CN112765891B - Prediction method for maximum value of disaster-causing factors of mine fire disaster - Google Patents
Prediction method for maximum value of disaster-causing factors of mine fire disaster Download PDFInfo
- Publication number
- CN112765891B CN112765891B CN202110108183.9A CN202110108183A CN112765891B CN 112765891 B CN112765891 B CN 112765891B CN 202110108183 A CN202110108183 A CN 202110108183A CN 112765891 B CN112765891 B CN 112765891B
- Authority
- CN
- China
- Prior art keywords
- disaster
- mine
- factor
- fire
- maximum value
- 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
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000004088 simulation Methods 0.000 claims abstract description 44
- 238000012360 testing method Methods 0.000 claims abstract description 27
- 238000012549 training Methods 0.000 claims abstract description 20
- 239000000779 smoke Substances 0.000 claims description 10
- 238000012795 verification Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 7
- 238000010200 validation analysis Methods 0.000 claims description 7
- 238000002485 combustion reaction Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 5
- 210000002569 neuron Anatomy 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 8
- 230000008859 change Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 4
- 238000009792 diffusion process Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 239000007789 gas Substances 0.000 description 3
- 231100000331 toxic Toxicity 0.000 description 3
- 230000002588 toxic effect Effects 0.000 description 3
- 238000009423 ventilation Methods 0.000 description 3
- 238000010304 firing Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- UGFAIRIUMAVXCW-UHFFFAOYSA-N Carbon monoxide Chemical compound [O+]#[C-] UGFAIRIUMAVXCW-UHFFFAOYSA-N 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000003546 flue gas Substances 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06Q50/10—Services
- G06Q50/26—Government or public services
- G06Q50/265—Personal security, identity or safety
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/10—Numerical modelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/08—Thermal analysis or thermal optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Evolutionary Computation (AREA)
- General Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Development Economics (AREA)
- General Engineering & Computer Science (AREA)
- Marketing (AREA)
- Life Sciences & Earth Sciences (AREA)
- Educational Administration (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Computer Security & Cryptography (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to the technical field of mine safety, and provides a method for predicting the maximum value of a mine fire disaster factor, which comprises the following steps: step 1: determining a disaster factor and a disaster factor influence factor of the mine fire disaster; step 2: establishing a mine fire numerical simulation model; step 3: obtaining data of maximum values of disaster causing factors under N value combinations of the disaster causing factor influencing factors by adopting a numerical simulation method to obtain a mine fire sample set; step 4: taking the influence factors of all disaster causing factors as input and the maximum value of all disaster causing factors as output, and constructing a mine fire disaster causing factor maximum value prediction model based on MLP; step 5: training and verifying a maximum value prediction model of a mine fire disaster factor to obtain an optimal prediction model; step 6: testing the optimal prediction model; step 7: and predicting the maximum value of the fire disaster factor of the mine to be predicted. The method can realize the rapid prediction of the maximum value of the disaster-causing factors of the mine fire disaster and improve the prediction accuracy and safety.
Description
Technical Field
The application relates to the technical field of mine safety, in particular to a method for predicting the maximum value of a mine fire disaster factor.
Background
The temperature, CO generation amount and CO of mine fire are controlled 2 Yield and flue gasThe generated quantity and fire wind pressure or thermal resistance are called as mine fire disaster factors, wherein a gallery is thermal resistance, and an inclined gallery is fire wind pressure. After the mine fire disaster occurs, the fire disaster of the fire disaster, the maximum generation amount of toxic and harmful gases and the influence of the fire disaster on a ventilation system are rapidly determined, and the method has extremely important significance for rapidly making an emergency disaster relief scheme.
The factors influencing the disaster factors of the mine fire disaster mainly comprise: the cross section of the tunnel, the length of the tunnel, the inclination angle of the tunnel, the wind speed of the entrance of the tunnel, the temperature, the relative humidity, the atmospheric pressure and the fire burning speed. The roadway environment is complicated, the values of the mine fire disaster factors under different conditions are studied, and the method has important significance for judging fire disaster danger, predicting the fire disaster influence range, emergency rescue and rapidly making an emergency disaster relief scheme, and can reduce injuries and losses caused by mine fires to a great extent.
At present, mine fire theory and simulation research are mature, and experiments and numerical simulation are taken as main research means. The numerical simulation of the mine fire involves the steps of modeling, meshing, operation and the like, the time period required for simulating one fire is long, the time required for one complete numerical simulation is up to several days, and the design operation of the mine fire experiment has special danger. When a mine fire disaster occurs, the highest temperature which can be reached by the fire disaster, the maximum generation amount of toxic and harmful gases and the influence of the fire disaster on a ventilation system cannot be quantified in time are difficult to obtain quickly by using numerical simulation and experimental means, and the determination of an emergency rescue decision scheme has time delay and cannot meet the requirements of emergency rescue.
Disclosure of Invention
Aiming at the problems existing in the prior art, the application provides a method for predicting the maximum value of the disaster causing factor of the mine fire disaster, which can realize the rapid prediction of the maximum value of the disaster causing factor of the mine fire disaster and improve the prediction accuracy and the safety.
The technical scheme of the application is as follows:
a prediction method for the maximum value of a disaster-causing factor of a mine fire disaster is characterized by comprising the following steps: comprising the following steps:
step 1: determining disaster factor and disaster factor influence factor of mine fire disaster
Five disaster causing factors a for determining mine fire disaster 1 、a 2 、...、a i 、...、a 5 The temperature, the CO generation amount and the CO of mine fire disaster are respectively 2 The generation amount, the smoke generation amount, the fire wind pressure or the thermal resistance, and eight influencing factors b of disaster factors are determined 1 、b 2 、...、b k 、...、b 8 The method comprises the steps of roadway cross-sectional area, roadway length, roadway inclination angle, roadway inlet wind speed, temperature, relative humidity, atmospheric pressure and fire combustion speed; wherein a is i I is the ith disaster causing factor, i is {1,2,.. The 5}, and a is the roadway is a gallery or an inclined roadway 5 Respectively, thermal resistance and fire wind pressure, b k K is {1,2,., 8};
step 2: establishing a numerical simulation model of mine fire disaster
Selecting an actual underground roadway as a prototype, and establishing a mine fire numerical simulation model with a fixed fire source position; the mine fire numerical simulation model is used for performing numerical simulation on mine fires under different tunnel cross-sectional areas, tunnel lengths, tunnel inclination angles, tunnel inlet wind speeds, temperatures, relative humidity, atmospheric pressure and fire burning speeds;
step 3: acquiring a mine fire sample set
The numerical simulation method is adopted to run a mine fire numerical simulation model, data of maximum values of all disaster causing factors under N value combinations of the disaster causing factor influence factors are obtained, and a mine fire sample set C= { C is obtained 1 ,c 2 ,...,c n ,...,c N -a }; wherein c n For the nth mine fire sample, N e {1,2,.. N }, N is the total number of mine fire samples, c n ＝(b 1n ,b 2n ,...,b kn ,...,b 8n ,a 1n ,a 2n ,...,a in ,...,a 5n )，b kn A is the value of the factor influencing the k in the fire disaster sample of the nth mine in The maximum value of the ith disaster causing factor in the nth mine fire sample;
step 4: establishing a maximum value prediction model of mine fire disaster factor
Taking the influence factors of all disaster causing factors as input and the maximum value of all disaster causing factors as output, and constructing a mine fire disaster causing factor maximum value prediction model based on MLP; the input end of the mine fire disaster factor maximum value prediction model based on the MLP is integrated with a normalization processing module, and the output end is integrated with an inverse normalization processing module;
step 5: training mine fire disaster factor maximum value prediction model
Step 5.1: randomly selecting N from a mine fire sample set C 1 The samples form a training set from the rest of the N-N 1 Randomly selecting N from each sample 2 The individual samples constitute a validation set;
step 5.2: training the maximum value prediction model of the mine fire disaster factor by using a training set, verifying the trained maximum value prediction model of the mine fire disaster factor by using a verification set, and calculating the relative error epsilon of the verification set 1 ；
Step 5.3: determining the relative error epsilon of the validation set 1 Whether or not it is below a preset threshold: if yes, outputting a maximum value prediction model of the mine fire disaster factor at the moment as an optimal prediction model; if not, adjusting parameters of a maximum value prediction model of the mine fire disaster factor, and returning to the step 5.1;
step 6: testing the optimal prediction model
Step 6.1: selecting the rest N-N in the mine fire disaster sample set C 1 -N 2 The samples form a test set;
step 6.2: testing the optimal prediction model by using a test set, and calculating the relative error epsilon of the test set 2 ；
Step 6.3: judging the relative error epsilon of the test set 2 Whether or not it is below a preset threshold: if yes, the optimal prediction model meets the accuracy requirement, and step 7 is carried out; if not, adjusting the parameters of the optimal prediction model, and returning to the step 5.1;
step 7: predicting the maximum value of disaster causing factors of mine to be predicted
Step 7.1: acquiring data of each disaster causing factor influence factor of a mine to be predicted;
step 7.2: inputting the data of the disaster causing factor influence factors of the mine to be predicted into the optimal prediction model, and outputting the predicted value of the maximum value of each disaster causing factor of the mine to be predicted.
Furthermore, the mine fire numerical simulation model is an underground trackless transportation fire numerical simulation model and is built through FDS fire power simulation software.
Further, in the step 5.3, parameters of a mine fire disaster factor maximum value prediction model are adjusted, which specifically include: adjusting the number N of randomly selected samples in the training set 1 The method comprises the steps of a weight optimizer, an activation function, regularization item parameters, an implicit layer number, an implicit layer neuron number and a maximum iteration number.
The beneficial effects of the application are as follows:
according to the application, the numerical simulation method is utilized to obtain the data of the maximum value of each disaster causing factor under various value combinations of the mine fire disaster causing factor influence factors, a mine fire disaster sample set is obtained, and the mine fire disaster causing factor maximum value prediction model is constructed based on the multilayer perceptron MLP, so that the rapid prediction of the mine fire disaster causing factor maximum value is realized, the prediction accuracy and the prediction safety are improved, the technical problems of high time consumption and high risk of an experimental method in the numerical simulation method in the prior art are effectively solved, a powerful technical support can be provided for the formulation of a mine fire disaster emergency rescue scheme so as to meet the requirements of emergency rescue, and the effect of universality is achieved.
Drawings
FIG. 1 is a flow chart of a method for predicting the maximum value of a mine fire disaster factor according to the present application.
Fig. 2 is a schematic diagram of a mine fire numerical simulation model of a certain inclined roadway in the specific embodiment.
Fig. 3 is a schematic diagram of internal fire source combustion and smoke diffusion when a mine fire disaster occurs in a certain inclined roadway in an embodiment.
Fig. 4 is a cloud chart of the mass fraction of the smoke in a mine in a certain inclined roadway in the specific embodiment.
Fig. 5 is an internal temperature cloud chart of a certain inclined roadway in the case of a fire disaster in a mine in the specific embodiment.
Fig. 6 is an internal CO diffusion cloud image of a certain inclined roadway in the case of a fire in a mine in an embodiment.
FIG. 7 is a schematic illustration of the internal CO of a roadway in an inclined roadway in the event of a fire in a mine 2 And diffusing the cloud image.
Fig. 8 is a schematic diagram of the change along the way of the maximum internal fire wind pressure of a certain inclined roadway when a fire disaster occurs in a mine in the specific embodiment.
Fig. 9 is a schematic structural diagram of a mine fire disaster factor maximum value prediction model based on MLP in the mine fire disaster factor maximum value prediction method according to the present application in a specific embodiment.
FIG. 10 is a graph of relative error scatter between predicted and simulated values for each disaster causing factor maximum of a validation set in an embodiment.
FIG. 11 is a graph of relative error scatter between predicted and simulated values for each disaster causing factor maximum of a test set according to an embodiment.
FIG. 12 is a graph showing the comparison of predicted and simulated values of the maximum disaster causing factors of a test set according to the embodiment.
FIG. 13 is a graph showing the comparison of the predicted value of each disaster causing factor maximum value and the simulated value of the prediction set according to the embodiment.
Detailed Description
The application will be further described with reference to the drawings and detailed description.
As shown in FIG. 1, the method for predicting the maximum value of the disaster causing factor of the mine fire disaster comprises the following steps:
step 1: determining disaster factor and disaster factor influence factor of mine fire disaster
Five disaster causing factors a for determining mine fire disaster 1 、a 2 、...、a i 、...、a 5 The temperature, the CO generation amount and the CO of mine fire disaster are respectively 2 The generation amount, the smoke generation amount, the fire wind pressure or the thermal resistance, and eight influencing factors of disaster factors are determinedb 1 、b 2 、...、b k 、...、b 8 The method comprises the steps of roadway cross-sectional area, roadway length, roadway inclination angle, roadway inlet wind speed, temperature, relative humidity, atmospheric pressure and fire combustion speed; wherein a is i I is the ith disaster causing factor, i is {1,2,.. The 5}, and a is the roadway is a gallery or an inclined roadway 5 Respectively, thermal resistance and fire wind pressure, b k K e {1, 2..8 }, which is the k-th disaster causing factor influencing factor.
Step 2: establishing a numerical simulation model of mine fire disaster
Selecting an actual underground roadway as a prototype, and establishing a mine fire numerical simulation model with a fixed fire source position; the mine fire numerical simulation model is used for performing numerical simulation on mine fires under different tunnel cross-sectional areas, tunnel lengths, tunnel inclination angles, tunnel inlet wind speeds, temperatures, relative humidity, atmospheric pressure and fire burning speeds.
Step 3: acquiring a mine fire sample set
The numerical simulation method is adopted to run a mine fire numerical simulation model, data of maximum values of all disaster causing factors under N value combinations of the disaster causing factor influence factors are obtained, and a mine fire sample set C= { C is obtained 1 ,c 2 ,...,c n ,...,c N -a }; wherein c n For the nth mine fire sample, N e {1,2,.. N }, N is the total number of mine fire samples, c n ＝(b 1n ,b 2n ,...,b kn ,...,b 8n ,a 1n ,a 2n ,...,a in ,...,a 5n )，b kn A is the value of the factor influencing the k in the fire disaster sample of the nth mine in Is the maximum value of the ith disaster causing factor in the nth mine fire sample.
In this embodiment, the mine fire numerical simulation model is an underground trackless transportation fire numerical simulation model, and is built by using FDS fire power simulation software. The change of the cross section area and the length of the roadway is embodied by the actual parameters of 7 different roadways; the roadway inclination angle change interval is from a downlink wind 14 degrees to an uplink wind 14 degrees, and the step length is 2 degrees; the tunnel inlet wind speed is set to be randomly selected within the interval of 1.5m/s to 10 m/s; ring(s)The temperature range is 0-50 ℃ and the step length is 5 ℃; the relative humidity interval of the environment is 10-100%, and the step length is 10%; the atmospheric pressure is expressed in the altitude interval-1500-4000 m, the step length is 300m between the altitude interval-500-1000 m, and the step length of the rest interval is 500m; the fire burning speed is divided into three steps according to the time of the fire source reaching the maximum, namely slow fire, fast fire and ultra-fast fire in sequence, and 1,2 and 3 dimensionless numbers are respectively used for replacing the fire burning speed in the neural network. Three layers of measuring points are arranged in the roadway and are 0.4m, 2m and 3.6m away from the bottom plate, and the temperatures, CO and CO in the roadway near the bottom plate, the middle part and the top plate in the roadway are respectively monitored 2 And the real-time change condition of the flue gas, each layer of measuring points are divided into an upwind side stage, a firing stage and a downwind side stage according to the difference of the distance from an air inlet to a firing point, the density degree of the measuring points in each stage is slightly different, and the specific positions of the measuring points are slightly adjusted due to the different sectional areas and lengths of the roadways.
And selecting the 7 different roadways to perform numerical simulation to obtain big data of the fire disaster factor. Fig. 2 is a schematic diagram of a mine fire numerical simulation model of one upwind inclined roadway. The full length of the roadway is 280m, the width of the roadway is 5m, the height of the roadway is 4m, the inclination angle is 14 degrees, the left end of the roadway is provided with an air inlet, the wind speed of the roadway is 4.85m/s, and the position of a fire source is 140m away from the air inlet. The tunnel is pre-ventilated for 50s, so that the combustion of the fire source begins after the environment in the tunnel is stabilized, and the combustion and smoke diffusion conditions of the fire source in the tunnel at different moments are shown in fig. 3. Specifically, the fire source starts to burn from 50s, the initial stage of the fire is 60s, the fire progresses to the maximum at 200s and the simulation ends. Sequentially selecting 60s, 80s, 130s, 200s, 300s, 500s, 650s as observation time to spread the smoke, temperature distribution in the tunnel, CO mass fraction and CO 2 And observing the mass fraction and the smoke mass fraction. As shown in fig. 4, 5, 6 and 7, the values of the disaster factors at each moment in the roadway in the simulation process are recorded and calculated to obtain big data of the disaster factors of the fire. The schematic diagram of the along-way change of the maximum internal fire wind pressure of the inclined roadway when the mine is in fire disaster is shown in fig. 8.
In this embodiment, according to the established mine fire numerical simulation model, the factors influencing the mine fire disaster factor, such as the cross-section area of the roadway, the roadway length, the roadway inclination angle, the roadway inlet wind speed, the temperature, the relative humidity, the atmospheric pressure, the fire burning speed, and the like, are changed by using the FDS numerical simulation mode, wherein the change of the cross-section area and the length of the roadway is realized by changing and simulating a specific roadway, and sample data of the n=257 groups of mine fire disaster factor influencing factors-disaster factor maximum values are obtained.
Step 4: establishing a maximum value prediction model of mine fire disaster factor
Taking the influence factors of all disaster causing factors as input and the maximum value of all disaster causing factors as output, and constructing a mine fire disaster causing factor maximum value prediction model based on MLP; the input end of the mine fire disaster factor maximum value prediction model based on the MLP is integrated with a normalization processing module, and the output end is integrated with an inverse normalization processing module. The normalization processing is performed on the sample big data to cancel the order-of-magnitude difference among the dimension data in the training sample big data, and improve the convergence speed and the convergence precision of iterative solution in the training process.
As shown in FIG. 9, the maximum temperature, the maximum CO generation amount and the CO reached by the fire disaster are determined by taking the cross-sectional area, the roadway length, the roadway inclination angle, the roadway inlet wind speed, the temperature, the relative humidity, the atmospheric pressure and the fire burning speed of each sample as input nodes of a mine fire disaster factor maximum value prediction model based on the MLP 2 The maximum generation amount, the maximum generation amount of smoke, the maximum fire wind pressure or the maximum thermal resistance is used as an output node of the mine fire disaster factor maximum value prediction model based on MLP.
Step 5: training mine fire disaster factor maximum value prediction model
Step 5.1: randomly selecting N from a mine fire sample set C 1 The samples form a training set from the rest of the N-N 1 Randomly selecting N from each sample 2 The individual samples constitute a validation set;
step 5.2: training the maximum value prediction model of the mine fire disaster factor by using a training set, and training the trained mine fire disaster by using a verification setVerifying the disaster factor maximum prediction model, and calculating the relative error epsilon of the verification set 1 ；
Step 5.3: determining the relative error epsilon of the validation set 1 Whether or not it is below a preset threshold: if yes, outputting a maximum value prediction model of the mine fire disaster factor at the moment as an optimal prediction model; if not, adjusting parameters of a maximum value prediction model of the mine fire disaster factor, and returning to the step 5.1.
In the step 5.3, parameters of a mine fire disaster factor maximum value prediction model are adjusted, and the method specifically comprises the following steps: adjusting the number N of randomly selected samples in the training set 1 The method comprises the steps of a weight optimizer, an activation function, regularization item parameters, an implicit layer number, an implicit layer neuron number and a maximum iteration number. Wherein when the maximum number of iterations is reached, if the relative error ε 1 And if the number of iterations is not lower than the preset threshold value, the maximum number of iterations is increased.
In this embodiment, in the process of constructing the mine fire disaster factor maximum value prediction model, a relative error scatter diagram of each disaster factor maximum value prediction value and a simulation value of the verification set of the obtained optimal MLP prediction model is shown in fig. 10. The parameters corresponding to the optimal MLP prediction model are respectively as follows: the weight optimizer is an optimizer of a quasi-Newton method, the activation function is a linear activation function, the regularization term parameter is 1e-6, the hidden layer number is 2, the hidden layer neuron number is 12, and the maximum iteration number is 25000. Training samples: verification sample: the number of test samples was 249:4:4.
step 6: testing the optimal prediction model
Step 6.1: selecting the rest N-N in the mine fire disaster sample set C 1 -N 2 The samples form a test set;
step 6.2: testing the optimal prediction model by using a test set, and calculating the relative error epsilon of the test set 2 ；
Step 6.3: judging the relative error epsilon of the test set 2 Whether or not it is below a preset threshold: if yes, the optimal prediction model meets the accuracy requirement, and step 7 is carried out; if not, adjusting the parameters of the optimal prediction model, and returning to the step5.1. The adjustment method of the parameters of the optimal prediction model is consistent with that in step 5.3.
In this example, 4 test samples were sample (1) (15.83, 1400, -8,2.305, 45, 81.9, 86223.26,2), sample (2) (17.32, 280, 14,1.265, 20, 90.4, 93247.94,3), sample (3) (17.32, 280,8,4.85, 20, 90.4, 93247.94,3), sample (4) (15.83, 1400, -8,1.8068, 20.2, 81.9, 86223.26,2), and data in the samples were roadway cross-section, roadway length, roadway inclination (positive for upwind and negative for downwind), roadway entry wind speed, temperature, relative humidity, atmospheric pressure, and fire burning rate. The test results of the test set are shown in table 1. The predicted value of the maximum value of the mine fire disaster factor outputted by the 4 groups of test sample big data through the optimal MLP prediction model and the simulated value of the maximum value of the mine fire disaster factor obtained by the 4 groups of test sample big data through FDS software simulation are given in the table 1. As can be seen from Table 1, the maximum relative error was-26.57%, the average relative error of each sample was (13.42%, 6.02%,7.23%, 6.3%), the average relative error of the five-term disaster factor maximum predictions was (2.34%, 3.65%,17.77%,8.647%, 8.81%), and the total average relative error was 8.24%. The relative error scatter diagram of the maximum predicted value and the simulated value of each disaster causing factor in the test set is shown in fig. 11, and the comparison between the predicted value and the simulated value is shown in fig. 12.
TABLE 1
Step 7: predicting the maximum value of disaster causing factors of mine to be predicted
Step 7.1: acquiring data of each disaster causing factor influence factor of a mine to be predicted;
step 7.2: inputting the data of the disaster causing factor influence factors of the mine to be predicted into the optimal prediction model, and outputting the predicted value of the maximum value of each disaster causing factor of the mine to be predicted.
In this embodiment, the optimal MLP prediction model verified by the verification set is used for predicting the maximum value of the mine fire disaster factor, and 3 prediction samples (17.32, 280, 14,4.85, 15, 90.4, 93247.94,4;17.32, 280, 14,4.85, 20, 100, 93247.94,3;17.32, 280, 14,4.85, 50, 90.4, 93247.94,3) are obtained to form the prediction set. A graph of the predicted value of each disaster causing factor maximum in the predicted set versus the simulated value is shown in FIG. 13. From the prediction result of the prediction set, the prediction value output by the optimal MLP prediction model is very close to the simulation value obtained by the FDS simulation model, the total average relative error is 9.76%, and the maximum value of the disaster causing factors related to mine fire disaster can be well predicted, so that the requirement of emergency rescue is met.
Therefore, the application can obtain different ventilation parameters and the highest temperature, CO and CO which can be reached when mine fire occurs under the roadway condition through FDS software by utilizing a numerical simulation method 2 And sample big data of maximum generation amount of smoke and maximum fire wind pressure or maximum thermal resistance and other disaster causing factors, and taking eight factors affecting the disaster causing factor value as input nodes of the MLP, taking the maximum of five disaster causing factors such as maximum temperature and maximum generation amount of toxic and harmful gas as output nodes, establishing a rapid prediction model of mine fire disaster causing factors, solving the problem of large consumption in processes such as numerical simulation modeling, calculation and data analysis processing, and the like, and under the condition of given related disaster causing factor affecting factors of assumed fire, the relative error of the maximum prediction of the five disaster causing factors can be controlled within 10%, thereby realizing rapid prediction of the maximum disaster causing factors when mine fire occurs in emergency rescue, and meeting the needs of emergency decision.
It should be apparent that the above-described embodiments are merely some, but not all, embodiments of the present application. The above examples are only for explaining the present application and do not limit the scope of the present application. Based on the above embodiments, all other embodiments obtained by those skilled in the art without making creative efforts, i.e., all modifications, equivalents, improvements etc., which are within the spirit and principles of the present application, fall within the protection scope of the present application as claimed.
Claims (3)
1. A prediction method for the maximum value of a disaster-causing factor of a mine fire disaster is characterized by comprising the following steps: comprising the following steps:
step 1: determining disaster factor and disaster factor influence factor of mine fire disaster
Five disaster causing factors a for determining mine fire disaster 1 、a 2 、...、a i 、...、a 5 The temperature, the CO generation amount and the CO of mine fire disaster are respectively 2 The generation amount, the smoke generation amount, the fire wind pressure or the thermal resistance, and eight influencing factors b of disaster factors are determined 1 、b 2 、...、b k 、...、b 8 The method comprises the steps of roadway cross-sectional area, roadway length, roadway inclination angle, roadway inlet wind speed, temperature, relative humidity, atmospheric pressure and fire combustion speed; wherein a is i I is the ith disaster causing factor, i is {1,2,.. The 5}, and a is the roadway is a gallery or an inclined roadway 5 Respectively, thermal resistance and fire wind pressure, b k K is {1,2,., 8};
step 2: establishing a numerical simulation model of mine fire disaster
Selecting an actual underground roadway as a prototype, and establishing a mine fire numerical simulation model with a fixed fire source position; the mine fire numerical simulation model is used for performing numerical simulation on mine fires under different tunnel cross-sectional areas, tunnel lengths, tunnel inclination angles, tunnel inlet wind speeds, temperatures, relative humidity, atmospheric pressure and fire burning speeds;
step 3: acquiring a mine fire sample set
The numerical simulation method is adopted to run a mine fire numerical simulation model, data of maximum values of all disaster causing factors under N value combinations of the disaster causing factor influence factors are obtained, and a mine fire sample set C= { C is obtained 1 ,c 2 ,...,c n ,...,c N -a }; wherein c n For the nth mine fire sample, N e {1,2,.. N }, N is the total number of mine fire samples, c n ＝(b 1n ,b 2n ,...,b kn ,...,b 8n ,a 1n ,a 2n ,...,a in ,...,a 5n )，b kn A is the value of the factor influencing the k in the fire disaster sample of the nth mine in The maximum value of the ith disaster causing factor in the nth mine fire sample;
step 4: establishing a maximum value prediction model of mine fire disaster factor
Taking the influence factors of all disaster causing factors as input and the maximum value of all disaster causing factors as output, and constructing a mine fire disaster causing factor maximum value prediction model based on MLP; the input end of the mine fire disaster factor maximum value prediction model based on the MLP is integrated with a normalization processing module, and the output end is integrated with an inverse normalization processing module;
step 5: training mine fire disaster factor maximum value prediction model
Step 5.1: randomly selecting N from a mine fire sample set C 1 The samples form a training set from the rest of the N-N 1 Randomly selecting N from each sample 2 The individual samples constitute a validation set;
step 5.2: training the maximum value prediction model of the mine fire disaster factor by using a training set, verifying the trained maximum value prediction model of the mine fire disaster factor by using a verification set, and calculating the relative error epsilon of the verification set 1 ；
Step 5.3: determining the relative error epsilon of the validation set 1 Whether or not it is below a preset threshold: if yes, outputting a maximum value prediction model of the mine fire disaster factor at the moment as an optimal prediction model; if not, adjusting parameters of a maximum value prediction model of the mine fire disaster factor, and returning to the step 5.1;
step 6: testing the optimal prediction model
Step 6.1: selecting the rest N-N in the mine fire disaster sample set C 1 -N 2 The samples form a test set;
step 6.2: testing the optimal prediction model by using a test set, and calculating the relative error epsilon of the test set 2 ；
Step 6.3: judgment testRelative error epsilon of the set 2 Whether or not it is below a preset threshold: if yes, the optimal prediction model meets the accuracy requirement, and step 7 is carried out; if not, adjusting the parameters of the optimal prediction model, and returning to the step 5.1;
step 7: predicting the maximum value of disaster causing factors of mine to be predicted
Step 7.1: acquiring data of each disaster causing factor influence factor of a mine to be predicted;
step 7.2: inputting the data of the disaster causing factor influence factors of the mine to be predicted into the optimal prediction model, and outputting the predicted value of the maximum value of each disaster causing factor of the mine to be predicted.
2. The method for predicting the maximum value of a disaster-causing factor of a mine fire according to claim 1, wherein the mine fire numerical simulation model is a downhole trackless transportation fire numerical simulation model and is established by FDS fire power simulation software.
3. The method for predicting the maximum value of a mine fire disaster factor according to claim 1, wherein in the step 5.3, parameters of a mine fire disaster factor maximum value prediction model are adjusted, and the method specifically comprises: adjusting the number N of randomly selected samples in the training set 1 The method comprises the steps of a weight optimizer, an activation function, regularization item parameters, an implicit layer number, an implicit layer neuron number and a maximum iteration number.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110108183.9A CN112765891B (en) | 2021-01-27 | 2021-01-27 | Prediction method for maximum value of disaster-causing factors of mine fire disaster |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110108183.9A CN112765891B (en) | 2021-01-27 | 2021-01-27 | Prediction method for maximum value of disaster-causing factors of mine fire disaster |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112765891A CN112765891A (en) | 2021-05-07 |
| CN112765891B true CN112765891B (en) | 2023-11-10 |
Family
ID=75706012
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110108183.9A Active CN112765891B (en) | 2021-01-27 | 2021-01-27 | Prediction method for maximum value of disaster-causing factors of mine fire disaster |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112765891B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101280389B1 (en) * | 2011-12-28 | 2013-07-01 | 한국지질자원연구원 | System and method for predicting ground disaster occurrence prediction |
| CN107729671A (en) * | 2017-10-25 | 2018-02-23 | 西安科技大学 | A kind of mine laneway coefficient of frictional resistance Forecasting Methodology based on SVMs |
| WO2018075995A1 (en) * | 2016-10-21 | 2018-04-26 | DataRobot, Inc. | Systems for predictive data analytics, and related methods and apparatus |
| CN109356652A (en) * | 2018-10-12 | 2019-02-19 | 深圳市翌日科技有限公司 | Adaptive fire grading forewarning system method and system under a kind of mine |
| CN111222238A (en) * | 2020-01-03 | 2020-06-02 | 辽宁工程技术大学 | Rapid prediction method for gas explosion shock wave propagation state for emergency rescue |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111489006A (en) * | 2019-01-29 | 2020-08-04 | 深圳富桂精密工业有限公司 | Fire development situation prediction method and device and computer-readable storage medium |
-
2021
- 2021-01-27 CN CN202110108183.9A patent/CN112765891B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101280389B1 (en) * | 2011-12-28 | 2013-07-01 | 한국지질자원연구원 | System and method for predicting ground disaster occurrence prediction |
| WO2018075995A1 (en) * | 2016-10-21 | 2018-04-26 | DataRobot, Inc. | Systems for predictive data analytics, and related methods and apparatus |
| CN107729671A (en) * | 2017-10-25 | 2018-02-23 | 西安科技大学 | A kind of mine laneway coefficient of frictional resistance Forecasting Methodology based on SVMs |
| CN109356652A (en) * | 2018-10-12 | 2019-02-19 | 深圳市翌日科技有限公司 | Adaptive fire grading forewarning system method and system under a kind of mine |
| CN111222238A (en) * | 2020-01-03 | 2020-06-02 | 辽宁工程技术大学 | Rapid prediction method for gas explosion shock wave propagation state for emergency rescue |
Non-Patent Citations (2)
| Title |
|---|
| 基于数据挖掘技术的火灾风险预警模型研究;何晨炜;张云明;;消防科学与技术(第12期);全文 * |
| 应急状态下矿井瓦斯爆炸致灾因子传播快速分类器研究;刘剑;曲敏;黄德;高科;邓立军;刘学;;中国安全生产科学技术(第08期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112765891A (en) | 2021-05-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wu et al. | Smart detection of fire source in tunnel based on the numerical database and artificial intelligence | |
| Liu et al. | Machine learning assisted prediction of exhaust gas temperature of a heavy-duty natural gas spark ignition engine | |
| Bastani et al. | Contaminant source identification within a building: toward design of immune buildings | |
| Liu et al. | An adaptive particle swarm optimization algorithm for fire source identification of the utility tunnel fire | |
| CN113361192A (en) | Boiler internal heating surface wall temperature safety monitoring and evaluating system | |
| Yu et al. | Asymmetric flow effect in a horizontal natural ventilated tunnel with different aspect ratios under the influence of longitudinal fire locations | |
| CN114021501A (en) | Fire temperature field reconstruction method, system, computer equipment, medium and terminal | |
| CN113094935B (en) | Method for predicting highest temperature of inner top plate of V-shaped slope tunnel and application | |
| CN110032747A (en) | A kind of coal-burning boiler NOxDischarge On-line Measuring Method | |
| CN112765891B (en) | Prediction method for maximum value of disaster-causing factors of mine fire disaster | |
| CN106250930A (en) | A kind of fire locating method based on flue gas concentration | |
| Karki et al. | CFD model for jet fan ventilation systems | |
| CN113962128A (en) | Prediction method of residual bending resistance bearing capacity of RC beam considering high-temperature bursting of concrete | |
| CN114019110A (en) | Work place gas detector end cloud integration platform based on big data | |
| Liu et al. | Prediction method of coal dust explosion flame propagation characteristics based on principal component analysis and BP neural network | |
| Johansson | Evaluation of a zone model for fire safety engineering in large spaces | |
| Wang et al. | A joint hydrogen and syngas chemical kinetic model optimized by particle swarm optimization | |
| Liu et al. | Smart prediction for tunnel fire state evolution based on an improved fire simulation curve through particle swarm optimization algorithm | |
| Wang et al. | Improvement of optimization methods in indoor time-variant source parameters estimation combining unsteady adjoint equations and flow field information | |
| Yang et al. | Prediction of Shanghai air quality index based on BP neural network optimized by genetic algorithm | |
| CN115526123A (en) | Tunnel fire smoke ceiling jet flow development forecasting method and system based on data assimilation | |
| CN1529291A (en) | Forest fire intelligent moritoring and alarming device | |
| Castle | Numerical modeling of laboratory-scale surface-to-crown fire transition | |
| Valger | Estimation of pollutant dispersion around a building within non-isothermal boundary-layer using detached eddy simulation | |
| CN116629141B (en) | Goaf natural ignition prediction method and system based on PSO-BP algorithm under coal mining condition |
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 |