CN110633504A - Prediction method for coal bed gas permeability - Google Patents
Prediction method for coal bed gas permeability Download PDFInfo
- Publication number
- CN110633504A CN110633504A CN201910774144.5A CN201910774144A CN110633504A CN 110633504 A CN110633504 A CN 110633504A CN 201910774144 A CN201910774144 A CN 201910774144A CN 110633504 A CN110633504 A CN 110633504A
- Authority
- CN
- China
- Prior art keywords
- prediction
- network
- data
- gas permeability
- coal bed
- 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 32
- 238000000034 method Methods 0.000 title claims abstract description 29
- 230000035699 permeability Effects 0.000 title claims abstract description 26
- 238000012549 training Methods 0.000 claims abstract description 24
- 230000002068 genetic effect Effects 0.000 claims abstract description 23
- 238000013528 artificial neural network Methods 0.000 claims abstract description 21
- 238000003062 neural network model Methods 0.000 claims abstract description 6
- 238000012545 processing Methods 0.000 claims abstract description 4
- 210000002569 neuron Anatomy 0.000 claims description 12
- 230000035772 mutation Effects 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 claims description 3
- 210000000349 chromosome Anatomy 0.000 claims description 2
- 238000011156 evaluation Methods 0.000 claims description 2
- 238000005457 optimization Methods 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 claims description 2
- 238000013138 pruning Methods 0.000 claims description 2
- 238000005070 sampling Methods 0.000 claims description 2
- 238000012163 sequencing technique Methods 0.000 claims description 2
- 238000011161 development Methods 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 4
- 238000005065 mining Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
Images
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
Abstract
The invention relates to the technical field of coal mine safety and coal bed gas development, in particular to a coal bed gas permeability prediction method. The method comprises the following steps: (1) collecting coal mine related data includes: the method comprises the following steps of (1) processing four parameters of gas pressure, reservoir temperature, compressive strength and effective stress and corresponding permeability; (2) setting relevant parameters of an Elman network, preliminarily establishing a network prediction model, and determining a network structure; (3) setting relevant parameters of a genetic algorithm and combining the relevant parameters with a neural network model; (4) training the established neural network prediction model optimized by the genetic algorithm by using training data; (5) and evaluating the accuracy of the established prediction model by using the prediction data. The method optimizes the neural network parameters by using the genetic algorithm, improves the model prediction precision, provides a new method for predicting the coal bed gas permeability, and greatly improves the prediction accuracy.
Description
Technical Field
The invention relates to the technical field of coal mine safety and coal bed gas development, in particular to a coal bed gas permeability prediction method.
Background
China has abundant coal resources, and reserves are ranked second in the world. The safety accidents frequently occur during mining, which causes casualties and a great deal of economic loss, wherein most accidents are caused by coal bed gas. In the process of mining, an original gas stress field and an original ground stress field are damaged, and gas moves unstably under the action of the two forces, so that mine disasters such as gas outburst are caused. Therefore, how to effectively prevent the gas outburst problem in the coal mining process is a problem to be solved urgently at present.
The Elman model is a network prediction model with strong practicability and high reliability, is generally applied to production practice, but has the defect that the Elman model is easy to fall into local minimum points and cannot achieve the optimal value, and influences the prediction precision.
Disclosure of Invention
In order to solve the problems of actual engineering and the defects in the prior art, the invention provides the prediction model for optimizing the neural network parameters by using the genetic algorithm, improves the model prediction precision, has strong practicability and operability, and is a coal bed gas permeability prediction method suitable for popularization.
In order to achieve the technical purpose, the invention provides a coal seam gas permeability prediction method, which comprises the following steps:
step 1, collecting related data of a coal mine, and preprocessing the related data;
step 2, setting relevant parameters of the Elman network, primarily establishing a network model, and determining a network structure;
step 3, setting relevant parameters of a genetic algorithm and combining the relevant parameters with a neural network model to establish a GA-Elman network model;
step 4, training the established neural network by using training data;
wherein, the coal mine related data in the step 1 comprises the following steps: in order to avoid the effect of large unit difference of data on four influencing factors of gas pressure, reservoir temperature, compressive strength and effective stress and corresponding permeability, normalization processing is carried out on the four influencing factors, so that the data value is between [0 and 1], and the normalization formula is as follows:
in the formula, xnFor the original sampling parameter, xminFor inputting the minimum value, x, of the same kind of parametersmaxThe maximum value among the same kind of parameters is input.
The relevant parameters in step 2 include: the method comprises the steps of inputting layer neuron node numbers, hiding layer neuron node numbers, outputting layer neuron node numbers, network establishing functions, network training functions, learning momentum parameters, learning rates, iteration targets, maximum iteration times and the like.
The number of input layer neuron nodes and the number of output layer neuron nodes in the step 2 are determined according to the input and output parameters and are respectively 4 and 1; the hidden layer neuron node number is selected according to an empirical formula, the selected node number is actually trained and evaluated by a pruning method, and finally 9 nodes are determined, wherein the empirical formula is as follows:
in the formula, n is the number of nodes of an input layer; m is the number of nodes of the output layer; a is an arbitrary number from 1 to 10.
The genetic algorithm optimization neural network in the step 3 mainly comprises the following steps:
step (1), initializing a population P, including a cross probability PcProbability of mutation PmAnd the termination algebra T, the algebra G and the like are adopted to encode the weight value and the threshold value of the Elman neural network by adopting binary coding, determine the initial scale M of the population and randomly generate an initial population.
Step (2), calculating the fitness of each individual and sequencing the fitnessAccording to the formulaSelecting the individual, wherein fiThe fitness value of the ith individual is measured by the sum of squared errors Ei, the Ei is the total error of the ith individual, and the calculation formula is shown as follows
In the formula: i is the chromosome number, i is 1, 2, …, n; k is the number of learning samples, k is 1, 2, …, m; ro is a target output value; d is the desired output value.
And (3) performing cross operation according to the cross probability Pc to obtain new excellent individuals, and performing self-replication operation on individuals without cross operation.
And (4) performing mutation operation by using the mutation probability Pm to keep the diversity of individuals so as to ensure the effectiveness of the genetic algorithm.
And (5) inserting the individuals generated in the steps (3) and (4) into the original population to form a new population, and then performing the second step of operation.
And (6) repeating the steps (2) to (5) until an individual meeting the requirements is found, and then bringing the finally determined optimal individual, namely the network weight and the threshold value into a neural network model for training and distinguishing.
The genetic algorithm related parameters in the step 3 comprise population size, maximum genetic algebra, gully, cross probability, mutation probability and individual length, and the specific values are set as shown in the following table:
TABLE 1 genetic Algorithm parameter set-ups
And 4, classifying the collected data into a training data set and a prediction data set, training the network model by the training data, and detecting the model precision by the prediction data.
And 5, evaluating the precision of the established prediction model by using the prediction data, and evaluating the index prediction precision.
The technical scheme provided by the invention has the beneficial effects that:
according to the coal bed gas permeability prediction method, the collected original data are normalized by the normalization formula through collecting the coal bed related data, and the influence of unit difference of the data on the result is reduced. An Elman neural network model is established by using MATLAB software, and the structure is simple and practical. The Elman neural network is optimized by utilizing the genetic algorithm, so that the prediction precision is improved, the adaptability of the model is improved, the model has strong practicability and operability, a new method is provided for predicting the coal bed gas permeability, and the accuracy of the prediction of the bed gas permeability is improved.
Drawings
FIG. 1: flow chart of the invention.
FIG. 2: elman neural network structure diagram.
FIG. 3: training effects of different intermediate node numbers.
FIG. 4: and (4) genetic evolution process map.
FIG. 5: and (5) training an error change diagram by the neural network.
FIG. 6: and (5) a prediction result graph.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in detail with reference to the following embodiments and the accompanying drawings.
As shown in fig. 1 to 6, the method for predicting the gas permeability of the coal seam specifically comprises the following steps:
1. data collection and processing
50 groups of coal mine information are collected, and each group of data comprises effective stress, MPa; gas pressure, MPa; reservoir temperature, deg.C; compressive strength, MPa; permeability, 10-15m2(ii) a And normalizing each factor by using the formula (1).
And selecting 40 processed groups of data as network training data, and 10 groups of data as network prediction data to predict the accuracy and generalization capability of the network training data.
2. Neural network setup
Determining the number of nodes of the neural network according to the characteristics of the data collected and processed in the step 1, wherein the number of the nodes of the neural network comprises four input nodes, hidden nodes and output nodes, the four input nodes are respectively gas pressure, reservoir temperature, compressive strength and effective stress, so that the number of the input nodes is 4, the number of the output nodes is one, the permeability is high, the number of the output nodes is 1, and the hidden nodes are selected through a formula (2). As shown in fig. 2, x is an input parameter; y is an output parameter; omegaijThe weight value between the ith input layer node and the jth hidden layer node is obtained; beta is ak1The weight between the kth hidden layer node and the output layer node; alpha is alphalThe weight from the l bearing layer node to the l hidden layer node. The result of calculating the mean square error is shown in fig. 3. Finally, 9 hidden nodes are determined.
Other parameters of the neural network, including the network creation function, training function, momentum parameters, learning rate, iteration goal, and maximum iteration number settings are shown in table 2.
Table 2 Elman network concrete parameter setting table
3. Genetic algorithms combined with neural networks
Relevant parameters of the genetic algorithm are set, and specific parameter settings of the example are shown in table 1.
TABLE 1 genetic Algorithm parameter set-ups
The genetic algorithm toolbox is utilized to establish the genetic algorithm according with the table 1, and a prediction model is established by combining with the neural network.
The established GA-Elman neural network is trained by utilizing 40 groups of training data, and the method specifically comprises the following steps: determining an initial weight and a threshold, calculating fitness, selecting operation, cross operation, mutation operation and the like, and finally obtaining the optimal network weight and threshold. The genetic evolution process is shown in fig. 4, and the neural network training error variation is shown in fig. 5.
4. Evaluation of model accuracy and generalization ability
10 groups of data which do not participate in network training are brought into a network model, the prediction precision, the mean square error and the correlation coefficient between the predicted permeability and the actual permeability are calculated, the performance of the model is evaluated, the prediction result is shown in figure 6, the prediction precision of the 10 groups of predicted data is 93.24 percent through calculation and is basically consistent with the actual numerical value, and the actual engineering requirement is met.
Claims (8)
1. The method for predicting the GAs permeability of the coal bed is based on a GA-Elman neural network, and comprises the following steps of:
step 1: collecting related data of a coal mine, and preprocessing the related data;
step 2: setting relevant parameters of an Elman network, initially establishing a network model, and determining a network structure;
and step 3: setting relevant parameters of a genetic algorithm and combining the relevant parameters with a neural network model to establish a GA-Elman network model;
and 4, step 4: training the established neural network by using training data;
and 5: and evaluating the accuracy of the established prediction model by using the prediction data.
2. The method for predicting coal seam gas permeability according to claim 1, wherein the coal mine related data in the step 1 comprises: four influencing factors of effective stress, gas pressure, reservoir temperature and compressive strength and corresponding permeability are adopted, in order to avoid the influence effect of large data unit difference, normalization processing is carried out on the effective stress, the gas pressure, the reservoir temperature and the compressive strength, the data value is enabled to be between [0 and 1], and the normalization formula is as follows:
in the formula, xnFor the original sampling parameter, xminFor inputting the minimum value, x, of the same kind of parametersmaxThe maximum value among the same kind of parameters is input.
3. The method for predicting coal bed gas permeability according to claim 1, wherein the relevant parameters in the step 2 include an input layer neuron node number, an implicit layer neuron node number, an output layer neuron node number, a network creation function, a network training function, a learning momentum parameter, a learning rate, an iteration target, and a maximum iteration number.
4. The method for predicting coal bed gas permeability according to claim 1, wherein the number of input layer neuron nodes and the number of output layer neuron nodes in the step 2 are determined according to input and output parameters, and are respectively 4 and 1; the hidden layer neuron node number is selected according to an empirical formula, the selected node number is actually trained and evaluated by a pruning method, and finally 9 nodes are determined, wherein the empirical formula is as follows:
in the formula, n is the number of nodes of an input layer; m is the number of nodes of the output layer; a is an arbitrary number from 1 to 10.
5. The method for predicting coal bed gas permeability according to claim 1, wherein the genetic algorithm optimization neural network in the step 3 mainly comprises the following steps:
step (1): initializing a population P, including a crossover probability PcProbability of mutation PmTerminating algebra T and algebra G, coding the weight and the threshold of the Elman neural network by adopting binary coding, determining the initial scale M of the population and randomly generating an initial population;
step (2): calculating the fitness of each individual and sequencing the fitness according to a formulaSelecting the individual, wherein fiThe fitness value of the ith individual is measured by the sum of squared errors Ei, the Ei is the total error of the ith individual, and the calculation formula is shown as follows
In the formula: i is the chromosome number, i is 1, 2, …, n; k is the number of learning samples, k is 1, 2, …, m; ro is a target output value; d is the desired output value;
and (3): carrying out cross operation according to the cross probability Pc to obtain new excellent individuals, and carrying out self-copying operation on individuals without cross operation;
and (4): performing mutation operation according to the mutation probability Pm to keep the diversity of individuals so as to ensure the effectiveness of a genetic algorithm;
and (5): inserting the individuals generated in the step (3) and the step (4) into the original population to form a new population, and then carrying out the next operation;
and (6): repeating the steps (2) to (5) until an individual meeting the requirements is found, and then bringing the finally determined optimal individual, namely the network weight and the threshold value into a neural network model for training and distinguishing.
6. The method for predicting coal bed gas permeability according to claim 1, wherein the parameters related to the genetic algorithm in the step 3 include population size, maximum genetic algebra, gully, cross probability, variation probability and individual length, and the specific values thereof are set as shown in the following table:
7. the method for predicting coal bed gas permeability according to claim 1, wherein the collected data in the step 4 are classified into a training data set and a prediction data set, the training data are used for training a network model, and the prediction data are used for detecting model accuracy.
8. The method for predicting coal bed gas permeability according to claim 1, wherein in the step 5, the accuracy of the established prediction model is evaluated by using the prediction data, and the evaluation index is the prediction accuracy.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910774144.5A CN110633504A (en) | 2019-08-21 | 2019-08-21 | Prediction method for coal bed gas permeability |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910774144.5A CN110633504A (en) | 2019-08-21 | 2019-08-21 | Prediction method for coal bed gas permeability |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110633504A true CN110633504A (en) | 2019-12-31 |
Family
ID=68970673
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910774144.5A Pending CN110633504A (en) | 2019-08-21 | 2019-08-21 | Prediction method for coal bed gas permeability |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110633504A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111950203A (en) * | 2020-08-13 | 2020-11-17 | 中核华辰建筑工程有限公司 | Blasting vibration speed prediction method based on adaptive neural fuzzy inference system |
CN112330435A (en) * | 2020-09-29 | 2021-02-05 | 百维金科(上海)信息科技有限公司 | Credit risk prediction method and system for optimizing Elman neural network based on genetic algorithm |
CN113033954A (en) * | 2021-02-18 | 2021-06-25 | 重庆大学 | Intelligent decision-making method for coordinated development of coal and coalbed methane |
CN113807025A (en) * | 2021-10-08 | 2021-12-17 | 浪潮云信息技术股份公司 | Method for constructing neural network force field model based on global optimization algorithm |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106869990A (en) * | 2017-03-02 | 2017-06-20 | 新疆大学 | Coal gas Permeability Prediction method based on LVQ CPSO BP algorithms |
CN108665095A (en) * | 2018-04-27 | 2018-10-16 | 东华大学 | Short term power prediction technique based on genetic algorithm optimization Elman neural networks |
-
2019
- 2019-08-21 CN CN201910774144.5A patent/CN110633504A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106869990A (en) * | 2017-03-02 | 2017-06-20 | 新疆大学 | Coal gas Permeability Prediction method based on LVQ CPSO BP algorithms |
CN108665095A (en) * | 2018-04-27 | 2018-10-16 | 东华大学 | Short term power prediction technique based on genetic algorithm optimization Elman neural networks |
Non-Patent Citations (1)
Title |
---|
尹光志等: "基于改进BP神经网络的媒体瓦斯渗透率预测模型", 《煤炭学报》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111950203A (en) * | 2020-08-13 | 2020-11-17 | 中核华辰建筑工程有限公司 | Blasting vibration speed prediction method based on adaptive neural fuzzy inference system |
CN112330435A (en) * | 2020-09-29 | 2021-02-05 | 百维金科(上海)信息科技有限公司 | Credit risk prediction method and system for optimizing Elman neural network based on genetic algorithm |
CN113033954A (en) * | 2021-02-18 | 2021-06-25 | 重庆大学 | Intelligent decision-making method for coordinated development of coal and coalbed methane |
CN113807025A (en) * | 2021-10-08 | 2021-12-17 | 浪潮云信息技术股份公司 | Method for constructing neural network force field model based on global optimization algorithm |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110633504A (en) | Prediction method for coal bed gas permeability | |
CN107169628B (en) | Power distribution network reliability assessment method based on big data mutual information attribute reduction | |
CN107194524B (en) | RBF neural network-based coal and gas outburst prediction method | |
CN109214026A (en) | A kind of shale gas horizontal well initial productivity prediction technique | |
CN105260599A (en) | Rockburst dynamic prediction method based on BP neural network modeling | |
CN112733417B (en) | Abnormal load data detection and correction method and system based on model optimization | |
CN103617147A (en) | Method for identifying mine water-inrush source | |
CN109934398A (en) | A kind of drill bursting construction tunnel gas danger classes prediction technique and device | |
CN112949202A (en) | Bayesian network-based rockburst probability prediction method | |
CN112364560B (en) | Intelligent prediction method for working hours of mine rock drilling equipment | |
CN108280289B (en) | Rock burst danger level prediction method based on local weighted C4.5 algorithm | |
CN115130375A (en) | Rock burst intensity prediction method | |
CN110633868A (en) | Method for predicting properties of oil and gas in exploration well oil testing layer by optimizing neural network through genetic algorithm | |
CN114021432A (en) | Stress corrosion cracking crack propagation rate prediction method and system | |
CN111984514B (en) | Log anomaly detection method based on Prophet-bLSTM-DTW | |
CN116992522A (en) | Deep foundation pit support structure deformation prediction method, device, equipment and storage medium | |
CN109116300B (en) | Extreme learning positioning method based on insufficient fingerprint information | |
CN103898890A (en) | Soil layer quantization layering method based on double-bridge static sounding data of BP neural network | |
CN116629111A (en) | Shale heat injection in-situ exploitation hydrocarbon discharge potential prediction method based on neural network | |
CN111310974A (en) | Short-term water demand prediction method based on GA-ELM | |
CN115577619A (en) | Method for predicting residual shear strength of liquefied soil based on machine learning | |
CN111340275B (en) | Tunnel support mode selection real-time prediction method based on detection while drilling technology | |
CN112348275A (en) | Regional ecological environment change prediction method based on online incremental learning | |
CN117522173B (en) | Natural gas hydrate depressurization exploitation productivity prediction method based on deep neural network | |
Chen et al. | The fault diagnosis of shield disc cutter based on neural network |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191231 |
|
RJ01 | Rejection of invention patent application after publication |