CN115856204B - Tunneling working face gas concentration prediction method based on three-dimensional echo state network - Google Patents
Tunneling working face gas concentration prediction method based on three-dimensional echo state network Download PDFInfo
- Publication number
- CN115856204B CN115856204B CN202211085414.XA CN202211085414A CN115856204B CN 115856204 B CN115856204 B CN 115856204B CN 202211085414 A CN202211085414 A CN 202211085414A CN 115856204 B CN115856204 B CN 115856204B
- Authority
- CN
- China
- Prior art keywords
- data
- gas concentration
- echo state
- state network
- working face
- 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.)
- Active
Links
Images
Classifications
-
- 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
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Abstract
The invention relates to the technical field of coal mine safety, in particular to a tunneling working face gas concentration prediction method based on a three-dimensional echo state network, which comprises the following steps of: s1, acquiring gas concentration data of a coal mine tunneling working face under a mine in real time, preprocessing the acquired gas data, and processing data complexity to obtain a tunneling working face gas concentration data set capable of being used for training an echo state network. The training method of the three-dimensional echo state network based on CEEMDAN-SE not only can effectively process random and large amount of data and generate a network learning result by distributing different weights, but also can effectively reduce time cost and improve operation efficiency, has considerable robustness on the result during training, is further beneficial to improving the accuracy of gas data prediction and recognition, can accurately find abnormal data characteristics in the predicted data, and performs safety measure intervention.
Description
Technical Field
The invention relates to the technical field of coal mine safety, in particular to a tunneling working face gas concentration prediction method based on a three-dimensional echo state network.
Background
The gas concentration superstration is the root of forming gas explosion accidents, is a very complex and dangerous dynamic disaster, is mainly used for strengthening gas management, strengthening gas monitoring and adherence to a gas inspection system, finding out that the gas concentration is in a range when the gas concentration exceeds the limit and reporting treatment is performed in time, and the gas mixture with the concentration lower than the lower limit or higher than the upper limit of explosion does not cause flame self-propagation when contacting with an ignition source. Therefore, how to predict and identify the occurrence of the phenomenon is particularly important, and the method is based on a deep learning algorithm, and through analyzing the gas concentration in real time and training an echo state network model, the abnormal data characteristics are searched in the predicted data, and the abnormal prediction monitoring data are transmitted to staff in an alarm mode for safety measure intervention. The traditional gas concentration data prediction method comprises a support vector machine, a decision tree, a traditional regression method and a statistical learning method with a time sequence decomposition algorithm.
At present, aiming at the characteristics of time variability, instability and nonlinearity of gas concentration data, and combining the reasons of large gas data volume and high complexity, the traditional prediction method cannot adapt to the characteristics of the data, so that the characteristics of the data are difficult to identify and process, and an ideal result is difficult to obtain. The echo state network in the deep learning algorithm is just suitable for processing data with large and complex data volume, and a large number of nodes and neurons in the echo state network can be used for training the data and remembering the characteristics of the data, so that the echo state network has certain advantages in data training and prediction realization.
However, when the first data is obtained, the data is still in high unbalance, the capability of the existing echo state network for preprocessing the data is limited, most of the echo state networks are one-dimensional echo state networks, the accuracy of predicting and identifying the gas data is still to be improved after the data transmitted by the underground sensor are processed in one dimension, and the abnormal data characteristics are not easy to accurately find in the data obtained through prediction, so that safety measures are interfered.
Disclosure of Invention
The invention aims to solve the defects in the prior art, and provides a tunneling working face gas concentration prediction method based on a three-dimensional echo state network.
In order to achieve the above purpose, the invention adopts the following technical scheme: the tunneling working face gas concentration prediction method based on the three-dimensional echo state network comprises the following steps of:
s1, acquiring gas concentration data of a tunneling working face under a mine in real time, preprocessing the acquired gas data, and processing data complexity to obtain a gas concentration data set of the tunneling working face which can be used for training an echo state network;
s2, learning the gas concentration data of the tunneling working face by adopting a CEEMDAN-SE three-dimensional echo state network, and then performing data training to obtain a prediction result;
s3, measuring a plurality of prediction models through different measurement indexes, comparing the correlation between the real value and the predicted value of the model with the loss, and selecting the model with the minimum error and the minimum loss;
s4, deploying the optimal model into a safety control server of the coal mine, and inputting the latest captured gas concentration data into the model; if the calculated result is not in the safe and controllable numerical value interval, judging that the safe environment of the gas is abnormal, and if not, judging that the safe environment of the gas is normal, and processing in time is needed;
preferably, the method for acquiring and processing the working face gas concentration data in S1 includes the following steps:
s101, acquiring the data of gas sensors of all working surfaces from a coal mine gas disaster wind direction management and control platform in real time;
s102, carrying out modal decomposition on the gas concentration data obtained from different sensor positions according to CEEMDAN pretreatment experience;
s103, carrying out sample entropy calculation data complexity classification processing on the data set after the modal decomposition processing;
s104, circularly traversing the whole gas concentration data to obtain the same sequence numbers of co-IMF1, co-IMF2 and co-IMF3 from the same sequence data but three different complexity.
Preferably, the empirical mode decomposition method for data preprocessing in S102 includes the following steps:
s102-1, reading data;
s102-2, setting the processing times K of the original data;
s102-3, respectively adding random white noise to the K pieces of original data to form a series of new data;
s102-4, performing EMD (empirical mode decomposition) on the series of new data to obtain a series of IMF (intrinsic mode function) components;
s102-5, respectively averaging IMF components of the corresponding modes to obtain EEMD decomposition results;
s102-6, carrying out ensemble average on the modal components after EEMD decomposition, carrying out ensemble average calculation on the obtained first-order IMF components after CEEMDAN decomposition, obtaining final first-order IMF components, and then repeating the operation on the residual parts.
Preferably, the processing method for complexity classification of the data set in S103 includes the following steps:
s103-1, performing sample entropy calculation on the collected K IMF component sequence data obtained by CEEMDAN decomposition;
s103-2, calculating the sequence complexity of each IMF component through sample entropy, and then dividing the K IMF components into three types according to the complexity range of different sequence data, namely a high-frequency intrinsic mode component co-IMF1, a medium-frequency mode component co-IMF2 and a low-frequency mode component co-IMF3.
Preferably, the data processing method of the three-dimensional echo state network of CEEMDAN-SE in S2 includes the following steps:
s201, dividing the wattage concentration degree co-IMF1, co-IMF2 and co-IMF3 into training sets D-co-IMF1, D-co-IMF2 and D-co-IMF3 and test sets T-co-IMF1, T-co-IMF2 and T-co-IMF3;
s201, selecting the architecture of an echo state network model, and then improving the echo state network into a three-dimensional echo state network capable of inputting three-dimensional data (in the model training process, determining alpha as leakage rate, and optimizing the network by an Adam optimizer);
s203, determining the number of nodes, the spectrum radius SR, the reserve pool scale N, the reserve pool input unit scale IS, the reserve pool sparseness degree SD and the regularization factor, wherein each change of the parameters can generate a new model, and meanwhile, the experimental result IS closely related to the parameters, so that the experimental result IS prevented from fitting or overfitting due to incorrect selection of the parameters.
Preferably, the gas concentration data set co-IMF in S201 is segmented into a training set D-co-IMF and a prediction set T-co-IMF according to a ratio of 7:3.
Preferably, in S202, the architecture of the echo state network model is selected, and then the one-dimensional echo state network is improved into a three-dimensional echo state network capable of inputting three-dimensional data;
the improved three-dimensional echo state network formula is as follows:
given three signals:
u(0),u(1),…,u(Nt-1)；
l(0),l(1),…,l(Nt-1)；
w(0),w(1),…,w(Nt-1)；
target value:
v(1),v(2),…,v(Nt)；
m(1),m(2),…,m(Nt)；
y(1),y(2),…,y(Nt)；
predicted value:
v(Nt+1),v(Nt+2),...；
m(Nt+1),m(Nt+2),...；
y(Nt+1),y(Nt+2),...；
wherein:
u∈R M*3 ，W IR ∈R N*M ，W res ∈R N*N ，r∈R N*3 ，W RO ∈R L*N ,v∈R L*3 ；W IR ∈R N*M ,W res ∈R N*N all are values given in advance, and only W needs to be calculated in the calculation process RO ∈R L*N And (3) obtaining the product.
From the operations input to the library:
W IR *u(t)，W IR *l(t)，W IR *w(t)；
updating r (t) in library:
r(t+Δt)＝f[W res *r(t)+W IR *u(t)]，
r(t+Δt)＝f[W res *r(t)+W IR *l(t)]，
r(t+Δt)＝f[W res *r(t)+W IR *w(t)]；
from library to output:
u(t+Δt)＝W RO *r(t+Δt)，
l(t+Δt)＝W RO *r(t+Δt)，
w(t+Δt)＝W RO *r(t+Δt)；
loss function:
preferably, the mode comparison method in S3 includes the steps of:
s301, respectively inputting the tested co-IMFs into the models obtained in the S2 to obtain model prediction co-IMFs';
s302, calculating evaluation index values of MAE and RMSE between real test values and predicted values of different models;
s303, screening out the model with the minimum loss of MAE and RMSE as the optimal model.
Wherein the formula for calculating MAE is:
the formula for calculating RMSE is:
wherein the method comprises the steps of
For model predictive value, +.>
Is the true predicted value.
Preferably, in order to ensure timeliness and high accuracy of the model, model training needs to be performed regularly, and an optimal model is selected.
Preferably, the prediction system for the gas concentration prediction method includes the following module units:
the working face gas concentration data acquisition unit is used for acquiring the gas concentration data of the coal mine tunneling working face under the mine in real time;
the working face gas concentration data processing unit is used for preprocessing the obtained gas data and processing the complexity of the data to obtain a tunneling working face gas concentration data set which can be used for training the echo state network;
the working face gas concentration data tunneling unit is used for learning and tunneling working face gas concentration data by adopting a CEEMDAN-SE three-dimensional echo state network, and then performing data training to obtain a prediction result;
the prediction model comparison unit is used for measuring a plurality of prediction models through different indexes, comparing the correlation between the real value and the predicted value of the model with the loss, and selecting the model with the minimum error and the minimum loss;
the latest captured gas concentration data processing judging unit is used for deploying the optimal model into a safety control server of the coal mine, inputting the latest captured gas concentration data into the model, judging that the safety environment of the gas is abnormal if the calculated result is not in a safe and controllable numerical value interval, and timely processing is needed, otherwise, the safety environment is considered as normal.
Compared with the prior art, the invention has the following beneficial effects:
the invention provides a tunneling working face gas concentration prediction method based on a three-dimensional echo state network, which processes data modal decomposition and calculation complexity and amplifies correlation and modal property between data; the training method of the three-dimensional echo state network based on CEEMDAN-SE not only can effectively process random and large amount of data and generate a network learning result by distributing different weights, but also can effectively reduce time cost and improve operation efficiency, has considerable robustness on the result during training, is further beneficial to improving the accuracy of gas data prediction and recognition, can accurately find abnormal data characteristics in the predicted data, and performs safety measure intervention.
Drawings
In the present invention, in order to more clearly describe the specific embodiments of the present invention, the following description will simply make the form of the drawings of the implementation steps.
FIG. 1 is a schematic flow chart of a tunneling working face gas concentration prediction method based on a three-dimensional echo state network;
FIG. 2 is a diagram of a gas concentration data intelligent system prediction framework of a tunneling working face gas concentration prediction method based on a three-dimensional echo state network;
FIG. 3 is an empirical mode decomposition diagram of a self-adaptive noise complete set for data preprocessing of a tunneling working face gas concentration prediction method based on a three-dimensional echo state network;
FIG. 4 is a prediction effect diagram of a tunneling working face gas concentration prediction method based on a three-dimensional echo state network, which is provided by the invention;
FIG. 5 is a table of predictive performance indexes of each predictive model for comparing heading face gas data based on the three-dimensional echo state network heading face gas concentration prediction method provided by the invention;
FIG. 6 is a system block diagram of a tunneling working face gas concentration prediction method based on a three-dimensional echo state network.
In the figure: 1. a working face gas concentration data acquisition unit; 2. a working face gas concentration data processing unit; 3. a working face gas concentration data tunneling unit; 4. a prediction model comparison unit; 5. and the latest captured gas concentration data processing and judging unit.
Detailed Description
The following description is presented to enable one of ordinary skill in the art to make and use the invention. The preferred embodiments in the following description are by way of example only and other obvious variations will occur to those skilled in the art.
In order to better understand the gas concentration prediction method in the present invention, the terms related to the embodiments of the present invention will be explained first.
CEEMDAN algorithm
The CEEMDAN method achieves a reconstruction error of almost 0 with a small average number of times by adding adaptive white noise at each decomposition stage.
EMD was proposed by Huang in 1998, and the method can adaptively decompose natural mode components (intrinsic model function, IMF) with different frequencies on the original signal, so that the method is strong in adaptability. EEMD is an improved method based on EMD, mainly comprising the steps of adding different Gaussian white noise into an original sequence for multiple times, then respectively carrying out EMD decomposition, and finally averaging the obtained IMF components to obtain a final result, thereby avoiding the occurrence of modal aliasing. However, in practice, white noise added by the EEMD method is not completely cancelled after being averaged multiple times. The algorithm depends on the magnitude and average number of white noise added. The CEEMDAN method achieves a reconstruction error of almost 0 with a small average number of times by adding adaptive white noise at each decomposition stage. Therefore, the CEEMDAN method can overcome the modal aliasing phenomenon existing in EMD, and solve the problems of incomplete decomposition of EEMD and low calculation efficiency caused by the need of reducing the reconstruction error by increasing the average frequency.
SE algorithm
Sample Entropy (SE) is a new method proposed by Richman et al in 2000 and capable of measuring the complexity of a time sequence, is an improvement of approximate entropy (approximate entropy, AE), reduces the dependence on the length of the time sequence, and can effectively reduce errors of the approximate entropy in the calculation process. Given a sequence of historical PM10 concentration values { x (i) |1+.i+.ltoreq.N }, m is the mode dimension, and r is the similarity margin.
Three-dimensional echo state network
As a novel neural network Echo State Network (ESN), the network is based on the basic principle of neural network in biology, and after understanding and abstracting the human brain structure and the external stimulus response mechanism, the network topology knowledge is used as a theoretical basis to simulate the neural system of the human brain to a complex information processing mechanism. The echo state network consists of an input layer, a reserve pool and an output layer. Neurons in the reservoir are interconnected to retain information left over at a previous time. The connection weights of the input layer of the echo state network to the reserve pool and the reserve pool are randomly initialized. In the training process, only the connection weight of the reserve pool to the output layer needs to be trained, which becomes a linear regression problem, so that the echo state network is trained very fast. It can overcome some drawbacks of the conventional neural network, such as low training efficiency, slow convergence speed of algorithm, etc., and is also commonly used for solving practical problems. However, in the research of the echo state network, there are many unsolved problems, for example, the full connection of the output synapse leads to the degradation of the network prediction performance, the random connection of the nerve synapse in the reserve pool leads to the degradation of the network prediction accuracy, and the random connection between the nerve cells in the reserve pool degrades the network prediction performance.
Improved three-dimensional echo state network formula
Given three signals:
u(0),u(1),…,u(Nt-1)；
l(0),l(1),…,l(Nt-1)；
w(0),w(1),…,w(Nt-1)；
and a target value:
v(1),v(2),…,v(Nt)；
m(1),m(2),…,m(Nt)；
y(1),y(2),…,y(Nt)；
predicted value:
v(Nt+1),v(Nt+2),...；
m(Nt+1),m(Nt+2),...；
y(Nt+1),y(Nt+2),...；
wherein:
u∈R M*3 ，W IR ∈R N*M ，W res ∈R N*N ，r∈R N*3 ，W RO ∈R L*N ,v∈RL*3；W IR ∈R N*M ,W res ∈R N*N all are values given in advance, and only W needs to be calculated in the calculation process RO ∈R L*N And (3) obtaining the product.
From the operations input to the library:
W IR *u(t)，W IR *l(t)，W IR *w(t)；
updating r (t) in library:
r(t+Δt)＝f[W res *r(t)+W IR *u(t)]，
r(t+Δt)＝f[W res *r(t)+W IR *l(t)]，
r(t+Δt)＝f[W res *r(t)+W IR *w(t)]；
from library to output:
u(t+Δt)＝W RO *r(t+Δt)，
l(t+Δt)＝W RO *r(t+Δt)，
w(t+Δt)＝W RO *r(t+Δt)；
loss function:
secondly, the method for predicting the gas concentration of the tunneling working face based on the three-dimensional echo state network is implemented in detail:
referring to fig. 1 to 6, a method for predicting gas concentration of a tunneling working face based on a three-dimensional echo state network is provided, which comprises the following steps:
in a gas safety system, original safety information is processed, historical gas concentration data is extracted from a database of a mine control center console, preprocessing analysis is carried out on the historical gas concentration data, and then an available historical tunneling working face gas concentration prediction data set is established;
dividing the obtained available historical tunneling working face gas concentration prediction data set into two types, wherein one type is a training set, the other type is a test set, training the constructed CEEMDAN-SE three-dimensional echo state network through the training set, and performing comparison test on the trained model through the test set;
inputting real-time gas concentration data into the trained echo state network model, and obtaining a result predicted by the model of the invention as a predicted value of gas concentration;
wherein processing the original security information comprises:
s1, acquiring underground gas concentration data in real time, preprocessing the acquired historical gas concentration data, and performing empirical mode decomposition to acquire a gas data set which can be used for training an echo state network in the invention;
real-time detection is carried out on each index data in the frame through a sensor installed in the mine, and the real-time detected data are transmitted to a line to acquire underground gas concentration data in real time;
s2, learning the gas concentration data of the tunneling working face by adopting a CEEMDAN-SE three-dimensional echo state network, and then performing data training to obtain a prediction result;
for a downhole gas data set, we process the data by adaptive noise complete set empirical mode decomposition (CEEMDAN) method, let E i (. Cndot.) is the ith eigen mode component obtained by EMD decomposition and the ith eigen mode component obtained by CEEMDAN decomposition
u j In order to satisfy the standard normally distributed gaussian white noise signal, j=1, 2, …, n. is the number of times white noise is added, epsilon is the standard table in which white noise is added, and y (t) is the signal to be decomposed. The CEEMDAN decomposition steps are as follows:
reading data to be decomposed;
adding Gaussian white noise into the number y (t) to be decomposed to obtain a new signal:
y(t)+(-1) q εv j (t)
wherein q=1, 2. EMD decomposition is performed on the new data to obtain a first order eigenmode component C 1 . The formula is:
the first eigenmode component of CEEMDAN decomposition is obtained by ensemble averaging the N modal components generated:
calculating a residual error after removing the first modal component:
at r 1 Adding positive and negative pair Gaussian white noise into (t) to obtain new data, and carrying out EMD (empirical mode decomposition) by taking the new data as a carrier to obtain a first-order modal component D 1 The 2 nd eigenmode component of the CEEMDAN decomposition can thus be obtained:
calculating a residual error after removing the second modal component:
repeating the steps until the obtained residual data is a monotonic function, and the decomposition cannot be continued, and ending the algorithm. When the number of the eigenvalue components obtained at this time is K, the raw data y (t) is decomposed into:
and evaluating the complexity of the time sequence corresponding to the obtained K eigenvalue components according to sample entropy (SampEn), and dividing the complexity into three types, namely a high-frequency eigenvalue component co-IMF1, a medium-frequency eigenvalue component co-IMF2 and a low-frequency modal component co-IMF3 according to different complexities.
The obtained co-IMF1, co-IMF2 and co-IMF3 are input into a three-dimensional echo state network for training, wherein the size of a library is required to be determined in the echo state network, the spectrum radius sr=1.5, the pool size n=2000, the average degree d=2, the regularization constant eta=1e-4 and the random seed feed=20000, and an Adam optimization function is used, and the network activation function is tanh.
And (3) obtaining training prediction result sequences co-IMF ', co-IMF', and carrying out reconstruction and addition on the results to obtain the prediction result which is finally needed by the user.
S3, measuring a plurality of different models through different measuring indexes, and comparing the correlation between the true value and the predicted value of the model to select the model with the minimum error and the minimum loss.
Respectively calculating average absolute error values and root mean square error values of a plurality of models, wherein models corresponding to the two parameter values as small as possible should be selected;
s4, applying the optimal model to the gas safety field for predicting the gas concentration of the tunneling working face, if the calculated result is not in a safe and controllable numerical value interval, judging that the gas safety environment is abnormal, and timely processing is needed, otherwise, the situation is normal.
In order to ensure that the model can obtain more accurate results, the model needs to be updated frequently, network data training is performed again at intervals to select the optimal model, namely, the data of the past four months are trained by using the past 36 days, and in the example, the training set time length is four months;
the prediction system for the gas concentration prediction method comprises the following module units:
the working face gas concentration data acquisition unit 1 is used for acquiring the gas concentration data of the coal mine tunneling working face under the mine in real time;
the working face gas concentration data processing unit 2 is used for preprocessing the obtained gas data and processing the complexity of the data to obtain a tunneling working face gas concentration data set which can be used for training an echo state network;
the working face gas concentration data tunneling unit 3 is used for learning and tunneling working face gas concentration data by adopting a three-dimensional echo state network of CEEMDAN-SE, and obtaining a plurality of models according to different model structures and different training methods;
the prediction model comparison unit 4 is used for measuring a plurality of prediction models through different indexes, comparing the correlation and the loss between the real value and the predicted value of the model, and selecting the model with the minimum error and the minimum loss;
the latest captured gas concentration data processing judging unit 5 is used for deploying the optimal model into a safety control server of the coal mine, inputting the latest captured gas concentration data into the model, judging that the safety environment of the gas is abnormal if the calculated result is not in a safe and controllable numerical value interval, and timely processing is needed, otherwise, the safety environment is considered as normal.
Finally, the result is demonstrated through experiments
The data used in the experiment are the data collected from the gas disaster intelligent accurate early warning system of five tiger mountain, including the data of the gas workbench between 2021 and 2022, wherein the data of 2021-11-01 to 2022-02-28 time period are adopted, 1 data are read every few minutes, and 25920 data are collected as the gas concentration data set.
Fig. 4 is a prediction effect diagram of a tunneling working face gas concentration prediction method based on a three-dimensional echo state network, and it can be seen that the prediction result has the characteristics of high prediction accuracy, small loss and high robustness. Through multiple predictions, the evaluation index RMSE of a prediction set and a training set of the coal mine tunneling working face gas concentration prediction method by combining an adaptive noise complete set empirical mode decomposition-sample entropy (CEEMDAN-SE) and a three-dimensional echo state network (3D-ESN) is 1.311, and the evaluation index MAE is 1.142. The evaluation index RMSE of the traditional ESN in the prediction set and the training set is 4.151, the evaluation index MAE is 3.543, and the results of multiple experiments tend to be stable and are higher than the evaluation indexes of the prediction set and the training set of the coal mine tunneling working face gas concentration prediction method combining the adaptive noise complete set empirical mode decomposition-sample entropy (CEEMDAN-SE) and the three-dimensional echo state network (3D-ESN);
therefore, the model disclosed by the invention can track the fluctuation trend of the gas concentration time sequence better, the predictive evaluation index value is minimum, and the predictive accuracy is highest. The method has the characteristics of practicability and accuracy in the field of gas safety, is favorable for improving the accuracy of gas data prediction and identification, can accurately find abnormal data characteristics in predicted data, and transmits the abnormal prediction monitoring data to staff in an alarm mode for safety measure intervention.
The foregoing has shown and described the basic principles, principal features and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the foregoing embodiments, but rather, the foregoing embodiments and description illustrate the principles of the invention, and that various changes and modifications may be effected therein without departing from the spirit and scope of the invention as defined by the appended claims and their equivalents.
Claims (9)
1. The tunneling working face gas concentration prediction method based on the three-dimensional echo state network is characterized by comprising the following steps of:
s1, acquiring gas concentration data of a coal mine tunneling working face under a mine in real time, preprocessing the acquired gas data, and processing data complexity to obtain a tunneling working face gas concentration data set which can be used for training an echo state network;
s2, learning the gas concentration data of the tunneling working face by adopting a CEEMDAN-SE three-dimensional echo state network, and then performing data training to obtain a prediction result;
s3, measuring related prediction models through different indexes, comparing the correlation between the real value and the predicted value of the models with the loss, and selecting the model with the minimum error and the minimum loss;
s4, deploying the optimal model into a safety control server of the coal mine, inputting the latest captured gas concentration data into the model, judging that the safety environment of the gas is abnormal if the calculated result is not in a safe and controllable numerical range, and treating in time if the safety environment is abnormal, otherwise, judging that the safety environment is normal.
2. The method for predicting the gas concentration of a tunneling working face based on a three-dimensional echo state network according to claim 1, wherein the method for acquiring and processing the gas concentration data of the working face in S1 comprises the following steps:
s101, acquiring the data of gas sensors of all working surfaces from a coal mine gas disaster wind direction management and control platform in real time;
s102, carrying out modal decomposition on the gas concentration data obtained from different sensor positions according to CEEMDAN pretreatment experience;
s103, carrying out sample entropy calculation data complexity classification processing on the data set after the modal decomposition processing;
s104, circularly traversing the whole gas concentration data to obtain the same sequence numbers of co-IMF1, co-IMF2 and co-IMF3 from the same sequence data but in three different complexity ranges.
3. The method for predicting the gas concentration of the tunneling working face based on the three-dimensional echo state network according to claim 2, wherein the empirical mode decomposition method for data preprocessing in S102 comprises the following steps:
s102-1, reading data;
s102-2, setting the processing times K of the original data;
s102-3, respectively adding random white noise to the K pieces of original data to form a series of new data;
s102-4, performing EMD (empirical mode decomposition) on the series of new data to obtain a series of IMF (intrinsic mode function) components;
s102-5, respectively averaging IMF components of the corresponding modes to obtain EEMD decomposition results;
s102-6, carrying out ensemble average on the modal components after EEMD decomposition, carrying out ensemble average calculation on the obtained first-order IMF components after CEEMDAN decomposition, obtaining final first-order IMF components, and then repeating the operation on the residual parts.
4. The method for predicting gas concentration of tunneling working face based on three-dimensional echo state network according to claim 3, wherein the processing method for classifying complexity of data set in S103 comprises the following steps:
s103-1, performing sample entropy calculation on the collected K IMF component sequence data obtained by CEEMDAN decomposition;
s103-2, calculating the sequence complexity of each IMF component through sample entropy, and then dividing the K IMF components into three types according to the complexity of different sequence data, namely a high-frequency intrinsic mode component co-IMF1, a medium-frequency mode component co-IMF2 and a low-frequency mode component co-IMF3.
5. The method for predicting gas concentration of tunneling working face based on three-dimensional echo state network according to claim 4, wherein the data processing method of the three-dimensional echo state network of CEEMDAN-SE in S2 comprises the following steps:
s201, dividing the wattage concentration degree co-IMF1, co-IMF2 and co-IMF3 into training sets D-co-IMF1, D-co-IMF2 and D-co-IMF3 and test sets T-co-IMF1, T-co-IMF2 and T-co-IMF3;
s202, selecting the architecture of a one-dimensional echo state network model, and then improving the one-dimensional echo state network into a three-dimensional echo state network capable of inputting three-dimensional data;
s203, determining the number of nodes, the spectrum radius SR, the reserve pool scale N, the reserve pool input unit scale IS, the reserve pool sparseness degree SD and the regularization factor, wherein each change of the parameters can generate a new model, and meanwhile, the experimental result IS closely related to the parameters, so that the experimental result IS prevented from fitting or overfitting due to incorrect selection of the parameters.
6. The method for predicting gas concentration of a tunneling working face based on a three-dimensional echo state network according to claim 4, wherein the gas concentration data co-IMF in S201 is segmented into a training set D-co-IMF and a prediction set T-co-IMF according to a ratio of 7:3.
7. The method for predicting gas concentration in a tunneling working surface based on a three-dimensional echo state network according to claim 5, wherein the improving the data processing formula of the three-dimensional echo state network in S202 comprises:
the improved three-dimensional echo state network formula is as follows:
given three signals:
target value:
predicted value:
wherein:
only calculation is needed in the calculation process
The preparation method is finished;
from the operations input to the library:
in a warehouse
Is updated by:
from library to output:
loss function:
8. the method for predicting the gas concentration of a tunneling working face based on a three-dimensional echo state network according to claim 7, wherein the model comparison method in S3 comprises the following steps:
s301, respectively inputting the test set D-co-IMF into the models obtained in the S2 to obtain model prediction results as co-IMF;
s302, calculating evaluation index values of MAE and RMSE between real test values and predicted values of different models;
s303, screening out the model with the minimum loss of MAE and RMSE as the optimal model.
9. The method for predicting the gas concentration of the tunneling working face based on the three-dimensional echo state network according to claim 8, wherein in order to ensure timeliness and high accuracy of the model, model training is required periodically, and an optimal model is selected.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211085414.XA CN115856204B (en) | 2022-09-06 | 2022-09-06 | Tunneling working face gas concentration prediction method based on three-dimensional echo state network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211085414.XA CN115856204B (en) | 2022-09-06 | 2022-09-06 | Tunneling working face gas concentration prediction method based on three-dimensional echo state network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115856204A CN115856204A (en) | 2023-03-28 |
| CN115856204B true CN115856204B (en) | 2023-06-23 |
Family
ID=85660756
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211085414.XA Active CN115856204B (en) | 2022-09-06 | 2022-09-06 | Tunneling working face gas concentration prediction method based on three-dimensional echo state network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115856204B (en) |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106503792B (en) * | 2016-10-25 | 2018-12-18 | 西安科技大学 | A kind of gas density prediction technique based on adaptation module neural network |
| CN108921359B (en) * | 2018-07-26 | 2022-03-11 | 安徽大学 | Distributed gas concentration prediction method and device |
| CN110334451B (en) * | 2019-07-09 | 2021-04-02 | 精英数智科技股份有限公司 | Method, device and system for determining factors and trends influencing gas concentration change |
| JP2023530436A (en) * | 2020-06-17 | 2023-07-18 | ネクセリス イノベーション ホールディングス，エルエルシー | Systems and methods for monitoring gas analytes |
| CN113887637A (en) * | 2021-10-09 | 2022-01-04 | 安徽理工大学 | Coal and gas outburst accident prediction and identification method based on deep cycle neural network |
-
2022
- 2022-09-06 CN CN202211085414.XA patent/CN115856204B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN115856204A (en) | 2023-03-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115578015B (en) | Sewage treatment whole process supervision method, system and storage medium based on Internet of things | |
| CN110442666B (en) | Mineral resource prediction method and system based on neural network model | |
| US20190227504A1 (en) | Computer System And Method For Monitoring Key Performance Indicators (KPIs) Online Using Time Series Pattern Model | |
| Zhu et al. | Dimensionality reduce-based for remaining useful life prediction of machining tools with multisensor fusion | |
| CN116341901B (en) | Integrated evaluation method for landslide surface domain-monomer hazard early warning | |
| CN112364560B (en) | Intelligent prediction method for working hours of mine rock drilling equipment | |
| CN116380445B (en) | Equipment state diagnosis method and related device based on vibration waveform | |
| CN112287988A (en) | Method for identifying water pollution source online monitoring data abnormity | |
| CN115828140A (en) | Neighborhood mutual information and random forest fusion fault detection method, system and application | |
| CN111179591A (en) | Road network traffic time sequence characteristic data quality diagnosis and restoration method | |
| CN111967308A (en) | Online road surface unevenness identification method and system | |
| CN111930601A (en) | Deep learning-based database state comprehensive scoring method and system | |
| Ferrer-Cid et al. | Volterra graph-based outlier detection for air pollution sensor networks | |
| CN115438897A (en) | Industrial process product quality prediction method based on BLSTM neural network | |
| CN115856204B (en) | Tunneling working face gas concentration prediction method based on three-dimensional echo state network | |
| Joslyn et al. | A supervised learning approach to water quality parameter prediction and fault detection | |
| CN116720743A (en) | Carbon emission measuring and calculating method based on data clustering and machine learning | |
| Hosseini et al. | The design of robust soft sensor using ANFIS network | |
| CN113988210A (en) | Method and device for restoring distorted data of structure monitoring sensor network and storage medium | |
| CN107220393B (en) | Atmospheric tower common line dry point prediction method for atmospheric and vacuum device | |
| CN107220252B (en) | Atmospheric tower normal line flash point prediction method for atmospheric and vacuum device | |
| CN111126694A (en) | Time series data prediction method, system, medium and device | |
| Dupree et al. | Time forecasting satellite light curve patterns using Neural Networks | |
| CN117792933B (en) | Network flow optimization method and system based on deep learning | |
| Hong et al. | Drainage network flow anomaly classification based on XGBoost |
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 |