b c )
internal state at present time: c. C t ＝f t ·c t-1 +i t ·c t
External state at present time: h is t ＝o t ·tanh(c t )
Where W is a weight matrix, b is an offset, σ (·) is a sigmoid function, and tanh (·) is a hyperbolic tangent function.
Preferably, in step 4, the LSTM self-encoding network includes an encoder and a decoder, both of which are built by the LSTM loop unit;
the input of the encoder corresponds to the original value of the CO concentration at each moment of the time series, the output of the encoder is the input of the decoder, and the output of the decoder corresponds to the reconstructed value of the CO concentration at each moment of the time series;
selecting a mean square error between an original value of the CO concentration at each moment of a time sequence input by an encoder and a reconstructed value of the CO concentration at each moment of the time sequence output by a decoder as a loss function, and optimizing the loss function by adopting an Adam self-adaptive algorithm until convergence by taking the minimum mean square error as a target to obtain a CO concentration time series data reconstruction model in an interference-free state;
preferably, a mean square error calculation formula of the original value of the CO concentration at each time of the time series input by the encoder and the reconstructed value of the CO concentration at each time of the time series output by the decoder is as follows:
wherein n is the number of data contained in the time series; x is the number of i The original value of the CO concentration data at the ith moment in the time series is obtained; y is i Is the reconstructed value of the CO concentration data at the ith time in the time series.
Preferably, in the step 5, the goaf coal spontaneous combustion abnormity early warning threshold value is the maximum reconstruction error in the CO concentration time series data reconstruction model training period in the non-interference state obtained in the step 4.
Preferably, in step 6, the spontaneous combustion real-time early warning process of the coal in the goaf is as follows:
inputting the generated CO concentration time series into the upper corner CO concentration time series classification model established in the step 3:
if the upper corner CO concentration time series classification model identifies that the interference sequence exists, continuing to monitor and repeating the step 6;
if the upper corner CO concentration time series classification model is identified as an interference-free sequence, inputting the generated CO concentration time series into the interference-free CO concentration time series data reconstruction model established in the step 4, and calculating a reconstruction error:
if the reconstruction error is less than or equal to the early warning threshold value set in the step 5, indicating that the goaf has no coal spontaneous combustion danger, continuing monitoring and repeating the step 6;
if the reconstruction error is larger than the early warning threshold set in the step 5, the coal spontaneous combustion danger exists in the goaf, and the goaf is in an abnormal state and early warning is given out.
The invention has the beneficial effects that:
(1) the method is driven by CO concentration data measured on site, adopts a method combining supervised learning and unsupervised learning to establish the goaf coal spontaneous combustion intelligent early warning model, provides double guarantee for goaf coal spontaneous combustion prevention, can effectively reduce the false alarm rate of goaf coal spontaneous combustion, reduces the blindness of goaf fire prevention and extinguishing strategy and the cost of fire prevention and extinguishing treatment, and is beneficial to realizing unmanned and intelligent coal mining.
(2) The spontaneous combustion of the coal in the gob is a gradual change process rather than an abrupt change process, the method processes the upper corner CO concentration data into a time sequence, fully considers the correlation of the CO concentration on a time dimension, utilizes the advantages of LSTM in processing the time sequence, is more suitable for the actual spontaneous combustion development process of the coal in the gob compared with the method for carrying out spontaneous combustion early warning of the coal in the gob by adopting a CO concentration single-point threshold in the prior art, and has higher accuracy.
(3) In the coal mine production process, most of the goafs are in a non-ignition state, and effective ignition samples are difficult to obtain.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic flow chart of a method for intelligently identifying abnormal disturbance of mine CO and determining a coal spontaneous combustion early warning value according to an embodiment of the present invention;
FIG. 2 is a diagram of an LSTM network architecture according to an embodiment of the present invention;
FIG. 3 is a block diagram of a circulation unit of the LSTM provided in an embodiment of the present invention;
fig. 4 is a diagram of an LSTM self-encoding network structure according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
As shown in fig. 1, a method for intelligently identifying abnormal disturbance of mine CO and determining a coal spontaneous combustion early warning value includes the following steps:
step 1: acquiring historical time series data of the CO concentration of the upper corner of the goaf in a state without coal spontaneous combustion signs during normal mining of a working face of a certain coal mine, wherein the time interval of the time series is 1min in the embodiment.
Step 2: and (2) dividing the historical time sequence data of the upper corner CO concentration obtained in the step (1) into a section of subsequence by adopting a sliding window technology, and marking the types of the subsequences into a sequence with interference and a sequence without interference one by one according to the corresponding time period of whether non-coal spontaneous combustion factors interfere in the actual production period of a coal mine, wherein the non-coal spontaneous combustion factors comprise working face pressure, blasting, coal cutting and trackless rubber-tyred vehicle aggregation.
If only the data points in the subsequence are in the non-coal spontaneous combustion factor interference time period, the subsequence is marked as an interference sequence, and if all the data points in the subsequence are not in the non-coal spontaneous combustion factor interference time period, the subsequence is marked as an interference-free sequence;
in this embodiment, the sliding window has a width of 20, that is, 20 data points are included in the sliding window, and the moving step of the sliding window is 1 time point.
And step 3: and (4) taking all the subsequences as training samples, and establishing an upper corner CO concentration time sequence classification model by adopting an LSTM network.
As shown in fig. 2, the LSTM network includes an input layer, a hidden layer, and an output layer, and the overall structure is built up by the circulation units of the LSTM. Wherein, the input layer neuron node corresponds to the CO concentration value of each time of the time sequence, the number of the hidden layer layers and the number of the neuron nodes are determined by adopting an experimental method,
in this embodiment, the number of hidden layers and the number of neuron nodes are 2 and 10, respectively, and a sigmoid function is selected as an output layer classifier. In the training process, the minimum cross entropy loss function is taken as a target, the Adam self-adaptive algorithm is adopted to optimize the cross entropy loss function until the cross entropy loss function is converged, and an upper corner CO concentration time series classification model is obtained.
As shown in fig. 3, the circulation cell of the LSTM includes a forgetting gate, an input gate, an output gate, and a cell state. The operation of the circulation unit of the LSTM is as follows: firstly, according to the input x of the current time t And an external state h of the previous time t-1 Calculating forgetting door f t And input gate i t Output gate o t And candidate state c t Then forget to gate f t And input gate i t And internal state c of the previous time t-1 Combined with the internal state c of the current time t The updating is carried out, and the updating is carried out,finally, the internal state c of the current moment is compared t And output gate o t Combining to obtain the external state h of the current time t . The calculation formulas of the forgetting gate, the input gate, the output gate and the unit state are as follows:
forget the door: f. of t ＝σ(W f *[h t-1 ,x t ]+b f )
An input gate: i.e. i t ＝σ(W i *[h t-1 ,x t ]+b i )
An output gate: o t ＝σ(W o *[h t-1 ,x t ]+b o )
Candidate states:c t ＝tanh(W c *[h t-1 ,x t ]+b c )
internal state at present time: c. C t ＝f t ·c t-1 +i t ·c t
External state at present time: h is t ＝o t ·tanh(c t )
Where W is a weight matrix, b is an offset, σ (·) is a sigmoid function, and tanh (·) is a hyperbolic tangent function.
And 4, step 4: and (3) taking the interference-free type subsequence as a training sample, and establishing a CO concentration time sequence data reconstruction model in the interference-free state by adopting an LSTM self-coding network.
As shown in fig. 4, the LSTM self-encoding network includes an encoder and a decoder, both built up from the cyclic units of the LSTM.
The input of the encoder corresponds to the original value of the CO concentration at each moment of the time series, the output of the encoder is the input of the decoder, and the output of the decoder corresponds to the reconstructed value of the CO concentration at each moment of the time series.
The method comprises the steps of selecting a mean square error (namely a reconstruction error) of an original value of the CO concentration at each moment of a time sequence input by an encoder and a reconstructed value of the CO concentration at each moment of the time sequence output by a decoder as a loss function, and optimizing the loss function by adopting an Adam self-adaptive algorithm until convergence by taking the minimum mean square error as a target to obtain a CO concentration time series data reconstruction model in an interference-free state.
The mean square error calculation formula of the original value of the CO concentration at each time of the time series input by the encoder and the reconstructed value of the CO concentration at each time of the time series output by the decoder is as follows:
wherein n is the number of data contained in the time series; x is the number of i The original value of the CO concentration data at the ith moment in the time sequence is obtained; y is i Is the reconstructed value of the CO concentration data at the ith time in the time series.
And 5: taking the maximum reconstruction error during the CO concentration time series data reconstruction model training period in an interference-free state as a spontaneous combustion abnormity early warning threshold value of the coal in the goaf, wherein the spontaneous combustion abnormity early warning threshold value of the coal in the goaf is 0.3226;
step 6: collecting CO concentration data of corners on a working surface in real time by using a CO sensor (the sampling frequency is the same as the time interval of the historical time sequence data of the CO concentration of the corners obtained in the step 1), generating a CO concentration time sequence by adopting the sliding window technology which is the same as that of the step 2,
the spontaneous combustion real-time early warning process of the coal in the goaf is as follows:
inputting the generated CO concentration time series into an upper corner CO concentration time series classification model:
if the upper corner CO concentration time series classification model identifies that the interference sequence exists, continuing to monitor and repeating the step 6;
if the upper corner CO concentration time series classification model is identified as an interference-free sequence, inputting the generated CO concentration time series into a CO concentration time series data reconstruction model under the interference-free state, and calculating a reconstruction error:
if the reconstruction error is less than or equal to 0.3226, the goaf is free of coal spontaneous combustion danger, and the step 6 is continuously monitored and repeated;
if the reconstruction error is larger than 0.3226, the coal spontaneous combustion danger exists in the goaf, and the goaf is in an abnormal state and gives out early warning.
The spontaneous combustion of the coal in the goaf is a process gradually evolving along with time, and the goaf is mostly in a non-ignition condition in the coal mine production process, so that the spontaneous combustion sample data of the coal in the goaf is difficult to obtain.
According to the method, only historical time series data of the CO concentration of the upper corner of the gob in a coal spontaneous combustion evidence-free state during normal advancing of a working face are used, the work face is divided into subsequences by adopting a sliding window technology, each subsequence is marked according to an actual mining condition, the advantage of LSTM in processing time series is utilized, the correlation of the CO concentration on a time dimension is fully considered, and an upper corner CO concentration time series classification model and a CO concentration time series data reconstruction model in an interference-free state are established and used for carrying out real-time early warning on spontaneous combustion of coal in the gob;
the principle is that the upper corner CO concentration time sequence classification model can effectively identify whether the upper corner CO concentration time sequence collected in real time is an interference sequence,
if yes, continuing monitoring;
because the samples used for establishing the early warning model are all data under the goaf safety state, if not, two possibilities exist: the goaf is in a safe state, and the goaf is in an abnormal state.
And the CO concentration time series data reconstruction model under the non-interference state can effectively reconstruct the CO concentration time series under the safe state of the goaf, but cannot reconstruct the CO concentration time series under the abnormal state of the goaf, then, the collected CO concentration time series data is input into the CO concentration time series data reconstruction model under the non-interference state, if the reconstruction error obtained by calculation is larger than an early warning threshold value, the goaf is proved to have coal spontaneous combustion danger, the goaf is in the abnormal state, early warning is sent out, otherwise, the goaf is proved to have no coal spontaneous combustion danger, the goaf is in the safe state, and monitoring is continued.
The method takes the time series data of the actually measured CO concentration on site as the drive, better accords with the characteristic that the spontaneous combustion of the coal in the goaf is a gradual change rather than a sudden change process, can effectively reduce the false alarm rate of the missed alarm of the spontaneous combustion of the coal in the goaf, and is beneficial to realizing the unmanned and intelligent mining of the coal.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. A mine CO abnormal disturbance intelligent identification and coal spontaneous combustion early warning value determination method is characterized by comprising the following steps:
step 1: acquiring historical time sequence data of the CO concentration of the upper corner of the goaf in a state without coal spontaneous combustion signs during normal mining of the working face;
step 2: dividing the historical time sequence data of the upper corner CO concentration obtained in the step 1 into a section of subsequence by adopting a sliding window technology, and marking the types of the subsequences as an interference sequence and a non-interference sequence one by one according to the corresponding time period of whether non-coal spontaneous combustion factors interfere in the actual production period of a coal mine;
and step 3: taking all the subsequences obtained in the step 2 as training samples, and establishing an upper corner CO concentration time sequence classification model by adopting an LSTM network;
and 4, step 4: taking the interference-free type subsequence obtained in the step 2 as a training sample, and establishing a CO concentration time series data reconstruction model in an interference-free state by adopting an LSTM self-coding network;
and 5: determining a coal spontaneous combustion abnormal early warning threshold value of the goaf according to the reconstruction error obtained in the step 4 during the CO concentration time series data reconstruction model training period in the non-interference state;
step 6: and (3) acquiring corner CO concentration data on a working face in real time by using a CO sensor, generating a CO concentration time sequence by adopting a sliding window technology the same as that in the step 2, and performing spontaneous combustion real-time early warning on the coal in the goaf by using the data obtained in the steps 3-5.
2. The method for intelligently identifying mine CO abnormal disturbance and determining the coal spontaneous combustion early warning value as claimed in claim 1, wherein in step 1, the time intervals of the historical time-series data of the upper corner CO concentration are equal to or less than 15 min.
3. The method for intelligently identifying mine CO abnormal disturbance and determining the coal spontaneous combustion early warning value as claimed in claim 3, wherein in the step 2, the width of a sliding window of the sliding window technology is greater than or equal to 20, that is, the sliding window at least comprises 20 data points, and the moving time step of the sliding window technology is 1 time point.
4. The method for intelligently identifying mine CO abnormal disturbance and determining the coal spontaneous combustion early warning value according to claim 1, wherein in the step 2, as long as a data point in the subsequence is in the non-coal spontaneous combustion factor interference period, the subsequence is marked as an interfered sequence, and all data points in the subsequence are not in the non-coal spontaneous combustion factor interference period, the subsequence is marked as an undisturbed sequence.
5. The mine CO abnormal disturbance intelligent identification and coal spontaneous combustion early warning value determination method as claimed in claim 1, wherein in step 3, the LSTM network comprises an input layer, a hidden layer and an output layer, and the whole structure is built by a circulation unit of the LSTM;
the input layer neuron node corresponds to the CO concentration value of each time of the time sequence, the number of hidden layers and the number of neuron nodes are determined by adopting an experimental method, and a sigmoid function is selected as an output layer classifier;
in the training process, the minimum cross entropy loss function is taken as a target, the Adam self-adaptive algorithm is adopted to optimize the cross entropy loss function until the cross entropy loss function is converged, and an upper corner CO concentration time sequence classification model is obtained.
6. The method for intelligently identifying mine CO abnormal disturbance and determining coal spontaneous combustion early warning value according to claim 5, wherein the LSTM circulation unit comprises a forgetting gate, an input gate, an output gate and a unit state;
the operation of the circulation unit of the LSTM is as follows:
firstly according to the input x of the current time t And the external state h of the previous moment t-1 Calculating forgetting door f t And an input gate i t And an output gate o t And candidate statesc t ；
Then forget the door f t And an input gate i t And internal state c of the previous time t-1 Combined with the internal state c of the current time t Updating is carried out;
finally, the internal state c of the current moment is compared t And output gate o t Combined to obtain the external state h of the current time t ；
The calculation formulas of the forgetting gate, the input gate, the output gate and the unit state are as follows:
forget the door: f. of t ＝σ(W f *[h t-1 ,x t ]+b f )
An input gate: i.e. i t ＝σ(W i *[h t-1 ,x t ]+b i )
An output gate: o t ＝σ(W o *[h t-1 ,x t ]+b o )
Candidate states:c t ＝tanh(W c *[h t-1 ,x t ]+b c )
internal state at present time: c. C t ＝f t ·c t-1 +i t ·c t
External state at present time: h is t ＝o t ·tanh(c t )
Where W is a weight matrix, b is an offset, σ (·) is a sigmoid function, and tanh (·) is a hyperbolic tangent function.
7. The method for intelligently identifying the abnormal disturbance of the mine CO and determining the coal spontaneous combustion early warning value according to claim 1, wherein in the step 4, the LSTM self-encoding network comprises an encoder and a decoder, and both the encoder and the decoder are built by an LSTM circulation unit;
the input of the encoder corresponds to the original value of the CO concentration at each moment of the time sequence, the output of the encoder is the input of the decoder, and the output of the decoder corresponds to the reconstructed value of the CO concentration at each moment of the time sequence;
and selecting the mean square error of the original value of the CO concentration at each moment of the time sequence input by the encoder and the reconstructed value of the CO concentration at each moment of the time sequence output by the decoder as a loss function, and optimizing the loss function by adopting an Adam adaptive algorithm until convergence to obtain a CO concentration time series data reconstruction model under the non-interference state.
8. The method for intelligently identifying mine CO abnormal disturbance and determining coal spontaneous combustion early warning value according to claim 7, wherein a mean square error calculation formula of the original value of CO concentration at each time of the time series input by the encoder and the reconstructed value of CO concentration at each time of the time series output by the decoder is as follows:
wherein n is the number of data contained in the time series; x is the number of i The original value of the CO concentration data at the ith moment in the time sequence is obtained; y is i Is the reconstructed value of the CO concentration data at the ith time in the time series.
9. The method for intelligently recognizing the CO abnormal disturbance and determining the coal spontaneous combustion early warning value in the mine according to claim 1, wherein in the step 5, the goaf coal spontaneous combustion abnormal early warning threshold value adopts the maximum reconstruction error in the training period of the CO concentration time series data reconstruction model in the non-interference state obtained in the step 4.
10. The method for intelligently identifying the abnormal disturbance of the mine CO and determining the coal spontaneous combustion early warning value according to claim 1, wherein in the step 6, the real-time coal spontaneous combustion early warning process in the goaf is as follows:
inputting the generated CO concentration time series into the upper corner CO concentration time series classification model established in the step 3:
if the upper corner CO concentration time series classification model identifies that the interference sequence exists, continuing to monitor and repeating the step 6;
if the upper corner CO concentration time series classification model is identified as an interference-free sequence, inputting the generated CO concentration time series into the interference-free CO concentration time series data reconstruction model established in the step 4, and calculating a reconstruction error:
if the reconstruction error is less than or equal to the early warning threshold set in the step 5, indicating that the goaf has no coal spontaneous combustion danger, continuing to monitor and repeating the step 6;
if the reconstruction error is larger than the early warning threshold set in the step 5, the coal spontaneous combustion danger exists in the goaf, and the goaf is in an abnormal state and sends out early warning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210537773.8A CN115017206A (en) | 2022-05-17 | 2022-05-17 | Mine CO abnormal disturbance intelligent identification and coal spontaneous combustion early warning value determination method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210537773.8A CN115017206A (en) | 2022-05-17 | 2022-05-17 | Mine CO abnormal disturbance intelligent identification and coal spontaneous combustion early warning value determination method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115017206A true CN115017206A (en) | 2022-09-06 |
Family
ID=83069189
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210537773.8A Pending CN115017206A (en) | 2022-05-17 | 2022-05-17 | Mine CO abnormal disturbance intelligent identification and coal spontaneous combustion early warning value determination method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115017206A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN116700213A (en) * | 2023-06-13 | 2023-09-05 | 无锡物联网创新中心有限公司 | Industrial equipment abnormality detection method and related device based on gating circulation unit |
-
2022
- 2022-05-17 CN CN202210537773.8A patent/CN115017206A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN116700213A (en) * | 2023-06-13 | 2023-09-05 | 无锡物联网创新中心有限公司 | Industrial equipment abnormality detection method and related device based on gating circulation unit |
| CN116700213B (en) * | 2023-06-13 | 2024-03-29 | 无锡物联网创新中心有限公司 | Industrial equipment abnormality detection method and related device based on gating circulation unit |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115017206A (en) | Mine CO abnormal disturbance intelligent identification and coal spontaneous combustion early warning value determination method | |
| CN110110740B (en) | Drilling process working condition identification method based on multi-time scale features and neural network | |
| CN113076975A (en) | Dam safety monitoring data anomaly detection method based on unsupervised learning | |
| CN115526422A (en) | Coal mine gas explosion risk prediction method | |
| CN112257917B (en) | Time sequence abnormal mode detection method based on entropy characteristics and neural network | |
| CN115638833B (en) | Monitoring data processing method and system | |
| CN115083123B (en) | Mine coal spontaneous combustion intelligent grading early warning method driven by measured data | |
| CN110956155B (en) | Fuzzy clustering identification method for fully mechanized mining face operation procedure based on CO data | |
| CN117473285B (en) | Intelligent operation and maintenance management system and method based on digital twinning | |
| CN110348603B (en) | Coal spontaneous combustion danger degree multi-source information fusion early warning method | |
| CN106246226B (en) | A kind of abnormal recognition methods of Mine Gas Gushing | |
| CN118037047A (en) | Mine safety monitoring system based on AI | |
| CN109902265B (en) | Underground early warning method based on hidden Markov model | |
| CN110991363B (en) | Method for extracting CO emission characteristics of coal mine safety monitoring system in different coal mining processes | |
| CN114611958A (en) | Petroleum drilling big data processing method and device based on machine learning | |
| CN111102006A (en) | Dynamic early warning intelligent decision method for outburst mine extraction and mining deployment imbalance | |
| CN115977736A (en) | Coal and gas outburst early warning method based on field real-time data drive | |
| CN113685166B (en) | Drilling accident early warning method and system | |
| CN117073918A (en) | Potential response real-time detection system and method for mine carbon dioxide sealing leakage source | |
| Fan et al. | Adaptive monitoring for geological drilling process using neighborhood preserving embedding and Jensen–Shannon divergence | |
| CN112443319B (en) | Kick monitoring method | |
| CN113190535B (en) | Gas concentration prediction method based on improved grey-long and short-term memory neural network combined model | |
| Chen et al. | Identification of Wellbore Flow Abnormal Working Conditions Based on Deep Learning | |
| CN114519923B (en) | Intelligent diagnosis and early warning method and system for power plant | |
| CN117846705B (en) | Multi-field coupling disaster-causing early warning method and system for close-range outburst coal seam group exploitation |
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 |