Mining laser methane telemetering system and method based on multispectral fusion

Abstract

The invention relates to the technical field of coal mine telemetering, in particular to a mining laser methane telemetering system based on multispectral fusion, which comprises a multispectral laser emission module, an optical receiving and signal conversion module, a multispectral fusion processing module, an intelligent algorithm compensation module, a data transmission and display module and a power supply module. According to the method, accurate and real-time monitoring of the methane concentration in the coal mine environment is achieved, reliable guarantee is provided for coal mine safety production, compared with a traditional neural network compensation model, the optimized neural network compensation model has the advantages that the methane concentration compensation precision is greatly improved, the generalization ability of the model to different environment conditions is obviously enhanced, and the method is suitable for popularization and application. Accurate measurement of methane concentration can be realized in a more complex and changeable coal mine environment. Meanwhile, due to the application of a dynamic weight compensation mechanism and an uncertainty quantification and compensation adjustment method, the reliability and the stability of a measurement result are further improved, and a more powerful guarantee is provided for safe production of a coal mine.

Description

Technical Field

[0001] The present invention relates to a mine-used laser methane telemetry system and method based on multi-spectral fusion, and in particular to a mine-used laser methane telemetry system and method based on multi-spectral fusion, belonging to the technical field of coal mine telemetry. Background Art

[0002] Methane is a common and dangerous gas in coal mining operations. It is flammable and explosive. When its concentration in the air reaches a certain level, it can explode when exposed to open flames or high temperatures, seriously threatening the lives of coal miners and the normal operation of the mine. Therefore, real-time and accurate monitoring of methane concentrations in coal mine environments is crucial.

[0003] Currently, traditional methods for monitoring methane in mines primarily include catalytic combustion and infrared absorption. Catalytic combustion sensors measure methane concentration by measuring heat changes generated by the flameless combustion of methane under the action of a catalyst. However, this method suffers from issues such as susceptibility to poisoning, short lifespan, and cross-sensitivity to other combustible gases. Furthermore, in harsh environments such as high humidity and high dust levels in coal mines, its measurement accuracy and stability are significantly affected. Infrared absorption sensors measure concentration by utilizing methane's absorption characteristics of infrared light of specific wavelengths. While these sensors offer advantages such as good selectivity and fast response, single-spectrum infrared absorption measurement methods often struggle to meet high-precision measurement requirements in the complex environment of coal mines due to the presence of multiple interfering gases and changes in environmental factors such as temperature and humidity.

[0004] With the development of laser technology, laser methane telemetry has gradually been applied to coal mine safety monitoring. Laser methane telemetry offers advantages such as long measurement distance, fast response speed, and no need for contact with the gas being measured. However, existing laser methane telemetry systems, which mostly use a single spectrum for measurement, also face the problem of low measurement accuracy due to environmental interference. Furthermore, the underground coal mine environment is complex and changeable. Changes in factors such as temperature, humidity, and dust concentration can affect laser transmission and methane absorption characteristics, further increasing the difficulty of measuring methane concentration. Furthermore, in existing neural network-based methane concentration compensation schemes, while traditional neural network models can fit the relationship between environmental factors and measurement errors to a certain extent, the underground coal mine environment is complex and changeable, with multiple nonlinear and dynamically changing environmental interference factors. When dealing with these complex conditions, traditional models often suffer from insufficient compensation accuracy and poor model generalization, making it difficult to meet the demand for high-precision methane concentration measurement.

[0005] Therefore, there is an urgent need for a mining laser methane telemetry system and method based on multi-spectral fusion to solve the above-mentioned problems. Summary of the Invention

[0006] The purpose of the present invention is to provide a mine-based laser methane telemetry system and method based on multispectral fusion to address the problems of low measurement accuracy and susceptibility to environmental interference in existing mine methane monitoring technologies in complex coal mine environments. This system achieves accurate, real-time monitoring of methane concentrations in coal mine environments, providing reliable protection for coal mine safety production. Compared with traditional neural network compensation models, the optimized neural network compensation model significantly improves the accuracy of methane concentration compensation, and the model's generalization ability for different environmental conditions is significantly enhanced, enabling accurate measurement of methane concentrations in more complex and variable coal mine environments. At the same time, the application of a dynamic weight compensation mechanism and uncertainty quantification and compensation adjustment methods further improves the reliability and stability of measurement results, providing a stronger guarantee for coal mine safety production.

[0007] In order to achieve the above objectives, the main technical solutions adopted by the present invention include: a mine laser methane telemetry system based on multi-spectral fusion, characterized by including a multi-spectral laser emission module, an optical receiving and signal conversion module, a multi-spectral fusion processing module, an intelligent algorithm compensation module, a data transmission and display module, and a power supply module; The multi-spectral laser emission module includes a plurality of lasers with different wavelengths and each laser emits laser light of a specific wavelength; The optical receiving and signal conversion module includes an optical receiving lens, a filter and a photodetector; The multi-spectral fusion processing module is used to receive the electrical signals corresponding to different wavelengths output by the optical receiving and signal conversion module, and perform pre-processing on these signals, such as amplification, filtering, etc. The intelligent algorithm compensation module is used to collect environmental parameters in coal mines in real time, mainly including temperature T, humidity H, dust concentration D, etc. The data transmission and display module is used to transmit the methane concentration data processed by the intelligent algorithm compensation module to the host computer or monitoring center; The power supply module is used to provide a stable power supply to the entire system.

[0008] Preferably, the multispectral fusion processing module uses a multispectral fusion algorithm to perform deep fusion of signals of different spectra. The specific algorithm process includes: Signal normalization processing: Let the electrical signal intensity corresponding to the i-th spectrum be Si. In order to eliminate the influence of the dimension difference between different spectral signals, it is normalized to obtain the normalized signal Si,norm:

[0009] Among them, Si,min and Si,max are the minimum and maximum values ​​of the i-th spectral signal, respectively, which can be obtained through experiments or historical data statistics.

[0010] Preferably, the multi-spectral fusion algorithm further includes a weighted fusion algorithm: taking into account factors such as the absorption coefficient αi of methane at different wavelengths and the signal-to-noise ratio SNRi of each spectral signal, corresponding weights wi are assigned to different spectral signals. The weight calculation formula is:

[0011] Where n is the number of spectra, j is the sequence number of the spectrum; The fused initial methane concentration measurement value Cinitial is the weighted sum of the measurement values ​​of each spectral signal:

[0012] Where Ci is the methane concentration value obtained by measuring the i-th spectral signal alone, which can be calculated using the Lambert-Beer law:

[0013] The deformation can be obtained:

[0014] Where I0 is the incident light intensity, I is the outgoing light intensity after absorption by methane, and L is the propagation path length of the laser in methane gas.

[0015] Preferably, the intelligent algorithm compensation module obtains accurate environmental data by installing corresponding sensors in the system, such as temperature sensors, humidity sensors and dust concentration sensors, and then uses intelligent algorithms such as neural networks to analyze and process the collected environmental parameters to establish a mathematical model between environmental factors and methane measurement results.

[0016] Preferably, the neural network model is constructed by adopting a three-layer feedforward neural network, including an input layer, a hidden layer and an output layer. The number of neurons in the input layer is the number of environmental parameters, that is, m=3 (temperature, humidity, dust concentration); the number of neurons in the hidden layer h can be determined according to an empirical formula or experiment, generally taking h=m+k+a (k is the number of neurons in the output layer, where k=1, and a is an integer between 1-10); the number of neurons in the output layer is 1, that is, the compensation value ΔC of the methane concentration.

[0017] Preferably, the neural network training is to collect a large amount of methane concentration measurement data under different environmental conditions and environmental parameter data (T, H, D) as training samples, let the training sample set be {(Tl, Hl, Dl, Cinitial, l, Δ )}= , where N is the number of samples, the weights and biases of the neural network are continuously adjusted through the back propagation algorithm so that the output of the network Δ C ^l and the actual compensation value Δ The error between them is the smallest, and the error function usually adopts the mean square error function: .

[0018] Preferably, the neural network model construction introduces a dynamic temporal neural network structure and constructs a multi-scale feature fusion network according to needs; the innovative training method for the neural network model construction includes an adaptive training strategy based on reinforcement learning and a combination of data enhancement and transfer learning; the innovative compensation calculation method for the neural network model construction adopts a dynamic weight compensation mechanism and uncertainty quantification and compensation adjustment; Based on the real-time collected environmental parameters, such as temperature, humidity, and dust concentration, an auxiliary weight calculation network (such as a small neural network or decision tree) dynamically calculates the compensation weight corresponding to each environmental factor. The formula for calculating the precise measurement value of methane concentration is: Δ

[0019] in, For the final accurate measurement of methane concentration, is the preliminary measurement value of methane concentration output by the multi-spectral fusion processing module, m is the number of environmental factors, wi is the dynamic weight corresponding to the i-th environmental factor, Δ is the compensation value of the i-th environmental factor calculated by the neural network model.

[0020] Preferably, the data transmission and display module adopts wired or wireless mode, which is selected according to the actual needs of the coal mine site; at the same time, the module is also equipped with a display device, including a liquid crystal display, for real-time display of current methane concentration value, environmental parameters and system working status and other information.

[0021] Preferably, the power module adopts an explosion-proof and intrinsically safe design, and has protection functions such as overvoltage, overcurrent, and short circuit, ensuring that the system can operate safely and reliably under various working conditions. The power module can be connected to the AC power supply underground in the coal mine, and can also provide backup power through the built-in battery pack to cope with sudden power outages and other situations.

[0022] A method for a mine-used laser methane telemetry system based on multi-spectral fusion, comprising the following steps: S1: Multi-spectral laser emission: The multi-spectral laser emission module emits multiple lasers of different wavelengths in sequence according to the preset timing and power. These lasers pass through the air under the coal mine and illuminate the measured area. Methane gas absorbs lasers of different wavelengths to varying degrees. S2: Optical reception and signal conversion: The optical reception and signal conversion module receives the multi-spectral laser signal reflected from the measured area. The optical receiving lens focuses the reflected light onto the filter, which selects the light signal of a specific wavelength. The photodetector then converts the light signal into an electrical signal. After pre-processing such as amplification and filtering, the electrical signal is transmitted to the multi-spectral fusion processing module. S3: Multispectral fusion processing: The multispectral fusion processing module analyzes and processes the received electrical signals corresponding to different wavelengths according to the above-mentioned normalization processing and weighted fusion algorithm to obtain the fused preliminary measurement value of methane concentration; S4: Environmental parameter collection and intelligent algorithm compensation: The intelligent algorithm compensation module collects environmental parameters such as temperature, humidity, and dust concentration in the coal mine in real time. The collected environmental parameters and the preliminary methane concentration measurement value output by the multi-spectral fusion processing module are input into a pre-trained neural network model. Based on the output of the neural network model, the preliminary methane concentration measurement value is dynamically compensated to obtain the final accurate methane concentration measurement value; S5: Data transmission and display: The data transmission and display module transmits the final precise measurement value of methane concentration and environmental parameters to the host computer or monitoring center, and displays them in real time on the on-site display device. At the same time, the system can also set a threshold alarm function. When the methane concentration exceeds the preset safety threshold, an audible and visual alarm signal will be issued in time to remind staff to take appropriate safety measures.

[0023] The present invention has at least the following beneficial effects: 1. By innovatively integrating multispectral technology and intelligent algorithms, accurate measurement of methane concentration in complex coal mine environments is achieved. This system and method have the advantages of high measurement accuracy, strong anti-interference ability, good real-time performance and strong adaptability. It can effectively improve the safety level of coal mine production and reduce the occurrence of methane explosion accidents. It has broad application prospects and important social and economic benefits. In actual application, through reasonable system installation and debugging, daily operation and maintenance, and grasp of the key points of method implementation, it can ensure the stable operation of the system and the accuracy of the measurement results, providing strong technical support for coal mine safety monitoring.

[0024] 2. Compared with the traditional neural network compensation model, the optimized model has greatly improved the accuracy of methane concentration compensation, and the model's generalization ability for different environmental conditions has been significantly enhanced. It can achieve accurate measurement of methane concentration in more complex and changeable coal mine environments. At the same time, the application of dynamic weight compensation mechanism and uncertainty quantification and compensation adjustment method further improves the reliability and stability of measurement results, providing stronger protection for coal mine safety production. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0026] The following will describe the implementation methods of the present application in detail with reference to the accompanying drawings and examples, so that the implementation process of how the present application applies technical means to solve technical problems and achieve technical effects can be fully understood and implemented accordingly.

[0027] like Figure 1 As shown, the mine laser methane telemetry system and method based on multi-spectral fusion provided in this embodiment include a multi-spectral laser emission module, an optical receiving and signal conversion module, a multi-spectral fusion processing module, an intelligent algorithm compensation module, a data transmission and display module, and a power supply module; The multi-spectral laser emission module includes multiple lasers with different wavelengths, each emitting laser light at a specific wavelength. These laser wavelengths are selected based on the absorption characteristics of methane at different spectra, making full use of the differences in methane absorption at different wavelengths to obtain more comprehensive methane concentration information. The laser uses a highly stable and long-life semiconductor laser to ensure continuous and stable operation in the harsh environment of coal mines. At the same time, the module is also equipped with a laser drive circuit and a temperature control circuit to precisely control the laser's output power and wavelength stability. The optical receiving and signal conversion module includes an optical receiving lens, a filter, and a photodetector. The optical receiving lens is used to collect the multi-spectral laser signal reflected from the measured area and focus it on the filter. The filter is used to filter out background light and other interfering light, allowing only light corresponding to the specific wavelength emitted by the laser emission module to pass through. The photodetector converts the optical signal passing through the filter into an electrical signal. Commonly used photodetectors include photodiodes and avalanche photodiodes, which have the characteristics of high sensitivity and low noise, and can accurately capture weak optical signal changes. The multi-spectral fusion processing module is used to receive the electrical signals corresponding to different wavelengths output by the optical receiving and signal conversion module, and perform pre-processing on these signals, such as amplification and filtering, to improve the signal quality. Then, the multi-spectral fusion algorithm is used to deeply fuse the signals of different spectra. The multispectral fusion processing module uses a multispectral fusion algorithm to deeply fuse signals from different spectra. The specific algorithm process includes: Signal normalization processing: Let the electrical signal intensity corresponding to the i-th spectrum be Si. In order to eliminate the influence of the dimension difference between different spectral signals, it is normalized to obtain the normalized signal Si,norm:

[0028] Among them, Si,min and Si,max are the minimum and maximum values ​​of the i-th spectral signal, respectively, which can be obtained through experiments or historical data statistics; The multi-spectral fusion algorithm also includes a weighted fusion algorithm: considering factors such as the absorption coefficient αi of methane at different wavelengths and the signal-to-noise ratio SNRi of each spectral signal, corresponding weights wi are assigned to different spectral signals. The weight calculation formula is:

[0029] Where n is the number of spectra, j is the sequence number of the spectrum; The fused initial methane concentration measurement value Cinitial is the weighted sum of the measurement values ​​of each spectral signal:

[0030] Where Ci is the methane concentration value obtained by measuring the i-th spectral signal alone, which can be calculated using the Lambert-Beer law:

[0031] The deformation can be obtained:

[0032] Where I0 is the incident light intensity, I is the outgoing light intensity after absorption by methane, and L is the propagation path length of the laser in methane gas; Through the above technical solution, the present invention has the characteristics of high measurement accuracy, strong anti-interference ability, good real-time performance and strong adaptability; Specifically: High measurement accuracy: Multi-spectral fusion technology fully utilizes the absorption characteristics of methane on different spectra to obtain more comprehensive methane concentration information, effectively reducing the error caused by single spectrum measurement. At the same time, the intelligent algorithm compensation module can take into account the impact of environmental factors on the measurement results in real time and perform dynamic compensation, further improving the accuracy of methane concentration measurement and meeting the demand for high-precision monitoring in coal mine safety production. Strong anti-interference ability: In complex coal mine environments, there are many interference factors, such as absorption by other gases, scattering of dust, and changes in temperature and humidity. The system and method of the present invention can effectively suppress the influence of these interference factors through multi-spectral fusion and intelligent algorithm compensation, ensuring the stability and reliability of the measurement results; Good real-time performance: The system uses high-speed data acquisition and processing technology to obtain methane concentration and environmental parameter information in real time, and quickly performs multi-spectral fusion and intelligent algorithm compensation processing to output the final measurement results in a timely manner. This enables staff to understand the safety status of the coal mine environment in the first place, providing strong support for taking corresponding safety measures; Strong adaptability: The system and method of the present invention are applicable to different types of underground coal mine environments, whether high-gas mines or low-gas mines, and can accurately measure methane concentrations. Furthermore, the system has good scalability, allowing for the addition or removal of monitoring points based on actual needs, facilitating integration with other coal mine safety monitoring systems. The intelligent algorithm compensation module is used to collect real-time environmental parameters in coal mines, including temperature T, humidity H, and dust concentration D. By installing appropriate sensors in the system, such as temperature sensors, humidity sensors, and dust concentration sensors, accurate environmental data is acquired. Then, intelligent algorithms such as neural networks are used to analyze and process the collected environmental parameters, establishing a mathematical model between environmental factors and methane measurement results. The neural network model is constructed using a three-layer feedforward neural network, consisting of an input layer, a hidden layer, and an output layer. The number of neurons in the input layer is the number of environmental parameters, i.e., m = 3 (temperature, humidity, and dust concentration). The number of neurons in the hidden layer, h, can be determined based on empirical formulas or experiments, and is generally taken as h = m + k + a (k is the number of neurons in the output layer, where k = 1 and a is an integer between 1 and 10). The number of neurons in the output layer is 1, which is the compensation value ΔC for methane concentration. Among them, neural network training: collecting a large amount of methane concentration measurement data under different environmental conditions and environmental parameter data (T, H, D) as training samples, let the training sample set be {(Tl, Hl, Dl, Cinitial, l, Δ )}= , where N is the number of samples, the weights and biases of the neural network are continuously adjusted through the back propagation algorithm so that the output of the network Δ C ^ l and the actual compensation value Δ The error between them is the smallest, and the error function usually adopts the mean square error function: ; Furthermore, the neural network model construction introduces dynamic temporal neural network structures and constructs multi-scale feature fusion networks according to needs. Dynamic temporal neural network structures such as long short-term memory networks (LSTMs) or gated recurrent units (GRUs) are introduced. These networks can capture the dependencies of environmental parameters in time series and better reflect the long-term and short-term impacts of environmental factors on methane concentration measurements. In Example 1, an LSTM is used as an example. It contains an input gate, a forget gate, and an output gate. It can dynamically adjust the retention and forgetting of information based on the current input and the hidden state of the previous moment, effectively processing time series data. In the methane concentration compensation model, environmental parameters such as temperature, humidity, and dust concentration at multiple consecutive moments, along with the corresponding preliminary methane concentration measurements, are used as input sequences. The LSTM network learns the inherent patterns between these time series data to obtain more accurate compensation values. In Example 2, a multi-scale feature fusion network extracts features from input environmental parameters and measurements through multiple convolutional layers or pooling layers of different scales. Each scale layer can capture feature information at different time or spatial scales. These features of different scales are then fused, and multi-scale features are integrated into a unified feature representation using methods such as splicing and weighted summation. Finally, the fused features are input into a subsequent fully connected layer or recurrent layer to calculate the compensation value. Through multi-scale feature fusion, the model can more comprehensively consider the multi-scale impact of environmental factors and improve the accuracy and robustness of compensation. Among them, innovative training methods for neural network model construction include adaptive training strategies based on reinforcement learning and the combination of data enhancement and transfer learning; In Example 3, a neural network model is used as an intelligent agent in a reinforcement learning framework, and the accuracy of methane concentration compensation is used as a reward signal. During training, the intelligent agent selects appropriate training parameters (such as learning rate and batch size) based on the current environmental parameters and model state to update the model. When the compensation value output by the model makes the final methane concentration measurement closer to the true value, the intelligent agent is given a positive reward; otherwise, a negative reward is given. Through continuous interaction with the environment and trial and error, the intelligent agent can learn the optimal training parameter adjustment strategy, so that the model can maintain good compensation performance under different environmental conditions. In Example 4, in terms of data enhancement, various transformations are performed on existing coal mine environmental data, such as adding noise, interpolating and resampling time series, and performing linear or nonlinear transformations on environmental parameters, to generate a large number of new training samples and enrich the diversity of the data set. At the same time, using transfer learning technology, methane concentration measurement data and environmental parameter data collected in other similar environments (such as chemical plants and underground pipelines) are used as source domain data to pre-train the model. Through transfer learning, the model can learn common environmental characteristics and methane concentration measurement patterns. It can then be fine-tuned on the target domain data in coal mines, quickly adapting to the special environment of coal mines, and improving the training efficiency and generalization ability of the model. The innovative compensation calculation method for neural network model construction adopts dynamic weight compensation mechanism as well as uncertainty quantification and compensation adjustment; Based on the real-time collected environmental parameters, such as temperature, humidity, and dust concentration, an auxiliary weight calculation network (such as a small neural network or decision tree) dynamically calculates the compensation weight corresponding to each environmental factor. The formula for calculating the precise measurement value of methane concentration is: Δ

[0033] in, For the final accurate measurement of methane concentration, is the preliminary measurement value of methane concentration output by the multi-spectral fusion processing module, m is the number of environmental factors, wi is the dynamic weight corresponding to the i-th environmental factor, Δ is the compensation value of the i-th environmental factor calculated by the neural network model; By introducing uncertainty quantification methods, the uncertainty of the compensation value output by the model is evaluated. Common uncertainty quantification methods include Bayesian neural networks and Monte Carlo dropout. Through uncertainty quantification, uncertainty estimates of the compensation value, such as mean and variance, can be obtained. Based on the uncertainty estimation results, the compensation value is dynamically adjusted. When the uncertainty is large, it means that the model's compensation prediction under the environmental conditions is not reliable enough. At this time, some conservative measures can be taken, such as reducing the weight of the compensation value or adding other auxiliary monitoring methods for verification. When the uncertainty is small, it means that the model prediction is relatively reliable, and the precise measurement value of methane concentration can be calculated according to the normal compensation calculation method. Through uncertainty quantification and compensation adjustment, the reliability and accuracy of methane concentration measurement can be further improved. The power module is used to provide a stable power supply for the entire system. It adopts an explosion-proof and intrinsically safe design and has protection functions such as overvoltage, overcurrent, and short circuit to ensure that the system can operate safely and reliably under various working conditions. The power module can be connected to the AC power supply in the coal mine, and can also provide backup power through the built-in battery pack to cope with sudden power outages. The above technical solution, through the optimization and innovation of the neural network model, has achieved remarkable results in actual testing in coal mines. Compared with the traditional neural network compensation model, the optimized model has greatly improved the accuracy of methane concentration compensation. The model's generalization ability to different environmental conditions has been significantly enhanced, and it can achieve accurate measurement of methane concentration in more complex and variable coal mine environments. At the same time, the application of a dynamic weight compensation mechanism and uncertainty quantification and compensation adjustment methods further improves the reliability and stability of measurement results, providing stronger protection for coal mine production safety. The data transmission and display module is used to transmit the methane concentration data processed by the intelligent algorithm compensation module to the host computer or monitoring center; the data transmission and display module adopts wired or wireless methods, which are selected according to the actual needs of the coal mine site; at the same time, the module is also equipped with a display device, including an LCD screen, which is used to display the current methane concentration value, environmental parameters, and system working status in real time, so that on-site staff can timely understand the safety status of the coal mine environment.

[0034] like Figure 1 As shown, a method for a mine laser methane telemetry system based on multi-spectral fusion includes the following steps: S1: Multi-spectral laser emission: The multi-spectral laser emission module emits multiple lasers of different wavelengths in sequence according to the preset timing and power. These lasers pass through the air under the coal mine and illuminate the measured area. Methane gas absorbs lasers of different wavelengths to varying degrees. S2: Optical reception and signal conversion: The optical reception and signal conversion module receives the multi-spectral laser signal reflected from the measured area. The optical receiving lens focuses the reflected light onto the filter, which selects the light signal of a specific wavelength. The photodetector then converts the light signal into an electrical signal. After pre-processing such as amplification and filtering, the electrical signal is transmitted to the multi-spectral fusion processing module. S3: Multispectral fusion processing: The multispectral fusion processing module analyzes and processes the received electrical signals corresponding to different wavelengths according to the above-mentioned normalization processing and weighted fusion algorithm to obtain the fused preliminary measurement value of methane concentration; S4: Environmental parameter collection and intelligent algorithm compensation: The intelligent algorithm compensation module collects environmental parameters such as temperature, humidity, and dust concentration in the coal mine in real time. The collected environmental parameters and the preliminary methane concentration measurement value output by the multi-spectral fusion processing module are input into a pre-trained neural network model. Based on the output of the neural network model, the preliminary methane concentration measurement value is dynamically compensated to obtain the final accurate methane concentration measurement value; S5: Data transmission and display: The data transmission and display module transmits the final precise measurement value of methane concentration and environmental parameters to the host computer or monitoring center, and displays them in real time on the on-site display device. At the same time, the system can also set a threshold alarm function. When the methane concentration exceeds the preset safety threshold, an audible and visual alarm signal will be issued in time to remind staff to take appropriate safety measures.

[0035] For example, certain words are used in the specification and claims to refer to specific components. Those skilled in the art should understand that hardware manufacturers may use different terms to refer to the same component. This specification and claims do not use differences in names as a way to distinguish components, but use differences in the functions of the components as the criteria for distinction. For example, "including" mentioned throughout the specification and claims is an open term and should be interpreted as "including but not limited to". "Approximately" means that within an acceptable error range, those skilled in the art can solve technical problems within a certain error range and basically achieve technical effects.

[0036] It should be noted that the terms "include," "comprises," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a product or system comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such product or system. In the absence of further limitations, an element defined by the phrase "comprising a..." does not exclude the presence of other identical elements in the product or system comprising the element.

[0037] The foregoing description shows and describes several preferred embodiments of the present invention. However, as previously stated, it should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. Rather, the present invention can be used in various other combinations, modifications, and environments and can be modified within the scope of the inventive concept described herein by the teachings above or by techniques or knowledge in the relevant art. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention are intended to be within the scope of the appended claims.

Claims

1. A mine laser methane telemetry system based on multi-spectral fusion, characterized in that: It includes a multi-spectral laser emission module, an optical receiving and signal conversion module, a multi-spectral fusion processing module, an intelligent algorithm compensation module, a data transmission and display module, and a power supply module; The multi-spectral laser emission module includes a plurality of lasers with different wavelengths and each laser emits laser light of a specific wavelength; The optical receiving and signal conversion module includes an optical receiving lens, a filter and a photodetector; The multi-spectral fusion processing module is used to receive the electrical signals corresponding to different wavelengths output by the optical receiving and signal conversion module, and perform pre-processing on these signals, such as amplification, filtering, etc. The intelligent algorithm compensation module is used to collect environmental parameters in coal mines in real time, mainly including temperature T, humidity H, dust concentration D, etc. The data transmission and display module is used to transmit the methane concentration data processed by the intelligent algorithm compensation module to the host computer or monitoring center; The power supply module is used to provide a stable power supply to the entire system.

2. The mine laser methane telemetry system based on multi-spectral fusion according to claim 1 is characterized by: The multispectral fusion processing module uses a multispectral fusion algorithm to deeply fuse signals of different spectra. The specific algorithm process includes: Signal normalization processing: Let the electrical signal intensity corresponding to the i-th spectrum be Si. In order to eliminate the influence of the dimension difference between different spectral signals, it is normalized to obtain the normalized signal Si,norm: ； Among them, Si,min and Si,max are the minimum and maximum values ​​of the i-th spectral signal, respectively, which can be obtained through experiments or historical data statistics.

3. The mine laser methane telemetry system based on multi-spectral fusion according to claim 2 is characterized by: The multi-spectral fusion algorithm also includes a weighted fusion algorithm: considering factors such as the absorption coefficient αi of methane at different wavelengths and the signal-to-noise ratio SNRi of each spectral signal, a corresponding weight wi is assigned to each spectral signal. The weight calculation formula is: ； Where n is the number of spectra, j is the sequence number of the spectrum; The fused initial methane concentration measurement value Cinitial is the weighted sum of the measurement values ​​of each spectral signal: ； Where Ci is the methane concentration value obtained by measuring the i-th spectral signal alone, which can be calculated using the Lambert-Beer law: ； The deformation can be obtained: ； Where I0 is the incident light intensity, I is the outgoing light intensity after absorption by methane, and L is the propagation path length of the laser in methane gas.

4. The mine laser methane telemetry system based on multi-spectral fusion according to claim 1 is characterized by: The intelligent algorithm compensation module obtains accurate environmental data by installing corresponding sensors in the system, such as temperature sensors, humidity sensors, and dust concentration sensors. It then uses intelligent algorithms such as neural networks to analyze and process the collected environmental parameters and establish a mathematical model between environmental factors and methane measurement results.

5. The mine laser methane remote sensing system based on multi-spectral fusion according to claim 4 is characterized by: The neural network model is constructed using a three-layer feedforward neural network, consisting of an input layer, a hidden layer, and an output layer. The number of neurons in the input layer is the number of environmental parameters, that is, m = 3 (temperature, humidity, and dust concentration). The number of neurons in the hidden layer, h, can be determined based on empirical formulas or experiments, and is generally taken as h = m + k + a (k is the number of neurons in the output layer, where k = 1 and a is an integer between 1 and 10). The number of neurons in the output layer is 1, which is the compensation value ΔC of methane concentration.

6. The mine laser methane telemetry system based on multi-spectral fusion according to claim 5 is characterized by: Among them, neural network training: collecting a large amount of methane concentration measurement data under different environmental conditions and environmental parameter data (T, H, D) as training samples, let the training sample set be {(Tl, Hl, Dl, Cinitial, l, Δ )}= , where N is the number of samples, the weights and biases of the neural network are continuously adjusted through the back propagation algorithm so that the output of the network Δ C ^ l and the actual compensation value Δ The error between them is the smallest, and the error function usually adopts the mean square error function: 。 7. The mine laser methane remote sensing system based on multi-spectral fusion according to claim 6 is characterized by: The neural network model construction introduces a dynamic time-series neural network structure and constructs a multi-scale feature fusion network according to needs; the innovative training method for the neural network model construction includes an adaptive training strategy based on reinforcement learning and a combination of data enhancement and transfer learning; the innovative compensation calculation method for the neural network model construction adopts a dynamic weight compensation mechanism and uncertainty quantification and compensation adjustment; Based on the real-time collected environmental parameters, such as temperature, humidity, and dust concentration, an auxiliary weight calculation network (such as a small neural network or decision tree) dynamically calculates the compensation weight corresponding to each environmental factor. The formula for calculating the precise measurement value of methane concentration is: D ; in, For the final accurate measurement of methane concentration, is the preliminary measurement value of methane concentration output by the multi-spectral fusion processing module, m is the number of environmental factors, wi is the dynamic weight corresponding to the i-th environmental factor, Δ is the compensation value of the i-th environmental factor calculated by the neural network model.

8. The mine laser methane telemetry system based on multi-spectral fusion according to claim 1 is characterized by: The data transmission and display module adopts wired or wireless mode, which is selected according to the actual needs of the coal mine site; at the same time, the module is also equipped with a display device, including a liquid crystal display, for real-time display of current methane concentration value, environmental parameters and system working status and other information.

9. The mine laser methane remote sensing system based on multi-spectral fusion according to claim 1 is characterized by: The power module adopts an explosion-proof and intrinsically safe design, and has protection functions such as overvoltage, overcurrent, and short circuit, ensuring that the system can operate safely and reliably under various working conditions. The power module can be connected to the AC power supply underground in the coal mine, and can also provide backup power through the built-in battery pack to cope with sudden power outages and other situations.

10. A method for a mine-used laser methane telemetry system based on multi-spectral fusion, characterized by: The steps include: S1: Multi-spectral laser emission: The multi-spectral laser emission module emits multiple lasers of different wavelengths in sequence according to the preset timing and power. These lasers pass through the air under the coal mine and illuminate the measured area. Methane gas absorbs lasers of different wavelengths to varying degrees. S2: Optical reception and signal conversion: The optical reception and signal conversion module receives the multi-spectral laser signal reflected from the measured area. The optical receiving lens focuses the reflected light onto the filter, which selects the light signal of a specific wavelength. The photodetector then converts the light signal into an electrical signal. After pre-processing such as amplification and filtering, the electrical signal is transmitted to the multi-spectral fusion processing module. S3: Multispectral fusion processing: The multispectral fusion processing module analyzes and processes the received electrical signals corresponding to different wavelengths according to the above-mentioned normalization processing and weighted fusion algorithm to obtain the fused preliminary measurement value of methane concentration; S4: Environmental parameter collection and intelligent algorithm compensation: The intelligent algorithm compensation module collects environmental parameters such as temperature, humidity, and dust concentration in the coal mine in real time. The collected environmental parameters and the preliminary methane concentration measurement value output by the multi-spectral fusion processing module are input into a pre-trained neural network model. Based on the output of the neural network model, the preliminary methane concentration measurement value is dynamically compensated to obtain the final accurate methane concentration measurement value; S5: Data transmission and display: The data transmission and display module transmits the final precise measurement value of methane concentration and environmental parameters to the host computer or monitoring center, and displays them in real time on the on-site display device. At the same time, the system can also set a threshold alarm function. When the methane concentration exceeds the preset safety threshold, an audible and visual alarm signal will be issued in time to remind staff to take appropriate safety measures.

Images