Mine water inrush disaster early warning system based on acousto-optic fusion perception

Abstract

The application discloses a mine water inrush disaster early warning system based on sound-light fusion perception and belongs to the technical field of mine safety management. The mine water inrush disaster early warning system based on sound-light fusion perception collects acoustic data, optical data and environmental parameters in a target mine through a preset acquisition device end, and preliminarily judges the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters. If the result of the preliminary risk judgment is that there is a risk, the acoustic data, the optical data and the environmental parameters are uploaded to an edge computing node. Through a risk prediction model pre-arranged at the edge computing node, the risk value of the water inrush disaster in the target mine is predicted according to the acoustic data, the optical data and the environmental parameters, and the risk value is fed back to a risk early warning end. According to the size of the risk value, corresponding early warning information is sent to prompt relevant personnel of the current risk situation of the target mine.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mine safety management, and particularly relates to a mine water inrush disaster early warning system based on sound-light fusion perception. BACKGROUND

[0002] Mine water inrush disaster is one of the major safety threats in underground mining operations such as coal mines and metal mines, and has the characteristics of strong suddenness, great destructive power and high prediction difficulty. Once it occurs, it often causes roadway flooding, equipment damage, even major casualties and huge economic losses.

[0003] At present, the monitoring and early warning of mine water inrush disaster mainly relies on the monitoring of hydrogeological parameters, surrounding rock stress and strain monitoring, geophysical exploration and artificial inspection. However, the above traditional monitoring methods often focus on the changes of a single or a few physical quantities. Before the occurrence of water inrush disaster, the precursor information is often multi-source, coupled and complex. Relying on single-dimensional information alone cannot fully and accurately capture the early signals of disaster incubation, which is easy to cause false negatives or false positives. SUMMARY

[0004] The main purpose of the present application is to provide a mine water inrush disaster early warning system based on sound-light fusion perception, which aims to solve the technical problem that current single-dimensional information in the mine cannot fully and accurately capture the early signals of disaster incubation, which is easy to cause false negatives or false positives.

[0005] To achieve the above purpose, the present application provides a mine water inrush disaster early warning system based on sound-light fusion perception, which comprises a data monitoring end, an edge computing node and a risk early warning end.

[0006] The data monitoring end is used to collect acoustic data, optical data and environmental parameters in the target mine through a preset acquisition device end, and to preliminarily judge the risk of water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters. If the result of the preliminary risk judgment is that there is a risk, the acoustic data, the optical data and the environmental parameters are uploaded to the edge computing node.

[0007] The edge computing node is used to predict the risk value of water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters through a risk prediction model pre-arranged at the edge computing node after receiving the acoustic data, the optical data and the environmental parameters sent by the data monitoring end, and to feed back the risk value to the risk early warning end.

[0008] The risk early warning end is used for receiving the risk value sent by the edge computing node, and issuing corresponding early warning information according to the size of the risk value to prompt relevant personnel about the current risk situation of the target mine.

[0009] In an embodiment, when the edge computing node predicts the risk value of the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters by using the risk prediction model pre-deployed at the edge computing node:

[0010] According to the acoustic data, the optical data and the environmental parameters, the dynamic weight corresponding to the water inrush stage of the corresponding water inrush disaster in the target mine is determined;

[0011] According to the acoustic data, the optical data and the dynamic weight, the risk value of the water inrush disaster in the target mine is predicted by using the risk prediction model pre-deployed at the edge computing node, wherein the dynamic weight is used to affect the tendency proportion of the acoustic data and the optical data when the risk prediction model performs risk prediction.

[0012] In an embodiment, when the edge computing node determines the dynamic weight corresponding to the water inrush stage of the corresponding water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters:

[0013] The first rule standard corresponding to the water inrush disaster in different water inrush stages in the target mine is obtained, and the second rule standard corresponding to different interference environments in the target mine is obtained;

[0014] According to the first rule standard, the first confidence parameter of the acoustic data and the optical data is determined;

[0015] According to the second rule standard, the second confidence parameter corresponding to the environmental data is determined;

[0016] According to the first confidence parameter, the second confidence parameter and the preset basic weight corresponding to different confidence parameters, the dynamic weight corresponding to the water inrush stage of the corresponding water inrush disaster in the target mine is calculated.

[0017] In an embodiment, after the edge computing node predicts the risk value of the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters:

[0018] The predicted historical risk value and the label information corresponding to the historical risk value are obtained, wherein the label information is obtained by relevant operation and maintenance personnel according to the historical risk value and the actual environmental situation of the target mine corresponding to the historical risk value;

[0019] According to the label information, the preset basic weight corresponding to the different confidence parameters is optimized.

[0020] In an embodiment, when the edge computing node determines the dynamic weight corresponding to the water inrush stage of the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters, the method further comprises:

[0021] The acoustic data includes sound event count and sound main frequency band energy;

[0022] If the sound event count is greater than a preset number and / or the change amount of the sound main frequency band energy is greater than a preset energy change threshold, a first preset weight group is selected as the dynamic weight, wherein the first preset weight group includes acoustic weight and optical weight, and the acoustic weight is greater than the optical weight.

[0023] In an embodiment, when the edge computing node determines the dynamic weight corresponding to the water inrush stage of the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters, the method further comprises:

[0024] The optical data includes infrared region maximum temperature difference and visible light crack length;

[0025] If the infrared region maximum temperature difference is greater than a preset temperature difference and / or the change amount of the visible light crack length is greater than a preset length change threshold, a second preset weight group is selected as the dynamic weight, wherein the second preset weight group includes acoustic weight and optical weight, and the optical weight is greater than the acoustic weight.

[0026] In an embodiment, when the data monitoring end determines the preliminary risk of the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters:

[0027] According to the influence of the environmental data on the acoustic-optical time difference, the acoustic data and the optical data are time-synchronized;

[0028] According to the synchronized data and a preset judgment rule, the preliminary risk of the water inrush disaster in the target mine is determined.

[0029] In an embodiment, after the data monitoring end determines the preliminary risk of the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters, the method further comprises:

[0030] The acoustic data includes sound event count and sound main frequency band energy, and the optical data includes infrared region maximum temperature difference and visible light crack length.

[0031] If any one of the sound event count and the visible light crack length is greater than a corresponding standard value, it is determined that there is a region to be patrolled in the target mine, and a corresponding report is generated according to a corresponding situation of the region to be patrolled.

[0032] If both the sound main frequency band energy and the infrared region highest temperature difference are greater than corresponding standard values, it is determined that the result of the preliminary risk judgment on the target mine is that there is a risk.

[0033] In an embodiment, when the risk warning end issues corresponding warning information according to the size of the risk value:

[0034] The range threshold interval corresponding to the risk value in the preset risk range value is determined.

[0035] According to the range threshold interval and the preset template information corresponding to the range threshold interval, corresponding warning information is generated.

[0036] In an embodiment, the data monitoring end is distributed and arranged in the target mine, wherein, according to the spatial range of the actual channel scene in the target mine, the data monitoring end is arranged at a preset distance between adjacent data monitoring ends in the spatial range.

[0037] The one or more technical solutions provided in the application have at least the following technical effects: the mine water inrush disaster early warning system based on sound-light fusion perception includes a data monitoring end, an edge computing node and a risk early warning end; the data monitoring end is configured to collect acoustic data, optical data and environmental parameters in a target mine through a preset acquisition device end, and perform preliminary risk judgment on a water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters, and if the preliminary risk judgment result is that there is a risk, the acoustic data, the optical data and the environmental parameters are uploaded to the edge computing node; the edge computing node is configured to, after receiving the acoustic data, the optical data and the environmental parameters sent by the data monitoring end, predict a risk value of the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters through a risk prediction model pre-arranged at the edge computing node, and feed back the risk value to the risk early warning end; the risk early warning end is configured to receive the risk value sent by the edge computing node, and send corresponding early warning information according to the size of the risk value to prompt relevant personnel about the current risk situation of the target mine, that is, in the application, a three-end interaction system composed of a data monitoring end, an edge computing node and a risk early warning end is constructed, through the data monitoring end, acoustic data, optical data and environmental data are collected at the same time, and preliminary risk judgment is performed on the three types of data, and after it is judged that there is a risk, the above multi-dimensional data is further sent to the edge computing node for accurate risk prediction through the risk prediction model arranged at the edge computing node, so as to realize multi-level risk supervision and prediction of multi-dimensional data in the mine through the construction of preliminary risk judgment of the data monitoring end and accurate prediction of the risk prediction model of the edge computing node, which not only ensures the reasonable actual application of multi-dimensional data, but also realizes the effect of accurate risk prediction through multi-dimensional data, thereby solving the technical problems that single-dimensional information in the mine is difficult to comprehensively and accurately capture early signals of disaster gestation, and is prone to cause false negatives or false positives. BRIEF DESCRIPTION OF DRAWINGS

[0038] The accompanying drawings incorporated in and forming a part of the specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the application.

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the accompanying drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, for those skilled in the art, other drawings can also be obtained based on these drawings without creative labor.

[0040] Figure 1A schematic diagram of the architecture of the mine water inrush disaster early warning system based on the sound-light fusion perception of the application;

[0041] Figure 2 A simplified process schematic diagram of the sound-light fusion data risk prediction of the application.

[0042] The implementation, functional features and advantages of the application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0043] It should be understood that the specific embodiments described herein are only used to explain the application and not to limit the application.

[0044] In an embodiment of the application, the mine water inrush disaster early warning system based on the sound-light fusion perception comprises a data monitoring end, an edge computing node and a risk early warning end;

[0045] The data monitoring end is configured to collect acoustic data, optical data and environmental parameters in the target mine through a preset acquisition device end, and make a preliminary risk judgment on the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters. If the result of the preliminary risk judgment is that there is a risk, the acoustic data, the optical data and the environmental parameters are uploaded to the edge computing node;

[0046] The edge computing node is configured to, after receiving the acoustic data, the optical data and the environmental parameters sent by the data monitoring end, predict the risk value of the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters through a risk prediction model pre-arranged at the edge computing node, and feed back the risk value to the risk early warning end;

[0047] The risk early warning end is configured to receive the risk value sent by the edge computing node, and issue corresponding early warning information according to the size of the risk value to prompt relevant personnel of the current risk situation of the target mine.

[0048] It should be noted that the mine water inrush disaster early warning system based on the sound-light fusion perception in the embodiment comprises a data monitoring end, an edge computing node and a risk early warning end. The interaction between the ends in the system and the arrangement of the ends in the target mine can be referred to Figure 1 .

[0049] The data monitoring end can be provided with a corresponding acoustic data monitoring end responsible for collecting sound signals or mechanical vibration signals and other data in the target mine. An optical data monitoring end can also be arranged correspondingly to collect crack, air turbidity and other data in the corresponding gallery of the target mine in a fixed point and range. An environmental monitoring end is also arranged to collect temperature and humidity data in the target mine.

[0050] The data monitoring end is responsible for collecting acoustic data, optical data and environmental data in the target mine, and is responsible for simple logical judgment through the above data, and preliminary judgment of whether the target mine represented by the above data exists corresponding risk, and after judging that there is risk, the data monitoring end will transmit the above three data to the edge computing node, and the edge computing node will make accurate risk prediction according to the above three data.

[0051] It should be noted that the above preliminary risk judgment process is mainly a simple logical judgment, and a corresponding multi-dimensional data judgment threshold is set, for example, when any two of the acoustic data and the optical data exceed the corresponding threshold standard at the same time, it can be judged that there is risk, but the risk degree is not within the judgment range of the data monitoring end, so as to ensure that the data monitoring end only needs a simple hardware chip to execute the corresponding data collection and logical judgment effect, and avoid setting a high-cost risk prediction model.

[0052] In addition, it should be noted that the range involved in the target mine is relatively large, so when the data monitoring end is laid out, a large number of monitoring ends need to be laid out according to the actual spatial position of the mine to ensure that all ranges of the target mine can be monitored in place, but the spatial position span is large, so a fixed number of data monitoring ends can be set to match an edge computing node to ensure the timeliness of the risk prediction of the edge computing node.

[0053] The edge computing node is pre-provided with a risk prediction model, which can take acoustic data, optical data and environmental data as input data and take risk value as output data. After receiving the corresponding data from the data monitoring end, the risk prediction model predicts the risk of the environment where the current data monitoring end is located, generates a corresponding risk value, and sends it to the corresponding risk warning end to ensure the overall interaction.

[0054] The risk prediction model is mainly a neural network model constructed according to the actual sample data in the target mine, which essentially fuses the physical meaning represented by the acoustic data, optical data and environmental data, learns the correlation between the above three data, and determines the risk degree of water inrush disaster represented by the three data through a rule setting method. The acoustic data, optical data and environmental data are extracted and the risk value is calculated through the corresponding weight, which reflects the risk of the actual environment composed of the above data in the current target mine through the risk value, wherein the greater the risk value, the greater the corresponding risk.

[0055] The risk warning end mainly includes a communication module or an alarm module. The risk warning end can be arranged in the target mine. After receiving the corresponding risk value, the risk warning end can send corresponding warning information according to the size of the corresponding risk value, prompt the current risk situation of the target mine to the relevant personnel, for example, the risk warning end and the terminal equipped by the relevant personnel can communicate with each other in the form of a text warning information, or the alarm model can send different color sound and light information to prompt the relevant personnel to focus on inspection or directly evacuate, and the like.

[0056] In summary, the simplified process diagram for predicting risks through data interaction can refer to Figure 2 .

[0057] It should be noted that the above process of predicting risks mainly includes two parts: simple logical judgment of fusion data of acoustic data, optical data and environmental data initiated by the data monitoring end, and simple logical operation by setting corresponding threshold values, on the one hand, to reduce the hardware cost of the data monitoring end, and on the other hand, as data screening, since the target mine is in a low-risk state most of the time under normal circumstances, it is not necessary to spend too much computing cost at this time, that is, it is not necessary to use a risk prediction model, the data monitoring end can filter out the low-risk data, and it is not necessary to send such data to the edge computing node, thereby directly reducing the frequency of using a high-cost risk prediction model, in addition, the edge node can make a precise judgment on the data that may exist in the high-risk state according to the logical judgment, so as to ensure the timeliness and accuracy of risk prediction, that is, the entire risk prediction process is divided into simple logical judgment and precise logical analysis, which not only reduces the high-cost investment of the risk prediction model, but also ensures the timeliness and accuracy of the entire risk prediction.

[0058] In the embodiment, when the edge computing node predicts the risk value of the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters through the risk prediction model pre-arranged at the edge computing node:

[0059] According to the acoustic data, the optical data and the environmental parameters, the dynamic weight corresponding to the water inrush stage of the water inrush disaster in the target mine is determined;

[0060] According to the acoustic data, the optical data and the dynamic weight, the risk value of the water inrush disaster in the target mine is predicted through the risk prediction model pre-arranged at the edge computing node, wherein the dynamic weight is used to affect the tendency proportion of the acoustic data and the optical data in the risk prediction model when predicting risks.

[0061] It should be noted that before the risk prediction model is used to predict the water inrush disaster existing in the target mine, it is necessary to consider that the phenomenon characteristics represented by different water inrush disaster stages are different when the water inrush disaster occurs. For example, a larger sound fluctuation may occur in the early stage of the water inrush disaster, and obvious crack changes and flowing water changes may occur on the monitoring image in the middle stage of the water inrush disaster. Therefore, different acoustic data and optical data are considered in different periods of the water inrush disaster. Therefore, in this embodiment, a dynamic weight is calculated according to the actually collected acoustic data, optical data and environmental data before the risk prediction model is used, and the dynamic weight affects the tendency proportion of the acoustic data and the optical data in the risk prediction of the risk prediction model, so as to ensure the accuracy of the risk result predicted by the risk prediction model.

[0062] The dynamic weight includes an acoustic weight and an optical weight which can be represented according to the actual data.

[0063] In this embodiment, when the edge computing node determines the dynamic weight corresponding to the water inrush stage of the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters:

[0064] The first rule standard corresponding to the different water inrush stages of the water inrush disaster in the target mine is obtained, and the second rule standard corresponding to the different interference environments in the target mine is obtained;

[0065] According to the first rule standard, the first confidence parameter of the acoustic data and the optical data is determined;

[0066] According to the second rule standard, the second confidence parameter corresponding to the environmental data is determined;

[0067] According to the first confidence parameter, the second confidence parameter and the preset basic weight corresponding to different confidence parameters, the dynamic weight corresponding to the water inrush stage of the water inrush disaster in the target mine is calculated.

[0068] It can be understood that when calculating the dynamic weight, the states represented by the acoustic data, the optical data and the environmental data need to be considered. For example, the sound energy frequency band corresponding to the acoustic data is high, or the crack growth of the tunnel wall in the target mine is monitored in the optical data. Different application scenarios need to be simulated and verified, and there will be a certain data influence tendency for different application scenarios. For example, high-frequency abnormal sound may occur in the early stage of water inrush, obvious crack changes may occur in the middle stage of water inrush, or the air in the tunnel is turbid during any water inrush disaster, which affects the judgment of optical data.

[0069] Therefore, in the embodiment, different rule criteria are designed, different confidence parameters are given corresponding to different rule criteria, and corresponding basic weights are preset for different rule criteria. For specific setting examples, refer to the following, and after the corresponding rule criteria are designed, the calculation formula of the corresponding dynamic weight is designed.

[0070] Specifically, the calculation formula of the dynamic weight is: wherein, the maximum weight of the acoustic data (0~1);

[0071] the weight of the optical data (complementary to the weight of the acoustic data, and the sum is 1);

[0072] the confidence of the ith rule criterion (indicating the credibility of the rule, 0~1);

[0073] the preset basic weight of the ith rule (0~1).

[0074] In the above formula, the physical meaning of the numerator is the acoustic weight of all rules weighted and considering the credibility of each rule criterion.

[0075] In the above formula, the sum of the credibility of all first rule criteria and second rule criteria is used for normalization.

[0076] In the dynamic weight finally calculated, the acoustic weight of each rule weight is a fuzzy weighted average, and the optical weight in the dynamic weight is its complement.

[0077] For example, assuming that the system calculates the dynamic adjustment weight according to the following three rule criteria:

[0078] Rule 1: High-frequency acoustic emission energy surge (μ=0.9, w=0.8);

[0079] Rule 2: Local cooling of infrared (μ=0.8, w=0.3);

[0080] Rule 3: Dust shielding of camera (μ=0.9, w=0.1).

[0081] The calculation steps are as follows:

[0082] Numerator (weighted sum): 0.9*0.8+0.8*0.3+0.9*0.1=0.72+0.24+0.09=1.05;

[0083] Denominator (membership sum): 0.9+0.8+0.9=2.6;

[0084] Acoustic weight in dynamic weight: = 1.05 / 2.6 ≈ 0.4;

[0085] Optical weight in dynamic weight: = 1−0.4 = 0.6.

[0086] It needs to be explained that although the acoustic energy suddenly increases (rule 1 suggests a weight of 0.8) after the above trigger rule standard, the camera is blocked by dust (rule 3 strongly inhibits optical data), and finally the optical weight is still high (0.6) because the infrared cooling (rule 2) provides reliable supplement.

[0087] In addition, corresponding examples are given for the first rule standard, the second rule standard, the first confidence parameter and the second confidence parameter, as follows:

[0088] (1) Acoustic dominant scene:

[0089] Rule 1: High-frequency acoustic emission energy suddenly increases, its trigger condition is that the energy in the frequency band of 50-80 kHz rises by 20 dB, its confidence parameter is 0.9, and its corresponding preset basic weight is 0.8;

[0090] Rule 2: The frequency of acoustic event count suddenly rises, its trigger condition is that the acoustic emission count is greater than 5 times per second for 30 seconds, its confidence parameter is 0.7, and its corresponding preset basic weight is 0.7;

[0091] Rule 3: The acoustic source positioning is close to the water-bearing layer, its trigger condition is that the positioning gushing point distance is less than 10 m, its confidence parameter is 0.6, and its corresponding preset basic weight is 0.6;

[0092] Its corresponding scene is the initial stage of water inrush, high-frequency sound waves are generated by rock mass rupture, but optical has not detected obvious seepage, at this time it is necessary to rely on acoustic data first (it is expected that the dynamic weight corresponding to acoustic data is greater than 0.7).

[0093] (2) Optical dominant scene:

[0094] Rule 1: Local cooling of infrared image, its trigger condition is that the temperature difference change of well wall temperature is greater than 2℃, its confidence parameter is 0.8, and its corresponding preset basic weight is 0.3;

[0095] Rule 2: Water turbidity suddenly increases, its trigger condition is that the laser scattering detects suspended solids concentration greater than 50, its confidence parameter is 0.7, and its corresponding preset basic weight is 0.2;

[0096] Rule 3: Visible light crack expansion, its trigger condition is that the crack growth rate is greater than 0.1 mm / s, its confidence parameter is 0.6, and its corresponding preset basic weight is 0.4;

[0097] The corresponding scene is the middle stage of water inrush. Water seepage causes temperature drop, turbidity rise, but acoustic signal is stable. At this time, optical data needs to be relied on preferentially (the dynamic weight corresponding to the optical data is expected to be increased to 0.6-0.8).

[0098] (3) Environmental interference scene:

[0099] Rule 1: The camera is blocked by dust, the triggering condition is that the image sharpness is less than the preset sharpness threshold, the confidence parameter is 0.9, and the corresponding preset basic weight is 0.1;

[0100] Rule 2: Mining machinery noise interference, the triggering condition is that the low-frequency energy ratio is greater than 60%, the confidence parameter is 0.8, and the corresponding preset basic weight is 0.3;

[0101] Rule 3: Infrared thermal imager fogging, the triggering condition is that the temperature field standard deviation is less than 0.5℃, the confidence parameter is 0.7, and the corresponding preset basic weight is 0.2;

[0102] The corresponding scene is that dust or mechanical noise causes a certain modal data to be unreliable. At this time, the weight of the affected modal is reduced (the dynamic weight corresponding to the optical data is expected to be reduced to 0.2).

[0103] (4) Balanced weight scene:

[0104] Rule 1: Acoustic and optical signal synchronization anomaly, the triggering condition is that the acoustic and optical time delay difference is greater than 10ms, the confidence parameter is 0.5, and the corresponding preset basic weight is 0.5;

[0105] Rule 2: Multi-modal weak correlation, the triggering condition is that the acoustic and optical feature correlation coefficient is less than 0.3, the confidence parameter is 0.4, and the corresponding preset basic weight is 0.5;

[0106] Rule 3: Infrared thermal imager fogging, the triggering condition is that the temperature field standard deviation is less than 0.5℃, the confidence parameter is 0.7, and the corresponding preset basic weight is 0.2;

[0107] The corresponding scene is that acoustic and optical data conflict, and the dominant modal cannot be determined (the dynamic weight corresponding to the optical data is expected to be 0.5, and the dynamic weight corresponding to the acoustic data is expected to be 0.5).

[0108] In the embodiment, when the edge computing node determines the dynamic weight corresponding to the water inrush stage of the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters, the method further comprises:

[0109] If the count of the sound generating events is greater than a preset number and / or the change in the sound main frequency band energy is greater than a preset energy change threshold, a first preset weight group is selected as the dynamic weight, wherein the first preset weight group includes an acoustic weight and an optical weight, and the acoustic weight is greater than the optical weight.

[0110] If the infrared region highest temperature difference is greater than a preset temperature difference and / or the change in the visible light crack length is greater than a preset length change threshold, a second preset weight group is selected as the dynamic weight, wherein the second preset weight group includes an acoustic weight and an optical weight, and the optical weight is greater than the acoustic weight.

[0111] It should be noted that according to the examples of the above rule standards, when calculating the dynamic weight, the corresponding preset basic weight is set in advance, and the dynamic weight actually used in the risk prediction model is calculated according to the preset basic weight and the corresponding calculation formula of the dynamic weight. However, this process is also an approximate calculation, and the main calculation principle is still to limit the actual effect according to the rules. Therefore, the process of calculating the dynamic weight can be further simplified, and the parameters mentioned in the above rules are directly mapped and connected. Only by setting the corresponding judgment conditions, when the event count is greater than the preset number and / or the change in the sound main frequency band energy is greater than the preset energy change threshold, the first preset weight group is directly selected as the dynamic weight. At this time, the acoustic weight in the first preset weight group is greater than the optical weight, which can be set to 0.7 and 0.3. When the infrared region temperature difference is greater than the preset temperature difference and / or the visible light crack length change is greater than the preset length change threshold, the second preset weight group is directly selected as the dynamic weight. At this time, the acoustic weight in the second preset weight group is less than the optical weight, which can be set to 0.3 and 0.7.

[0112] The first preset weight group corresponds to the early stage of the water inrush disaster, and the second preset weight group corresponds to the middle and late stages of the water inrush disaster.

[0113] In the embodiment, after the edge computing node predicts the risk value of the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters:

[0114] The historical risk value predicted and the label information corresponding to the historical risk value are obtained, wherein the label information is obtained by a relevant operation and maintenance personnel according to the historical risk value and the actual environmental situation of the target mine corresponding to the historical risk value; and the preset basic weight corresponding to the different confidence parameters is optimized according to the label information.

[0115] It should be noted that the risk value predicted by the risk prediction model does not necessarily fully meet the actual needs of the changing situation in the target mine, therefore, after the risk prediction model is used for a period of time, the risk values predicted in this period of time are collected and summarized into historical risk values, the historical risk values and the label information corresponding to the historical risk values are processed, and the parameters related to the dynamic weight required by the risk prediction model are further optimized, that is, different confidence parameters and their corresponding preset basic weights are optimized, the size of the confidence parameters and the preset basic weights of different rules are adjusted, so that the accuracy in subsequent calculation of the dynamic weight is ensured.

[0116] The label information corresponding to the historical risk value is obtained by the relevant operation and maintenance personnel according to the historical risk value and the actual environmental situation in the target mine corresponding to the historical risk value, so as to establish the corresponding relationship between the risk value and the actual environmental situation, and ensure the accuracy of the subsequent model use.

[0117] In the present embodiment, when the data monitoring end preliminarily judges the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters:

[0118] According to the influence of the environmental data on the time difference between sound and light, the acoustic data and the optical data are time-synchronized; according to the data after synchronization processing and the preset judgment rule, the water inrush disaster in the target mine is preliminarily judged.

[0119] It can be understood that the environmental data includes the smoke condition and the temperature and humidity condition inside the target mine tunnel, and under such environmental conditions, the collection time corresponding to the acoustic data and the optical data collected by the data monitoring end cannot be corresponded, therefore, before the acoustic data and the optical data are preliminarily predicted, the two types of data need to be time-synchronized to ensure that different data for the same event have the same timestamp, so as to facilitate subsequent logical reasoning to judge whether there is a risk in the current target mine.

[0120] Among them, the time synchronization processing of acoustic data and optical data mainly includes the following two schemes:

[0121] One of them is that the data monitoring end needs certain hardware configuration for data synchronization, among which the acoustic end adopts a distributed high-frequency microphone array (20-200 kHz), one set is arranged every 50 m, and the sound source position is calculated by TDOA (time difference positioning) algorithm; the optical end adopts an infrared thermal imager (monitors temperature field) and a visible light camera (crack identification), which are coaxially installed with the acoustic sensor.

[0122] The synchronization mechanism of the two can adopt any of the following schemes:

[0123] Scheme 1: Hardware level PTP protocol (Precision Time Protocol), synchronization error less than 1 μs (requires wired network support).

[0124] Scheme 2: Software level dynamic time warping, aligning non-uniform sampling data (such as acoustic 100 kHz or optical 30 fps).

[0125] Another, according to the environmental data to select a reasonable time difference, that is, considering the difference between the sound and light propagation speed, setting a sound propagation fixed delay parameter, and combining different environmental scenarios, adjusting the sound propagation fixed delay parameter, for example, in the scene where the humidity is larger or the visibility is lower, the time of collecting light will be slightly later than that in the scene where the visibility is high, and the sound propagation fixed delay parameter needs to be reduced.

[0126] Among them, the preset judgment rule refers to the linkage of multidimensional data, and the flexible judgment condition is set by the relevant personnel, which needs to be combined with the data in the acoustic data and the optical data that exist in the mutual influence relationship to make joint judgment, for example, in the acoustic data, a high frequency sound segment is generated, and the optical data collects that the crack of the wall in the target mine increases, which can be used as a judgment basis to judge whether there is a risk in the target mine.

[0127] In the present embodiment, after the data monitoring end preliminarily judges the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters, the method further comprises:

[0128] The acoustic data includes sound event count and sound main frequency band energy, and the optical data includes infrared region maximum temperature difference and visible light crack length;

[0129] If any one of the sound event count and the visible light crack length is greater than the corresponding standard value, it is determined that there is a to-be-inspected area in the target mine, and a corresponding report is generated according to the corresponding situation of the to-be-inspected area;

[0130] If the sound main frequency band energy and the infrared region maximum temperature difference are both greater than the corresponding standard value, it is determined that the result of the preliminary risk judgment of the target mine is that there is a risk.

[0131] It can be understood that the acoustic data at least includes the event, the sound main frequency band energy, and can also include the sound maximum decibel and the like, and the optical data at least includes the infrared region maximum temperature difference and the visible light crack length, and can also include the visibility caused by the environmental dust in the space and the like.

[0132] In the embodiment, the acoustic data includes sound event count and sound dominant band energy, and the optical data includes infrared region maximum temperature difference and visible light crack length.

[0133] The acoustic data and the optical data are combined to set a corresponding preliminary risk judgment process. Specifically, when the sound dominant band energy and the infrared region maximum temperature difference are greater than the corresponding standard values, it is determined that the target mine has a preliminary risk judgment result of risk. When any one of the sound event count and the visible light crack length is greater than the corresponding standard value, it is determined that the target mine has a to-be-inspected area, and a corresponding report needs to be generated according to the to-be-inspected area.

[0134] That is, in the preliminary risk judgment process, the result of the judgment is divided into two categories, including risk and to-be-inspected area. When it is determined that there is a risk, risk prediction is needed through the edge computing node. When it is determined that there is a to-be-inspected area, a corresponding report needs to be generated to prompt a corresponding person to inspect.

[0135] It should be noted that through the above preliminary risk judgment process, it can be determined whether there is a risk in the current target mine through simple logical rules, thereby reducing the cost of deploying a risk prediction model in the target mine. At the same time, this judgment method can rely only on low-cost hardware devices to perform corresponding preliminary risk judgment, greatly reducing the cost demand of risk prediction. At the same time, the combination of the judgment method and the risk prediction model for risk prediction can reduce the amount of data needed to be processed by the risk prediction model. Through preliminary risk judgment, most of the data is filtered out, thereby ensuring the running energy consumption cost under the condition of continuous monitoring.

[0136] In the embodiment, when the risk value is greater than the preset risk range value, the corresponding warning information is sent.

[0137] The range threshold interval corresponding to the risk value in the preset risk range value is determined. According to the range threshold interval and the preset template information corresponding to the range threshold interval, the corresponding warning information is generated.

[0138] It can be understood that the risk value predicted by the risk prediction model represents the probability of the possibility of water inrush risk. The greater the risk value, the greater the risk of water inrush disaster under the current circumstances. The risk value is in the interval of 0-1. In this interval, according to the size of the risk value, a plurality of value range intervals can be further divided.

[0139] For example, three interval sections of 0-0.4, 0.4-0.7, and 0.7-1, and each interval corresponds to a risk warning state. Specifically, the first interval 0-0.4 (including the risk value equal to 0) corresponds to a low risk or no risk interval, in which only relevant personnel need to be prompted for normal daily inspection. The second interval 0.4-0.6 (including the risk value equal to 0.4) corresponds to a medium risk interval, which has a certain tendency of water inrush disaster, but no water inrush disaster risk for the time being, and needs to be focused on defense and monitoring. The third interval 0.7-1 (including the risk value equal to 0.7 and 1) corresponds to a high risk interval, which has a water inrush disaster risk, and the water inrush disaster risk is expected to occur at a time close to the current time, and personnel need to be immediately dispersed, and the corresponding equipment needs to be removed.

[0140] The first interval, the second interval, and the third interval can be used as the range threshold interval corresponding to the preset risk range value. When the risk value falls into the corresponding range threshold interval, the preset template information corresponding to the range threshold interval is selected, and the corresponding warning information is generated according to the preset template information.

[0141] In the embodiment, the data monitoring end is distributed and arranged in the target mine, wherein, according to the space range of the actual channel scene in the target mine, the preset distance is arranged between adjacent data monitoring ends in the space range.

[0142] It can be understood that, in order to monitor the entire target mine, the data monitoring end needs to be reasonably arranged according to the actual tunnel trend of the target mine and the high-risk area in the target mine. The main arrangement method of the data monitoring end is distributed arrangement, which is arranged according to the preset distance between adjacent data monitoring ends in the space range of the actual channel scene in the target mine. In the high-risk area that needs to be monitored, the size of the preset distance can be appropriately reduced.

[0143] The mine water inrush disaster early warning system based on the sound-light fusion perception comprises a data monitoring end, an edge computing node and a risk early warning end. The data monitoring end is used for collecting acoustic data, optical data and environmental parameters in a target mine through a preset acquisition device end, and performing preliminary risk judgment on water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters. If the result of the preliminary risk judgment is that there is a risk, the acoustic data, the optical data and the environmental parameters are uploaded to the edge computing node. The edge computing node is used for predicting the risk value of the water inrush disaster in the target mine according to the acoustic data, the optical data and the environmental parameters through a risk prediction model pre-arranged at the edge computing node after receiving the acoustic data, the optical data and the environmental parameters sent by the data monitoring end, and feeding back the risk value to the risk early warning end. The risk early warning end is used for receiving the risk value sent by the edge computing node, and issuing corresponding early warning information according to the size of the risk value to prompt relevant personnel about the current risk situation of the target mine. In this application, a three-end interaction system comprising a data monitoring end, an edge computing node and a risk early warning end is constructed. Through the data monitoring end, preliminary risk judgment is performed on three types of data while collecting acoustic data, optical data and environmental data, and after judging that there is a risk, the above multi-dimensional data is further sent to the edge computing node to perform accurate risk prediction through the risk prediction model arranged at the edge computing node, so as to realize multi-level risk supervision and prediction of multi-dimensional data in the mine through the construction of preliminary risk judgment of the data monitoring end and accurate prediction of the risk prediction model of the edge computing node, which not only ensures the reasonable actual application of multi-dimensional data, but also realizes the effect of accurate risk prediction through multi-dimensional data, so as to solve the technical problems that single dimension information in the mine is difficult to capture early signals of disaster gestation comprehensively and accurately, and is easy to cause missed report or false report.

[0144] The above only describes some embodiments of the present application, and does not limit the patent scope of the present application. Any equivalent structural transformation made by using the contents of the present application specification and drawings, or direct / indirect application in other related technical fields is included in the patent protection scope of the present application.

[0145] It should be noted that, in this text, the terms "comprising", "containing" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or includes elements inherent to such process, method, article or system. Without more limitations, the element defined by the statement "including a" does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0146] The above-mentioned sequence numbers of the embodiments of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments.

[0147] The above is only the preferred embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent process transformation using the content of the present application specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A mine water inrush disaster early warning system based on acoustic-optical fusion sensing, characterized in that, The mine water inrush disaster early warning system based on acoustic-optical fusion perception includes: a data monitoring terminal, an edge computing node, and a risk early warning terminal. The data monitoring terminal is used to collect acoustic data, optical data, and environmental parameters from the target mine through a pre-set acquisition device. Based on the acoustic data, optical data, and environmental parameters, it performs a preliminary risk assessment of the water inrush disaster in the target mine. If the preliminary risk assessment indicates a risk, the acoustic data, optical data, and environmental parameters are uploaded to the edge computing node. The edge computing node, after receiving the acoustic data, optical data, and environmental parameters from the data monitoring terminal, uses a pre-deployed risk prediction model at the edge computing node to predict the risk value of the water inrush disaster in the target mine based on the acoustic data, optical data, and environmental parameters, and feeds back the risk value to the risk early warning terminal. When the edge computing node predicts the risk value of a water inrush disaster in the target mine based on the acoustic data, optical data, and environmental parameters using a risk prediction model pre-deployed at the edge computing node: Based on the acoustic data, optical data, and environmental parameters, the dynamic weights corresponding to the water inrush stage of the water inrush disaster in the target mine are determined; the risk value of the water inrush disaster in the target mine is predicted based on the acoustic data, optical data, and dynamic weights using the risk prediction model pre-deployed at the edge computing node, wherein the dynamic weights are used to influence the tendency proportion of the acoustic data and optical data when the risk prediction model performs risk prediction; the risk warning terminal is used to receive the risk value sent by the edge computing node and issue corresponding warning information based on the magnitude of the risk value to alert relevant personnel to the current risk situation of the target mine.

2. The system as described in claim 1, characterized in that, When the edge computing node determines the dynamic weight corresponding to the water inrush stage of the water inrush disaster in the target mine based on the acoustic data, the optical data, and the environmental parameters: it acquires a first rule standard corresponding to the water inrush disaster in the target mine at different water inrush stages, and acquires a second rule standard corresponding to different interference environments in the target mine; it determines a first confidence parameter of the acoustic data and the optical data based on the first rule standard; and it determines a second confidence parameter corresponding to the environmental data based on the second rule standard. Based on the first confidence parameter, the second confidence parameter, and the preset basic weights corresponding to different confidence parameters, the dynamic weights corresponding to the water inrush stage of the water inrush disaster in the target mine are calculated.

3. The system as described in claim 1, characterized in that, After the edge computing node predicts the risk value of water inrush disaster in the target mine based on the acoustic data, the optical data, and the environmental parameters: it obtains the predicted historical risk value and the corresponding tag information, wherein the tag information is obtained by relevant operation and maintenance personnel based on the historical risk value and the actual environmental conditions of the target mine corresponding to the historical risk value; and it optimizes the preset basic weights corresponding to the different confidence parameters based on the tag information.

4. The system as described in claim 1, characterized in that, When the edge computing node determines the dynamic weight corresponding to the water inrush stage of the water inrush disaster in the target mine based on the acoustic data, the optical data, and the environmental parameters, the method further includes: the acoustic data includes a count of sound events and the energy of the main frequency band of sound; if the count of sound events is greater than a preset number and / or the change in the energy of the main frequency band of sound is greater than a preset energy change threshold, then a first preset weighting set is selected as the dynamic weight, wherein the first preset weighting set includes acoustic weight and optical weight, and the acoustic weight is greater than the optical weight.

5. The system as described in claim 1, characterized in that, When the edge computing node determines the dynamic weight corresponding to the water inrush stage of the water inrush disaster in the target mine based on the acoustic data, the optical data, and the environmental parameters, the method further includes: the optical data includes the maximum temperature difference in the infrared region and the visible light crack length; if the maximum temperature difference in the infrared region is greater than a preset temperature difference and / or the change in the visible light crack length is greater than a preset length change threshold, then a second preset weighting is selected as the dynamic weight, wherein the second preset weighting includes acoustic weight and optical weight, and the optical weight is greater than the acoustic weight.

6. The system as described in claim 1, characterized in that, When the data monitoring terminal makes a preliminary risk assessment of water inrush disaster in the target mine based on the acoustic data, the optical data and the environmental parameters: the acoustic data and the optical data are time-synchronized according to the influence of the environmental data on the acoustic-optical time difference; Based on the synchronously processed data and preset judgment rules, a preliminary risk assessment of water inrush disaster in the target mine is conducted.

7. The system as described in claim 6, characterized in that, After the data monitoring terminal performs a preliminary risk assessment of water inrush disaster in the target mine based on the acoustic data, the optical data, and the environmental parameters, the method further includes: the acoustic data includes the count of sound events and the energy of the dominant sound frequency band; the optical data includes the maximum temperature difference in the infrared region and the visible light crack length; if either the count of sound events or the visible light crack length is greater than the corresponding standard value, it is determined that there is an area to be inspected in the target mine, and a corresponding report is generated based on the situation of the area to be inspected; if both the energy of the dominant sound frequency band and the maximum temperature difference in the infrared region are greater than the corresponding standard values, it is determined that the preliminary risk assessment result of the target mine is that there is a risk.

8. The system as described in claim 1, characterized in that, When the risk warning terminal issues a corresponding warning message based on the magnitude of the risk value: it determines the range threshold interval corresponding to the risk value within a preset risk range value; and generates the corresponding warning message based on the range threshold interval and the preset template information corresponding to the range threshold interval.

9. The system as described in claim 1, characterized in that, The data monitoring terminals are distributed throughout the target mine, and are deployed according to a preset distance between adjacent data monitoring terminals within the spatial range of the actual passageway within the target mine.

Images