Mine fire monitoring method and system based on multi-dimensional perception

Abstract

This invention belongs to the field of mine fire monitoring, specifically involving a mine fire monitoring method and system based on multi-dimensional perception. It addresses the technical problems of inaccurate underground fire monitoring and the inability to identify different fire modes in existing technologies. The monitoring method includes the following steps: S1, obtaining the environmental smoke background concentration over time by performing median filtering on the smoke concentration sequence; S2, calculating the smoldering characteristic value and flame composite parameters; S3, calculating the energy entropy of the infrared radiation intensity sequence within the time window; and weighting the flame composite parameters and smoldering characteristic value to obtain a comprehensive fire index; S4, outputting a first fire alarm signal and a second fire alarm signal. This invention introduces the environmental smoke background concentration, which can filter out interference from non-fire factors such as dust and extract the actual smoke increment caused by the fire.

Description

Technical Field

[0001] This invention belongs to the field of mine fire monitoring, specifically relating to a mine fire monitoring method and system based on multi-dimensional perception. Background Technology

[0002] The underground working environment is complex, and rescue operations are extremely difficult once an accident occurs. Currently, most commonly used mine fire detectors rely on single physical quantities such as smoke, infrared radiation, or ultraviolet radiation to determine the fire situation. However, single sensors have the following limitations: smoke detectors are easily interfered with by non-fire factors such as underground dust, water vapor, and locomotive exhaust, resulting in a high false alarm rate; infrared detectors are sensitive to changes in ambient temperature and can easily misidentify normal heat sources such as electromechanical equipment, welding, and lighting as fire sources; while ultraviolet detectors can effectively detect open flames, they are ineffective in the initial smoldering stage (without open flames) of fires such as spontaneous combustion of coal, and are easily interfered with by non-fire ultraviolet sources such as arc welding and corona discharge from high-voltage equipment. These limitations lead to high false alarm and false negative rates for existing single-sensor fire monitoring methods, making it difficult to meet the safety requirements of complex mine environments.

[0003] While comprehensive analysis of smoke, infrared, and ultraviolet data can improve the accuracy of fire detection, existing simple fusion methods struggle to effectively distinguish between open flame and coal smoldering fire modes. Furthermore, they cannot dynamically adjust the weights of various parameters based on different stages of fire development (e.g., smoke dominates during coal smoldering, while radiation dominates during open flame). Therefore, there is an urgent need for a monitoring method capable of analyzing multi-source monitoring data, comprehensively assessing fire development, effectively identifying the early smoldering stage, and differentiating between different fire modes for alarm purposes. Summary of the Invention

[0004] This invention provides a mine fire monitoring method and system based on multi-dimensional perception to solve the technical problems of inaccurate mine fire monitoring and inability to identify different fire modes in the prior art.

[0005] In a first aspect, the present invention provides a method for monitoring mine fires based on multi-dimensional perception, comprising the following steps: S1, acquire the infrared radiation intensity sequence, ultraviolet radiation intensity sequence, and smoke concentration sequence collected in real time at the monitoring point; obtain the environmental smoke background concentration that changes over time by performing median filtering on the smoke concentration sequence; S2, Calculate the smoldering characteristic value and flame composite parameters: The smoldering characteristic value is obtained by dividing the difference between the current smoke concentration and the ambient smoke background concentration by the current infrared radiation intensity; the flame composite parameters are calculated based on the current infrared radiation intensity, the first-order difference of the infrared radiation intensity, and the ultraviolet radiation intensity. S3. Set a time window based on the smoldering characteristic value and calculate the energy entropy of the infrared radiation intensity sequence within the time window; determine the weight of the smoldering characteristic value by using the number of peak values ​​of the ultraviolet radiation intensity sequence counted within the first preset time window; and use this weight to weight the flame composite parameters and the smoldering characteristic value to obtain the comprehensive fire index. S4. When the comprehensive fire index is greater than the first fire threshold and the energy entropy is greater than the entropy threshold, output the first fire alarm signal; when the comprehensive fire index is between the first fire threshold and the second fire threshold and no ultraviolet radiation is detected within the second preset time window, output the second fire alarm signal.

[0006] Furthermore, based on the current infrared radiation intensity, the first-order difference value of the infrared radiation intensity, and the ultraviolet radiation intensity, the flame composite parameters are calculated, including: The flame composite parameters are obtained by multiplying or weighting the current infrared radiation intensity, the first-order difference of the current infrared radiation intensity, and the current ultraviolet radiation intensity.

[0007] Furthermore, a time window is set based on the braising characteristic values, including: Set the base window length to The simmering coefficient is Minimum window length is Number of sampling points in the time window Calculated using the following formula: ; in, This represents the current simmering characteristic value. For the floor function, the base window length is... The simmering coefficient represents the baseline value used when calculating the length of the time window. Minimum window length is used to adjust the influence of simmering characteristic values ​​on the time window length. It is the lower limit of the time window length, ensuring that the system always has enough data for analysis.

[0008] Further, the energy entropy of the infrared radiation intensity sequence within this time window is calculated, including: Each infrared radiation intensity value within the time window Divide by the sum of all infrared radiation intensity values ​​within the time window to obtain the energy percentage at each moment. ; The energy entropy is calculated using the following formula. : .

[0009] Furthermore, a comprehensive fire risk index will be established, including: The number of times the ultraviolet radiation intensity exceeded a preset ultraviolet threshold within the first preset time window is recorded as the peak value. ; The weights of the braising characteristic values ​​are calculated using the following formula. : ; Weighting of flame composite parameters equal ; The flame composite parameters and simmering characteristic values ​​were normalized separately to obtain the normalized flame composite parameters. Normalized braised characteristic value ; The comprehensive fire index is calculated using the following formula. : .

[0010] Secondly, the present invention provides a mine fire monitoring system based on multi-dimensional perception, including a memory and a processor. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned mine fire monitoring method based on multi-dimensional perception is implemented.

[0011] The beneficial effects are as follows: Compared with existing technologies, this invention introduces the background concentration of ambient smoke, which can filter out interference from non-fire factors such as dust and extract the actual smoke increment caused by fire. The constructed smoldering feature value and flame composite parameters enable targeted identification of different fire modes. Simultaneously, by using the energy entropy of the infrared radiation intensity sequence as a criterion and leveraging the unstable characteristics of the fire source signal, it distinguishes between real fires and stable heat sources, reducing false alarms caused by electromechanical equipment, welding heat sources, etc. The comprehensive fire index established by this invention adjusts the weighting coefficients according to the number of peak values ​​in the ultraviolet radiation intensity sequence, focusing on the current fire mode. Combined with a dual-threshold and graded alarm mechanism, it can not only distinguish between smoldering and open flame conditions and output different levels of alarms, but also provide early warning of early smoldering fires, buying valuable time for mine fire prevention. Attached Figure Description

[0012] Figure 1 A flowchart of a mine fire monitoring method based on multi-dimensional perception; Figure 2 The sequence diagrams show infrared radiation intensity, ultraviolet radiation intensity, and smoke / fog density. Figure 3 This is a schematic diagram showing the original smoke concentration and the background smoke concentration in the environment. Detailed Implementation

[0013] An embodiment of the mine fire monitoring method based on multi-dimensional perception provided by this invention: like Figure 1As shown, a mine fire monitoring method based on multi-dimensional perception includes the following steps: S1, acquire the infrared radiation intensity sequence, ultraviolet radiation intensity sequence, and smoke concentration sequence collected in real time at the monitoring point; obtain the environmental smoke background concentration that changes over time by performing median filtering on the smoke concentration sequence.

[0014] In one embodiment, any fire in a mine generates a significant amount of heat. Infrared detectors, with their high sensitivity and fast response, can detect potential thermal anomalies. However, infrared detectors are also susceptible to interference from non-fire heat sources such as high-temperature equipment, leading to false alarms. Ultraviolet radiation, originating from high-energy chemical reactions during combustion, is extremely resistant to interference and rarely causes false alarms. However, ultraviolet signals are easily blocked by smoke and dust and cannot detect smoldering fires without open flames.

[0015] Combining the two methods allows for the use of the high sensitivity of infrared radiation to detect potential thermal anomalies, while the high specificity of ultraviolet radiation confirms the presence of open flames, ensuring rapid alarm while reducing false alarm rates. Specifically, an integrated fire detector is deployed in the monitoring area of ​​the mine roadway. This integrated fire detector incorporates infrared, ultraviolet, and smoke sensors and is controlled by a microprocessor. It simultaneously acquires analog signals from the three sensors, converts them into digital signals, and generates a discrete sequence of infrared radiation intensity. Ultraviolet radiation intensity sequence and smoke concentration sequence , Figure 2 This is a sequence diagram of the collected infrared radiation intensity, ultraviolet radiation intensity, and smoke concentration.

[0016] Set a length of Sliding time window (e.g.) =120, which represents the data from the past two minutes), indicating 120 smoke concentration samples within this sliding time window. to Perform numerical sorting, and select the value at the middle position after sorting, i.e., the 60th or 61st value, as the current moment. Background concentration of ambient smoke ,like Figure 3 As shown. Background concentration of ambient smoke. It can smooth out short-term concentration fluctuations caused by non-fire factors such as dust and water vapor, and reflect the baseline smoke level of the environment.

[0017] S2, calculate the smoldering characteristic value and flame composite parameters: obtain the smoldering characteristic value by dividing the difference between the current smoke concentration and the ambient smoke background concentration by the current infrared radiation intensity; calculate the flame composite parameters based on the current infrared radiation intensity, the first-order difference of the infrared radiation intensity, and the ultraviolet radiation intensity.

[0018] In one embodiment, the simmering characteristic value The calculation formula is Coal produces a large amount of smoke during the simmering stage, but the heat radiation intensity is low; therefore, the molecules... - The value is very large, while the denominator is large. The value is small, resulting in a low characteristic value for simmering. Increased molecular density can effectively identify early-stage smoldering fires. Conversely, for smokeless heat sources such as mechanical and electrical equipment, the molecular density is close to zero, resulting in a smoldering characteristic value. It will be very small.

[0019] In one embodiment, flame composite parameters The calculation formula is a weighted sum, and its formula is: .in, The rate of increase of infrared radiation intensity (i.e., the first-order difference value). , , These are preset weighting coefficients. The flame composite parameters combine the characteristics of high-intensity thermal radiation, ultraviolet radiation, and rapid radiation growth generated by an open flame. When an open flame is present, the flame composite parameters... The value will increase.

[0020] In an optional embodiment, flame composite parameters are calculated based on the current infrared radiation intensity, the first-order difference value of the infrared radiation intensity, and the ultraviolet radiation intensity, including: The flame composite parameters are obtained by multiplying or weighting the current infrared radiation intensity, the first-order difference of the current infrared radiation intensity, and the current ultraviolet radiation intensity.

[0021] Specifically, when an open flame appears, it simultaneously generates infrared and ultraviolet radiation. Furthermore, due to the flickering and spread of the flame, the intensity of its infrared radiation changes over time. Multiplying this information across all three dimensions amplifies the signal of a real flame while suppressing interference from single characteristics. For example, a stable, high-temperature object (such as machinery or lighting) may produce strong infrared radiation, but its ultraviolet radiation is weak and its infrared radiation changes gradually, resulting in a first-order difference value for its infrared radiation intensity approaching zero. In this case, even if the infrared radiation intensity is high, the combined flame parameter after multiplication will be very small because the other two parameters are also small, thus avoiding false alarms.

[0022] Taking a real-world example, the sensor in The infrared radiation intensity collected at any given time was 1500 units, and the ultraviolet radiation intensity was 1200 units, while... If the infrared radiation intensity collected at any given time is 1300 units, then at the current time... The first-order difference value of the infrared radiation intensity is 1500 - 1300 = 200 units. If the flame composite parameters are calculated using a multiplication method, the result is... =360000000.

[0023] To prevent a small multiplication result due to a small value for a single parameter, in an alternative embodiment, a weighted summation method is used to calculate the flame composite parameters. In addition, to prevent excessively large data from causing memory overflow or anomalies, all values ​​can be reduced by a factor of 100 for processing.

[0024] S3. Set a time window based on the smoldering characteristic value and calculate the energy entropy of the infrared radiation intensity sequence within the time window; determine the weight of the smoldering characteristic value by using the number of peak values ​​of the ultraviolet radiation intensity sequence counted within the first preset time window; and use this weight to weight the flame composite parameters and the smoldering characteristic value to obtain the comprehensive fire index.

[0025] In an optional embodiment, a time window is set based on the simmering characteristic value, including: Set the base window length to The simmering coefficient is Minimum window length is Number of sampling points in the time window Calculated using the following formula: ; in, This represents the current simmering characteristic value. For the floor function, the base window length is... The simmering coefficient represents the baseline value used when calculating the length of the time window. Minimum window length is used to adjust the influence of simmering characteristic values ​​on the time window length. It is the lower limit of the time window length, ensuring that the system always has enough data for analysis.

[0026] Base window length This is the baseline analysis time used by the system when fire characteristics are not obvious. Its principle is to ensure sufficient time to confirm the random fluctuation characteristics of smokeless flames, preventing stable heat sources from being mistaken for fires. The acquisition steps are as follows: The experiment simulates a smokeless open flame scenario, while simultaneously simulating various intense non-fire interference sources (such as electric welding). By adjusting the base window length The length of the system was used to observe the energy entropy discrimination results under smokeless open flame and interference sources. Choose a time length that provides the best discrimination accuracy. This length should be long enough to capture the long-term random fluctuation characteristics of the flame signal, thereby avoiding misjudgment.

[0027] Simmering coefficient It is to adjust the stewing characteristic value The sensitivity coefficient to the degree of influence of the time window length determines the number of sampling points. With the characteristic value of stewing The principle behind obtaining the speed of increase and decrease is to find the optimal balance between response speed and accuracy. The steps to obtain this are as follows: Simulate various smoldering fire scenarios and record smoldering characteristic values. Typical variation curves of the values ​​were observed, and different simmering coefficients were tested. ; By calculating different simmering coefficients Choose a value that ensures the system can respond quickly and accurately in typical simmering scenarios, considering both response time and discrimination accuracy.

[0028] Minimum window length is This is the absolute lower limit of the time window length. It must ensure that the amount of data within the window is sufficient for reliable energy entropy calculation. The principle for obtaining it is to determine the minimum effective sample size for the energy entropy algorithm through experiments. The steps for obtaining it are as follows: In the laboratory, using a known and stable flame signal, the number of infrared radiation intensity sampling points used to calculate energy entropy is gradually reduced; Find a critical point below which the calculated energy entropy value begins to become unstable or unreliable; Set the minimum window length to Set the value slightly above the critical point to ensure that the system's discrimination accuracy is not lost at the fastest response speed.

[0029] When the characteristic value of simmering A smaller value indicates the presence of a smokeless open flame, and the fire is developing rapidly. A longer time window is needed to confirm the random fluctuations in the flame to prevent misjudgment. When the smoldering characteristic value... A larger smoke level indicates the presence of a significant amount of smoke, which is typical of a smoldering fire. In this case, the time window can be shortened to expedite confirmation and alarm response.

[0030] For example, setting the base window length For 200 sampling points, the simmering coefficient The minimum window length is 0.5. There were 50 sampling points. In the early stages of a fire, the smoldering characteristics were not obvious, and the smoldering characteristic values ​​were... The value is 0.2, which represents the number of sampling points in the time window. The calculation results were obtained from 181 sampling points, and the system used a relatively long window for analysis. When the fire develops into a state of smoldering with thick smoke, the smoldering characteristic value... The value increased to 5.0, and the number of sampling points in the time window increased. The calculation result is 57 sampling points, and the window is shortened, which enables the system to respond quickly to smoldering fires.

[0031] In an optional embodiment, calculating the energy entropy of the infrared radiation intensity sequence within the time window includes: Each infrared radiation intensity value within the time window Divide by the sum of all infrared radiation intensity values ​​within the time window to obtain the energy percentage at each moment. ; The energy entropy is calculated using the following formula. : .

[0032] The actual combustion process of a flame is irregular, with its infrared radiation intensity exhibiting disordered jumps. This disorder leads to a highly uneven distribution of energy within a time window, with the energy proportion at certain moments being much higher than at other moments, resulting in a lower calculated energy entropy value. Conversely, for non-fire interference sources such as electric welding and incandescent lamps, although their infrared radiation may also fluctuate, it is usually more regular or stable, resulting in a relatively uniform energy distribution within the time window, thus leading to a higher calculated energy entropy value.

[0033] For example, within a time window containing 5 sampling points, the infrared radiation intensity sequence of a real flame is 100, 800, 150, 900, and 120. The energy is concentrated at the second and fourth points, showing uneven distribution. The calculated energy percentage sequences are 0.05, 0.39, 0.07, 0.43, and 0.06, respectively, resulting in an energy entropy value of approximately 1.8, which is relatively low. In contrast, the infrared radiation intensity sequence of an infrared heater is 400, 410, 405, 415, and 408. The energy distribution is very even, with the energy percentage at each point close to 0.2. The calculated energy entropy will be close to its maximum value, approximately 2.3, which is relatively high. By setting an appropriate threshold (e.g., 2.0), signals with entropy values ​​below this threshold can be identified as fires.

[0034] In an optional embodiment, a comprehensive fire index is established, including: The number of times the ultraviolet radiation intensity exceeded a preset ultraviolet threshold within the first preset time window is recorded as the peak value. ; The weights of the braising characteristic values ​​are calculated using the following formula. : ; Weighting of flame composite parameters equal ; The flame composite parameters and simmering characteristic values ​​were normalized separately to obtain the normalized flame composite parameters. Normalized braised characteristic value ; The comprehensive fire index is calculated using the following formula. : .

[0035] The first preset time window is an empirical parameter, and its acquisition steps are as follows: Determine the typical response time of the ultraviolet signal: In a controlled experimental environment, simulate typical open flames of different intensities. From the moment the open flame starts to burn, record the time required for the ultraviolet radiation intensity sequence to reach the preset ultraviolet noise threshold (i.e., form the first effective peak value) for the first time. Through multiple measurements, take the average value or the statistical minimum value to determine the typical response time of the open flame signal from zero to stable appearance in a mining environment. Determine the typical fluctuation period of the ultraviolet signal: Record the ultraviolet radiation intensity sequence during the stable combustion stage of the open flame, analyze the random fluctuation of the signal, and identify the typical frequency or period of the ultraviolet peak. The first preset time window must contain at least several complete fluctuation periods in order to count the number of representative peaks. Calculate and calibrate the first preset time window: Set the first preset time window to a multiple of the typical response time.

[0036] The number of times the ultraviolet signal peaks (number of peaks) This can be used as a basis for judging the probability of open flames occurring. When When no obvious open flame is detected, the fire situation should be judged based on the smoldering characteristic value. Therefore, the weight of the smoldering characteristic value is... It is set to its maximum value of 1. With... The increase in the number of open flames indicates a rise in the likelihood of open flames, gradually reducing the weight of the simmering characteristic. At the same time, increase the weight of flame composite parameters. This shifts the basis for judgment towards the characteristics of open flame.

[0037] For example, normalized flame composite parameters Normalized braised characteristic value The values ​​range from 0 to 1. In one detection, no ultraviolet peak was detected, meaning... =0, at this time =1, =0. If the normalized characteristic value of simmering is... =0.8, normalized flame composite parameter =0.1, then the comprehensive fire index =0.8, indicating the current state is simmering. In another detection, four ultraviolet peaks were detected. =4, then =0.8, =0.2. If at this time =0.9, =0.3, then The value of 0.78 indicates that the fire situation has shifted to being dominated by open flames.

[0038] S4. When the comprehensive fire index is greater than the first fire threshold and the energy entropy is greater than the entropy threshold, output the first fire alarm signal; when the comprehensive fire index is between the first fire threshold and the second fire threshold and no ultraviolet radiation is detected within the second preset time window, output the second fire alarm signal.

[0039] Comprehensive fire index The calculated energy entropy is compared with a preset first fire threshold (e.g., 80). Compare with a preset entropy threshold (e.g., 0.6). If Greater than 80 and If both conditions are met simultaneously (greater than 0.6), it indicates that the overall fire situation index is high and the heat source signal has the fluctuating characteristics of flames. This is considered a confirmed fire, and the first fire alarm signal will be immediately output, which will immediately activate the fire protection system or notify the main control room.

[0040] Comprehensive fire index Compare with a first fire threshold (80) and a second fire threshold (e.g., 50). Simultaneously, examine all UV radiation intensity samples within a second preset time window (e.g., the past 30 seconds). If If the value is between 50 and 80 and no ultraviolet radiation is detected within 30 seconds, it indicates that the fire is at a moderate level and there are no open flame characteristics, which is consistent with the characteristics of a smoldering fire. At this time, a second fire alarm signal is output as a warning and requires personnel to confirm.

[0041] The system employs a first fire threshold to determine whether a first fire alarm signal has been triggered, and a second fire threshold to define the lower limit of the fire warning range. Both the first and second fire thresholds are set by the system designer based on historical data, sensor sensitivity, and false alarm tolerance. The entropy threshold is used to distinguish between real flames and interference from stable heat sources, and it is set based on experimental data or theoretical calculations.

[0042] An embodiment of the mine fire monitoring system based on multi-dimensional perception provided by this invention: A mine fire monitoring system based on multi-dimensional perception includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the aforementioned mine fire monitoring method based on multi-dimensional perception.

[0043] The mine fire monitoring system based on multi-dimensional perception also includes other components well known to those skilled in the art, such as communication interfaces. Their settings and functions are known in the art and will not be described in detail here.

Claims

1. A mine fire monitoring method based on multi-dimensional perception, characterized in that, Includes the following steps: S1, acquire the infrared radiation intensity sequence, ultraviolet radiation intensity sequence, and smoke concentration sequence collected in real time at the monitoring point; obtain the environmental smoke background concentration that changes over time by performing median filtering on the smoke concentration sequence; S2, Calculate the smoldering characteristic value and flame composite parameters: The smoldering characteristic value is obtained by dividing the difference between the current smoke concentration and the ambient smoke background concentration by the current infrared radiation intensity; the flame composite parameters are calculated based on the current infrared radiation intensity, the first-order difference of the infrared radiation intensity, and the ultraviolet radiation intensity. S3, set a time window based on the simmering characteristic value, and calculate the energy entropy of the infrared radiation intensity sequence within the time window; By using the number of peak values ​​of the ultraviolet radiation intensity sequence statistically obtained within the first preset time window, the weight of the smoldering characteristic value is determined. The flame composite parameters and the smoldering characteristic value are weighted using this weight to obtain the comprehensive fire index. S4. When the comprehensive fire index is greater than the first fire threshold and the energy entropy is greater than the entropy threshold, output the first fire alarm signal; when the comprehensive fire index is between the first fire threshold and the second fire threshold and no ultraviolet radiation is detected within the second preset time window, output the second fire alarm signal.

2. The mine fire monitoring method based on multi-dimensional perception according to claim 1, characterized in that, The flame composite parameters are calculated based on the current infrared radiation intensity, the first-order difference of the infrared radiation intensity, and the ultraviolet radiation intensity, including: The flame composite parameters are obtained by multiplying or weighting the current infrared radiation intensity, the first-order difference of the current infrared radiation intensity, and the current ultraviolet radiation intensity.

3. The mine fire monitoring method based on multi-dimensional perception according to claim 1, characterized in that, A time window is set based on the characteristics of simmering, including: Set the base window length to The simmering coefficient is Minimum window length is Number of sampling points in the time window Calculated using the following formula: ; in, This represents the current simmering characteristic value. For the floor function, the base window length is... The simmering coefficient represents the baseline value used when calculating the length of the time window. Minimum window length is used to adjust the influence of simmering characteristic values ​​on the time window length. It is the lower limit of the time window length, ensuring that the system always has enough data for analysis.

4. The mine fire monitoring method based on multi-dimensional perception according to claim 3, characterized in that, Calculate the energy entropy of the infrared radiation intensity sequence within this time window, including: Each infrared radiation intensity value within the time window Divide by the sum of all infrared radiation intensity values ​​within the time window to obtain the energy percentage at each moment. ; The energy entropy is calculated using the following formula. : .

5. The mine fire monitoring method based on multi-dimensional perception according to claim 4, characterized in that, Establish a comprehensive fire index, including: The number of times the ultraviolet radiation intensity exceeded a preset ultraviolet threshold within the first preset time window is recorded as the peak value. ; The weights of the braising characteristic values ​​are calculated using the following formula. : ; Weighting of flame composite parameters equal ; The flame composite parameters and simmering characteristic values ​​were normalized separately to obtain the normalized flame composite parameters. Normalized braised characteristic value ; The comprehensive fire index is calculated using the following formula. : .

6. A mine fire monitoring system based on multi-dimensional perception, characterized in that, It includes a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the mine fire monitoring method based on multi-dimensional perception as described in any one of claims 1-5 is implemented.

Images