Prediction method and system of spontaneous combustion "three zones" distribution in goaf based on WOA-BP

Abstract

The present application relates to the field of mining technology, and provides a method and system for predicting the distribution of "three zones" of spontaneous combustion in goaf based on WOA-BP. In the method, based on the multi-field coupling model of spontaneous combustion in goaf, the distribution cloud map of oxygen concentration field and solid temperature field in goaf is obtained; based on the distribution cloud map of oxygen concentration field and solid temperature field, a simulation data set is constructed to train the distribution prediction model of "three zones" of spontaneous combustion in goaf under pre-constructed coal mining conditions; based on the trained distribution prediction model of "three zones" of spontaneous combustion, the oxygen concentration of the boundaries on both sides of goaf inlet and return air at different preset distances from the coal mining working face is predicted according to the measured multiple groups of boundary measured temperatures at different preset distances from the coal mining working face on both sides of the goaf inlet and return air, as well as the working parameters of the corresponding coal mining working face and the distance of the measuring point from the corresponding coal mining working face. In this way, the oxygen concentration of the boundaries on both sides of the inlet and return air is predicted in real time to divide the "three zones" of spontaneous combustion in goaf.

Description

Technical Field

[0001] The present application relates to the field of mining technology, and in particular to a method and system for predicting the distribution of "three zones" of spontaneous combustion in goaf based on WOA-BP. Background Art

[0002] Under coal mining conditions, the destructiveness of spontaneous combustion accidents in coal mines is far greater than that of external fires. This is because spontaneous combustion accidents are not easy to detect. After the fire is formed, it will cause the loss of coal resources and threaten the safety of workers, causing production interruptions. Especially in goaf areas, spontaneous combustion fires frequently occur with the oxidation reaction between the recovered coal and the air leaking in. Because the area is large and difficult to detect, it brings hidden dangers to coal production safety. Clarifying the internal distribution of goaf areas and proposing reasonable fire prevention and extinguishing measures are mainly reflected in the specific division of the "three zones" of spontaneous combustion in goaf areas, and being able to analyze the effects achieved after taking measures.

[0003] Oxygen concentration is an important indicator for dividing the "three zones" of spontaneous combustion in goaf under coal mining conditions. Direct pipe laying tests require a lot of manpower and time, and the test results have certain deviations. Understanding the distribution range of the "three zones" of spontaneous combustion in goaf is not only the main basis for prevention and control measures such as nitrogen injection and air leakage plugging in goaf, but also an important reference for determining the reasonable advancement speed of working face recovery. Therefore, the testing and division of the "three zones" of spontaneous combustion in goaf are of great significance in fire prevention and extinguishing work. Summary of the invention

[0004] The purpose of this application is to provide a method and system for predicting the "three zones" distribution of spontaneous combustion in goaf based on WOA-BP, so as to solve or alleviate the problems existing in the above-mentioned prior art.

[0005] In order to achieve the above objectives, this application provides the following technical solutions:

[0006] The present application provides a method for predicting the "three zones" distribution of spontaneous combustion in goaf based on WOA-BP, comprising: step S101, based on the multi-field coupling model of natural combustion in goaf, obtaining the distribution cloud map of the oxygen concentration field and the solid temperature field in the goaf; step S102, based on the distribution cloud map of the oxygen concentration field and the solid temperature field in the goaf, constructing a simulation data set to train the pre-constructed "three zones" distribution prediction model of spontaneous combustion in goaf under coal mining conditions; wherein the simulation data set includes: the simulated working parameters of the coal mining face, the simulated distance of the measuring point from the coal mining face, and the boundary simulated temperature on both sides of the goaf inlet and return air and the corresponding simulated oxygen concentration; step S103, based on the trained spontaneous combustion "three zones" distribution prediction model, according to the measured multiple groups of boundary measured temperatures at different preset distances from the coal mining face on both sides of the goaf inlet and return air, and the corresponding working parameters of the coal mining face, and the distance of the measuring point from the coal mining face, predict the oxygen concentration at the boundaries on both sides of the goaf inlet and return air.

[0007] Preferably, in step S101, based on the multi-field coupling model of spontaneous combustion in the goaf, according to the working parameters of the coal mining working face and the coal sample parameters, the distribution cloud diagrams of the oxygen concentration field and the solid temperature field in the goaf are determined.

[0008] Preferably, in step S102, based on the WOA whale optimization algorithm, the initial network weights of the spontaneous combustion "three-band" distribution prediction model based on the BP neural network architecture are determined.

[0009] Preferably, the WOA whale optimization algorithm is used to determine the initial network weights of the spontaneous combustion "three-band" distribution prediction model based on the BP neural network architecture, including: based on the WOA whale optimization algorithm, according to the prediction results of the spontaneous combustion "three-band" distribution prediction model and the mean square error of the simulated oxygen concentration corresponding to the prediction results in the simulation data set, to determine the feature weights of the input features in the simulation data set at each iteration of the spontaneous combustion "three-band" distribution prediction model; according to the feature weights of the input features in the simulation data set at each iteration of the spontaneous combustion "three-band" distribution prediction model and their corresponding fitness, to determine the initial network weights of the spontaneous combustion "three-band" distribution prediction model.

[0010] Preferably, according to the formula:

[0011]

[0012] Determine the feature weight p(i) of the input feature in the simulation data set at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model;

[0013] Among them, X i(t|G) is the feature weight of the input feature evolved to the tth time when the spontaneous combustion "three-band" distribution prediction model is iterated for the i-th time; f[X i (t|G)] is the feature weight X i (t|G); t∈G, G is the total number of evolutions of the input features in each iteration of the self-ignition "three-band" distribution prediction model; t and G are both positive integers; n is the number of training samples in the simulation data set, and each group of training samples includes the corresponding input features and the oxygen simulation concentration, y k is the simulated oxygen concentration in the kth group of training samples in the simulated data set; It is the prediction result obtained by the spontaneous combustion "three-band" distribution prediction model based on the input features in the kth group of training samples; 1≤k≤n, k and n are both positive integers.

[0014] Preferably, the evolution random number p in response to the input feature is less than 0.5 and According to the formula:

[0015]

[0016] Determine the optimal feature weight of the input feature in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model

[0017] Among them, X i (t|G) is the feature weight of the input feature evolved to the tth time when the spontaneous combustion "three-band" distribution prediction model is iterated for the i-th time; X i (t+1|G) is the feature weight of the input feature evolved to the t+1th time when the spontaneous combustion "three-band" distribution prediction model is iterated for the i-th time; X i (m|t) is the mth feature weight of the input feature in the tth evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model; f[X i (m|t)] is the feature weight X i (m|t) is the fitness; t is the current evolution times of the input feature; G is the total evolution times of the input feature in each iteration of the self-ignition "three-band" distribution prediction model; r 1 and r 2 Represents two random numbers between 0 and 1;

[0018] In response to the evolution random number p corresponding to the input feature < 0.5 and |2×2-2×tG·r1≥1, according to the formula:

[0019]

[0020] Determine the optimal feature weight of the input feature in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model

[0021] In response to the evolution random number p≥0.5 corresponding to the input feature, according to the formula:

[0022]

[0023] Determine the optimal feature weight of the input feature in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model Where b is a constant, r 3 A random number between 0 and 1.

[0024] Preferably, according to the formula:

[0025]

[0026] Determine the initial network weight P of the spontaneous combustion "three-band" distribution prediction model g (pop);

[0027] Wherein, p(i) is the feature weight of the input feature in the simulated data set at the i-th iteration of the spontaneous combustion “three-band” distribution prediction model; pop is the total number of iterations of the spontaneous combustion “three-band” distribution prediction model, and i and pop are both positive integers; G is the total number of evolutions of the input feature at each iteration of the spontaneous combustion “three-band” distribution prediction model; f[p(i)] is the fitness of the feature weight p(i) of the input feature in the simulated data set at the i-th iteration of the spontaneous combustion “three-band” distribution prediction model; X i (t|G) is the feature weight of the input feature evolved to the tth time when the spontaneous combustion "three-band" distribution prediction model is iterated for the i-th time; f[X i (t|G)] is the feature weight X i (t|G) corresponds to the fitness.

[0028] Preferably, in step S103, distributed optical fiber is used to collect boundary temperatures at different preset distances from the coal mining face on both sides of the goaf air inlet and return air at preset intervals to obtain multiple groups of measured boundary temperatures.

[0029] The embodiment of the present application provides a "three-zone" distribution prediction system for spontaneous combustion in goaf based on WOA-BP under coal mining conditions, including: a simulation unit, configured to obtain a distribution cloud map of the oxygen concentration field and the solid temperature field of the goaf based on a multi-field coupling model of natural combustion in the goaf; a training set unit, configured to construct a simulation data set based on the distribution cloud map of the oxygen concentration field and the solid temperature field of the goaf, so as to train a pre-constructed "three-zone" distribution prediction model for spontaneous combustion in goaf under coal mining conditions; wherein the simulation data set includes: simulated working parameters of the coal mining face, simulated distance of the measuring point from the coal mining face, and simulated boundary temperatures on both sides of the goaf inlet and return air and corresponding simulated oxygen concentrations; a concentration prediction unit, configured to predict the oxygen concentrations on both sides of the goaf inlet and return air based on the trained spontaneous combustion "three-zone" distribution prediction model according to multiple groups of boundary measured temperatures at different preset distances from the coal mining face on both sides of the goaf inlet and return air, as well as the corresponding working parameters of the coal mining face and the distance of the measuring point from the coal mining face.

[0030] Beneficial effects:

[0031] The WOA-BP-based goaf spontaneous combustion "three-zone" distribution prediction method provided in the embodiment of the present application first obtains the distribution cloud map of the oxygen concentration field and the solid temperature field of the goaf based on the multi-field coupling model of natural combustion in the goaf; then, based on the distribution cloud map of the oxygen concentration field and the solid temperature field in the goaf, a simulation data set is constructed to train the pre-constructed goaf spontaneous combustion "three-zone" distribution prediction model under coal mining conditions; wherein the simulation data set includes: the simulated working parameters of the coal mining face, the simulated distance of the measuring point from the coal mining face, and the boundary simulated temperature on both sides of the goaf inlet and return air and the corresponding simulated oxygen concentration; finally, based on the trained spontaneous combustion "three-zone" distribution prediction model, the oxygen concentration on both sides of the goaf inlet and return air boundaries is predicted according to multiple groups of boundary measured temperatures at different preset distances from the coal mining face on both sides of the goaf inlet and return air, as well as the corresponding working parameters of the coal mining face and the distance of the measuring point from the corresponding coal mining face. Therefore, a mapping relationship between the boundary temperature and oxygen concentration on both sides of the goaf inlet and return air was established, which effectively improved the prediction accuracy and precision. The oxygen concentration on both sides of the goaf inlet and return air boundaries was predicted in real time to divide the goaf into the "three zones" of spontaneous combustion. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The drawings constituting part of the present application are used to provide a further understanding of the present application. The exemplary embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. Among them:

[0033] Figure 1A flow chart of a method for predicting the distribution of "three zones" of spontaneous combustion in goaf based on WOA-BP according to some embodiments of the present application;

[0034] Figure 2 A distribution cloud diagram of oxygen concentration field in a goaf provided according to some embodiments of the present application;

[0035] Figure 3 A distribution cloud diagram of the solid temperature field in the goaf provided according to some embodiments of the present application;

[0036] Figure 4 A neural network topology diagram of a spontaneous combustion "three-band" distribution prediction model provided according to some embodiments of the present application;

[0037] Figure 5 A logic block diagram of a spontaneous combustion "three-band" distribution prediction model provided according to some embodiments of the present application;

[0038] Figure 6 A schematic diagram of the arrangement of distributed optical fibers according to some embodiments of the present application;

[0039] Figure 7 A schematic diagram of the division of the "three zones" of spontaneous combustion in goaf areas provided according to some embodiments of the present application;

[0040] Figure 8 A schematic diagram of the structure of a "three-zone" distribution prediction system for spontaneous combustion in goaf based on WOA-BP according to some embodiments of the present application. DETAILED DESCRIPTION

[0041] The present application will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments. Each example is provided by way of explanation of the present application and does not limit the present application. In fact, it will be clear to those skilled in the art that modifications and variations may be made in the present application without departing from the scope or spirit of the present application. For example, a feature shown or described as a part of an embodiment may be used in another embodiment to produce yet another embodiment. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the embodiments of the present invention should belong to the scope of protection of the embodiments of the present invention.

[0042] Under the existing coal mining conditions, the simulation and construction of the physical field model of the goaf are limited. The physical field model of the goaf is mostly determined by considering the flow field velocity, ignoring the factor of the boundary temperature on both sides of the goaf inlet and return air, and lacks the relationship between the boundary temperature on both sides of the goaf inlet and return air and the oxygen concentration, which affects the accuracy of the prediction results of the physical field model of the goaf.

[0043] Based on this, Figures 1 to 7As shown, the embodiment of the present application provides a method for predicting the distribution of "three zones" of spontaneous combustion in goaf based on WOA-BP, including:

[0044] Step S101, based on the multi-field coupling model of spontaneous combustion in the goaf, obtain the distribution cloud map of the oxygen concentration field and the solid temperature field in the goaf.

[0045] In this application, based on the constructed multi-field coupling model of spontaneous combustion in goaf under coal mining conditions, the working parameters of the coal mining face (advance speed, thickness of residual coal, ventilation resistance) and coal sample parameters (oxygen consumption rate, heat release intensity) are comprehensively considered to simulate the oxygen concentration field and solid temperature field in the goaf, and a nonlinear mapping of the oxygen concentration field and solid temperature field on both sides of the goaf's inlet and return air boundaries is established.

[0046] Specifically, based on the multi-field coupling model of spontaneous combustion in the goaf, the distribution cloud map of the oxygen concentration field and the solid temperature field in the goaf is determined according to the working parameters of the coal mining face and the coal sample parameters. In other words, the simulation parameters (simulated working parameters, simulated coal sample parameters) are input into the multi-field coupling model of spontaneous combustion, and the distribution simulation cloud map of the oxygen concentration field and the solid temperature field on both sides of the goaf's inlet and return air boundaries is obtained through numerical simulation, and the nonlinear mapping between the temperature and oxygen concentration on both sides of the goaf's inlet and return air boundaries is characterized by the simulation cloud map.

[0047] Among them, the multi-field coupling model of spontaneous combustion in goaf under coal mining conditions is:

[0048]

[0049] Where K is the permeability coefficient of the porous medium, g is the gravitational acceleration, P is the sum of the static pressure and velocity pressure of the gas in the goaf; ρ g is the gas density in the goaf, h is the elevation with the preset horizontal plane as the reference plane, S is the area of ​​any microelement on the surface of the control body; D is the surface area of ​​the control body; is the unit normal vector on the boundary surface of the control volume; the control volume represents an arbitrary tetrahedron obtained by spatial discretization in the goaf.

[0050] n is the porosity of floating coal in the goaf, v is the seepage velocity at the micro-element area S, is the diffusion coefficient constant of oxygen in the goaf, is the molar concentration of oxygen in the goaf, u(t) is the oxygen consumption per unit volume of coal per unit time, and V is the volume of the control body.

[0051] λ g is the thermal conductivity of gas in the goaf, T g is the gas temperature in the goaf, K e is the convection heat transfer coefficient between coal rock in goaf and gas in goaf, S eis the surface area of ​​heat exchange between coal rock and gas in the goaf, T s is the coal rock temperature in the goaf, t g is the temperature distribution function of the boundary.

[0052] λ s is the thermal conductivity of the coal and rock in the goaf, q(t) is the heat released by the coal in the control body per unit time t, v 0 is the advancing speed of the coal mining face, ρ s is the density of coal rock, C s is the specific heat capacity of coal rock, and α is the inclination angle of the coal seam.

[0053] Step S102: construct a simulation data set based on the distribution cloud map of the oxygen concentration field and the solid temperature field in the goaf to train the pre-constructed "three-zone" distribution prediction model of spontaneous combustion in the goaf under coal mining conditions.

[0054] Through numerical simulation, the distribution simulation cloud map of the oxygen concentration field and the solid temperature field on both sides of the goaf air inlet and return air boundaries is established. In this process, the simulation parameters input into the multi-field coupling model of the spontaneous combustion in the goaf are: simulation working parameters, simulation coal sample parameters, and the output of the multi-field coupling model of the spontaneous combustion is the simulated oxygen concentration and the corresponding simulated temperature on both sides of the goaf air inlet and return air boundaries. In this application, the corresponding input simulation parameters and output simulation results in the distribution simulation cloud map of the oxygen concentration field and the solid temperature field are used as sample data to construct a simulation data set. In other words, the simulation data set includes: the simulated working parameters of the coal mining face (advance speed (simulation), residual coal thickness (simulation), ventilation resistance (simulation)), the simulated distance of the measuring point from the coal mining face, and the simulated temperature of the boundaries on both sides of the goaf air inlet and return air and the corresponding simulated oxygen concentration.

[0055] Here, it should be noted that the measuring point refers to the measurement position corresponding to the boundary simulation temperature and oxygen simulation concentration on both sides of the goaf inlet and return air. In the actual coal mining process, it is the buried position of the sensor probe. As the coal mining face advances, the distance between the measuring point and the coal mining face continues to change, and the corresponding collected temperature and oxygen concentration also change accordingly.

[0056] In this application, a pre-constructed prediction model of the "three zones" distribution of spontaneous combustion in goaf under coal mining conditions is trained by simulating a data set. Specifically, multiple sets of data sets with one-to-one correspondence between input and output are normalized. According to the formula:

[0057]

[0058] Normalize multiple sets of input-output corresponding data sets to generate simulated data sets. In the formula, X′ is the normalized value of the original data to be normalized, X is the original value of the data to be normalized, and Xmin is the minimum value of the original data to be normalized; X maz is the maximum value of the original data to be normalized.

[0059] In this application, based on the WOA whale optimization algorithm, the initial network weights of the spontaneous combustion "three-band" distribution prediction model based on the BP neural network architecture are determined to improve the prediction accuracy and generalization ability of the spontaneous combustion "three-band" distribution prediction model. Specifically, based on the WOA whale optimization algorithm, the feature weights of the input features in the simulated data set at each iteration of the spontaneous combustion "three-band" distribution prediction model are determined according to the mean square error between the prediction results of the spontaneous combustion "three-band" distribution prediction model and the simulated oxygen concentration corresponding to the prediction results in the simulated data set.

[0060] In this application, according to the formula:

[0061]

[0062] Determine the feature weight p(i) of the input feature in the simulated data set at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model. i (t|G) is the feature weight of the input feature evolution to the tth time in the i-th iteration of the spontaneous combustion “three-band” distribution prediction model; f[X i (t|G)] is the feature weight X i (t|G); t∈G, G is the total number of evolutions of the input features in each iteration of the self-ignition "three-band" distribution prediction model; t and G are both positive integers; n is the number of training samples in the simulated data set, each group of training samples includes the corresponding input features and oxygen simulation concentration, yk is the oxygen simulation concentration in the kth group of training samples in the simulated data set; is the prediction result of the spontaneous combustion “three-band” distribution prediction model based on the input features in the kth group of training samples; 1≤k≤n, k and n are all positive integers.

[0063] The feature weight p(i) of the input feature represents the optimal feature weight corresponding to the G-time evolution of the input feature in the i-th iteration of the spontaneous combustion "three-band" distribution prediction model. In this application, the optimal feature weight of the input feature in the t-time evolution of the spontaneous combustion "three-band" distribution prediction model is defined as The optimal feature weight is It is continuously updated as the input features evolve until the optimal feature weight obtained by the input feature evolution G times is That is the feature weight p(i).

[0064] In the process of optimizing the spontaneous combustion "three-zone" distribution prediction model through the WOA whale optimization algorithm, the surrounding prey method, the spiral attack prey method, and the random search prey method can be used to determine the optimal feature weights of the input features in the t-time evolution process of the spontaneous combustion "three-zone" distribution prediction model at the i-th iteration.

[0065] The evolution random number p of the input feature is randomly selected in the range of [0, 1], and the evolution random number p of the input feature is less than 0.5 and The optimal feature weight of the input feature in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-zone" distribution prediction model is determined by encircling the prey. Specifically, according to the formula:

[0066]

[0067] Determine the optimal feature weights of the input features in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model Among them, X i (t|G) is the feature weight of the input feature evolution to the tth time in the i-th iteration of the spontaneous combustion “three-band” distribution prediction model; X i (t+1|G) is the feature weight of the input feature evolution to the t+1th iteration of the spontaneous combustion “three-band” distribution prediction model; X i (m|t) is the feature weight of the input feature at the mth time in the tth evolution process during the i-th iteration of the spontaneous combustion “three-band” distribution prediction model; f[X i (m|t)] is the feature weight X i (m|t) is the fitness; t is the current evolution times of the input feature; G is the total evolution times of the input feature in each iteration of the spontaneous combustion "three-band" distribution prediction model; r 1 and r 2 Represents two random numbers between 0 and 1.

[0068] In response to the evolutionary random number p corresponding to the input feature < 0.5 and The optimal feature weight of the input feature in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-zone" distribution prediction model is determined by randomly searching for prey. Specifically, according to the formula:

[0069]

[0070] Determine the optimal feature weights of the input features in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model

[0071] In response to the evolutionary random number p ≥ 0.5 corresponding to the input feature, the spiral attack method is adopted to determine the optimal feature weight of the input feature in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model. Specifically, according to the formula:

[0072]

[0073] Determine the optimal feature weights of the input features in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model In the formula, b is a constant, r 3 A random number between 0 and 1.

[0074] Then, according to the feature weights of the input features in the simulated data set and their corresponding fitness at each iteration of the spontaneous combustion "three-band" distribution prediction model, the initial network weights of the spontaneous combustion "three-band" distribution prediction model are determined. Specifically, according to the fitness f[p(i)] of the optimal position of the input feature, i.e., the feature weight p(i), during each iteration of the spontaneous combustion "three-band" distribution prediction model (i.e., the optimal feature weight obtained by the evolution of the input feature G times is The fitness of the spontaneous combustion "three-zone" distribution prediction model is determined in the entire iteration process. G (pop). pop is the total number of iterations of the spontaneous combustion “three-band” distribution prediction model.

[0075] Specifically, according to the formula:

[0076]

[0077] Determine the initial network weights P of the spontaneous combustion "three-zone" distribution prediction model g (pop);

[0078] Wherein, p(i) is the feature weight of the input feature in the simulated data set at the i-th iteration of the spontaneous combustion “three-band” distribution prediction model; pop is the total number of iterations of the spontaneous combustion “three-band” distribution prediction model, i and pop are both positive integers; G is the total number of evolutions of the input feature in each iteration of the spontaneous combustion “three-band” distribution prediction model; f[p(i)] is the fitness of the feature weight p(i) of the input feature in the simulated data set at the i-th iteration of the spontaneous combustion “three-band” distribution prediction model; X i (t|G) is the feature weight of the input feature evolution to the tth time in the i-th iteration of the spontaneous combustion “three-band” distribution prediction model; f[X i (t|G)] is the feature weight X i (t|G) corresponds to the fitness.

[0079] After determining the initial network weights of the goaf spontaneous combustion "three zones" distribution prediction model, the BP neural network model is trained according to the simulated data set of the goaf spontaneous combustion "three zones" distribution under coal mining conditions. Specifically, 75% of the sample data in the simulated data set are used as training sets (training sample data), and the remaining 25% of the sample data are used as test sets (test sample data). The spontaneous combustion "three zones" distribution prediction model based on the BP neural network architecture is trained until the error accuracy of the goaf spontaneous combustion "three zones" distribution prediction model is less than the preset error threshold, or the spontaneous combustion "three zones" distribution prediction model reaches the preset maximum number of training times, and the goaf spontaneous combustion "three zones" distribution prediction model is successfully trained.

[0080] Step S103, based on the trained spontaneous combustion "three-zone" distribution prediction model, predict the oxygen concentration on both sides of the goaf inlet and return air boundaries according to multiple groups of boundary measured temperatures at different preset distances from the coal mining working face on both sides of the goaf inlet and return air, as well as the corresponding working parameters of the coal mining working face and the distance of the measuring point from the corresponding coal mining working face.

[0081] After the training of the spontaneous combustion "three zones" distribution prediction model of the goaf is completed, the oxygen concentration on both sides of the goaf's inlet and return air boundaries can be predicted through measured parameters. Specifically, the boundary temperatures at different preset distances from the coal mining face on both sides of the goaf's inlet and return air are collected using distributed optical fibers at preset intervals to obtain multiple sets of measured boundary temperatures. Combined with the advancement speed of the coal mining face, the thickness of the remaining coal, the ventilation resistance, and the distance of the measuring point from the coal mining face corresponding to the measured boundary temperatures, the trained spontaneous combustion "three zones" distribution prediction model outputs the oxygen concentration (prediction result) on both sides of the goaf's inlet and return air boundaries.

[0082] In a specific example, two temperature measuring optical cables are arranged along the air inlet and return chute of the working face, and buried in the goaf as the coal mining working face is pushed forward, the boundary temperature on both sides of the goaf is collected, and the temperature data at 0 meters, 7 meters, 15 meters, 22 meters, 30 meters, 38 meters, 46 meters, 54 meters, 63 meters, 71 meters, 80 meters, 89 meters, and 98 meters away from the coal mining working face are recorded. The measured boundary temperature on both sides of the goaf air inlet and return, the working face advancement speed, the thickness of the remaining coal, the ventilation resistance, and the distance from the measuring point to the coal mining working face are input into the goaf spontaneous combustion "three zones" distribution prediction model to realize the prediction of the oxygen concentration on both sides of the goaf air inlet and return boundaries.

[0083] Spontaneous combustion is the phenomenon that coal seams with a tendency to self-ignite are oxidized after being mined and crushed and come into contact with air at room temperature, generating heat that raises their temperature, resulting in ignition and smoke. For coal to spontaneously ignite, three conditions must be met at the same time: (1) the coal seams have a tendency to self-ignite and are piled up in a crushed state, with a pile thickness generally greater than 0.4m; (2) continuous ventilation and oxygen supply conditions; and (3) a continuous heat storage environment.

[0084] In the goaf of the working face of the mine where the coal seam has a tendency to spontaneously combust, the three necessary conditions are specifically manifested as the leakage wind speed in the goaf and the oxygen concentration in the air in the goaf. The leakage wind speed directly affects the supply and heat storage conditions of the coal, while the oxygen concentration determines the oxidation and spontaneous combustion capacity of the coal seam. Therefore, according to the leakage and oxygen concentration in the goaf, the coal oxidation and spontaneous combustion area in the goaf can be divided into: heat dissipation zone, oxidation zone and asphyxiation zone, that is, the "three zones" of spontaneous combustion in the goaf.

[0085] For the heat dissipation zone, the oxygen concentration is greater than 18%, the leakage wind speed is large, there is oxidation but no heat storage conditions, and the heat of oxidation is carried away by the leakage; for the oxidation zone, the oxygen concentration is greater than or equal to 5% and less than or equal to 18%. Due to the fall and gradual compaction of the overlying rock strata in the goaf, there is less leakage in the goaf. At this time, there are both sufficient oxygen supply conditions and a good heat storage environment, and the coal is easy to spontaneously combust; for the asphyxiation zone, the oxygen concentration is less than 5%. After the oxidation zone, the rocks that fall from the goaf are continuously compacted, the leakage is weak, and the oxygen concentration drops to the asphyxiation concentration below 5%. The coal that has spontaneously combusted in the oxidation zone is suffocated and extinguished in the asphyxiation zone due to the decrease in oxygen concentration.

[0086] Therefore, according to the working face advancement speed, coal thickness, ventilation resistance, distance of the measuring point from the working face and boundary temperature data measured during real-time mining, the oxygen concentration on both sides of the goaf's inlet and return air boundaries is predicted in real time. By predicting the oxygen concentration on both sides of the goaf's inlet and return air boundaries in real time, the goaf's spontaneous combustion "three zones" are divided, effectively improving the prediction precision and accuracy.

[0087] The embodiment of the present application also provides a prediction system for the "three zones" distribution of spontaneous combustion in goaf areas based on WOA-BP under coal mining conditions, such as Figure 8 As shown, the prediction system includes:

[0088] The simulation unit 801 is configured to obtain a distribution cloud diagram of the oxygen concentration field and the solid temperature field in the goaf based on a multi-field coupling model of spontaneous combustion in the goaf;

[0089] The training set unit 802 is configured to construct a simulation data set based on the distribution cloud map of the oxygen concentration field and the solid temperature field of the goaf, so as to train the pre-constructed "three-zone" distribution prediction model of the goaf spontaneous combustion under coal mining conditions; wherein the simulation data set includes: the simulated working parameters of the coal mining face, the simulated distance of the measuring point from the coal mining face, and the boundary simulated temperature on both sides of the goaf inlet and return air and the corresponding simulated oxygen concentration;

[0090] The concentration prediction unit 803 is configured to predict the oxygen concentration on both sides of the goaf's inlet and return air boundaries based on the trained spontaneous combustion "three-zone" distribution prediction model, according to multiple groups of boundary measured temperatures at different preset distances from the coal mining working face on both sides of the goaf's inlet and return air, as well as the corresponding working parameters of the coal mining working face and the distance of the measuring point from the corresponding coal mining working face.

[0091] The "three-zone" distribution prediction system for spontaneous combustion of goaf based on WOA-BP under coal mining conditions provided in the embodiment of the present application can implement the steps and processes of the "three-zone" distribution prediction method for spontaneous combustion of goaf based on WOA-BP in any of the above-mentioned embodiments, and achieve the same technical effects, which will not be repeated here one by one.

[0092] In the description of the present invention, the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" etc. mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0093] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for predicting the distribution of "three zones" of spontaneous combustion in goaf based on WOA-BP, characterized in that: include: Step S101, based on the multi-field coupling model of spontaneous combustion in the goaf, obtaining the distribution cloud map of the oxygen concentration field and the solid temperature field in the goaf; Step S102: construct a simulation data set based on the distribution cloud map of the oxygen concentration field and the solid temperature field of the goaf, and train the pre-constructed "three-zone" distribution prediction model of spontaneous combustion in the goaf under coal mining conditions based on the WOA whale optimization algorithm; wherein the simulation data set includes: simulated working parameters of the coal mining face, simulated distance of the measuring point from the coal mining face, and boundary simulated temperature on both sides of the goaf inlet and return air and the corresponding simulated oxygen concentration; In response to each iteration of the self-ignition "three-band" distribution prediction model, the evolutionary random number p of the input feature in the simulated data set is less than 0.5 and According to the formula: Determine the optimal feature weight of the input feature in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model Among them, X i (t|G) is the feature weight of the input feature evolved to the tth time when the spontaneous combustion "three-band" distribution prediction model is iterated for the i-th time; X i (t+1|G) is the feature weight of the input feature evolved to the t+1th time when the spontaneous combustion "three-band" distribution prediction model is iterated for the i-th time; X i (m|t) is the mth feature weight of the input feature in the tth evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model; f[X i (m|t)] is the feature weight X i (m|t) is the fitness; t is the current evolution times of the input feature; G is the total evolution times of the input feature in each iteration of the self-ignition "three-band" distribution prediction model; r1 and r2 represent two random numbers between 0 and 1; In response to the evolutionary random number corresponding to the input feature p<0.5 and |2× According to the formula: Determine the optimal feature weight of the input feature in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model In response to the evolution random number p≥0.5 corresponding to the input feature, according to the formula: Determine the optimal feature weight of the input feature in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model In the formula, b is a constant, r3 is a random number between 0 and 1; Step S103, based on the trained spontaneous combustion "three-zone" distribution prediction model, predict the oxygen concentration on both sides of the goaf inlet and return air boundaries according to multiple groups of measured boundary temperatures at different preset distances from the coal mining working face on both sides of the goaf inlet and return air, and the corresponding working parameters of the coal mining working face, and the distance of the measuring point from the coal mining working face.

2. The WOA-BP-based goaf spontaneous combustion "three zones" distribution prediction method according to claim 1 is characterized in that: In step S101, Based on the multi-field coupling model of spontaneous combustion in the goaf, and according to the working parameters and coal sample parameters of the coal mining working face, the distribution cloud diagrams of the oxygen concentration field and the solid temperature field in the goaf are determined.

3. The WOA-BP-based goaf spontaneous combustion "three zones" distribution prediction method according to claim 1 is characterized in that: In step S102, Based on the WOA whale optimization algorithm, the initial network weights of the spontaneous combustion "three-band" distribution prediction model based on the BP neural network architecture are determined.

4. The WOA-BP-based goaf spontaneous combustion "three zones" distribution prediction method according to claim 3 is characterized in that: The WOA whale optimization algorithm is used to determine the initial network weights of the spontaneous combustion "three-band" distribution prediction model based on the BP neural network architecture, including: Based on the WOA whale optimization algorithm, according to the prediction result of the spontaneous combustion "three-zone" distribution prediction model and the mean square error of the simulated oxygen concentration corresponding to the prediction result in the simulation data set, the feature weight of the input feature in the simulation data set at each iteration of the spontaneous combustion "three-zone" distribution prediction model is determined; The initial network weights of the spontaneous combustion "three-zone" distribution prediction model are determined according to the feature weights of the input features in the simulation data set and their corresponding fitness at each iteration of the spontaneous combustion "three-zone" distribution prediction model.

5. The WOA-BP-based goaf spontaneous combustion "three zones" distribution prediction method according to claim 4 is characterized in that: According to the formula: Determine the feature weight p(i) of the input feature in the simulation data set at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model; Among them, X i (t|G) is the feature weight of the input feature evolved to the tth time when the spontaneous combustion "three-band" distribution prediction model is iterated for the i-th time; f[X i (t|G)] is the feature weight X i (t|G); t∈G, G is the total number of evolutions of the input features in each iteration of the spontaneous combustion "three-band" distribution prediction model; t and G are both positive integers; n is the number of training samples in the simulated data set, and each group of training samples includes the corresponding input features and the simulated oxygen concentration, y k is the simulated oxygen concentration in the kth group of training samples in the simulated data set; is the prediction result obtained by the spontaneous combustion "three-band" distribution prediction model based on the input features in the kth group of training samples; 1≤k≤n, k, n are both positive integers.

6. The WOA-BP-based goaf spontaneous combustion "three zones" distribution prediction method according to claim 5 is characterized in that: According to the formula: Determine the initial network weight P of the spontaneous combustion "three-zone" distribution prediction model g (pop); Wherein, p(i) is the feature weight of the input feature in the simulation data set at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model; pop is the total number of iterations of the spontaneous combustion "three-band" distribution prediction model, and both i and pop are positive integers; G is the total number of evolutions of the input feature at each iteration of the spontaneous combustion "three-band" distribution prediction model; f[p(i)] is the fitness of the feature weight p(i) of the input feature in the simulation data set at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model; i (t|G) is the feature weight of the input feature evolved to the tth time when the spontaneous combustion "three-band" distribution prediction model is iterated for the i-th time; f[X i (t|G)] is the feature weight X i (t|G) corresponds to the fitness.

7. The WOA-BP-based goaf spontaneous combustion "three zones" distribution prediction method according to claim 1 is characterized in that: In step S103, Distributed optical fiber is used to collect the boundary temperatures at different preset distances from the coal mining face on both sides of the goaf air inlet and return air at preset intervals to obtain multiple groups of measured boundary temperatures.

8. A "three-zone" distribution prediction system for spontaneous combustion in goaf based on WOA-BP under coal mining conditions, characterized in that: include: The simulation unit is configured to obtain a distribution cloud diagram of the oxygen concentration field and the solid temperature field of the goaf based on a multi-field coupling model of spontaneous combustion in the goaf; The training set unit is configured to construct a simulation data set based on the distribution cloud map of the oxygen concentration field and the solid temperature field of the goaf, and train the pre-constructed "three-zone" distribution prediction model of spontaneous combustion in the goaf under coal mining conditions based on the WOA whale optimization algorithm; wherein the simulation data set includes: simulated working parameters of the coal mining face, simulated distance of the measuring point from the coal mining face, and boundary simulated temperature on both sides of the inlet and return air of the goaf and the corresponding simulated oxygen concentration; In response to each iteration of the self-ignition "three-band" distribution prediction model, the evolutionary random number p of the input feature in the simulated data set is less than 0.5 and According to the formula: Determine the optimal feature weight of the input feature in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model Among them, X i (t|G) is the feature weight of the input feature evolved to the tth time when the spontaneous combustion "three-band" distribution prediction model is iterated for the i-th time; X i (t+1|G) is the feature weight of the input feature evolved to the t+1th time when the spontaneous combustion "three-band" distribution prediction model is iterated for the i-th time; X i (m|t) is the mth feature weight of the input feature in the tth evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model; f[X i (m|t)] is the feature weight X i (m|t) is the fitness; t is the current evolution times of the input feature; G is the total evolution times of the input feature in each iteration of the self-ignition "three-band" distribution prediction model; r1 and r2 represent two random numbers between 0 and 1; In response to the evolutionary random number p corresponding to the input feature < 0.5 and According to the formula: Determine the optimal feature weight of the input feature in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model In response to the evolution random number p≥0.5 corresponding to the input feature, according to the formula: Determine the optimal feature weight of the input feature in the t-time evolution process at the i-th iteration of the spontaneous combustion "three-band" distribution prediction model In the formula, b is a constant, r3 is a random number between 0 and 1; The concentration prediction unit is configured to predict the oxygen concentration on both sides of the goaf's inlet and return air boundaries based on the trained spontaneous combustion "three-zone" distribution prediction model, according to multiple groups of measured boundary temperatures at different preset distances from the coal mining working face on both sides of the goaf's inlet and return air, as well as the corresponding working parameters of the coal mining working face and the distance of the measuring point from the coal mining working face.

Images